Application List

This is a list of applications you can create experiments for using Ramble. It is automatically generated based on the applications in this Ramble version.

Ramble currently has 52 mainline applications:

architecture-check¶

Ramble applications:

architecture-check/application.py

Description:

Application used to check the default architecture from various tools

Workloads:

standard

Executables:: spack_test, gcc_version, gcc_test, lscpu_test, free_test

babelstream¶

Ramble applications:

babelstream/application.py

Description:

Babelstream is a benchmark application to measure memory bandwidth See https://github.com/UoB-HPC/BabelStream for more information The pkg_spec for babelstream controls the resulting binary name. This is then controlled with the exec_name variable.

Workloads:

cpp-models

Executables:

stage-bin, execute_template

Variables:

exec_pre: Pre Command for execution (e.g. numatcl)
exec_name: Name of babelstream executable. Depends on spec used to installed babelstream
array_size: Number of elements in array that will be streamed
num_times: Number of times to repeat stream test
additional_args: Additional arguments to babelstream executable
partition: Define which partition we are going to use for execution

fortran-models

Executables:

stage-bin-fortran, execute_fortran

Variables:

f_model: Fortran option name
exec_pre: Pre Command for execution (e.g. numatcl)
exec_name: Name of babelstream executable. Depends on spec used to installed babelstream
array_size: Number of elements in array that will be streamed
num_times: Number of times to repeat stream test
additional_args: Additional arguments to babelstream executable
partition: Define which partition we are going to use for execution

cloverleaf¶

Ramble applications:

cloverleaf/application.py

Description:

Define CLOVERLEAF application

Workloads:

bm_short_c

Executables:: stage-input, execute

bm_short

Executables:: stage-input, execute

qa

Executables:: stage-input, execute

sodbig

Executables:: stage-input, execute

sodx

Executables:: stage-input, execute

sodxy

Executables:: stage-input, execute

sody

Executables:: stage-input, execute

clover

Executables:: stage-default-input, execute

bm2

Executables:: stage-input, execute

bm4

Executables:: stage-input, execute

bm8

Executables:: stage-input, execute

bm16

Executables:: stage-input, execute

bm32

Executables:: stage-input, execute

bm64

Executables:: stage-input, execute

bm128

Executables:: stage-input, execute

bm256

Executables:: stage-input, execute

bm512

Executables:: stage-input, execute

bm1024

Executables:: stage-input, execute

bm2048

Executables:: stage-input, execute

bm4096

Executables:: stage-input, execute

bm8192

Executables:: stage-input, execute

bm2_short

Executables:: stage-input, execute

bm4_short

Executables:: stage-input, execute

bm8_short

Executables:: stage-input, execute

bm16_short

Executables:: stage-input, execute

bm32_short

Executables:: stage-input, execute

bm64_short

Executables:: stage-input, execute

bm128_short

Executables:: stage-input, execute

bm256_short

Executables:: stage-input, execute

bm512_short

Executables:: stage-input, execute

bm1024_short

Executables:: stage-input, execute

bm2048_short

Executables:: stage-input, execute

bm4096_short

Executables:: stage-input, execute

bm8192_short

Executables:: stage-input, execute

bm1s_short

Executables:: stage-input, execute

bm2s_short

Executables:: stage-input, execute

bm4s_short

Executables:: stage-input, execute

bm8s_short

Executables:: stage-input, execute

bm16s_short

Executables:: stage-input, execute

bm32s_short

Executables:: stage-input, execute

bm64s_short

Executables:: stage-input, execute

bm128s_short

Executables:: stage-input, execute

bm256s_short

Executables:: stage-input, execute

bm512s_short

Executables:: stage-input, execute

bm1024s_short

Executables:: stage-input, execute

bm2048s_short

Executables:: stage-input, execute

bm4096s_short

Executables:: stage-input, execute

bm8192s_short

Executables:: stage-input, execute

bm16384s_short

Executables:: stage-input, execute

elk¶

Ramble applications:

elk/application.py

Description:

An all-electron full-potential linearised augmented-plane wave (LAPW) code with many advanced features. Written originally at Karl-Franzens- Universität Graz as a milestone of the EXCITING EU Research and Training Network, the code is designed to be as simple as possible so that new developments in the field of density functional theory (DFT) can be added quickly and reliably. The code is freely available under the GNU General Public License. https://elk.sourceforge.io/ This package intentionally does not use spack, and instead relies on the user to manage both the package install (eg via yum), and the environment through explicit paths. By default, yum installs to /usr/lib64/openmpi/bin/elk_openmpi

Workloads:

Cu

Executables:

stage-input, update_input, execute

Inputs:

examples: Main installer which contains various example input decks and species

Variables:

install_prefix: Install path to main executable
input_path: Path to input deck

fio¶

Ramble applications:

fio/application.py

Description:

Flexible I/O Tester. Fio spawns a number of threads or processes doing a particular type of I/O action as specified by the user. fio takes a number of global parameters, each inherited by the thread unless otherwise parameters given to them overriding that setting is given.

Workloads:

standard

Executables:

run, cleanup

Variables:

job_file: Job file to run. If a job file is not specified, one will be generated from variables.
job_file_path: Path to job file.
out_file: File to write results
out_format: Format to write results. Defaults to 'json' for Ramble to analyze results.
job_name: Job name
directory: Used to place files in a different location than ./
ioengine: I/O engine
direct: If true, use non-buffered I/O
buffered: If true, use buffered I/O. This is the opposite of the direct option
time_based: If set, fio will run for the duration of the runtime specified even if the file(s) are completely read or written.
runtime: Limit runtime. The test will run until it completes the configured I/O workload or until it has run for this specified amount of time, whichever occurs first.
refill_buffers: If this option is given, fio will refill the I/O buffers on every submit.
norandommap: Normally fio will cover every block of the file when doing random I/O. If this option is given, fio will just get a new random offset without looking at past I/O history.
randrepeat: Seed all random number generators in a predictable way so the pattern is repeatable across runs.
group_reporting: To see the final report per-group instead of per-job, use group_reporting. Jobs in a file will be part of the same reporting group, unless if separated by a stonewall, or by using new_group.
size: The total size of file I/O for each thread of this job.
rw: Type of I/O pattern
bs: The block size in bytes used for I/O units.
iodepth: Number of I/O units to keep in flight against the file.
numjobs: Create the specified number of clones of this job.

multinode

Executables:

multinode-run, cleanup

Variables:

job_file: Job file to run. If a job file is not specified, one will be generated from variables.
job_file_path: Path to job file.
out_file: File to write results
out_format: Format to write results. Defaults to 'json' for Ramble to analyze results.
job_name: Job name
directory: Used to place files in a different location than ./
ioengine: I/O engine
direct: If true, use non-buffered I/O
buffered: If true, use buffered I/O. This is the opposite of the direct option
time_based: If set, fio will run for the duration of the runtime specified even if the file(s) are completely read or written.
runtime: Limit runtime. The test will run until it completes the configured I/O workload or until it has run for this specified amount of time, whichever occurs first.
refill_buffers: If this option is given, fio will refill the I/O buffers on every submit.
norandommap: Normally fio will cover every block of the file when doing random I/O. If this option is given, fio will just get a new random offset without looking at past I/O history.
randrepeat: Seed all random number generators in a predictable way so the pattern is repeatable across runs.
group_reporting: To see the final report per-group instead of per-job, use group_reporting. Jobs in a file will be part of the same reporting group, unless if separated by a stonewall, or by using new_group.
size: The total size of file I/O for each thread of this job.
rw: Type of I/O pattern
bs: The block size in bytes used for I/O units.
iodepth: Number of I/O units to keep in flight against the file.
numjobs: Create the specified number of clones of this job.

gromacs¶

Ramble applications:

gromacs/application.py

Description:

Define a Gromacs application

Workloads:

water_gmx50

Executables:

print-binary-info, pre-process, execute-gen

Inputs:

water_gmx50_bare

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
size: Workload size
type: Workload type.
input_path: Input path for workload

water_bare

Executables:

print-binary-info, pre-process, execute-gen

Inputs:

water_bare_hbonds

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
size: Workload size
type: Workload type.
input_path: Input path for workload

lignocellulose

Executables:

print-binary-info, execute

Inputs:

lignocellulose: A model of cellulose and lignocellulosic biomass in an aqueous solution. This system of 3.3M atoms is inhomogeneous, at least with GROMACS 4.5. This system uses reaction-fieldelectrostatics instead of PME and therefore should scale well.

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
input_path: Input path for workload

hecbiosim

Executables:

print-binary-info, execute

Inputs:

HECBioSim: https://www.hecbiosim.ac.uk/access-hpc/benchmarks

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
input_path: Input path for workload
type: Workload type. Valid values are Crambin, Glutamine-Binding-Protein, hEGFRDimer, hEGFRDimerPair, hEGFRDimerSmallerPL, hEGFRtetramerPair

benchpep

Executables:

print-binary-info, execute

Inputs:

BenchPEP: 12M Atoms, Peptides in Water, 2fs time step, all bonds constrained. https://www.mpinat.mpg.de/grubmueller/bench

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
input_path: Input path for workload

benchpep_h

Executables:

print-binary-info, execute

Inputs:

BenchPEP_h: 12M Atoms, Peptides in Water, 2fs time step, h-bonds constrained. https://www.mpinat.mpg.de/grubmueller/bench

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
input_path: Input path for workload

benchmem

Executables:

print-binary-info, execute

Inputs:

BenchMEM: 82k atoms, protein in membrane surrounded by water, 2 fs time step. https://www.mpinat.mpg.de/grubmueller/bench

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
input_path: Input path for workload

benchrib

Executables:

print-binary-info, execute

Inputs:

BenchRIB: 2 M atoms, ribosome in water, 4 fs time step. https://www.mpinat.mpg.de/grubmueller/bench

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
input_path: Input path for workload

stmv_rf

Executables:

print-binary-info, pre-process, execute-gen

Inputs:

JCP_benchmarks: GROMACS_heterogeneous_parallelization_benchmark_info_and_systems_JCP

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
input_path: Input path for workload
type: Workload type for JCP_benchmarks
input_path: Input path for JCP_benchmark {workload_name}

stmv_pme

Executables:

print-binary-info, pre-process, execute-gen

Inputs:

JCP_benchmarks: GROMACS_heterogeneous_parallelization_benchmark_info_and_systems_JCP

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
input_path: Input path for workload
type: Workload type for JCP_benchmarks
input_path: Input path for JCP_benchmark {workload_name}

rnase_cubic

Executables:

print-binary-info, pre-process, execute-gen

Inputs:

JCP_benchmarks: GROMACS_heterogeneous_parallelization_benchmark_info_and_systems_JCP

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
input_path: Input path for workload
type: Workload type for JCP_benchmarks
input_path: Input path for JCP_benchmark {workload_name}

ion_channel

Executables:

print-binary-info, pre-process, execute-gen

Inputs:

JCP_benchmarks: GROMACS_heterogeneous_parallelization_benchmark_info_and_systems_JCP

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
input_path: Input path for workload
type: Workload type for JCP_benchmarks
input_path: Input path for JCP_benchmark {workload_name}

adh_dodec

Executables:

print-binary-info, pre-process, execute-gen

Inputs:

JCP_benchmarks: GROMACS_heterogeneous_parallelization_benchmark_info_and_systems_JCP

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
input_path: Input path for workload
input_path: Input path for JCP_benchmark {workload_name}
type: Workload type for JCP_benchmarks

water_33m

Executables:

print-binary-info, execute

Inputs:

water_33m: A cubic box with 33 million water molecules (~100 million atoms). This deck is provided by Dr. Carsten Kutzner.

Variables:

additional_args: Additiaonl Exec Args
gmx: Name of the gromacs binary
grompp: How to run grompp
mdrun: How to run mdrun
nsteps: Simulation steps
resetstep: Reset performance counters at this step
verbose: Set to empty string to run without verbose mode
notunepme: Whether to set -notunepme for mdrun
dlb: Whether to use dynamic load balancing for mdrun
input_path: Input path for workload

hmmer¶

Ramble applications:

hmmer/application.py

Description:

HMMER is used for searching sequence databases for sequence homologs, and for making sequence alignments. It implements methods using probabilistic models called profile hidden Markov models (profile HMMs). It is often used with profile databases such as Pfam (for protein families), Rfam (for non-coded RNA families), Dfam (for repetitive DNA based), etc. Homepage: www.hmmer.org

Workloads:

fasta_pfam

Executables:

execute, tail_hmmsearch_out

Inputs:

Pfam_A: The Pfam database is a large collection of protein families, each represented by multiple sequence alignments and hidden Markov models (HMMs).
uniprot_sprot_fasta: Uniprot Swiss Prot fasta search input

Variables:

database_path: Database path for Pfam-A
input_path: Input path for uniprot_sprot.fasta

hostname¶

Ramble applications:

hostname/application.py

Description:

This is an example application that will simply run the hostname command

Workloads:

local

Executables:: local

local_bg

Executables:: local_bg, local_bg2

serial

Executables:: serial

parallel

Executables:: parallel

hp2p¶

Ramble applications:

hp2p/application.py

Description:

HP2P (Heavy Peer To Peer) benchmark is a test which performs MPI Point- to-Point non-blocking communications between all MPI processes. Its goal is to measure the bandwidths and the latencies in a situation where the network is busy. This benchmark can help to detect network problems like congestions or problems with switches or links.

Workloads:

standard

Executables:

execute

Variables:

iter: Number of iterations
msg_size: Message size
nb_msg: Number of msg per comm

hpcc¶

Ramble applications:

hpcc/application.py

Description:

Define the HPCC application HPCC is a collection of multiple benchmarks, which include: - HPL - DGEMM - STREAM - PTRANS - MPIRandomAccess - FFT - LatencyBandwidth

Workloads:

standard

Executables:

stage-config, execute

Inputs:

hpccinf: Input/Config file for HPCC benchmark

Variables:

config_file: Default config file
out_file: Output file for results

hpcg¶

Ramble applications:

hpcg/application.py

Description:

The High Performance Conjugate Gradients (HPCG) Benchmark project is an effort to create a new metric for ranking HPC systems. HPCG is intended as a complement to the High Performance LINPACK (HPL) benchmark, currently used to rank the TOP500 computing systems. The computational and data access patterns of HPL are still representative of some important scalable applications, but not all. HPCG is designed to exercise computational and data access patterns that more closely match a different and broad set of important applications, and to give incentive to computer system designers to invest in capabilities that will have impact on the collective performance of these applications.

Workloads:

standard

Executables:

execute, move-log

Variables:

matrix_size: Dimensions of the matrix to use
iterations: Number of iterations to perform
out_file: Output file for results

hpl¶

Ramble applications:

hpl/application.py

Description:

Define HPL application

Workloads:

standard

Executables:

execute

Variables:

output_file: Output file name (if any)
device_out: Output device
N-Ns: Number of problems sizes
Ns: Problem sizes
N-NBs: Number of NBs
NBs: NB values
PMAP: PMAP Process mapping. (0=Row-, 1=Column-Major)
N-Grids: Number of process grids (P x Q)
Ps: P values
Qs: Q values
threshold: Residual threshold
NPFACTs: Number of PFACTs
PFACTs: PFACT Values
N-NBMINs: Number of NBMINs
NBMINs: NBMIN values
N-NDIVs: Number of NDIVs
NDIVs: NDIV values
N-RFACTs: Number of RFACTs
RFACTs: RFACT values
N-BCASTs: Number of BCASTs
BCASTs: BCAST values
N-DEPTHs: Number of DEPTHs
DEPTHs: DEPTH values
SWAP: Swapping algorithm
swapping_threshold: Swapping threshold
L1: Storage for upper triangular portion of columns
U: Storage for the rows of U
Equilibration: Determines if equilibration should be enabled or disabled.
mem_alignment: Sets the alignment in doubles for memory addresses
used_percent_comment: Comment about used percent of memory for workload

calculator

Executables:

execute

Variables:

used_percent_comment: Comment about used percent of memory for workload
percent_mem: Percent of memory to use (default 85)
memory_per_node: Memory per node in GB
block_size: Size of each block
pfact: PFACT for optimized calculator
nbmin: NBMIN for optimized calculator
rfact: RFACT for optimized calculator
bcast: BCAST for optimized calculator
depth: DEPTH for optimized calculator
NPFACTs: Number of PFACTs
PFACTs: PFACT Values
N-NBMINs: Number of NBMINs
NBMINs: NBMIN values
N-RFACTs: Number of RFACTs
RFACTs: RFACT values

intel-hpcg¶

Ramble applications:

intel-hpcg/application.py

Description:

Define HPCG application using Intel MKL optimized binary from intel- oneapi-mkl package

Workloads:

standard

Executables:

set_vars, execute, move-log, reformat-output

Variables:

matrix_size: Dimensions of the matrix to use
iterations: Number of iterations to perform
out_file: Output file for results
exec_name: Name of executable to use for Intel HPCG
exec_name: Name of executable to use for Intel HPCG
hpcg_exec_path: Path to HPCG executable
hpcg_exec_path: Path to HPCG executable

intel-hpl¶

Ramble applications:

intel-hpl/application.py

Description:

Define HPL application using Intel MKL optimized binary from intel- oneapi-mkl package

Workloads:

standard

Executables:

stage-binary, execute

Variables:

hpl_run_script: Path to the HPL run script.
output_file: Output file name (if any)
device_out: Output device
N-Ns: Number of problems sizes
Ns: Problem sizes
N-NBs: Number of NBs
NBs: NB values
PMAP: PMAP Process mapping. (0=Row-, 1=Column-Major)
N-Grids: Number of process grids (P x Q)
Ps: P values
Qs: Q values
threshold: Residual threshold
NPFACTs: Number of PFACTs
PFACTs: PFACT Values
N-NBMINs: Number of NBMINs
NBMINs: NBMIN values
N-NDIVs: Number of NDIVs
NDIVs: NDIV values
N-RFACTs: Number of RFACTs
RFACTs: RFACT values
N-BCASTs: Number of BCASTs
BCASTs: BCAST values
N-DEPTHs: Number of DEPTHs
DEPTHs: DEPTH values
SWAP: Swapping algorithm
swapping_threshold: Swapping threshold
L1: Storage for upper triangular portion of columns
U: Storage for the rows of U
Equilibration: Determines if equilibration should be enabled or disabled.
mem_alignment: Sets the alignment in doubles for memory addresses
used_percent_comment: Comment about used percent of memory for workload
hpl_bench_dir: Path to Intel HPL benchmarks directory
numa_per_mpi: numa per mpi process

calculator

Executables:

stage-binary, execute

Variables:

hpl_run_script: Path to the HPL run script.
used_percent_comment: Comment about used percent of memory for workload
percent_mem: Percent of memory to use (default 85)
memory_per_node: Memory per node in GB
block_size: Size of each block
pfact: PFACT for optimized calculator
nbmin: NBMIN for optimized calculator
rfact: RFACT for optimized calculator
bcast: BCAST for optimized calculator
depth: DEPTH for optimized calculator
NPFACTs: Number of PFACTs
PFACTs: PFACT Values
N-NBMINs: Number of NBMINs
NBMINs: NBMIN values
N-RFACTs: Number of RFACTs
RFACTs: RFACT values
hpl_bench_dir: Path to Intel HPL benchmarks directory
numa_per_mpi: numa per mpi process
bcast: BCAST for Intel MKL optimized calculator
bcast: BCAST for Intel MKL optimized calculator

standard

Executables:

stage-binary, stage-runme, execute

Variables:

hpl_run_script: Path to the HPL run script.
output_file: Output file name (if any)
device_out: Output device
N-Ns: Number of problems sizes
Ns: Problem sizes
N-NBs: Number of NBs
NBs: NB values
PMAP: PMAP Process mapping. (0=Row-, 1=Column-Major)
N-Grids: Number of process grids (P x Q)
Ps: P values
Qs: Q values
threshold: Residual threshold
NPFACTs: Number of PFACTs
PFACTs: PFACT Values
N-NBMINs: Number of NBMINs
NBMINs: NBMIN values
N-NDIVs: Number of NDIVs
NDIVs: NDIV values
N-RFACTs: Number of RFACTs
RFACTs: RFACT values
N-BCASTs: Number of BCASTs
BCASTs: BCAST values
N-DEPTHs: Number of DEPTHs
DEPTHs: DEPTH values
SWAP: Swapping algorithm
swapping_threshold: Swapping threshold
L1: Storage for upper triangular portion of columns
U: Storage for the rows of U
Equilibration: Determines if equilibration should be enabled or disabled.
mem_alignment: Sets the alignment in doubles for memory addresses
used_percent_comment: Comment about used percent of memory for workload
hpl_bench_dir: Path to Intel HPL benchmarks directory
numa_per_mpi: numa per mpi process

calculator

Executables:

stage-binary, stage-runme, execute

Variables:

hpl_run_script: Path to the HPL run script.
used_percent_comment: Comment about used percent of memory for workload
percent_mem: Percent of memory to use (default 85)
memory_per_node: Memory per node in GB
block_size: Size of each block
pfact: PFACT for optimized calculator
nbmin: NBMIN for optimized calculator
rfact: RFACT for optimized calculator
bcast: BCAST for optimized calculator
depth: DEPTH for optimized calculator
NPFACTs: Number of PFACTs
PFACTs: PFACT Values
N-NBMINs: Number of NBMINs
NBMINs: NBMIN values
N-RFACTs: Number of RFACTs
RFACTs: RFACT values
hpl_bench_dir: Path to Intel HPL benchmarks directory
numa_per_mpi: numa per mpi process
bcast: BCAST for Intel MKL optimized calculator
bcast: BCAST for Intel MKL optimized calculator

intel-mlc¶

Ramble applications:

intel-mlc/application.py

Description:

Workloads:

max_bandwidth

Executables:

configure_hugepage, execute_bw

Variables:

exec_name: Name of executable to use for Intel MLC
mlc_exec_path: Path to MLC executable
cpu_list: Comma delimited list of CPUs to run tests with. Defined by application if omitted from an experiment definition.
additional_args: Additional command line arguments for the mlc binary
cores_per_node: Maximum number of cores to use per node when generating cpu_list
thread_distribution: Thread distribution method when generating cpu_list
isa_flag: Flag for controlling the vector instructions used.
buffer_size: Size of buffer per thread
spread_divisions: Number of blocks to spread threads over
nr_hugepages: Number of hugepages to configure

intel-mpi-benchmarks¶

Ramble applications:

intel-mpi-benchmarks/application.py

Description:

Intel MPI Benchmark application. https://www.intel.com/content/www/us/en /developer/articles/technical/intel-mpi-benchmarks.html https://github.com/intel/mpi-benchmarks https://www.intel.com/content/www/us/en/develop/documentation/imb-user- guide/top.html

Workloads:

pingpong

Executables:

pingpong

Variables:

install_path: User configurable dir to executables
pingpong_type: Pingpong Algorithm to Use
num_iterations: Number of iterations to test over
msglog_min: Min Message Size (power of 2)
msglog_max: Max Message Size (power of 2)
additional_args: Additional cmd-line arguments

multi-pingpong

Executables:

multi-pingpong

Variables:

num_cores: Number of cores
multi_val: Value to pass to the multi arg
map_args: Args to map by
install_path: User configurable dir to executables
num_iterations: Number of iterations to test over
msglog_min: Min Message Size (power of 2)
msglog_max: Max Message Size (power of 2)
additional_args: Additional cmd-line arguments

collective

Executables:

collective

Variables:

install_path: User configurable dir to executables
collective_type: Collective type to test
min_collective_ranks: Minimum number of ranks to use in a collective operation
num_iterations: Number of iterations to test over
msglog_min: Min Message Size (power of 2)
msglog_max: Max Message Size (power of 2)
additional_args: Additional cmd-line arguments

transfer

Executables:

transfer

Variables:

install_path: User configurable dir to executables
transfer_type: (Non-collective) Transfer pattern to benchmark
num_iterations: Number of iterations to test over
msglog_min: Min Message Size (power of 2)
msglog_max: Max Message Size (power of 2)
additional_args: Additional cmd-line arguments

ior¶

Ramble applications:

ior/application.py

Description:

Define the IOR parallel IO benchmark. Also includes

Workloads:

multi-file

Executables:

ior-prep, ior

Variables:

transfer-size: Transfer Size
block-size: Block Size
segment-count: Segment Count
iterations: Iterations
file_args: FilePerProc flag
target_directory: Target directory for the r/w test. This can be used to target different file systems.
additional_args: Additional args to pass. The default aims to suppress the use of page cache.

single-file

Executables:

ior-prep, ior

Variables:

transfer-size: Transfer Size
block-size: Block Size
segment-count: Segment Count
iterations: Iterations
file_args: FilePerProc flag, default to empty
target_directory: Target directory for the r/w test. This can be used to target different file systems.
additional_args: Additional args to pass. The default aims to suppress the use of page cache.

write-once-read-many

Executables:

ior-prep, ior-write, ior-read, ior-finalize

Variables:

transfer-size: Transfer Size
block-size: Block Size
segment-count: Segment Count
iterations: Iterations
file_args: FilePerProc flag
target_directory: Target directory for the r/w test. This can be used to target different file systems.
write_args: Additional args to pass for write. Used to apply Lustre Stripe count
read_args: Additional args to pass for read.
file_stat_cmd: Command to check file distribution or storage status, e.g., 'lfs getstripe' or 'lfs df -h'.

iozone¶

Ramble applications:

iozone/application.py

Description:

Define the Iozone file system benchmark.

Workloads:

cluster

Executables:

gen_clientlist, execute

Variables:

test_args: Specify the tests to run
file_size: Size of the test file
record_size: Record size
working_directory: Working directory for the test
additional_args
rsh_alternative: Alternative remote access mechanism of rsh
hostfile_path
clientlist_path
iozone_bin_path

iperf2¶

Ramble applications:

iperf2/application.py

Description:

Define the iperf2 application

Workloads:

iperf2_server

Executables:

iperf2

Variables:

input_flags: Input flags to start iperf2 in server mode
additional_flags: Allow users to pass additional flags

iperf2_client

Executables:

iperf2

Variables:

input_flags: Input flags to start iperf2 in server mode
time: time in seconds to listen for new connections as well as to receive traffic (default not set)
interval: seconds between periodic bandwidth reports
host: run in client mode, connecting to <host>
num_threads: number of parallel client threads to run
additional_flags: Allow users to pass additional flags

iperf2_custom

Executables:

iperf2

Variables:

input_flags: Input flags in custom mode
additional_flags: Allow users to pass additional flags

lammps¶

Ramble applications:

lammps/application.py

Description:

Define LAMMPS application

Workloads:

lj

Executables:

stage-input, configure-size-scale, configure-run-timesteps, execute

Inputs:

leonard-jones: Atomic fluid. 32k atoms. 100 timesteps. https://www.lammps.org/bench.html#lj

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path for the workload input file.
timesteps: Number of timesteps
xx: xx value
yy: yy value
zz: zz value

eam

Executables:

stage-input, configure-size-scale, configure-run-timesteps, execute

Inputs:

eam: Cu metallic solid with embedded atom method potential. 32k atoms. https://www.lammps.org/bench.html#eam

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path for the workload input file.
timesteps: Number of timesteps
xx: xx value
yy: yy value
zz: zz value

chain

Executables:

stage-input, set-data-path, configure-run-timesteps, execute

Inputs:

polymer-chain-melt: Bead-spring polymer melt with 100-mer chains and FENE bonds. 32k atoms. 100 timesteps. https://www.lammps.org/bench.html#chain
lammps-stage: Stage of lammps source from release

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path for the workload input file.
timesteps: Number of timesteps

chute

Executables:

stage-input, configure-run-timesteps, execute

Inputs:

chute: Chute flow of packed granular particles with frictional history potential. 32k atoms. 100 timeteps. https://www.lammps.org/bench.html#chute

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path for the workload input file.
timesteps: Number of timesteps

rhodo

Executables:

stage-input, set-data-path, execute

Inputs:

rhodo: All-atom rhodopsin protein in solvated lipid bilayer with CHARMM force field, long-range Coulombics via PPPM (particle-particle particle mesh), SHAKE constraints. This model contains counter-ions and a reduced amount of water to make a 32K atom system. 32k atoms. 100 timesteps. https://www.lammps.org/bench.html#rhodo
lammps-stage: Stage of lammps source from release

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path for the workload input file.
timesteps: Number of timesteps

intel.airebo

Executables:

stage-input, change-root, configure-size-scale, configure-timestep-variables, execute

Inputs:

lammps-stage: Stage of lammps source from release

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path to input file
input_path: Path to input file for workload
warmup_timesteps: Number of warmup timesteps
main_timesteps: Number of main timesteps
timestep_multiplier: Multiplier for main timesteps
xx: xx value
yy: yy value
zz: zz value

intel.dpd

Executables:

stage-input, change-root, configure-size-scale, configure-timestep-variables, execute

Inputs:

lammps-stage: Stage of lammps source from release

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path to input file
input_path: Path to input file for workload
warmup_timesteps: Number of warmup timesteps
main_timesteps: Number of main timesteps
timestep_multiplier: Multiplier for main timesteps
xx: xx value
yy: yy value
zz: zz value

intel.eam

Executables:

stage-input, change-root, configure-size-scale, configure-timestep-variables, execute

Inputs:

lammps-stage: Stage of lammps source from release

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path to input file
input_path: Path to input file for workload
warmup_timesteps: Number of warmup timesteps
main_timesteps: Number of main timesteps
timestep_multiplier: Multiplier for main timesteps
xx: xx value
yy: yy value
zz: zz value

intel.lc

Executables:

stage-input, change-root, configure-timestep-variables, execute

Inputs:

lammps-stage: Stage of lammps source from release

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path to input file
input_path: Path to input file for workload
warmup_timesteps: Number of warmup timesteps
main_timesteps: Number of main timesteps
timestep_multiplier: Multiplier for main timesteps

intel.lj

Executables:

stage-input, change-root, configure-size-scale, configure-timestep-variables, execute

Inputs:

lammps-stage: Stage of lammps source from release

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path to input file
input_path: Path to input file for workload
warmup_timesteps: Number of warmup timesteps
main_timesteps: Number of main timesteps
timestep_multiplier: Multiplier for main timesteps
xx: xx value
yy: yy value
zz: zz value

intel.rhodo

Executables:

stage-input, change-root, configure-timestep-variables, execute

Inputs:

lammps-stage: Stage of lammps source from release

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path to input file
input_path: Path to input file for workload
warmup_timesteps: Number of warmup timesteps
main_timesteps: Number of main timesteps
timestep_multiplier: Multiplier for main timesteps

intel.sw

Executables:

stage-input, change-root, configure-size-scale, configure-timestep-variables, execute

Inputs:

lammps-stage: Stage of lammps source from release

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path to input file
input_path: Path to input file for workload
warmup_timesteps: Number of warmup timesteps
main_timesteps: Number of main timesteps
timestep_multiplier: Multiplier for main timesteps
xx: xx value
yy: yy value
zz: zz value

intel.tersoff

Executables:

stage-input, change-root, configure-size-scale, configure-timestep-variables, execute

Inputs:

lammps-stage: Stage of lammps source from release

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path to input file
input_path: Path to input file for workload
warmup_timesteps: Number of warmup timesteps
main_timesteps: Number of main timesteps
timestep_multiplier: Multiplier for main timesteps
xx: xx value
yy: yy value
zz: zz value

intel.water

Executables:

stage-input, change-root, configure-timestep-variables, execute

Inputs:

lammps-stage: Stage of lammps source from release

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path to input file
input_path: Path to input file for workload
warmup_timesteps: Number of warmup timesteps
main_timesteps: Number of main timesteps
timestep_multiplier: Multiplier for main timesteps

hns-reaxff

Executables:

stage-contents, stage-input-file, configure-reaxff, execute

Inputs:

lammps-stage: Stage of lammps source from release

Variables:

intel_test_path: Path for Intel Test files
examples_path: Path for example files
input_stage: Stage name of LAMMPS input archive
lammps_flags: Additional execution flags for lammps
input_path: Path to input file for workload
input_file: hns-reaxff input file name
timesteps: Number of timesteps
x: x value
y: y value
z: z value

lulesh¶

Ramble applications:

lulesh/application.py

Description:

Define LULESH application

Workloads:

standard

Executables:

execute

Variables:

size_flag: Problem size in a single dimension. Real problem is size^3. Needs to be prefixed by -s
iteration_flag: Fixed number of iterations to perform. Needs to be prefixed by -i
flags: Flags to pass in to LULESH

maxtext¶

Ramble applications:

maxtext/application.py

Description:

MaxText is a high performance, highly scalable, open-source LLM written in pure Python/Jax and targeting Google Cloud TPUs and GPUs for training and inference. MaxText achieves high MFUs and scales from single host to very large clusters while staying simple and "optimization-free" thanks to the power of Jax and the XLA compiler.

Workloads:

train

Executables:

train

Inputs:

base_config: Base configuration base.yml
model_config: Additional config to override base.yml

Variables:

maxtext_path: Path to maxtext root
train_path: Path to train.py
config_path: Path to config
base_config_source: Path or URL to base config
model_config_source: path or URL to secondary model config
run_name: Run name
metrics_file: For testing, local file that stores scalar metrics. If empty, no metrics are written.
maxtext_mount: Container mount for maxtext root
container_mounts: All container mounts in a ramble variable

md-test¶

Ramble applications:

md-test/application.py

Description:

Define the MDTest parallel IO benchmark

Workloads:

multi-file

Executables:

ior

Variables:

num-objects: Number of files and dirs to create (per rank)
iterations: Number of iterations
additional-args: Pass additional args, such as working directiroy (-d)

minixyce¶

Ramble applications:

minixyce/application.py

Description:

Define miniXyce application

Workloads:

cir1

Executables:

stage-network-file, execute

Variables:

t_start: Start time
t_step: Time step
t_stop: Stop time
tol: Tolerance
k: Tells GMRES how often to restart

cir2

Executables:

stage-network-file, execute

Variables:

t_start: Start time
t_step: Time step
t_stop: Stop time
tol: Tolerance
k: Tells GMRES how often to restart

cir3

Executables:

stage-network-file, execute

Variables:

t_start: Start time
t_step: Time step
t_stop: Stop time
tol: Tolerance
k: Tells GMRES how often to restart

cir4

Executables:

stage-network-file, execute

Variables:

t_start: Start time
t_step: Time step
t_stop: Stop time
tol: Tolerance
k: Tells GMRES how often to restart

cir5

Executables:

stage-network-file, execute

Variables:

t_start: Start time
t_step: Time step
t_stop: Stop time
tol: Tolerance
k: Tells GMRES how often to restart

RC_ladder

Executables:

generate_RC_ladder, execute

Variables:

t_start: Start time
t_step: Time step
t_stop: Stop time
tol: Tolerance
k: Tells GMRES how often to restart
num_ladder_stages: Number of Ladder stages for RC|RLC ladder|ladder2 test generation

RLC_ladder

Executables:

generate_RLC_ladder, execute

Variables:

t_start: Start time
t_step: Time step
t_stop: Stop time
tol: Tolerance
k: Tells GMRES how often to restart
num_ladder_stages: Number of Ladder stages for RC|RLC ladder|ladder2 test generation

RC_ladder2

Executables:

generate_RC_ladder2, execute

Variables:

t_start: Start time
t_step: Time step
t_stop: Stop time
tol: Tolerance
k: Tells GMRES how often to restart
num_ladder_stages: Number of Ladder stages for RC|RLC ladder|ladder2 test generation

RLC_ladder2

Executables:

generate_RLC_ladder2, execute

Variables:

t_start: Start time
t_step: Time step
t_stop: Stop time
tol: Tolerance
k: Tells GMRES how often to restart
num_ladder_stages: Number of Ladder stages for RC|RLC ladder|ladder2 test generation

namd¶

Ramble applications:

namd/application.py

Description:

Define NAMD application

Workloads:

stmv

Executables:

stage-input, execute

Inputs:

stmv: STMV (virus) benchmark (1,066,628 atoms, periodic, PME

Variables:

input_file: Input file for namd
namd_flags: Flags for running NAMD
input_path: Path to the workload inputs

ApoA1

Executables:

stage-input, execute

Inputs:

ApoA1: ApoA1 benchmark (92,224 atoms, periodic, PME)

Variables:

input_file: Input file for namd
namd_flags: Flags for running NAMD
input_path: Path to the workload inputs

ATPase

Executables:

stage-input, execute

Inputs:

ATPase: ATPase benchmark (327,506 atoms, periodic, PME)

Variables:

input_file: Input file for namd
namd_flags: Flags for running NAMD
input_path: Path to the workload inputs

20STMV

Executables:

stage-input, execute

Inputs:

20STMV: 20STMV and 210STMV benchmarks from SC14 paper

Variables:

input_file: Input file for namd
namd_flags: Flags for running NAMD
input_path: Path to the workload inputs

tiny

Executables:

stage-input, execute

Inputs:

tiny: tiny benchmark (507 atoms, periodic, PME)

Variables:

input_file: Input file for namd
namd_flags: Flags for running NAMD
input_path: Path to the workload inputs

interactive_BPTI

Executables:

stage-input, execute

Inputs:

interactive_BPTI: Interactive BPTI (882 atoms, small)

Variables:

input_file: Input file for namd
namd_flags: Flags for running NAMD
input_path: Path to the workload inputs

ER-GRE

Executables:

stage-input, execute

Inputs:

ER-GRE: ER-GRE benchmark (36573 atoms, spherical)

Variables:

input_file: Input file for namd
namd_flags: Flags for running NAMD
input_path: Path to the workload inputs

decalanin

Executables:

stage-input, execute

Inputs:

decalanin: decalanin (66 atoms, tiny, ancient)

Variables:

input_file: Input file for namd
namd_flags: Flags for running NAMD
input_path: Path to the workload inputs

tcl-forces

Executables:

stage-input, execute

Inputs:

tcl-forces: Tcl forces (decalanin)

Variables:

input_file: Input file for namd
namd_flags: Flags for running NAMD
input_path: Path to the workload inputs

nccl-tests¶

Ramble applications:

nccl-tests/application.py

Description:

These tests check both the performance and the correctness of NCCL operations. https://github.com/NVIDIA/nccl-tests

Workloads:

all-to-all

Executables:

all-to-all-execute

Variables:

num_gpus_per_task: Number of GPUs per task
begin_message_size: Beginning message size
end_message_size: Ending message size
message_size_factor: Factor used in size sweep
begin_message_size: Beginning message size
warmup_iters: Number of warmup iterations
num_iters: Number of iterations to perform
result_check: 0 to skip checking, 1 to enable checking
additional_args: Arguments for all reduce
nccl_tests_split_mask: How NCCL communicators should be split, if at all. "0x7" for rail-aligned, "0x0" for world-level.

all-reduce

Executables:

all-reduce-execute

Variables:

num_gpus_per_task: Number of GPUs per task
begin_message_size: Beginning message size
end_message_size: Ending message size
message_size_factor: Factor used in size sweep
begin_message_size: Beginning message size
warmup_iters: Number of warmup iterations
num_iters: Number of iterations to perform
result_check: 0 to skip checking, 1 to enable checking
additional_args: Arguments for all reduce
nccl_tests_split_mask: How NCCL communicators should be split, if at all. "0x7" for rail-aligned, "0x0" for world-level.

all-gather

Executables:

all-gather-execute

Variables:

num_gpus_per_task: Number of GPUs per task
begin_message_size: Beginning message size
end_message_size: Ending message size
message_size_factor: Factor used in size sweep
begin_message_size: Beginning message size
warmup_iters: Number of warmup iterations
num_iters: Number of iterations to perform
result_check: 0 to skip checking, 1 to enable checking
additional_args: Arguments for all reduce
nccl_tests_split_mask: How NCCL communicators should be split, if at all. "0x7" for rail-aligned, "0x0" for world-level.

reduce-scatter

Executables:

reduce-scatter-execute

Variables:

num_gpus_per_task: Number of GPUs per task
begin_message_size: Beginning message size
end_message_size: Ending message size
message_size_factor: Factor used in size sweep
begin_message_size: Beginning message size
warmup_iters: Number of warmup iterations
num_iters: Number of iterations to perform
result_check: 0 to skip checking, 1 to enable checking
additional_args: Arguments for all reduce
nccl_tests_split_mask: How NCCL communicators should be split, if at all. "0x7" for rail-aligned, "0x0" for world-level.

send-recv

Executables:

send-recv-execute

Variables:

num_gpus_per_task: Number of GPUs per task
begin_message_size: Beginning message size
end_message_size: Ending message size
message_size_factor: Factor used in size sweep
begin_message_size: Beginning message size
warmup_iters: Number of warmup iterations
num_iters: Number of iterations to perform
result_check: 0 to skip checking, 1 to enable checking
additional_args: Arguments for all reduce
nccl_tests_split_mask: How NCCL communicators should be split, if at all. "0x7" for rail-aligned, "0x0" for world-level.

ngspice¶

Ramble applications:

ngspice/application.py

Description:

Define NgSpice application

Workloads:

c1355

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c1355 workload

c1908

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c1908 workload

c2670

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c2670 workload

c3540

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c3540 workload

c432

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c432 workload

c499

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c499 workload

c6288

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c6288 workload

c7552

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c7552 workload

c880

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c880 workload

c1355_ann

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c1355 workload

c1908_ann

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c1908 workload

c2670_ann

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c2670 workload

c3540_ann

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c3540 workload

c499_ann

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c499 workload

c6288_ann

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c6288 workload

c880_ann

Executables:

execute

Inputs:

Ngspice_BenchMark_Inputs: Input files for the ngspice workload

Variables:

input_file: Path to input file for c880 workload

nvbandwidth¶

Ramble applications:

nvbandwidth/application.py

Description:

Define the nvbandwidth benchmark

Workloads:

all_benchmarks

Executables:

nvbandwidth

Variables:

transfer-size: Transfer Size

nvidia-dcgm¶

Ramble applications:

nvidia-dcgm/application.py

Description:

NVIDIA Data Center GPU Manager (DCGM). DCGM is a suite of tools for managing and monitoring NVIDIA GPUs in cluster environments. https://docs.nvidia.com/datacenter/dcgm/latest/user-guide/dcgm- diagnostics.html # Example Ramble configuration for the DCGM application ramble: env_vars: set: OMP_NUM_THREADS: '{n_threads}' variants: package_manager: user-managed workflow_manager: user-managed variables: n_ranks: 1 processes_per_node: 1 mpi_command: '' # DCGM commands don't need MPI batch_submit: '{execute_experiment}' # Simple direct execution applications: nvidia-dcgm: workloads: diag: experiments: dcgm_r{diag_level}: # Parameterize dcgmi diag level variables: diag_level: [1,2,3] # No specific software packages needed if dcgmi is in the PATH software: packages: {} environments: {}

Workloads:

diag

Executables:

diag

Variables:

diag_level: Level for dcgmi diagnostics

nvidia-hpcg¶

Ramble applications:

nvidia-hpcg/application.py

Description:

NVIDIA's HPCG benchmark accelerates the High Performance Conjugate Gradients (HPCG) Benchmark. HPCG is a software package that performs a fixed number of multigrid preconditioned (using a symmetric Gauss-Seidel smoother) conjugate gradient (PCG) iterations using double precision (64-bit) floating point values.

Workloads:

standard

Executables:

execute

Variables:

internal_mpi_command: MPI Command for execution using container built-in MPI
nvshmem_disable_cuda_vmm
pmix_mca_gds: PMIX MCA gds
ompi_mca_btl: OpenMPI MCA btl
ompi_mca_pml: OpenMPI MCA pml
ucx_net_devices: UCX Net Devices
ucx_max_rndv_rails: UCX MAximum RNDV Rails
matrix_size: Dimensions of the matrix to use
iterations: Number of iterations to perform
out_file: Output file for results

nvidia-hpl¶

Ramble applications:

nvidia-hpl/application.py

Description:

This application defines how to run NVIDIA's optimized version of HPL, which is contained in NVIDIA's HPC-Benchmarks collection. The NVIDIA HPC-Benchmarks collection provides four benchmarks (HPL, HPL-MxP, HPCG, and STREAM) widely used in the HPC community optimized for performance on NVIDIA accelerated HPC systems. NVIDIA's HPL and HPL-MxP benchmarks provide software packages to solve a (random) dense linear system in double precision (64-bit) arithmetic and in mixed precision arithmetic using Tensor Cores, respectively, on distributed-memory computers equipped with NVIDIA GPUs, based on the Netlib HPL benchmark and HPL-MxP benchmark. https://catalog.ngc.nvidia.com/orgs/nvidia/containers/hpc- benchmarks

Workloads:

standard

Executables:

execute

Variables:

output_file: Output file name (if any)
device_out: Output device
N-Ns: Number of problems sizes
Ns: Problem sizes
N-NBs: Number of NBs
NBs: NB values
PMAP: PMAP Process mapping. (0=Row-, 1=Column-Major)
N-Grids: Number of process grids (P x Q)
Ps: P values
Qs: Q values
threshold: Residual threshold
NPFACTs: Number of PFACTs
PFACTs: PFACT Values
N-NBMINs: Number of NBMINs
NBMINs: NBMIN values
N-NDIVs: Number of NDIVs
NDIVs: NDIV values
N-RFACTs: Number of RFACTs
RFACTs: RFACT values
N-BCASTs: Number of BCASTs
BCASTs: BCAST values
N-DEPTHs: Number of DEPTHs
DEPTHs: DEPTH values
SWAP: Swapping algorithm
swapping_threshold: Swapping threshold
L1: Storage for upper triangular portion of columns
U: Storage for the rows of U
Equilibration: Determines if equilibration should be enabled or disabled.
mem_alignment: Sets the alignment in doubles for memory addresses
used_percent_comment: Comment about used percent of memory for workload
internal_mpi_command: MPI Command for execution using container built-in MPI
nvshmem_disable_cuda_vmm
pmix_mca_gds: PMIX MCA gds
ompi_mca_btl: OpenMPI MCA btl
ompi_mca_pml: OpenMPI MCA pml
ucx_net_devices: UCX Net Devices
ucx_max_rndv_rails: UCX MAximum RNDV Rails
hpl_fct_comm_policy: Which communication library to use in the panel factorization. 0 = NVSHMEM, 1 = Host MPI
hpl_use_nvshmem: Whether to use NVSHMEM or not. 0 = Disable, 1 = Enable.
hpl_p2p_as_bcast: 0 = ncclBcast, 1 = ncclSend/Recv, 2 = CUDA-aware MPI, 3 = host MPI, 4 = NVSHMEM
hpl_nvshmem_swap: Performs row swaps using NVSHMEM instead of NCCL. 0 = Disable, 1 = Enable.
hpl_chunk_size_nbs: Number of matrix blocks to group for computations. Needs to be > 0
hpl_dist_trsm_flag: Perform the solve step (TRSM) in parallel, rather than on only the ranks that own part of the matrix.
hpl_cta_per_fct: Sets the number of CTAs (thread blocks) for factorization. Needs to be > 0.
hpl_alloc_hugepages: Use 2MB hugepages for host-side allocations. Done through the madvise syscall.
warmup_end_prog: Runs the main loop once before the 'real' run. Stops the warmup at x%. Values can be 1 - 100.
test_loops: Runs the main loop X many times
hpl_cusolver_mp_tests: Runs several tests of individual components of HPL (GEMMS, comms, etc.)
hpl_cusolver_mp_tests_gemm_iters: Number of repeat GEMM calls in tests. Needs to be > 0.
hpl_ooc_mode: Enables / disales out-of-core mode
hpl_ooc_max_gpu_mem: Limits the amount of GPU memory used for OOC. In GiB. Needs to be >= -1.
hpl_ooc_tile_m: Row blocking factor. Needs to be > 0
hpl_ooc_tile_n: Column blocking factor. Needs to be > 0
hpl_ooc_num_streams: Number of streams used for OCC operations
hpl_ooc_safe_size: GPU memory (in GiB) needed for driver. This amount will not be used by HPL OCC

calculator

Executables:

execute

Variables:

used_percent_comment: Comment about used percent of memory for workload
percent_mem: Percent of memory to use (default 85)
memory_per_node: Memory per node in GB
block_size: Size of each block
pfact: PFACT for optimized calculator
nbmin: NBMIN for optimized calculator
rfact: RFACT for optimized calculator
bcast: BCAST for optimized calculator
depth: DEPTH for optimized calculator
NPFACTs: Number of PFACTs
PFACTs: PFACT Values
N-NBMINs: Number of NBMINs
NBMINs: NBMIN values
N-RFACTs: Number of RFACTs
RFACTs: RFACT values
internal_mpi_command: MPI Command for execution using container built-in MPI
nvshmem_disable_cuda_vmm
pmix_mca_gds: PMIX MCA gds
ompi_mca_btl: OpenMPI MCA btl
ompi_mca_pml: OpenMPI MCA pml
ucx_net_devices: UCX Net Devices
ucx_max_rndv_rails: UCX MAximum RNDV Rails
hpl_fct_comm_policy: Which communication library to use in the panel factorization. 0 = NVSHMEM, 1 = Host MPI
hpl_use_nvshmem: Whether to use NVSHMEM or not. 0 = Disable, 1 = Enable.
hpl_p2p_as_bcast: 0 = ncclBcast, 1 = ncclSend/Recv, 2 = CUDA-aware MPI, 3 = host MPI, 4 = NVSHMEM
hpl_nvshmem_swap: Performs row swaps using NVSHMEM instead of NCCL. 0 = Disable, 1 = Enable.
hpl_chunk_size_nbs: Number of matrix blocks to group for computations. Needs to be > 0
hpl_dist_trsm_flag: Perform the solve step (TRSM) in parallel, rather than on only the ranks that own part of the matrix.
hpl_cta_per_fct: Sets the number of CTAs (thread blocks) for factorization. Needs to be > 0.
hpl_alloc_hugepages: Use 2MB hugepages for host-side allocations. Done through the madvise syscall.
warmup_end_prog: Runs the main loop once before the 'real' run. Stops the warmup at x%. Values can be 1 - 100.
test_loops: Runs the main loop X many times
hpl_cusolver_mp_tests: Runs several tests of individual components of HPL (GEMMS, comms, etc.)
hpl_cusolver_mp_tests_gemm_iters: Number of repeat GEMM calls in tests. Needs to be > 0.
hpl_ooc_mode: Enables / disales out-of-core mode
hpl_ooc_max_gpu_mem: Limits the amount of GPU memory used for OOC. In GiB. Needs to be >= -1.
hpl_ooc_tile_m: Row blocking factor. Needs to be > 0
hpl_ooc_tile_n: Column blocking factor. Needs to be > 0
hpl_ooc_num_streams: Number of streams used for OCC operations
hpl_ooc_safe_size: GPU memory (in GiB) needed for driver. This amount will not be used by HPL OCC
block_size: Size of each block
block_size: Size of each block
block_size: Size of each block

nvidia-hpl-mxp¶

Ramble applications:

nvidia-hpl-mxp/application.py

Description:

This application defines how to run NVIDIA's optimized version of HPL, which is contained in NVIDIA's HPC-Benchmarks collection. The NVIDIA HPC-Benchmarks collection provides four benchmarks (HPL, HPL-MxP, HPCG, and STREAM) widely used in the HPC community optimized for performance on NVIDIA accelerated HPC systems. NVIDIA's HPL and HPL-MxP benchmarks provide software packages to solve a (random) dense linear system in double precision (64-bit) arithmetic and in mixed precision arithmetic using Tensor Cores, respectively, on distributed-memory computers equipped with NVIDIA GPUs, based on the Netlib HPL benchmark and HPL-MxP benchmark. https://catalog.ngc.nvidia.com/orgs/nvidia/containers/hpc- benchmarks

Workloads:

standard

Executables:

execute

Variables:

output_file: Output file name (if any)
device_out: Output device
N-Ns: Number of problems sizes
Ns: Problem sizes
N-NBs: Number of NBs
NBs: NB values
PMAP: PMAP Process mapping. (0=Row-, 1=Column-Major)
N-Grids: Number of process grids (P x Q)
Ps: P values
Qs: Q values
threshold: Residual threshold
NPFACTs: Number of PFACTs
PFACTs: PFACT Values
N-NBMINs: Number of NBMINs
NBMINs: NBMIN values
N-NDIVs: Number of NDIVs
NDIVs: NDIV values
N-RFACTs: Number of RFACTs
RFACTs: RFACT values
N-BCASTs: Number of BCASTs
BCASTs: BCAST values
N-DEPTHs: Number of DEPTHs
DEPTHs: DEPTH values
SWAP: Swapping algorithm
swapping_threshold: Swapping threshold
L1: Storage for upper triangular portion of columns
U: Storage for the rows of U
Equilibration: Determines if equilibration should be enabled or disabled.
mem_alignment: Sets the alignment in doubles for memory addresses
used_percent_comment: Comment about used percent of memory for workload
internal_mpi_command: MPI Command for execution using container built-in MPI
nvshmem_disable_cuda_vmm
pmix_mca_gds: PMIX MCA gds
ompi_mca_btl: OpenMPI MCA btl
ompi_mca_pml: OpenMPI MCA pml
ucx_net_devices: UCX Net Devices
ucx_max_rndv_rails: UCX MAximum RNDV Rails
nporder: Major order to use for matrix
gpu_affinity: Colon delimited list of GPU IDs

calculator

Executables:

execute

Variables:

used_percent_comment: Comment about used percent of memory for workload
percent_mem: Percent of memory to use (default 85)
memory_per_node: Memory per node in GB
block_size: Size of each block
pfact: PFACT for optimized calculator
nbmin: NBMIN for optimized calculator
rfact: RFACT for optimized calculator
bcast: BCAST for optimized calculator
depth: DEPTH for optimized calculator
NPFACTs: Number of PFACTs
PFACTs: PFACT Values
N-NBMINs: Number of NBMINs
NBMINs: NBMIN values
N-RFACTs: Number of RFACTs
RFACTs: RFACT values
internal_mpi_command: MPI Command for execution using container built-in MPI
nvshmem_disable_cuda_vmm
pmix_mca_gds: PMIX MCA gds
ompi_mca_btl: OpenMPI MCA btl
ompi_mca_pml: OpenMPI MCA pml
ucx_net_devices: UCX Net Devices
ucx_max_rndv_rails: UCX MAximum RNDV Rails
block_size: Size of each block
block_size: Size of each block
block_size: Size of each block
block_size: Size of each block
nporder: Major order to use for matrix
gpu_affinity: Colon delimited list of GPU IDs

openfoam¶

Ramble applications:

openfoam/application.py

Description:

Define the Openfoam application

Workloads:

motorbike

Executables:

clean, stage_input, stage_trisurface, stage_geometry, stage_0, configure_mesh, surfaceFeatures, blockMesh, decomposePar1, snappyHexMesh, configure_simplefoam, redistributePar, decomposePar2, patchSummary, potentialFoam, checkMesh, simpleFoam, post-exec-clean

Variables:

input_path: Path to the tutorial input
geometry_path: Path to the geometry resource
decomposition_path: Path to decomposition files
control_path: Path to control file
block_mesh_path: Path to block mesh file
hex_mesh_path: Path to hexh mesh file
decomposition_method: Set the decomposition method. Empty string means no change
file_handler: Control how processes write output
end_time: End time for simulation
write_interval: Interval to write output files
start_from: How to start a new simulation
mesh_size: Mesh size for simulation
max_local_cells: Max local cells for simulation
max_global_cells: Max global cells for simulation
n_ranks_hex: Number of ranks to use for snappyHexMesh
hex_flags: Flags for snappyHexMesh
potential_flags: Flags for potentialFoam
simple_flags: Flags for simpleFoam
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_prefix: Prefix for exporting an environment variable with mpirun
workload_exports: Placeholder variable which holds all variable exports. Defined during setup.
export_variables: Comma separated list of all env-var names that need to be exported
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_variables: Comma separated list of all env-var names that need to be exported

motorbike_20m

Executables:

clean, stage_input, stage_trisurface, stage_geometry, stage_0, configure_mesh, surfaceFeatures, blockMesh, decomposePar1, snappyHexMesh, configure_simplefoam, redistributePar, decomposePar2, patchSummary, potentialFoam, checkMesh, simpleFoam, post-exec-clean

Variables:

input_path: Path to the tutorial input
geometry_path: Path to the geometry resource
decomposition_path: Path to decomposition files
control_path: Path to control file
block_mesh_path: Path to block mesh file
hex_mesh_path: Path to hexh mesh file
decomposition_method: Set the decomposition method. Empty string means no change
file_handler: Control how processes write output
end_time: End time for simulation
write_interval: Interval to write output files
start_from: How to start a new simulation
mesh_size: Mesh size for simulation
max_local_cells: Max local cells for simulation
max_global_cells: Max global cells for simulation
n_ranks_hex: Number of ranks to use for snappyHexMesh
hex_flags: Flags for snappyHexMesh
potential_flags: Flags for potentialFoam
simple_flags: Flags for simpleFoam
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_prefix: Prefix for exporting an environment variable with mpirun
workload_exports: Placeholder variable which holds all variable exports. Defined during setup.
export_variables: Comma separated list of all env-var names that need to be exported
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_variables: Comma separated list of all env-var names that need to be exported

motorbike_42m

Executables:

clean, stage_input, stage_trisurface, stage_geometry, stage_0, configure_mesh, surfaceFeatures, blockMesh, decomposePar1, snappyHexMesh, configure_simplefoam, redistributePar, decomposePar2, patchSummary, potentialFoam, checkMesh, simpleFoam, post-exec-clean

Variables:

input_path: Path to the tutorial input
geometry_path: Path to the geometry resource
decomposition_path: Path to decomposition files
control_path: Path to control file
block_mesh_path: Path to block mesh file
hex_mesh_path: Path to hexh mesh file
decomposition_method: Set the decomposition method. Empty string means no change
file_handler: Control how processes write output
end_time: End time for simulation
write_interval: Interval to write output files
start_from: How to start a new simulation
mesh_size: Mesh size for simulation
max_local_cells: Max local cells for simulation
max_global_cells: Max global cells for simulation
n_ranks_hex: Number of ranks to use for snappyHexMesh
hex_flags: Flags for snappyHexMesh
potential_flags: Flags for potentialFoam
simple_flags: Flags for simpleFoam
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_prefix: Prefix for exporting an environment variable with mpirun
workload_exports: Placeholder variable which holds all variable exports. Defined during setup.
export_variables: Comma separated list of all env-var names that need to be exported
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_variables: Comma separated list of all env-var names that need to be exported

motorbike_100m

Executables:

clean, stage_input, stage_trisurface, stage_geometry, stage_0, configure_mesh, surfaceFeatures, blockMesh, decomposePar1, snappyHexMesh, configure_simplefoam, redistributePar, decomposePar2, patchSummary, potentialFoam, checkMesh, simpleFoam, post-exec-clean

Variables:

input_path: Path to the tutorial input
geometry_path: Path to the geometry resource
decomposition_path: Path to decomposition files
control_path: Path to control file
block_mesh_path: Path to block mesh file
hex_mesh_path: Path to hexh mesh file
decomposition_method: Set the decomposition method. Empty string means no change
file_handler: Control how processes write output
end_time: End time for simulation
write_interval: Interval to write output files
start_from: How to start a new simulation
mesh_size: Mesh size for simulation
max_local_cells: Max local cells for simulation
max_global_cells: Max global cells for simulation
n_ranks_hex: Number of ranks to use for snappyHexMesh
hex_flags: Flags for snappyHexMesh
potential_flags: Flags for potentialFoam
simple_flags: Flags for simpleFoam
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_prefix: Prefix for exporting an environment variable with mpirun
workload_exports: Placeholder variable which holds all variable exports. Defined during setup.
export_variables: Comma separated list of all env-var names that need to be exported
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_variables: Comma separated list of all env-var names that need to be exported

motorbike_200m

Executables:

clean, stage_input, stage_trisurface, stage_geometry, stage_0, configure_mesh, surfaceFeatures, blockMesh, decomposePar1, snappyHexMesh, configure_simplefoam, redistributePar, decomposePar2, patchSummary, potentialFoam, checkMesh, simpleFoam, post-exec-clean

Variables:

input_path: Path to the tutorial input
geometry_path: Path to the geometry resource
decomposition_path: Path to decomposition files
control_path: Path to control file
block_mesh_path: Path to block mesh file
hex_mesh_path: Path to hexh mesh file
decomposition_method: Set the decomposition method. Empty string means no change
file_handler: Control how processes write output
end_time: End time for simulation
write_interval: Interval to write output files
start_from: How to start a new simulation
mesh_size: Mesh size for simulation
max_local_cells: Max local cells for simulation
max_global_cells: Max global cells for simulation
n_ranks_hex: Number of ranks to use for snappyHexMesh
hex_flags: Flags for snappyHexMesh
potential_flags: Flags for potentialFoam
simple_flags: Flags for simpleFoam
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_prefix: Prefix for exporting an environment variable with mpirun
workload_exports: Placeholder variable which holds all variable exports. Defined during setup.
export_variables: Comma separated list of all env-var names that need to be exported
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_variables: Comma separated list of all env-var names that need to be exported

openfoam-org¶

Ramble applications:

openfoam-org/application.py

Description:

Define the Openfoam-org application

Workloads:

motorbike

Executables:

clean, stage_input, stage_trisurface, stage_geometry, stage_0, configure_mesh, surfaceFeatures, blockMesh, decomposePar1, snappyHexMesh, configure_simplefoam, redistributePar, decomposePar2, patchSummary, potentialFoam, checkMesh, simpleFoam, post-exec-clean

Variables:

input_path: Path to the tutorial input
geometry_path: Path to the geometry resource
decomposition_path: Path to decomposition files
control_path: Path to control file
block_mesh_path: Path to block mesh file
hex_mesh_path: Path to hexh mesh file
decomposition_method: Set the decomposition method. Empty string means no change
file_handler: Control how processes write output
end_time: End time for simulation
write_interval: Interval to write output files
start_from: How to start a new simulation
mesh_size: Mesh size for simulation
max_local_cells: Max local cells for simulation
max_global_cells: Max global cells for simulation
n_ranks_hex: Number of ranks to use for snappyHexMesh
hex_flags: Flags for snappyHexMesh
potential_flags: Flags for potentialFoam
simple_flags: Flags for simpleFoam
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_prefix: Prefix for exporting an environment variable with mpirun
workload_exports: Placeholder variable which holds all variable exports. Defined during setup.
export_variables: Comma separated list of all env-var names that need to be exported
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_variables: Comma separated list of all env-var names that need to be exported

motorbike_20m

Executables:

clean, stage_input, stage_trisurface, stage_geometry, stage_0, configure_mesh, surfaceFeatures, blockMesh, decomposePar1, snappyHexMesh, configure_simplefoam, redistributePar, decomposePar2, patchSummary, potentialFoam, checkMesh, simpleFoam, post-exec-clean

Variables:

input_path: Path to the tutorial input
geometry_path: Path to the geometry resource
decomposition_path: Path to decomposition files
control_path: Path to control file
block_mesh_path: Path to block mesh file
hex_mesh_path: Path to hexh mesh file
decomposition_method: Set the decomposition method. Empty string means no change
file_handler: Control how processes write output
end_time: End time for simulation
write_interval: Interval to write output files
start_from: How to start a new simulation
mesh_size: Mesh size for simulation
max_local_cells: Max local cells for simulation
max_global_cells: Max global cells for simulation
n_ranks_hex: Number of ranks to use for snappyHexMesh
hex_flags: Flags for snappyHexMesh
potential_flags: Flags for potentialFoam
simple_flags: Flags for simpleFoam
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_prefix: Prefix for exporting an environment variable with mpirun
workload_exports: Placeholder variable which holds all variable exports. Defined during setup.
export_variables: Comma separated list of all env-var names that need to be exported
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_variables: Comma separated list of all env-var names that need to be exported

motorbike_42m

Executables:

clean, stage_input, stage_trisurface, stage_geometry, stage_0, configure_mesh, surfaceFeatures, blockMesh, decomposePar1, snappyHexMesh, configure_simplefoam, redistributePar, decomposePar2, patchSummary, potentialFoam, checkMesh, simpleFoam, post-exec-clean

Variables:

input_path: Path to the tutorial input
geometry_path: Path to the geometry resource
decomposition_path: Path to decomposition files
control_path: Path to control file
block_mesh_path: Path to block mesh file
hex_mesh_path: Path to hexh mesh file
decomposition_method: Set the decomposition method. Empty string means no change
file_handler: Control how processes write output
end_time: End time for simulation
write_interval: Interval to write output files
start_from: How to start a new simulation
mesh_size: Mesh size for simulation
max_local_cells: Max local cells for simulation
max_global_cells: Max global cells for simulation
n_ranks_hex: Number of ranks to use for snappyHexMesh
hex_flags: Flags for snappyHexMesh
potential_flags: Flags for potentialFoam
simple_flags: Flags for simpleFoam
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_prefix: Prefix for exporting an environment variable with mpirun
workload_exports: Placeholder variable which holds all variable exports. Defined during setup.
export_variables: Comma separated list of all env-var names that need to be exported
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_variables: Comma separated list of all env-var names that need to be exported

motorbike_100m

Executables:

clean, stage_input, stage_trisurface, stage_geometry, stage_0, configure_mesh, surfaceFeatures, blockMesh, decomposePar1, snappyHexMesh, configure_simplefoam, redistributePar, decomposePar2, patchSummary, potentialFoam, checkMesh, simpleFoam, post-exec-clean

Variables:

input_path: Path to the tutorial input
geometry_path: Path to the geometry resource
decomposition_path: Path to decomposition files
control_path: Path to control file
block_mesh_path: Path to block mesh file
hex_mesh_path: Path to hexh mesh file
decomposition_method: Set the decomposition method. Empty string means no change
file_handler: Control how processes write output
end_time: End time for simulation
write_interval: Interval to write output files
start_from: How to start a new simulation
mesh_size: Mesh size for simulation
max_local_cells: Max local cells for simulation
max_global_cells: Max global cells for simulation
n_ranks_hex: Number of ranks to use for snappyHexMesh
hex_flags: Flags for snappyHexMesh
potential_flags: Flags for potentialFoam
simple_flags: Flags for simpleFoam
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_prefix: Prefix for exporting an environment variable with mpirun
workload_exports: Placeholder variable which holds all variable exports. Defined during setup.
export_variables: Comma separated list of all env-var names that need to be exported
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_variables: Comma separated list of all env-var names that need to be exported

motorbike_200m

Executables:

clean, stage_input, stage_trisurface, stage_geometry, stage_0, configure_mesh, surfaceFeatures, blockMesh, decomposePar1, snappyHexMesh, configure_simplefoam, redistributePar, decomposePar2, patchSummary, potentialFoam, checkMesh, simpleFoam, post-exec-clean

Variables:

input_path: Path to the tutorial input
geometry_path: Path to the geometry resource
decomposition_path: Path to decomposition files
control_path: Path to control file
block_mesh_path: Path to block mesh file
hex_mesh_path: Path to hexh mesh file
decomposition_method: Set the decomposition method. Empty string means no change
file_handler: Control how processes write output
end_time: End time for simulation
write_interval: Interval to write output files
start_from: How to start a new simulation
mesh_size: Mesh size for simulation
max_local_cells: Max local cells for simulation
max_global_cells: Max global cells for simulation
n_ranks_hex: Number of ranks to use for snappyHexMesh
hex_flags: Flags for snappyHexMesh
potential_flags: Flags for potentialFoam
simple_flags: Flags for simpleFoam
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_prefix: Prefix for exporting an environment variable with mpirun
workload_exports: Placeholder variable which holds all variable exports. Defined during setup.
export_variables: Comma separated list of all env-var names that need to be exported
dict_delim: Delimiter for dictionary entries
coeffs_dict: Coeffs dictionary name
export_variables: Comma separated list of all env-var names that need to be exported

orca¶

Ramble applications:

orca/application.py

Description:

Define the ORCA application. For more details, see https://www.faccts.de/orca/.

Workloads:

standard

Executables:

stage-input, configure_nprocs, run_orca

Inputs:

orca_in: Input deck archive for ORCA

Variables:

scratch_dir: Scratch directory
input_path: Path for the fetched input deck
main_input_file: Leaf name of the main input file
ompi_args: Extra openmpi args supplied to orca

osu-micro-benchmarks¶

Ramble applications:

osu-micro-benchmarks/application.py

Description:

Define an OSU micro benchmarks application

Workloads:

osu_bibw

Executables:

execute-osu_bibw

Variables:

additional_args: Additional arguments for benchmark

osu_bw

Executables:

execute-osu_bw

Variables:

additional_args: Additional arguments for benchmark

osu_latency

Executables:

execute-osu_latency

Variables:

additional_args: Additional arguments for benchmark

osu_latency_mp

Executables:

execute-osu_latency_mp

Variables:

additional_args: Additional arguments for benchmark

osu_latency_mt

Executables:

execute-osu_latency_mt

Variables:

additional_args: Additional arguments for benchmark

osu_mbw_mr

Executables:

execute-osu_mbw_mr

Variables:

additional_args: Additional arguments for benchmark

osu_multi_lat

Executables:

execute-osu_multi_lat

Variables:

additional_args: Additional arguments for benchmark

osu_allgather

Executables:

execute-osu_allgather

Variables:

additional_args: Additional arguments for benchmark

osu_allreduce_persistent

Executables:

execute-osu_allreduce_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_alltoallw

Executables:

execute-osu_alltoallw

Variables:

additional_args: Additional arguments for benchmark

osu_bcast_persistent

Executables:

execute-osu_bcast_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_iallgather

Executables:

execute-osu_iallgather

Variables:

additional_args: Additional arguments for benchmark

osu_ialltoallw

Executables:

execute-osu_ialltoallw

Variables:

additional_args: Additional arguments for benchmark

osu_ineighbor_allgather

Executables:

execute-osu_ineighbor_allgather

Variables:

additional_args: Additional arguments for benchmark

osu_ireduce

Executables:

execute-osu_ireduce

Variables:

additional_args: Additional arguments for benchmark

osu_neighbor_allgatherv

Executables:

execute-osu_neighbor_allgatherv

Variables:

additional_args: Additional arguments for benchmark

osu_reduce_persistent

Executables:

execute-osu_reduce_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_scatterv

Executables:

execute-osu_scatterv

Variables:

additional_args: Additional arguments for benchmark

osu_allgather_persistent

Executables:

execute-osu_allgather_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_alltoall

Executables:

execute-osu_alltoall

Variables:

additional_args: Additional arguments for benchmark

osu_alltoallw_persistent

Executables:

execute-osu_alltoallw_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_gather

Executables:

execute-osu_gather

Variables:

additional_args: Additional arguments for benchmark

osu_iallgatherv

Executables:

execute-osu_iallgatherv

Variables:

additional_args: Additional arguments for benchmark

osu_ibarrier

Executables:

execute-osu_ibarrier

Variables:

additional_args: Additional arguments for benchmark

osu_ineighbor_allgatherv

Executables:

execute-osu_ineighbor_allgatherv

Variables:

additional_args: Additional arguments for benchmark

osu_ireduce_scatter

Executables:

execute-osu_ireduce_scatter

Variables:

additional_args: Additional arguments for benchmark

osu_neighbor_alltoall

Executables:

execute-osu_neighbor_alltoall

Variables:

additional_args: Additional arguments for benchmark

osu_reduce_scatter

Executables:

execute-osu_reduce_scatter

Variables:

additional_args: Additional arguments for benchmark

osu_scatterv_persistent

Executables:

execute-osu_scatterv_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_allgatherv

Executables:

execute-osu_allgatherv

Variables:

additional_args: Additional arguments for benchmark

osu_alltoall_persistent

Executables:

execute-osu_alltoall_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_barrier

Executables:

execute-osu_barrier

Variables:

additional_args: Additional arguments for benchmark

osu_gather_persistent

Executables:

execute-osu_gather_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_iallreduce

Executables:

execute-osu_iallreduce

Variables:

additional_args: Additional arguments for benchmark

osu_ibcast

Executables:

execute-osu_ibcast

Variables:

additional_args: Additional arguments for benchmark

osu_ineighbor_alltoall

Executables:

execute-osu_ineighbor_alltoall

Variables:

additional_args: Additional arguments for benchmark

osu_iscatter

Executables:

execute-osu_iscatter

Variables:

additional_args: Additional arguments for benchmark

osu_neighbor_alltoallv

Executables:

execute-osu_neighbor_alltoallv

Variables:

additional_args: Additional arguments for benchmark

osu_reduce_scatter_persistent

Executables:

execute-osu_reduce_scatter_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_allgatherv_persistent

Executables:

execute-osu_allgatherv_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_alltoallv

Executables:

execute-osu_alltoallv

Variables:

additional_args: Additional arguments for benchmark

osu_barrier_persistent

Executables:

execute-osu_barrier_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_gatherv

Executables:

execute-osu_gatherv

Variables:

additional_args: Additional arguments for benchmark

osu_ialltoall

Executables:

execute-osu_ialltoall

Variables:

additional_args: Additional arguments for benchmark

osu_igather

Executables:

execute-osu_igather

Variables:

additional_args: Additional arguments for benchmark

osu_ineighbor_alltoallv

Executables:

execute-osu_ineighbor_alltoallv

Variables:

additional_args: Additional arguments for benchmark

osu_iscatterv

Executables:

execute-osu_iscatterv

Variables:

additional_args: Additional arguments for benchmark

osu_neighbor_alltoallw

Executables:

execute-osu_neighbor_alltoallw

Variables:

additional_args: Additional arguments for benchmark

osu_scatter

Executables:

execute-osu_scatter

Variables:

additional_args: Additional arguments for benchmark

osu_allreduce

Executables:

execute-osu_allreduce

Variables:

additional_args: Additional arguments for benchmark

osu_alltoallv_persistent

Executables:

execute-osu_alltoallv_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_bcast

Executables:

execute-osu_bcast

Variables:

additional_args: Additional arguments for benchmark

osu_gatherv_persistent

Executables:

execute-osu_gatherv_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_ialltoallv

Executables:

execute-osu_ialltoallv

Variables:

additional_args: Additional arguments for benchmark

osu_igatherv

Executables:

execute-osu_igatherv

Variables:

additional_args: Additional arguments for benchmark

osu_ineighbor_alltoallw

Executables:

execute-osu_ineighbor_alltoallw

Variables:

additional_args: Additional arguments for benchmark

osu_neighbor_allgather

Executables:

execute-osu_neighbor_allgather

Variables:

additional_args: Additional arguments for benchmark

osu_reduce

Executables:

execute-osu_reduce

Variables:

additional_args: Additional arguments for benchmark

osu_scatter_persistent

Executables:

execute-osu_scatter_persistent

Variables:

additional_args: Additional arguments for benchmark

osu_acc_latency

Executables:

execute-osu_acc_latency

Variables:

additional_args: Additional arguments for benchmark

osu_cas_latency

Executables:

execute-osu_cas_latency

Variables:

additional_args: Additional arguments for benchmark

osu_fop_latency

Executables:

execute-osu_fop_latency

Variables:

additional_args: Additional arguments for benchmark

osu_get_acc_latency

Executables:

execute-osu_get_acc_latency

Variables:

additional_args: Additional arguments for benchmark

osu_get_bw

Executables:

execute-osu_get_bw

Variables:

additional_args: Additional arguments for benchmark

osu_get_latency

Executables:

execute-osu_get_latency

Variables:

additional_args: Additional arguments for benchmark

osu_put_bibw

Executables:

execute-osu_put_bibw

Variables:

additional_args: Additional arguments for benchmark

osu_put_bw

Executables:

execute-osu_put_bw

Variables:

additional_args: Additional arguments for benchmark

osu_put_latency

Executables:

execute-osu_put_latency

Variables:

additional_args: Additional arguments for benchmark

osu_hello

Executables:

execute-osu_hello

Variables:

additional_args: Additional arguments for benchmark

osu_init

Executables:

execute-osu_init

Variables:

additional_args: Additional arguments for benchmark

pip-test¶

Ramble applications:

pip-test/application.py

Description:

This is an example application that will run a python import

Workloads:

import

Executables:: import

py-cosmoflow¶

Ramble applications:

py-cosmoflow/application.py

Description:

This is a an implementation of the CosmoFlow 3D convolutional neural network for benchmarking. It is written in TensorFlow with the Keras API and uses Horovod for distributed training.

Workloads:

cosmoUniverse_mini

Executables:

execute

Inputs:

cosmoUniverse_mini: Cosmoflow Universe - Mini input
mlperf-hpc: MLPerf HPC Repo

Variables:

dockerfile_path: Dockerfile for cosmoflow from the MLPerf-HPC repo
docker_tag_name: Name of docker image tag
docker_tag_version: Version of docker image tag
docker_tensorflow_version: Version of tensorflow base container to use
cosmoflow_config: Name of generated input for cosmoflow
data.data_dir: Cosmoflow Data Directory
data.n_train: Number of training data sets
data.n_valid: Number of valid data sets
mlperf.org: Organization for reporting MLPerf results
mlperf.division: Division for reporting MLPerf results
mlperf.status: Cluster status for reporting MLPerf results
mlperf.platform: Platform name for reporting MLPerf results
output_dir: Experiment output directory
cosmoflow_base_config: Base configuration file to generate cosmoflow inputs from

cosmoUniverse

Executables:

execute

Inputs:

cosmoUniverse: Cosmoflow Universe - Main input
mlperf-hpc: MLPerf HPC Repo

Variables:

dockerfile_path: Dockerfile for cosmoflow from the MLPerf-HPC repo
docker_tag_name: Name of docker image tag
docker_tag_version: Version of docker image tag
docker_tensorflow_version: Version of tensorflow base container to use
cosmoflow_config: Name of generated input for cosmoflow
data.data_dir: Cosmoflow Data Directory
data.n_train: Number of training data sets
data.n_valid: Number of valid data sets
mlperf.org: Organization for reporting MLPerf results
mlperf.division: Division for reporting MLPerf results
mlperf.status: Cluster status for reporting MLPerf results
mlperf.platform: Platform name for reporting MLPerf results
output_dir: Experiment output directory
cosmoflow_base_config: Base configuration file to generate cosmoflow inputs from

py-espresso¶

Ramble applications:

py-espresso/application.py

Description:

Define py-espresso (pypresso) application ESPResSo is a highly versatile software package for performing and analyzing scientific Molecular Dynamics many-particle simulations of "coarse-grained" bead-spring models as they are used in soft-matter research in physics, chemistry and molecular biology. It can be used to simulate systems as for example polymers, liquid crystals, colloids, ferrofluids and biological systems such as DNA and lipid membranes.

Workloads:

p3m

Executables:

execute

Inputs:

particle_sample_input: Sample p3m input

Variables:

input_file: Input file to run

py-nemo¶

Ramble applications:

py-nemo/application.py

Description:

A scalable generative AI framework built for researchers and developers working on Large Language Models, Multimodal, and Speech AI (Automatic Speech Recognition and Text-to-Speech) model.ffn_hidden_size gets a default value of {4*{model.hidden_size}} if it can not be cast to an integer from the default config.

Workloads:

pretraining

Executables:

create_logs, setup_transformer_cache, pretraining_exec

Inputs:

nemo_fetched_config: Base config for NeMo experiments

Variables:

custom_injected_string: Custom string to inject before execution NeMo workload
nemo_container_version: Version for NeMo container
nemo_stage: Stage to run in NeMo
cuda_visible_devices: Comma delimited list of CUDA device IDs.
transformers_offline: Whether transformers are offline (0) or not (1)
torch_nccl_avoid_record_streams: Avoid (1) recording streams for Torch NCCL, or not (0)
nccl_nvls_enable: Enable (1) NCCL NVLS or not (0)
results_mount: Container mount for results data
container_mounts: All container mounts in a ramble variable
processed_log_file: Path to store processed NeMo output
model_inputs: NeMo model input directory
nemo_launcher_tag: Tag of NeMo-Framework-Launcher repo to extract inputs from (1.0 only)
nemo_model: Model to run in NeMo
nemo_config_name: Configuration name to run in NeMo. This is the name of the input yaml file without the extension. e.g. 5b.yaml -> 5b, while mixtral_8x22b.yaml -> mixtral_8x22b
nemo_base_config: Path to base config used for generating experiments. Defaults to the fetched input, but can refer to a provided input.
nemo_generated_config_name: Name of nemo config file
nemo_generated_config_path: Path where nemo config file is contained
nemo_remove_variables: Name of variables to remove from the base nemo config
logs_mount: Container mount for results data
results_mount: Container mount for results data
container_mounts: All container mounts in a ramble variable
run.name: Name of run
run.results_dir: Experiment results directory
run.time_limit: Experiment time limit
run.dependency: Experiment dependency type
trainer.num_nodes: Number of nodes
trainer.devices: Number of devices per node
trainer.accelerator: Accelerator to use as device
exp_manager.explicit_log_dir: Log directory for exp manager
exp_manager.exp_dir: Experiment directory for exp manager
exp_manager.name: Exp manager name
exp_manager.wandb_logger_kwargs.project: wandb logger project
exp_manager.wandb_logger_kwargs.name: wandb logger name
exp_manager.checkpoint_callback_params.model_parallel_size: Parallel size

py-nemo-2¶

Ramble applications:

py-nemo-2/application.py

Description:

A scalable generative AI framework built for researchers and developers working on Large Language Models, Multimodal, and Speech AI (Automatic Speech Recognition and Text-to-Speech) NeMo 2.0 requires NeMo container version >= 24.12.

Workloads:

pretraining

Executables:

setup_transformer_cache, pretraining_exec

Variables:

custom_injected_string: Custom string to inject before execution NeMo workload
nemo_container_version: Version for NeMo container
nemo_stage: Stage to run in NeMo
cuda_visible_devices: Comma delimited list of CUDA device IDs.
transformers_offline: Whether transformers are offline (0) or not (1)
torch_nccl_avoid_record_streams: Avoid (1) recording streams for Torch NCCL, or not (0)
nccl_nvls_enable: Enable (1) NCCL NVLS or not (0)
results_mount: Container mount for results data
container_mounts: All container mounts in a ramble variable
processed_log_file: Path to store processed NeMo output
nemo_config_dir_path: Path to folder containing Nemo 2.0 python config to be used.
nemo_config_name: Name of NeMo 2.0 config under {nemo_config_dir_path}.
nemo_cli_args: CLI args to pass to append to the Python Nemo command.
results_mount: Container mount for results data
container_mounts: All container mounts in a ramble variable

quantum-espresso¶

Ramble applications:

quantum-espresso/application.py

Description:

Define Quantum-Espresso application.

Workloads:

AUSURF112

Executables:

copy_inputs, execute

Inputs:

AUSURF112: Input file for AUSURF112 benchmark

Variables:

flags: Flags for Quantum Espresso
input_path: Path to inputs for workload
input_file: Name of input file for workload

CNT10POR8

Executables:

copy_inputs, execute

Inputs:

CNT10POR8: Input file for CNT10POR8 benchmark

Variables:

flags: Flags for Quantum Espresso
input_path: Path to inputs for workload
input_file: Name of input file for workload

GRIR443

Executables:

copy_inputs, execute

Inputs:

GRIR443: Input file for GRIR443 benchmark

Variables:

flags: Flags for Quantum Espresso
input_path: Path to inputs for workload
input_file: Name of input file for workload

GRIR686

Executables:

copy_inputs, execute

Inputs:

GRIR686: Input file for GRIR686 benchmark

Variables:

flags: Flags for Quantum Espresso
input_path: Path to inputs for workload
input_file: Name of input file for workload

PSIWAT

Executables:

copy_inputs, execute

Inputs:

PSIWAT: Input file for PSIWAT benchmark

Variables:

flags: Flags for Quantum Espresso
input_path: Path to inputs for workload
input_file: Name of input file for workload

WATER_EXX

Executables:

copy_potential1, copy_potential2, execute

Inputs:

WATER_EXX: Input file for WATER_EXX benchmark
WATER_EXX_H: H Pseudopotential for WATER_EXX benchmark
WATER_EXX_O: O Pseudopotential for WATER_EXX benchmark

Variables:

flags: Flags for Quantum Espresso
input_file: Name of input file for workload
potential1: Path to WATER_EXX H potential file
potential2: Path to WATER_EXX O potential file

roms¶

Ramble applications:

roms/application.py

Description:

The Regional Ocean Modeling System (ROMS) is a free-surface, terrain- following, primitive equations ocean model widely used by the scientific community for a diverse range of applications. ROMS includes accurate and efficient physical and numerical algorithms and several coupled models for biogeochemical, bio-optical, sediment, and sea ice applications. https://www.myroms.org/

Workloads:

benchmark_1

Executables:

stage-input, stage-varinfo, execute

Inputs:

bm1: Simple test benchmark (small)
varinfo: Metadata dict for benchmark inputs

Variables:

input_deck: Name of input deck
input_path: Path to input deck

sleep¶

Ramble applications:

sleep/application.py

Description:

This is an example application that will simply run sleep for a controlled random amount of time

Workloads:

sleep

Executables:

define_sleep_time, echo, sleep

Variables:

sleep_seconds: Number of seconds to sleep for

rand_sleep

Executables:

define_sleep_time, echo, sleep

Variables:

rand_min: Minimum of the random range
rand_max: Maximum of the random range
sleep_seconds: Number of seconds to sleep for

spack-stack¶

Ramble applications:

spack-stack/application.py

Description:

Application definition for creating a spack software stack This application definition is used solely to create spack software stacks. As such, compiler installation and concretization are handled by `ramble workspace setup` but environment installation is handled as part of the experiment. The `spack install` phase happens with the '{mpi_command}' prefix to accelerate package installation. The experiments are considered successful if the installation completed. This application should be used with the `spack-lightweight` package manager.

Workloads:

create

Executables:

builtin::remove_env_files, builtin::find_externals, configure, install, builtin::remove_packages

Variables:

install_flags: Flags to use for `spack install`
external_packages: List of packages to `spack external find` and mark not buildable
removed_packages: List of packages to remove from the environment after installation is complete
padded_length: Length to pad install prefixes with

remove

Executables:

uninstall

Variables:

uninstall_flags: Flags to use for `spack uninstall`

streamc¶

Ramble applications:

streamc/application.py

Description:

Define STREAMC application

Workloads:

streamc

Executables:: execute_c

su2¶

Ramble applications:

su2/application.py

Description:

Define SU2 (Stanford University Unstructured) application. See https://su2code.github.io/ for more details.

Workloads:

inv_channel

Executables:

stage-files, execute

Inputs:

inv_channel_in: input deck used in https://su2code.github.io/tutorials/Inviscid_Bump
inv_mesh_in: input deck used in https://su2code.github.io/tutorials/Inviscid_Bump

Variables:

input_path: Path to the input for experiments
input_config: Name of the input configuration file
su2_executable: Path to the SU2 executable

tpu-device-count¶

Ramble applications:

tpu-device-count/application.py

Description:

This is an example application that runs on TPUs

Workloads:

count_devices

Executables:: count_devices

ufs-weather-model¶

Ramble applications:

ufs-weather-model/application.py

Description:

Define FV3 application via ufs-weather-model

Workloads:

simple_test_case

Executables:

stage-input, execute

Inputs:

simple_test_case: Simple test case for ufs-weather-model

Variables:

input_path: extracted simple-test-case tarfile path

control_c48_intel

Executables:

stage-input, execute

Inputs:

20210323.060000.coupler.res: Restart Part for c48_intel
20210323.060000.fv_core.res.nc: Restart Part for c48_intel
20210323.060000.fv_core.res.tile1.nc: Restart Part for c48_intel
20210323.060000.fv_core.res.tile2.nc: Restart Part for c48_intel
20210323.060000.fv_core.res.tile3.nc: Restart Part for c48_intel
20210323.060000.fv_core.res.tile4.nc: Restart Part for c48_intel
20210323.060000.fv_core.res.tile5.nc: Restart Part for c48_intel
20210323.060000.fv_core.res.tile6.nc: Restart Part for c48_intel
20210323.060000.fv_srf_wnd.res.tile1.nc: Restart Part for c48_intel
20210323.060000.fv_srf_wnd.res.tile2.nc: Restart Part for c48_intel
20210323.060000.fv_srf_wnd.res.tile3.nc: Restart Part for c48_intel
20210323.060000.fv_srf_wnd.res.tile4.nc: Restart Part for c48_intel
20210323.060000.fv_srf_wnd.res.tile5.nc: Restart Part for c48_intel
20210323.060000.fv_srf_wnd.res.tile6.nc: Restart Part for c48_intel
20210323.060000.fv_tracer.res.tile1.nc: Restart Part for c48_intel
20210323.060000.fv_tracer.res.tile2.nc: Restart Part for c48_intel
20210323.060000.fv_tracer.res.tile3.nc: Restart Part for c48_intel
20210323.060000.fv_tracer.res.tile4.nc: Restart Part for c48_intel
20210323.060000.fv_tracer.res.tile5.nc: Restart Part for c48_intel
20210323.060000.fv_tracer.res.tile6.nc: Restart Part for c48_intel
20210323.060000.phy_data.tile1.nc: Restart Part for c48_intel
20210323.060000.phy_data.tile2.nc: Restart Part for c48_intel
20210323.060000.phy_data.tile3.nc: Restart Part for c48_intel
20210323.060000.phy_data.tile4.nc: Restart Part for c48_intel
20210323.060000.phy_data.tile5.nc: Restart Part for c48_intel
20210323.060000.phy_data.tile6.nc: Restart Part for c48_intel
20210323.060000.sfc_data.tile1.nc: Restart Part for c48_intel
20210323.060000.sfc_data.tile2.nc: Restart Part for c48_intel
20210323.060000.sfc_data.tile3.nc: Restart Part for c48_intel
20210323.060000.sfc_data.tile4.nc: Restart Part for c48_intel
20210323.060000.sfc_data.tile5.nc: Restart Part for c48_intel
20210323.060000.sfc_data.tile6.nc: Restart Part for c48_intel

Variables:

input_path: c48 input path

wrf¶

Ramble applications:

wrf/application.py

Description:

Define Wrf application

Workloads:

CONUS_2p5km

Executables:

stage-files, stage-namelist, cleanup, define_nproc_y, define_nproc_x, execute, copy-logs, post-exec-clean

Inputs:

CONUS_2p5km: 2.5 km resolution mesh of the continental United States.

Variables:

num_tiles: Number of tiles to use in WRF domain
nproc_x: Number of process in the x dimension
nproc_y: Number of process in the y dimension
nproc_y: Number of process in the y dimension
input_path: Path for workload inputs.

CONUS_12km

Executables:

stage-files, stage-namelist, cleanup, define_nproc_y, define_nproc_x, fix_12km, execute, copy-logs, post-exec-clean

Inputs:

CONUS_12km: 12 km resolution mesh of the continental United States.

Variables:

num_tiles: Number of tiles to use in WRF domain
nproc_x: Number of process in the x dimension
nproc_y: Number of process in the y dimension
nproc_y: Number of process in the y dimension
input_path: Path for workload inputs.

Maria_1km

Executables:

stage-files, stage-namelist, cleanup, define_nproc_y, define_nproc_x, execute, copy-logs, post-exec-clean

Inputs:

Maria_1km: 1 km Maria workload input

Variables:

num_tiles: Number of tiles to use in WRF domain
nproc_x: Number of process in the x dimension
nproc_y: Number of process in the y dimension
nproc_y: Number of process in the y dimension
input_path: Path for workload inputs.

CONUS_2p5km

Executables:

stage-files, stage-namelist, setup, execute, copy-logs

Inputs:

CONUS_2p5km: 2.5 km resolution mesh of the continental United States.

Variables:

input_path: Path for workload inputs.

CONUS_12km

Executables:

stage-files, stage-namelist, setup, execute, copy-logs

Inputs:

CONUS_12km: 12 km resolution mesh of the continental United States.

Variables:

input_path: Path for workload inputs.

wrfv3¶

Ramble applications:

wrfv3/application.py

Description:

Define Wrf version 3 application

Workloads:

CONUS_2p5km

Executables:

stage-files, stage-namelist, setup, execute

Inputs:

CONUS_2p5km: 2.5 km resolution mesh of the continental United States.

Variables:

input_path: Path for workload inputs.

CONUS_12km

Executables:

stage-files, stage-namelist, setup, execute

Inputs:

CONUS_12km: 12 km resolution mesh of the continental United States.

Variables:

input_path: Path for workload inputs.

wrfv4¶

Ramble applications:

wrfv4/application.py

Description:

Define Wrf version 4 application

Workloads:

CONUS_2p5km

Executables:

stage-files, stage-namelist, cleanup, define_nproc_y, define_nproc_x, execute, copy-logs, post-exec-clean

Inputs:

CONUS_2p5km: 2.5 km resolution mesh of the continental United States.

Variables:

num_tiles: Number of tiles to use in WRF domain
nproc_x: Number of process in the x dimension
nproc_y: Number of process in the y dimension
nproc_y: Number of process in the y dimension
input_path: Path for workload inputs.

CONUS_12km

Executables:

stage-files, stage-namelist, cleanup, define_nproc_y, define_nproc_x, fix_12km, execute, copy-logs, post-exec-clean

Inputs:

CONUS_12km: 12 km resolution mesh of the continental United States.

Variables:

num_tiles: Number of tiles to use in WRF domain
nproc_x: Number of process in the x dimension
nproc_y: Number of process in the y dimension
nproc_y: Number of process in the y dimension
input_path: Path for workload inputs.

Maria_1km

Executables:

stage-files, stage-namelist, cleanup, define_nproc_y, define_nproc_x, execute, copy-logs, post-exec-clean

Inputs:

Maria_1km: 1 km Maria workload input

Variables:

num_tiles: Number of tiles to use in WRF domain
nproc_x: Number of process in the x dimension
nproc_y: Number of process in the y dimension
nproc_y: Number of process in the y dimension
input_path: Path for workload inputs.