[QAAS]
timestamp = gmz12.benchmarkcenter.megware.com
app_name = llama.cpp
git_commit = 
dataset_name = Llama-3.1-8B-Q8_0 PP=128 NPL=4
run_cmd = <binary> -m meta-llama-3.1-8b-instruct-Q8_0.gguf -t <OMP_NUM_THREADS> -b 2048 -ub 512 -npp 128 -ntg 0 -npl 4 -c 16384 --seed 0 --output-format jsonl
LANG = C/CXX

[REPORTS]
figure_of_merit_type = RATE
figure_of_merit_unit = tokens/s
compiler_default = aocc
multicompiler_report = qaas_compilers.csv
mpi_scaling = no
openmp_scaling = strong
scalability_report = qaas_multicore.csv
scalability_reference_line = aocc:2

[SYSTEM]
machine = gmz12.benchmarkcenter.megware.com
model_name = AMD EPYC 9655 96-Core Processor
ISA = x86_64
architecture = ZEN_V5
number_of_cpus = 384
number_of_cores = 192
number_of_sockets = 2
number_of_cores_per_socket = 96
number_of_numa_domains = 8
frequency_driver = acpi-cpufreq
frequency_governor = performance
scaling_max_frequency = 2600000
scaling_min_frequency = 1500000
advertized_frequency = unsupported
maximal_frequency = 4509375
huge_pages = 
icx_version = 2025.2.0.20250806
aocc_version = 5.0.0
mpi_provider = OpenMPI
mpi_version = 5.0.8

[TIME]
initial_profile = 00H04M33S
build_binaries = 00H11M43S
multicompiler = 00H15M12S