Dataset Viewer
metrics.accuracy
large_stringlengths 65
120
| metrics.result
float64 64
129k
| metrics.result_per_accelerator
float64 23.6
16.1k
| metrics.units
large_stringclasses 1
value | model.architecture
large_stringclasses 1
value | model.mlperf_name
large_stringclasses 8
values | model.name
large_stringclasses 19
values | model.number_of_parameters
float64 6
671
| model.weight_data_types
large_stringclasses 7
values | software.framework
large_stringclasses 28
values | software.framework.ai
large_stringclasses 0
values | software.framework.cuda
large_stringclasses 6
values | software.framework.cudnn
large_stringclasses 0
values | software.framework.jetpack
large_stringclasses 1
value | software.framework.pytorch
large_stringclasses 2
values | software.framework.rocm
large_stringclasses 5
values | software.framework.sima
large_stringclasses 0
values | software.framework.tensorrt
large_stringclasses 9
values | software.framework.vllm
large_stringclasses 4
values | software.operating_system
large_stringclasses 31
values | submission.availability
large_stringclasses 2
values | submission.division
large_stringclasses 2
values | submission.organization
large_stringclasses 24
values | submission.scenario
large_stringclasses 2
values | system.accelerator.count_per_node
int64 1
8
| system.accelerator.memory_capacity
float64 23.6
2.05k
⌀ | system.accelerator.memory_config
large_stringclasses 9
values | system.accelerator.name
large_stringclasses 20
values | system.accelerator.total_count
int64 1
72
| system.accelerator.vendor
large_stringclasses 2
values | system.cpu.caches
large_stringclasses 15
values | system.cpu.core_count
int64 4
384
⌀ | system.cpu.count_per_node
int64 1
4
| system.cpu.frequency
float64 2
4.8k
⌀ | system.cpu.model
large_stringclasses 56
values | system.cpu.vcpu_count
large_stringclasses 2
values | system.cpu.vendor
large_stringclasses 4
values | system.interconnect.accelerator
large_stringclasses 7
values | system.interconnect.accelerator_host
large_stringclasses 4
values | system.memory.capacity
float64 64
8.19k
⌀ | system.memory.configuration
large_stringclasses 66
values | system.name
large_stringclasses 109
values | system.number_of_nodes
int64 1
18
| system.type
large_stringclasses 2
values |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
ROUGEL: 21.658590925735947 exact_match: 90.0034579748511 TOKENS_PER_SAMPLE: 648.3 | 428.895 | 53.611875 | Tokens/s | LLM | llama3.1-405b | llama3_1-405b | 405 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | open | HPE | Offline | 8 | 141 | HBM3e | NVIDIA H200-NVL-141GB | 8 | NVIDIA | null | 192 | 2 | null | Intel(R) Xeon(R) 6740E | null | Intel | NVLink-4 | PCIe-5 | 3,072 | 16x 128GB MTC40F2047S1RC64BB1 MWFF; 16x 64GB HMCG94AHBRA480N | HPE ProLiant DL380a Compute Gen12 (8x H200-NVL-141GB, TensorRT) | 1 | datacenter |
ROUGE1: 30.6202 ROUGE2: 13.9221 ROUGEL: 18.9101 TOKENS_PER_SAMPLE: 581.8 | 2,631.93 | 2,631.93 | Tokens/s | LLM | llama2-70b-99 | DeepSeek-R1-Distill-Llama-8B | 8 | bfloat16 | vLLM v0.7.3 | null | null | null | null | null | null | null | null | null | Ubuntu 22.04.5 LTS (5.15.0-131-generic) | available | open | FlexAI | Server | 1 | null | HBM3 | NVIDIA H100 80GB HBM3 | 1 | NVIDIA | L1d cache: 6.3 MiB (200 instances), L1i cache: 6.3 MiB (200 instances), L2 cache: 800 MiB (200 instances), L3 cache: 3.1 GiB (200 instances) | 52 | 2 | null | Intel Xeon Processor (SapphireRapids) | null | Intel | NVLink | PCIe | null | undefined | flexbench test node 0ef307db09d34a91 with 8xH100 | 1 | datacenter |
ROUGE1: 44.246 ROUGE2: 22.4512 ROUGEL: 27.9398 TOKENS_PER_SAMPLE: 335.5 | 134.714 | 134.714 | Tokens/s | LLM | llama2-70b-99 | Llama-3.3-70B-Instruct-FP8-dynamic | 70 | bfloat16 | vLLM v0.7.3 | null | null | null | null | null | null | null | null | null | Ubuntu 22.04.5 LTS (5.15.0-131-generic) | available | open | FlexAI | Server | 1 | null | HBM3 | NVIDIA H100 80GB HBM3 | 1 | NVIDIA | L1d cache: 6.3 MiB (200 instances), L1i cache: 6.3 MiB (200 instances), L2 cache: 800 MiB (200 instances), L3 cache: 3.1 GiB (200 instances) | 52 | 2 | null | Intel Xeon Processor (SapphireRapids) | null | Intel | NVLink | PCIe | null | undefined | flexbench test node 0ef307db09d34a91 with 8xH100 | 1 | datacenter |
ROUGE1: 44.4619 ROUGE2: 22.0574 ROUGEL: 28.6452 TOKENS_PER_SAMPLE: 295.9 | 1,482.31 | 370.5775 | Tokens/s | LLM | llama2-70b-99 | Llama-2-70b-chat-hf-FP8 | 70 | fp8 | vLLM v0.7.3 | null | null | null | null | null | null | null | null | null | Ubuntu 22.04.5 LTS (5.15.0-131-generic) | available | open | FlexAI | Server | 4 | null | HBM3 | NVIDIA H100 80GB HBM3 | 4 | NVIDIA | L1d cache: 6.3 MiB (200 instances), L1i cache: 6.3 MiB (200 instances), L2 cache: 800 MiB (200 instances), L3 cache: 3.1 GiB (200 instances) | 52 | 2 | null | Intel Xeon Processor (SapphireRapids) | null | Intel | NVLink | PCIe | null | undefined | flexbench test node 0ef307db09d34a91 with 8xH100 | 1 | datacenter |
ROUGE1: 44.8294 ROUGE2: 22.3814 ROUGEL: 29.1878 TOKENS_PER_SAMPLE: 271.8 | 573.765 | 573.765 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp16 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Jetson L4T | preview | open | NVIDIA | Offline | 1 | null | LPDDR5X | NVIDIA Jetson AGX Thor 128G | 1 | NVIDIA | null | 14 | 1 | null | 14-core ARM Poseidon-AE CPU | null | ARM | null | null | 128 | 128GB 256-bit LPDDR5X | NVIDIA Jetson AGX Thor Developer Kit 128G (TensorRT) | 1 | datacenter |
ROUGE1: 44.8294 ROUGE2: 22.3814 ROUGEL: 29.1878 TOKENS_PER_SAMPLE: 271.8 | 573.765 | 573.765 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp16 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Jetson L4T | preview | open | NVIDIA | Offline | 1 | null | LPDDR5X | NVIDIA Jetson AGX Thor 128G | 1 | NVIDIA | null | 14 | 1 | null | 14-core ARM Poseidon-AE CPU | null | ARM | null | null | 128 | 128GB 256-bit LPDDR5X | NVIDIA Jetson AGX Thor Developer Kit 128G (TensorRT) | 1 | datacenter |
ROUGE1: 47.6186 ROUGE2: 25.2204 ROUGEL: 31.4595 TOKENS_PER_SAMPLE: 296.9 | 16,162.8 | 2,020.35 | Tokens/s | LLM | llama2-70b-99.9 | llama3_1-70b-fp8_dyn | 70 | fp16 | SGLang v0.4.3 | null | null | null | null | null | null | null | null | null | Ubuntu 22.04.4 LTS (Jammy Jellyfish) | available | open | Krai | Offline | 8 | 192 | HBM3 | AMD Instinct MI300X 192GB HBM3 | 8 | AMD | null | 40 | 2 | null | Intel(R) Xeon(R) Platinum 8460Y+ | null | Intel | XGMI | PCIe-5 | 2,048 | 32x 64GB DDR5 | Dell PowerEdge XE9680 (8x MI300X-192GB, SGLang v0.4.3) | 1 | datacenter |
ROUGEL: 18.9480793556203 exact_match: 90.14579086697552 TOKENS_PER_SAMPLE: 641.1 | 243.449 | 30.431125 | Tokens/s | LLM | llama3.1-405b | deepseek-v3 | 671 | UNSET | SGLang v0.4.3 | null | null | null | null | null | null | null | null | null | Ubuntu 22.04.4 LTS (Jammy Jellyfish) | available | open | Krai | Offline | 8 | 192 | HBM3 | AMD Instinct MI300X 192GB HBM3 | 8 | AMD | null | 40 | 2 | null | Intel(R) Xeon(R) Platinum 8460Y+ | null | Intel | XGMI | PCIe-5 | 2,048 | 32x 64GB DDR5 | Dell PowerEdge XE9680 (8x MI300X-192GB, SGLang v0.4.3) | 1 | datacenter |
ROUGE1: 44.5375 ROUGE2: 22.202 ROUGEL: 28.7796 TOKENS_PER_SAMPLE: 299.2 | 16,229 | 2,028.625 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-fp8_dyn | 70 | UNSET | vLLM v0.7.3 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, vLLM v0.7.3) | 1 | datacenter |
ROUGE1: 44.6113 ROUGE2: 22.2456 ROUGEL: 28.8393 TOKENS_PER_SAMPLE: 298.1 | 14,426.4 | 1,803.3 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-fp8_dyn | 70 | UNSET | vLLM v0.7.3 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Server | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, vLLM v0.7.3) | 1 | datacenter |
ROUGEL: 21.59129621595158 exact_match: 90.09433818663138 TOKENS_PER_SAMPLE: 641.0 | 188.604 | 23.5755 | Tokens/s | LLM | llama3.1-405b | llama3_1-405b | 405 | fp16 | vLLM v0.7.3 | null | null | null | null | null | null | null | null | null | Ubuntu 22.04.4 LTS (Jammy Jellyfish) | available | open | Krai | Offline | 8 | 192 | HBM3 | AMD Instinct MI300X 192GB HBM3 | 8 | AMD | null | 40 | 2 | null | Intel(R) Xeon(R) Platinum 8460Y+ | null | Intel | XGMI | PCIe-5 | 2,048 | 32x 64GB DDR5 | Dell PowerEdge XE9680 (8x MI300X-192GB, vLLM v0.7.3) | 1 | datacenter |
ROUGEL: 21.66444823059425 exact_match: 90.10726671078757 TOKENS_PER_SAMPLE: 632.1 | 278.649 | 34.831125 | Tokens/s | LLM | llama3.1-405b | llama3_1-405b-fp8_dyn | 405 | fp16 | vLLM v0.7.3 | null | null | null | null | null | null | null | null | null | Ubuntu 22.04.4 LTS (Jammy Jellyfish) | available | open | Krai | Offline | 8 | 192 | HBM3 | AMD Instinct MI300X 192GB HBM3 | 8 | AMD | null | 40 | 2 | null | Intel(R) Xeon(R) Platinum 8460Y+ | null | Intel | XGMI | PCIe-5 | 2,048 | 32x 64GB DDR5 | Dell PowerEdge XE9680 (8x MI300X-192GB, vLLM v0.7.3) | 1 | datacenter |
ROUGE1: 44.4974 ROUGE2: 22.1155 ROUGEL: 28.7195 TOKENS_PER_SAMPLE: 298.1 | 17,128.9 | 2,141.1125 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-fp8_dyn | 70 | fp16 | vLLM v0.7.3 | null | null | null | null | null | null | null | null | null | Ubuntu 22.04.4 LTS (Jammy Jellyfish) | available | open | Krai | Offline | 8 | 192 | HBM3 | AMD Instinct MI300X 192GB HBM3 | 8 | AMD | null | 40 | 2 | null | Intel(R) Xeon(R) Platinum 8460Y+ | null | Intel | XGMI | PCIe-5 | 2,048 | 32x 64GB DDR5 | Dell PowerEdge XE9680 (8x MI300X-192GB, vLLM v0.7.3) | 1 | datacenter |
ROUGE1: 44.5066 ROUGE2: 22.1409 ROUGEL: 28.7316 TOKENS_PER_SAMPLE: 297.1 | 12,989 | 1,623.625 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-fp8_dyn | 70 | fp16 | vLLM v0.7.3 | null | null | null | null | null | null | null | null | null | Ubuntu 22.04.4 LTS (Jammy Jellyfish) | available | open | Krai | Server | 8 | 192 | HBM3 | AMD Instinct MI300X 192GB HBM3 | 8 | AMD | null | 40 | 2 | null | Intel(R) Xeon(R) Platinum 8460Y+ | null | Intel | XGMI | PCIe-5 | 2,048 | 32x 64GB DDR5 | Dell PowerEdge XE9680 (8x MI300X-192GB, vLLM v0.7.3) | 1 | datacenter |
ROUGE1: 44.5183 ROUGE2: 22.1357 ROUGEL: 28.7267 TOKENS_PER_SAMPLE: 297.0 | 21,372.3 | 2,671.5375 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-fp8_dyn | 70 | fp16 | vLLM v0.6.4 FlashInfer | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, vLLM v0.6.4) | 1 | datacenter |
ROUGE1: 44.5421 ROUGE2: 22.1749 ROUGEL: 28.8061 TOKENS_PER_SAMPLE: 295.1 | 19,618.1 | 2,452.2625 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-fp8_dyn | 70 | fp16 | vLLM v0.6.4 FlashInfer | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Server | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, vLLM v0.6.4) | 1 | datacenter |
ROUGE1: 44.5823 ROUGE2: 22.2287 ROUGEL: 28.8178 TOKENS_PER_SAMPLE: 296.5 | 15,906.1 | 1,988.2625 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-fp8_dyn | 70 | fp16 | vLLM v0.6.4 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, vLLM v0.6.4) | 1 | datacenter |
ROUGE1: 44.5362 ROUGE2: 22.176 ROUGEL: 28.773 TOKENS_PER_SAMPLE: 295.7 | 13,716.7 | 1,714.5875 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-fp8_dyn | 70 | fp16 | vLLM v0.6.4 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Server | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, vLLM v0.6.4) | 1 | datacenter |
ROUGE1: 44.4973 ROUGE2: 22.1394 ROUGEL: 28.7255 TOKENS_PER_SAMPLE: 301.3 | 18,561.4 | 2,320.175 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-fp8_dyn | 70 | fp16 | vLLM v0.6.4 | null | null | null | null | null | null | null | null | null | Ubuntu 22.04.4 LTS (Jammy Jellyfish) | available | open | Krai | Offline | 8 | 192 | HBM3 | AMD Instinct MI300X 192GB HBM3 | 8 | AMD | null | 40 | 2 | null | Intel(R) Xeon(R) Platinum 8460Y+ | null | Intel | XGMI | PCIe-5 | 2,048 | 32x 64GB DDR5 | Dell PowerEdge XE9680 (8x MI300X-192GB, vLLM v0.6.4) | 1 | datacenter |
ROUGE1: 44.5398 ROUGE2: 22.1484 ROUGEL: 28.725 TOKENS_PER_SAMPLE: 299.5 | 13,916.2 | 1,739.525 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-fp8_dyn | 70 | fp16 | vLLM v0.6.4 | null | null | null | null | null | null | null | null | null | Ubuntu 22.04.4 LTS (Jammy Jellyfish) | available | open | Krai | Server | 8 | 192 | HBM3 | AMD Instinct MI300X 192GB HBM3 | 8 | AMD | null | 40 | 2 | null | Intel(R) Xeon(R) Platinum 8460Y+ | null | Intel | XGMI | PCIe-5 | 2,048 | 32x 64GB DDR5 | Dell PowerEdge XE9680 (8x MI300X-192GB, vLLM v0.6.4) | 1 | datacenter |
ROUGE1: 47.5723 ROUGE2: 25.2003 ROUGEL: 31.4338 TOKENS_PER_SAMPLE: 299.6 | 24,624.1 | 3,078.0125 | Tokens/s | LLM | llama2-70b-99.9 | llama3_1-70b-fp8_dyn | 70 | fp16 | SGLang v0.4.3 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, SGLang v0.4.3) | 1 | datacenter |
ROUGE1: 47.589 ROUGE2: 25.2681 ROUGEL: 31.545 TOKENS_PER_SAMPLE: 292.1 | 27,687.9 | 3,460.9875 | Tokens/s | LLM | llama2-70b-99.9 | llama3_1-70b-fp8_pre | 70 | fp8 | SGLang v0.4.3 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, SGLang v0.4.3) | 1 | datacenter |
ROUGE1: 47.6531 ROUGE2: 25.2981 ROUGEL: 31.5596 TOKENS_PER_SAMPLE: 291.3 | 26,074.4 | 3,259.3 | Tokens/s | LLM | llama2-70b-99.9 | llama3_1-70b-fp8_pre | 70 | fp8 | SGLang v0.4.3 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Server | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, SGLang v0.4.3) | 1 | datacenter |
ROUGE1: 47.5684 ROUGE2: 25.2712 ROUGEL: 31.533 TOKENS_PER_SAMPLE: 292.8 | 27,949.8 | 3,493.725 | Tokens/s | LLM | llama2-70b-interactive-99.9 | llama3_1-70b-interactive-fp8_pre | 70 | fp8 | SGLang v0.4.3 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, SGLang v0.4.3) | 1 | datacenter |
ROUGE1: 47.6193 ROUGE2: 25.2818 ROUGEL: 31.5351 TOKENS_PER_SAMPLE: 291.8 | 12,192.4 | 1,524.05 | Tokens/s | LLM | llama2-70b-interactive-99.9 | llama3_1-70b-interactive-fp8_pre | 70 | fp8 | SGLang v0.4.3 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Server | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, SGLang v0.4.3) | 1 | datacenter |
ROUGEL: 21.634266581485935 exact_match: 90.12628722700197 TOKENS_PER_SAMPLE: 656.3 | 326.537 | 40.817125 | Tokens/s | LLM | llama3.1-405b | llama3_1-405b-fp8_dyn | 405 | fp16 | SGLang v0.4.3 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, SGLang v0.4.3) | 1 | datacenter |
ROUGE1: 46.5248 ROUGE2: 24.2311 ROUGEL: 30.2319 TOKENS_PER_SAMPLE: 323.9 | 30,530.1 | 3,816.2625 | Tokens/s | LLM | llama2-70b-interactive-99.9 | llama3_1-70b-interactive-fp8_nim | 70 | fp16 | NIM v1.5.0 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, NIM v1.5.0) | 1 | datacenter |
ROUGE1: 46.5348 ROUGE2: 24.2605 ROUGEL: 30.246 TOKENS_PER_SAMPLE: 324.7 | 15,960.1 | 1,995.0125 | Tokens/s | LLM | llama2-70b-interactive-99.9 | llama3_1-70b-interactive-fp8_nim | 70 | fp16 | NIM v1.5.0 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Server | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, NIM v1.5.0) | 1 | datacenter |
ROUGE1: 46.5273 ROUGE2: 24.2307 ROUGEL: 30.2329 TOKENS_PER_SAMPLE: 323.8 | 30,042.8 | 3,755.35 | Tokens/s | LLM | llama2-70b-99.9 | llama3_1-70b-fp8_nim | 70 | fp16 | NIM v1.5.0 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, NIM v1.5.0) | 1 | datacenter |
ROUGE1: 46.5201 ROUGE2: 24.2273 ROUGEL: 30.2319 TOKENS_PER_SAMPLE: 323.9 | 28,421 | 3,552.625 | Tokens/s | LLM | llama2-70b-99.9 | llama3_1-70b-fp8_nim | 70 | fp16 | NIM v1.5.0 | null | null | null | null | null | null | null | null | null | RHEL 9.4 | available | open | Krai | Server | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 (8x H200-SXM-141GB, NIM v1.5.0) | 1 | datacenter |
ROUGE1: 44.7869 ROUGE2: 22.383 ROUGEL: 29.1434 TOKENS_PER_SAMPLE: 272.9 | 97,403.4 | 12,175.425 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp4 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04 | available | open | Lambda | Offline | 8 | 180 | HBM3e | NVIDIA B200-SXM-180GB | 8 | NVIDIA | null | 56 | 2 | null | Intel(R) Xeon(R) Platinum 8570 | null | Intel | NVLink-5 | PCIe-5 | 2,048 | 32x 64GB HMCG94AGBRA179N | NVIDIA DGX B200 (8x B200-SXM-180GB, TensorRT) | 1 | datacenter |
ROUGE1: 45.6315 ROUGE2: 23.4063 ROUGEL: 30.5125 TOKENS_PER_SAMPLE: 144.1 gsm8k_accuracy: 73.06 mbxp_accuracy: 60.38 | 121,886 | 15,235.75 | Tokens/s | LLM | mixtral-8x7b | mixtral-8x7b | 47 | fp4 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04 | available | open | Lambda | Offline | 8 | 180 | HBM3e | NVIDIA B200-SXM-180GB | 8 | NVIDIA | null | 56 | 2 | null | Intel(R) Xeon(R) Platinum 8570 | null | Intel | NVLink-5 | PCIe-5 | 2,048 | 32x 64GB HMCG94AGBRA179N | NVIDIA DGX B200 (8x B200-SXM-180GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.7869 ROUGE2: 22.383 ROUGEL: 29.1434 TOKENS_PER_SAMPLE: 272.9 | 97,403.4 | 12,175.425 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp4 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04 | available | open | Lambda | Offline | 8 | 180 | HBM3e | NVIDIA B200-SXM-180GB | 8 | NVIDIA | null | 56 | 2 | null | Intel(R) Xeon(R) Platinum 8570 | null | Intel | NVLink-5 | PCIe-5 | 2,048 | 32x 64GB HMCG94AGBRA179N | NVIDIA DGX B200 (8x B200-SXM-180GB, TensorRT) | 1 | datacenter |
ROUGEL: 21.738071137617286 exact_match: 90.02873593646592 TOKENS_PER_SAMPLE: 660.4 | 564.696 | 70.587 | Tokens/s | LLM | llama3.1-405b | llama3_1-405b | 405 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04 | available | open | Supermicro | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | null | 48 | 2 | null | Intel(R) Xeon(R) Platinum 8568Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x 64GB DDR5 4800MT/s @ 4400MT/s | SYS-821GE-TNHR (8x H200-SXM-141GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1186 ROUGE2: 20.1722 ROUGEL: 30.0185 GEN_LEN: 4156655 | 2,791.38 | 2,791.38 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | CUDA 12.8 | null | 12.8 | null | null | null | null | null | null | null | Ubuntu 24.04 | available | open | Supermicro | Offline | 1 | 96 | HBM3 | NVIDIA GH200 Grace Hopper Superchip 96GB | 1 | NVIDIA | null | 72 | 1 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 480 | LPDDR5x | ARS-111GL-NHR (1x GH200-96GB_aarch64) | 1 | datacenter |
ROUGE1: 43.1186 ROUGE2: 20.1722 ROUGEL: 30.0185 GEN_LEN: 4156655 | 2,791.38 | 2,791.38 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | CUDA 12.8 | null | 12.8 | null | null | null | null | null | null | null | Ubuntu 24.04 | available | open | Supermicro | Offline | 1 | 96 | HBM3 | NVIDIA GH200 Grace Hopper Superchip 96GB | 1 | NVIDIA | null | 72 | 1 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 480 | LPDDR5x | ARS-111GL-NHR (1x GH200-96GB_aarch64) | 1 | datacenter |
ROUGE1: 44.5498 ROUGE2: 22.134 ROUGEL: 28.7842 TOKENS_PER_SAMPLE: 291.4 | 8,231.9 | 4,115.95 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Red Hat 9.4 + OpenShift 4.15 | available | open | Supermicro | Offline | 2 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 2 | NVIDIA | null | 144 | 2 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 960 | 15x 16DP (32GB) LPDDR5x | ARS-221GL-NHIR (2x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 44.5491 ROUGE2: 22.1346 ROUGEL: 28.7871 TOKENS_PER_SAMPLE: 292.3 | 7,538.8 | 3,769.4 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Red Hat 9.4 + OpenShift 4.15 | available | open | Supermicro | Server | 2 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 2 | NVIDIA | null | 144 | 2 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 960 | 15x 16DP (32GB) LPDDR5x | ARS-221GL-NHIR (2x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 44.5498 ROUGE2: 22.134 ROUGEL: 28.7842 TOKENS_PER_SAMPLE: 291.4 | 8,231.9 | 4,115.95 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Red Hat 9.4 + OpenShift 4.15 | available | open | Supermicro | Offline | 2 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 2 | NVIDIA | null | 144 | 2 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 960 | 15x 16DP (32GB) LPDDR5x | ARS-221GL-NHIR (2x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 44.5491 ROUGE2: 22.1346 ROUGEL: 28.7871 TOKENS_PER_SAMPLE: 292.3 | 7,538.8 | 3,769.4 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Red Hat 9.4 + OpenShift 4.15 | available | open | Supermicro | Server | 2 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 2 | NVIDIA | null | 144 | 2 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 960 | 15x 16DP (32GB) LPDDR5x | ARS-221GL-NHIR (2x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 44.4322 ROUGE2: 22.0403 ROUGEL: 28.6357 TOKENS_PER_SAMPLE: 294.3 | 31,238 | 3,904.75 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with HPE GreenLake for File Storage (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.4297 ROUGE2: 22.0409 ROUGEL: 28.6367 TOKENS_PER_SAMPLE: 295.4 | 30,457.7 | 3,807.2125 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with HPE GreenLake for File Storage (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 20,796 | 2,599.5 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with HPE GreenLake for File Storage (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 19,363.3 | 2,420.4125 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with HPE GreenLake for File Storage (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 20,993.9 | 2,624.2375 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with HPE GreenLake for File Storage (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 19,363.2 | 2,420.4 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with HPE GreenLake for File Storage (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.4546 ROUGE2: 22.0578 ROUGEL: 28.6551 TOKENS_PER_SAMPLE: 294.3 | 23,389.3 | 2,923.6625 | Tokens/s | LLM | llama2-70b-interactive-99 | llama2-70b-interactive-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with HPE GreenLake for File Storage (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5336 ROUGE2: 22.0916 ROUGEL: 28.7043 TOKENS_PER_SAMPLE: 294.3 | 14,008.5 | 1,751.0625 | Tokens/s | LLM | llama2-70b-interactive-99 | llama2-70b-interactive-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with HPE GreenLake for File Storage (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.4546 ROUGE2: 22.0578 ROUGEL: 28.6551 TOKENS_PER_SAMPLE: 294.3 | 23,385 | 2,923.125 | Tokens/s | LLM | llama2-70b-interactive-99.9 | llama2-70b-interactive-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with HPE GreenLake for File Storage (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5144 ROUGE2: 22.0732 ROUGEL: 28.686 TOKENS_PER_SAMPLE: 294.4 | 13,987.6 | 1,748.45 | Tokens/s | LLM | llama2-70b-interactive-99.9 | llama2-70b-interactive-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with HPE GreenLake for File Storage (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.4347 ROUGE2: 22.0435 ROUGEL: 28.6357 TOKENS_PER_SAMPLE: 294.3 | 31,252.1 | 3,906.5125 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with HPE GreenLake for File Storage (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.4184 ROUGE2: 22.033 ROUGEL: 28.6265 TOKENS_PER_SAMPLE: 295.6 | 30,500.8 | 3,812.6 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with HPE GreenLake for File Storage (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5458 ROUGE2: 22.1332 ROUGEL: 28.7834 TOKENS_PER_SAMPLE: 291.5 | 30,297.5 | 3,787.1875 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 141 | HBM3e | NVIDIA H200-NVL-141GB | 8 | NVIDIA | null | 192 | 2 | null | Intel(R) Xeon(R) 6740E | null | Intel | NVLink-4 | PCIe-5 | 3,072 | 16x 128GB MTC40F2047S1RC64BB1 MWFF; 16x 64GB HMCG94AHBRA480N | HPE ProLiant DL380a Compute Gen12 (8x H200-NVL-141GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5621 ROUGE2: 22.1437 ROUGEL: 28.7928 TOKENS_PER_SAMPLE: 292.3 | 27,638.9 | 3,454.8625 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 141 | HBM3e | NVIDIA H200-NVL-141GB | 8 | NVIDIA | null | 192 | 2 | null | Intel(R) Xeon(R) 6740E | null | Intel | NVLink-4 | PCIe-5 | 3,072 | 16x 128GB MTC40F2047S1RC64BB1 MWFF; 16x 64GB HMCG94AHBRA480N | HPE ProLiant DL380a Compute Gen12 (8x H200-NVL-141GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 17,882.9 | 2,235.3625 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 141 | HBM3e | NVIDIA H200-NVL-141GB | 8 | NVIDIA | null | 192 | 2 | null | Intel(R) Xeon(R) 6740E | null | Intel | NVLink-4 | PCIe-5 | 3,072 | 16x 128GB MTC40F2047S1RC64BB1 MWFF; 16x 64GB HMCG94AHBRA480N | HPE ProLiant DL380a Compute Gen12 (8x H200-NVL-141GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 18,065.1 | 2,258.1375 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 141 | HBM3e | NVIDIA H200-NVL-141GB | 8 | NVIDIA | null | 192 | 2 | null | Intel(R) Xeon(R) 6740E | null | Intel | NVLink-4 | PCIe-5 | 3,072 | 16x 128GB MTC40F2047S1RC64BB1 MWFF; 16x 64GB HMCG94AHBRA480N | HPE ProLiant DL380a Compute Gen12 (8x H200-NVL-141GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 17,882.9 | 2,235.3625 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 141 | HBM3e | NVIDIA H200-NVL-141GB | 8 | NVIDIA | null | 192 | 2 | null | Intel(R) Xeon(R) 6740E | null | Intel | NVLink-4 | PCIe-5 | 3,072 | 16x 128GB MTC40F2047S1RC64BB1 MWFF; 16x 64GB HMCG94AHBRA480N | HPE ProLiant DL380a Compute Gen12 (8x H200-NVL-141GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 18,065.1 | 2,258.1375 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 141 | HBM3e | NVIDIA H200-NVL-141GB | 8 | NVIDIA | null | 192 | 2 | null | Intel(R) Xeon(R) 6740E | null | Intel | NVLink-4 | PCIe-5 | 3,072 | 16x 128GB MTC40F2047S1RC64BB1 MWFF; 16x 64GB HMCG94AHBRA480N | HPE ProLiant DL380a Compute Gen12 (8x H200-NVL-141GB, TensorRT) | 1 | datacenter |
ROUGE1: 45.3505 ROUGE2: 23.2113 ROUGEL: 30.312 TOKENS_PER_SAMPLE: 144.9 gsm8k_accuracy: 73.02 mbxp_accuracy: 59.82 | 52,205.9 | 6,525.7375 | Tokens/s | LLM | mixtral-8x7b | mixtral-8x7b | 47 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 141 | HBM3e | NVIDIA H200-NVL-141GB | 8 | NVIDIA | null | 192 | 2 | null | Intel(R) Xeon(R) 6740E | null | Intel | NVLink-4 | PCIe-5 | 3,072 | 16x 128GB MTC40F2047S1RC64BB1 MWFF; 16x 64GB HMCG94AHBRA480N | HPE ProLiant DL380a Compute Gen12 (8x H200-NVL-141GB, TensorRT) | 1 | datacenter |
ROUGE1: 45.4369 ROUGE2: 23.2825 ROUGEL: 30.4229 TOKENS_PER_SAMPLE: 145.9 gsm8k_accuracy: 73.48 mbxp_accuracy: 59.94 | 50,123.9 | 6,265.4875 | Tokens/s | LLM | mixtral-8x7b | mixtral-8x7b | 47 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 141 | HBM3e | NVIDIA H200-NVL-141GB | 8 | NVIDIA | null | 192 | 2 | null | Intel(R) Xeon(R) 6740E | null | Intel | NVLink-4 | PCIe-5 | 3,072 | 16x 128GB MTC40F2047S1RC64BB1 MWFF; 16x 64GB HMCG94AHBRA480N | HPE ProLiant DL380a Compute Gen12 (8x H200-NVL-141GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5458 ROUGE2: 22.1332 ROUGEL: 28.7834 TOKENS_PER_SAMPLE: 291.5 | 30,297.5 | 3,787.1875 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 141 | HBM3e | NVIDIA H200-NVL-141GB | 8 | NVIDIA | null | 192 | 2 | null | Intel(R) Xeon(R) 6740E | null | Intel | NVLink-4 | PCIe-5 | 3,072 | 16x 128GB MTC40F2047S1RC64BB1 MWFF; 16x 64GB HMCG94AHBRA480N | HPE ProLiant DL380a Compute Gen12 (8x H200-NVL-141GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5621 ROUGE2: 22.1437 ROUGEL: 28.7928 TOKENS_PER_SAMPLE: 292.3 | 27,638.9 | 3,454.8625 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 141 | HBM3e | NVIDIA H200-NVL-141GB | 8 | NVIDIA | null | 192 | 2 | null | Intel(R) Xeon(R) 6740E | null | Intel | NVLink-4 | PCIe-5 | 3,072 | 16x 128GB MTC40F2047S1RC64BB1 MWFF; 16x 64GB HMCG94AHBRA480N | HPE ProLiant DL380a Compute Gen12 (8x H200-NVL-141GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.4331 ROUGE2: 22.0455 ROUGEL: 28.6365 TOKENS_PER_SAMPLE: 294.2 | 31,174.7 | 3,896.8375 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.4303 ROUGE2: 22.039 ROUGEL: 28.6356 TOKENS_PER_SAMPLE: 295.3 | 30,498.9 | 3,812.3625 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 20,716.5 | 2,589.5625 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 20,795.3 | 2,599.4125 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 20,925.3 | 2,615.6625 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 20,991.3 | 2,623.9125 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5218 ROUGE2: 22.1082 ROUGEL: 28.731 TOKENS_PER_SAMPLE: 293.1 | 23,435.4 | 2,929.425 | Tokens/s | LLM | llama2-70b-interactive-99 | llama2-70b-interactive-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5203 ROUGE2: 22.0752 ROUGEL: 28.6917 TOKENS_PER_SAMPLE: 294.2 | 14,004.9 | 1,750.6125 | Tokens/s | LLM | llama2-70b-interactive-99 | llama2-70b-interactive-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5218 ROUGE2: 22.1082 ROUGEL: 28.731 TOKENS_PER_SAMPLE: 293.1 | 23,653.7 | 2,956.7125 | Tokens/s | LLM | llama2-70b-interactive-99.9 | llama2-70b-interactive-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5263 ROUGE2: 22.0802 ROUGEL: 28.6949 TOKENS_PER_SAMPLE: 294.6 | 14,014.1 | 1,751.7625 | Tokens/s | LLM | llama2-70b-interactive-99.9 | llama2-70b-interactive-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.4346 ROUGE2: 22.0435 ROUGEL: 28.6357 TOKENS_PER_SAMPLE: 294.3 | 31,147.4 | 3,893.425 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.4347 ROUGE2: 22.0423 ROUGEL: 28.6376 TOKENS_PER_SAMPLE: 295.4 | 30,480.6 | 3,810.075 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 80 | HBM3 | NVIDIA H100-SXM-80GB | 8 | NVIDIA | L1d: 5.3 MiB (112 instances), L1i: 3.5 MiB (112 instances), L2: 224 MiB (112 instances), L3: 210 MiB (2 instances) | 56 | 2 | 3.8 | Intel(R) Xeon(R) Platinum 8480+ | null | Intel | NVLink-4 | null | 1,024 | 16x 64GB DDR5-4800 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H100-SXM-80GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5024 ROUGE2: 22.0807 ROUGEL: 28.7026 TOKENS_PER_SAMPLE: 293.0 | 3,655.89 | 456.98625 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp16 | TensorRT 9.0, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 9.0 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 48 | HBM2e | NVIDIA L40S | 8 | NVIDIA | L1d: 6 MiB, L1i: 12 MiB, L2: 192 MiB, L3: 192 MiB | 96 | 2 | 2.4 | Intel(R) Xeon(R) 6740E | null | Intel | null | PCIe-5 | 1,024 | 16x 64GB 36ASF8G72PZ-3G2E1 | HPE ProLiant Compute DL380a Gen12 (8x L40S-PCIe-48GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5208 ROUGE2: 22.0821 ROUGEL: 28.7224 TOKENS_PER_SAMPLE: 291.9 | 3,181.65 | 397.70625 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp16 | TensorRT 9.0, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 9.0 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 48 | HBM2e | NVIDIA L40S | 8 | NVIDIA | L1d: 6 MiB, L1i: 12 MiB, L2: 192 MiB, L3: 192 MiB | 96 | 2 | 2.4 | Intel(R) Xeon(R) 6740E | null | Intel | null | PCIe-5 | 1,024 | 16x 64GB 36ASF8G72PZ-3G2E1 | HPE ProLiant Compute DL380a Gen12 (8x L40S-PCIe-48GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5024 ROUGE2: 22.0807 ROUGEL: 28.7026 TOKENS_PER_SAMPLE: 293.0 | 3,655.89 | 456.98625 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp16 | TensorRT 9.0, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 9.0 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 48 | HBM2e | NVIDIA L40S | 8 | NVIDIA | L1d: 6 MiB, L1i: 12 MiB, L2: 192 MiB, L3: 192 MiB | 96 | 2 | 2.4 | Intel(R) Xeon(R) 6740E | null | Intel | null | PCIe-5 | 1,024 | 16x 64GB 36ASF8G72PZ-3G2E1 | HPE ProLiant Compute DL380a Gen12 (8x L40S-PCIe-48GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5208 ROUGE2: 22.0821 ROUGEL: 28.7224 TOKENS_PER_SAMPLE: 291.9 | 3,181.65 | 397.70625 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp16 | TensorRT 9.0, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 9.0 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 48 | HBM2e | NVIDIA L40S | 8 | NVIDIA | L1d: 6 MiB, L1i: 12 MiB, L2: 192 MiB, L3: 192 MiB | 96 | 2 | 2.4 | Intel(R) Xeon(R) 6740E | null | Intel | null | PCIe-5 | 1,024 | 16x 64GB 36ASF8G72PZ-3G2E1 | HPE ProLiant Compute DL380a Gen12 (8x L40S-PCIe-48GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5446 ROUGE2: 22.1336 ROUGEL: 28.7828 TOKENS_PER_SAMPLE: 291.5 | 4,609.05 | 4,609.05 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.7 | null | 12.7 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 1 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 1 | NVIDIA | null | 72 | 1 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 480 | 15x 16DP (32GB) LPDDR5x | HPE ProLiant Compute DL384 Gen12 (1x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 44.5531 ROUGE2: 22.1392 ROUGEL: 28.7903 TOKENS_PER_SAMPLE: 291.5 | 4,327.68 | 4,327.68 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.7 | null | 12.7 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 1 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 1 | NVIDIA | null | 72 | 1 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 480 | 15x 16DP (32GB) LPDDR5x | HPE ProLiant Compute DL384 Gen12 (1x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 43.1186 ROUGE2: 20.1722 ROUGEL: 30.0185 GEN_LEN: 4156655 | 2,845.43 | 2,845.43 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | TensorRT 10.8, CUDA 12.7 | null | 12.7 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 1 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 1 | NVIDIA | null | 72 | 1 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 480 | 15x 16DP (32GB) LPDDR5x | HPE ProLiant Compute DL384 Gen12 (1x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 43.1186 ROUGE2: 20.1722 ROUGEL: 30.0185 GEN_LEN: 4156655 | 2,526.03 | 2,526.03 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | TensorRT 10.8, CUDA 12.7 | null | 12.7 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 1 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 1 | NVIDIA | null | 72 | 1 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 480 | 15x 16DP (32GB) LPDDR5x | HPE ProLiant Compute DL384 Gen12 (1x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 43.1186 ROUGE2: 20.1722 ROUGEL: 30.0185 GEN_LEN: 4156655 | 2,845.43 | 2,845.43 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.7 | null | 12.7 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 1 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 1 | NVIDIA | null | 72 | 1 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 480 | 15x 16DP (32GB) LPDDR5x | HPE ProLiant Compute DL384 Gen12 (1x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 43.1186 ROUGE2: 20.1722 ROUGEL: 30.0185 GEN_LEN: 4156655 | 2,526.03 | 2,526.03 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.7 | null | 12.7 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 1 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 1 | NVIDIA | null | 72 | 1 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 480 | 15x 16DP (32GB) LPDDR5x | HPE ProLiant Compute DL384 Gen12 (1x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 45.2956 ROUGE2: 23.2043 ROUGEL: 30.3791 TOKENS_PER_SAMPLE: 144.9 gsm8k_accuracy: 73.0 mbxp_accuracy: 59.96 | 8,324.43 | 8,324.43 | Tokens/s | LLM | mixtral-8x7b | mixtral-8x7b | 47 | fp8 | TensorRT 10.8, CUDA 12.7 | null | 12.7 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 1 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 1 | NVIDIA | null | 72 | 1 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 480 | 15x 16DP (32GB) LPDDR5x | HPE ProLiant Compute DL384 Gen12 (1x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 45.2751 ROUGE2: 23.1818 ROUGEL: 30.3401 TOKENS_PER_SAMPLE: 145.5 gsm8k_accuracy: 73.2 mbxp_accuracy: 60.1 | 7,735.98 | 7,735.98 | Tokens/s | LLM | mixtral-8x7b | mixtral-8x7b | 47 | fp8 | TensorRT 10.8, CUDA 12.7 | null | 12.7 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 1 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 1 | NVIDIA | null | 72 | 1 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 480 | 15x 16DP (32GB) LPDDR5x | HPE ProLiant Compute DL384 Gen12 (1x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 44.5446 ROUGE2: 22.1336 ROUGEL: 28.7828 TOKENS_PER_SAMPLE: 291.5 | 4,609.05 | 4,609.05 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.7 | null | 12.7 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 1 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 1 | NVIDIA | null | 72 | 1 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 480 | 15x 16DP (32GB) LPDDR5x | HPE ProLiant Compute DL384 Gen12 (1x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 44.5531 ROUGE2: 22.1392 ROUGEL: 28.7903 TOKENS_PER_SAMPLE: 291.5 | 4,327.68 | 4,327.68 | Tokens/s | LLM | llama2-70b-99.9 | llama2-70b-99.9 | 70 | fp8 | TensorRT 10.8, CUDA 12.7 | null | 12.7 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 1 | 144 | HBM3e | NVIDIA GH200 Grace Hopper Superchip 144GB | 1 | NVIDIA | null | 72 | 1 | null | NVIDIA Grace CPU | null | NVIDIA | InfiniBand | NVLink | 480 | 15x 16DP (32GB) LPDDR5x | HPE ProLiant Compute DL384 Gen12 (1x GH200-144GB_aarch64, TensorRT) | 1 | datacenter |
ROUGE1: 43.1295 ROUGE2: 20.1909 ROUGEL: 30.052 GEN_LEN: 4153653 | 6,823.91 | 852.98875 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 48 | HBM2e | NVIDIA L40S | 8 | NVIDIA | L1d: 6 MiB, L1i: 12 MiB, L2: 192 MiB, L3: 192 MiB | 96 | 2 | 2.4 | Intel(R) Xeon(R) 6740E | null | Intel | null | PCIe-5 | 1,024 | 16x 64GB 36ASF8G72PZ-3G2E1 | HPE ProLiant Compute DL380a Gen12 (8x L40S-PCIe-48GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1396 ROUGE2: 20.1904 ROUGEL: 30.0525 GEN_LEN: 4153772 | 6,821.11 | 852.63875 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 48 | HBM2e | NVIDIA L40S | 8 | NVIDIA | L1d: 6 MiB, L1i: 12 MiB, L2: 192 MiB, L3: 192 MiB | 96 | 2 | 2.4 | Intel(R) Xeon(R) 6740E | null | Intel | null | PCIe-5 | 1,024 | 16x 64GB 36ASF8G72PZ-3G2E1 | HPE ProLiant Compute DL380a Gen12 (8x L40S-PCIe-48GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1295 ROUGE2: 20.1909 ROUGEL: 30.052 GEN_LEN: 4153653 | 6,823.91 | 852.98875 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 8 | 48 | HBM2e | NVIDIA L40S | 8 | NVIDIA | L1d: 6 MiB, L1i: 12 MiB, L2: 192 MiB, L3: 192 MiB | 96 | 2 | 2.4 | Intel(R) Xeon(R) 6740E | null | Intel | null | PCIe-5 | 1,024 | 16x 64GB 36ASF8G72PZ-3G2E1 | HPE ProLiant Compute DL380a Gen12 (8x L40S-PCIe-48GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.1396 ROUGE2: 20.1904 ROUGEL: 30.0525 GEN_LEN: 4153772 | 6,821.11 | 852.63875 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 8 | 48 | HBM2e | NVIDIA L40S | 8 | NVIDIA | L1d: 6 MiB, L1i: 12 MiB, L2: 192 MiB, L3: 192 MiB | 96 | 2 | 2.4 | Intel(R) Xeon(R) 6740E | null | Intel | null | PCIe-5 | 1,024 | 16x 64GB 36ASF8G72PZ-3G2E1 | HPE ProLiant Compute DL380a Gen12 (8x L40S-PCIe-48GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.0194 ROUGE2: 20.0851 ROUGEL: 29.9991 GEN_LEN: 4155554 | 7,240.14 | 1,810.035 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Offline | 4 | 94 | HBM3 | NVIDIA H100-NVL-94GB | 4 | NVIDIA | L1d: 9 MiB (192 instances), L1i: 6 MiB (192 instances), L2: 384 MiB (192 instances), L3: 420 MiB (4 instances) | 192 | 4 | 3.8 | INTEL(R) XEON(R) PLATINUM 8468 | null | Intel | null | null | 8,192 | 1TB total, 32x 32GB | HPE Compute Scale-up Server 3200 (4x H100-NVL-94GB, TensorRT) | 1 | datacenter |
ROUGE1: 43.0194 ROUGE2: 20.0851 ROUGEL: 29.9991 GEN_LEN: 4155554 | 7,179.86 | 1,794.965 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | Ubuntu 24.04.1 LTS | available | closed | HPE | Server | 4 | 94 | HBM3 | NVIDIA H100-NVL-94GB | 4 | NVIDIA | L1d: 9 MiB (192 instances), L1i: 6 MiB (192 instances), L2: 384 MiB (192 instances), L3: 420 MiB (4 instances) | 192 | 4 | 3.8 | INTEL(R) XEON(R) PLATINUM 8468 | null | Intel | null | null | 8,192 | 1TB total, 32x 32GB | HPE Compute Scale-up Server 3200 (4x H100-NVL-94GB, TensorRT) | 1 | datacenter |
ROUGE1: 44.5453 ROUGE2: 22.1331 ROUGEL: 28.7834 TOKENS_PER_SAMPLE: 291.6 | 33,781.8 | 4,222.725 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | RHEL 9.4 | available | closed | HPE | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H200-SXM-141GB) | 1 | datacenter |
ROUGE1: 44.5663 ROUGE2: 22.1467 ROUGEL: 28.7958 TOKENS_PER_SAMPLE: 292.3 | 31,281.1 | 3,910.1375 | Tokens/s | LLM | llama2-70b-99 | llama2-70b-99 | 70 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | RHEL 9.4 | available | closed | HPE | Server | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H200-SXM-141GB) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 21,014.6 | 2,626.825 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | RHEL 9.4 | available | closed | HPE | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H200-SXM-141GB) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 21,019.9 | 2,627.4875 | Tokens/s | LLM | gptj-99.9 | gptj-99.9 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | RHEL 9.4 | available | closed | HPE | Server | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H200-SXM-141GB) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 21,324.9 | 2,665.6125 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | RHEL 9.4 | available | closed | HPE | Offline | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H200-SXM-141GB) | 1 | datacenter |
ROUGE1: 43.1118 ROUGE2: 20.1071 ROUGEL: 29.9676 GEN_LEN: 4153194 | 21,025.5 | 2,628.1875 | Tokens/s | LLM | gptj-99 | gptj-99 | 6 | fp8 | TensorRT 10.8, CUDA 12.8 | null | 12.8 | null | null | null | null | null | 10.8 | null | RHEL 9.4 | available | closed | HPE | Server | 8 | 141 | HBM3e | NVIDIA H200-SXM-141GB | 8 | NVIDIA | L1d: 3 MiB (64 instances), L1i: 2 MiB (64 instances), L2: 128 MiB (64 instances), L3: 120 MiB (2 instances) | 32 | 2 | 2.8 | Intel(R) Xeon(R) Platinum 8562Y+ | null | Intel | NVLink-4 | PCIe-5 | 2,048 | 32x HPE Cray Supercomputing XD 64GB (1x64GB) Dual Rank x4 DDR5-5600 Registered Standard Memory | HPE Cray XD670 with Cray ClusterStor (8x H200-SXM-141GB) | 1 | datacenter |
End of preview. Expand
in Data Studio
Preparing OpenMLPerf dataset
To process the semi-raw MLPerf data into the OpenMLPerf dataset, run the following command:
# Untar raw files
bzip2 -d semi-raw-mlperf-data.tar.bz2
tar xvf semi-raw-mlperf-data.tar
# Create a virtual environment
python -m venv .venv
# Activate the virtual environment
source .venv/bin/activate
# Install the required packages
pip install -r requirements.txt
# Run the processing script
python process.py
The processed dataset will be saved both as data.json and data.parquet in the OpenMLPerf-dataset directory.
The data.json file is a JSON file containing the processed data, while the data.parquet file is a Parquet file containing the same data in a more efficient format for storage and processing.
Preprocessing raw MLPerf results using MLCommons CMX
We preprocess official raw MLPerf data, such as inference v5.0,
into semi-raw format compatible with the process.py script, using the MLCommons CM/CMX automation framework.
This is done using through the "import mlperf results"
automation action, which we plan to document in more detail soon.
License and Copyright
This project is licensed under the Apache License 2.0.
© 2025 FlexAI
Portions of the data were adapted from the following MLCommons repositories, which are also licensed under the Apache 2.0 license:
- mlcommons@inference_results_v5.0
- mlcommons@inference_results_v4.1
- mlcommons@inference_results_v4.0
- mlcommons@inference_results_v3.1
Authors and maintaners
Daniel Altunay and Grigori Fursin (FCS Labs)
- Downloads last month
- 59