Erste SchritteLeaderboardDecode calculatorModelleOrganizationsReportsMatrix runsHardwareBenchmarksMarktplatzVermietungenProAPI-Doku
Sprache
Actual Computer — Every computer, one endpoint

Speed test matrix

Laguna-XS.2 on H200 NVL 141GB

vllm · 0.26.0 · BF166 contexts (8K–251K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 128K, 251K, 256K
by @keennaylmx dev importMeasured

Hardware

Type
Discrete GPU
GPU
H200 NVL
VRAM
141 GB
OS
linux

Engine

Engine
vllm
Version
0.26.0
Quantization
BF16
Backend
cuda
Build
cuda-13.2

Run

Served model
poolside
Revision
main
Decode test
Fixed duration
Prefill mode
Measured during decode warm-up
Time per measurement
30 s
Max output
4,096 tokens

Launch command

"$HOME/env_h200-poolside-laguna-xs-vllm/bin/vllm" serve poolside/Laguna-XS.2 --served-model-name poolside --trust-remote-code --tensor-parallel-size 1 --reasoning-parser poolside_v1 --enable-auto-tool-choice --tool-call-parser poolside_v1 --max-model-len 262144 --gpu-memory-utilization 0.876036 --host 0.0.0.0 --port 8000 --api-key <redacted>

Prefill

prompt tok/s by context length
8K context43.2K190 ms TTFT
16K context40.1K404 ms TTFT
32K context37K872 ms TTFT
64K context29.9K2,151 ms TTFT
128K context20.9K6,146 ms TTFT
251K context13.1K19,135 ms TTFT
256K context13.1K19,615 ms TTFT

Sustained decode

total tok/s · rows are context length, columns are simultaneous requests
Context1248163264128
8K208.637 ms TTFT355.567 ms TTFT589.076 ms TTFT905.7159 ms TTFT1,449.5199 ms TTFT2,158.1317 ms TTFT3,232.3479 ms TTFTOver capacity
16K204.663 ms TTFT347.0120 ms TTFT570.1155 ms TTFT864.4218 ms TTFT1,362.3262 ms TTFT1,967.5475 ms TTFT2,800.2681 ms TTFTOver capacity
32K197.7115 ms TTFT332.5132 ms TTFT537.4203 ms TTFT795.7357 ms TTFT1,207.7510 ms TTFT1,666.6842 ms TTFTOver capacityOver capacity
64K184.9168 ms TTFT306.5208 ms TTFT483.5249 ms TTFT690.1565 ms TTFT988.9965 ms TTFTOver capacityOver capacityOver capacity
128K165.2300 ms TTFT264.2375 ms TTFT406.0558 ms TTFT540.8878 ms TTFTOver capacityOver capacityOver capacityOver capacity
251K135.7576 ms TTFT213.5769 ms TTFT308.71,159 ms TTFTOver capacityOver capacityOver capacityOver capacityOver capacity

Over capacity: not run because this many requests at this context would exceed the server's KV cache.

TTFT is measured after the benchmark warms the server's prompt cache, so it mostly excludes prefill.

All measurements

Latencies are medians. Total tok/s is output across all simultaneous requests.
PhaseContextRequests at onceStatusTotal tok/sPer request tok/sTTFTTime between tokensAverage active requestsCompleted requestsOutput tokensMeasured for
Prefill8K1Valid43.2K—190 ms——1——
Prefill16K1Valid40.1K—404 ms——1——
Prefill32K1Valid37K—872 ms——1——
Prefill64K1Valid29.9K—2,151 ms——1——
Prefill128K1Valid20.9K—6,146 ms——1——
Prefill251K1Valid13.1K—19,135 ms——1——
Prefill256K1Valid13.1K—19,615 ms——1——
Sustained decode8K1Valid208.6209.137 ms4.8 ms1.016,26230.0 s
Sustained decode8K2Valid355.5178.567 ms5.6 ms2.0210,66330.0 s
Sustained decode8K4Valid589.0148.376 ms6.7 ms4.0417,66930.0 s
Sustained decode8K8Valid905.7113.6159 ms8.8 ms8.0027,16830.0 s
Sustained decode8K16Valid1,449.591.4199 ms10.9 ms16.0043,48830.0 s
Sustained decode8K32Valid2,158.168.1317 ms14.7 ms32.0064,73630.0 s
Sustained decode8K64Valid3,232.351.1479 ms19.6 ms64.0096,96030.0 s
Sustained decode8K128Over capacity————————
Sustained decode16K1Valid204.6205.363 ms4.9 ms1.016,13730.0 s
Sustained decode16K2Valid347.0174.7120 ms5.7 ms2.0210,41030.0 s
Sustained decode16K4Valid570.1144.0155 ms6.9 ms4.0417,10230.0 s
Sustained decode16K8Valid864.4108.4218 ms9.2 ms8.0025,92830.0 s
Sustained decode16K16Valid1,362.385.8262 ms11.7 ms16.0040,86430.0 s
Sustained decode16K32Valid1,967.561.9475 ms16.1 ms32.0059,00830.0 s
Sustained decode16K64Valid2,800.244.2681 ms22.6 ms64.0083,96830.0 s
Sustained decode16K128Over capacity————————
Sustained decode32K1Valid197.7198.8115 ms5.0 ms1.015,93230.0 s
Sustained decode32K2Valid332.5167.4132 ms6.0 ms2.029,97430.0 s
Sustained decode32K4Valid537.4136.1203 ms7.3 ms4.0416,11930.0 s
Sustained decode32K8Valid795.799.7357 ms10.0 ms8.0023,86430.0 s
Sustained decode32K16Valid1,207.776.0510 ms13.2 ms16.0036,22430.0 s
Sustained decode32K32Valid1,666.652.4842 ms19.1 ms32.0049,98430.0 s
Sustained decode32K64Over capacity————————
Sustained decode32K128Over capacity————————
Sustained decode64K1Valid184.9186.3168 ms5.4 ms1.015,54730.0 s
Sustained decode64K2Valid306.5155.0208 ms6.5 ms2.029,19530.0 s
Sustained decode64K4Valid483.5123.0249 ms8.1 ms3.9414,50130.0 s
Sustained decode64K8Valid690.186.4565 ms11.6 ms8.0020,69630.0 s
Sustained decode64K16Valid988.962.0965 ms16.1 ms16.0029,66430.0 s
Sustained decode64K32Over capacity————————
Sustained decode64K64Over capacity————————
Sustained decode64K128Over capacity————————
Sustained decode128K1Valid165.2167.7300 ms6.0 ms1.014,95530.0 s
Sustained decode128K2Valid264.2135.3375 ms7.4 ms1.927,92530.0 s
Sustained decode128K4Valid406.0101.7558 ms9.8 ms4.0012,18030.0 s
Sustained decode128K8Valid540.867.6878 ms14.8 ms8.0016,22430.0 s
Sustained decode128K16Over capacity————————
Sustained decode128K32Over capacity————————
Sustained decode128K64Over capacity————————
Sustained decode128K128Over capacity————————
Sustained decode251K1Valid135.7141.0576 ms7.1 ms1.014,07430.0 s
Sustained decode251K2Valid213.5107.1769 ms9.3 ms2.006,40630.0 s
Sustained decode251K4Valid308.777.41,159 ms12.9 ms4.009,26030.0 s
Sustained decode251K8Over capacity————————
Sustained decode251K16Over capacity————————
Sustained decode251K32Over capacity————————
Sustained decode251K64Over capacity————————
Sustained decode251K128Over capacity————————

Over capacity: not run because this many requests at this context would exceed the server's KV cache.