|
|
|
@@ -7,6 +7,8 @@ export const config = {
|
|
|
|
|
supportedHardware: [
|
|
|
|
|
"h100", "h200", "b200", "b300", "gb200", "gb300",
|
|
|
|
|
"rtx6000",
|
|
|
|
|
// AMD ROCm — MI300X (Flash FP8) + MI355X (Flash/Pro, FP4/FP8).
|
|
|
|
|
"mi300x", "mi355x",
|
|
|
|
|
],
|
|
|
|
|
|
|
|
|
|
// Model-specific GPUs the shared HARDWARE_CATALOG doesn't carry — the engine
|
|
|
|
@@ -43,6 +45,10 @@ export const config = {
|
|
|
|
|
// H200 FP8 needs the sgl-project repackaging (Hopper can't run FP4-mixed Instruct).
|
|
|
|
|
"h200|flash|fp8": "sgl-project/DeepSeek-V4-Flash-FP8",
|
|
|
|
|
"h200|pro|fp8": "sgl-project/DeepSeek-V4-Pro-FP8",
|
|
|
|
|
// AMD FP8 uses the sgl-project repackaging.
|
|
|
|
|
"mi300x|flash|fp8": "sgl-project/DeepSeek-V4-Flash-FP8",
|
|
|
|
|
"mi355x|flash|fp8": "sgl-project/DeepSeek-V4-Flash-FP8",
|
|
|
|
|
"mi355x|pro|fp8": "sgl-project/DeepSeek-V4-Pro-FP8",
|
|
|
|
|
},
|
|
|
|
|
|
|
|
|
|
placeholders: {
|
|
|
|
@@ -147,6 +153,10 @@ sgl-eval run aime25 \\
|
|
|
|
|
b300: "lmsysorg/sglang:latest",
|
|
|
|
|
gb200: "lmsysorg/sglang:latest",
|
|
|
|
|
gb300: "lmsysorg/sglang:latest",
|
|
|
|
|
// AMD daily-updated lmsysorg/sglang-rocm images. Bump the dated tag when you
|
|
|
|
|
// re-verify on a newer build.
|
|
|
|
|
mi300x: "lmsysorg/sglang-rocm:v0.5.13.post1-rocm720-mi30x-20260615",
|
|
|
|
|
mi355x: "lmsysorg/sglang-rocm:v0.5.13.post1-rocm720-mi35x-20260615",
|
|
|
|
|
},
|
|
|
|
|
|
|
|
|
|
// Pre-selects the issue template's `model` dropdown on "Submit verified cell".
|
|
|
|
@@ -1200,5 +1210,719 @@ sgl-eval run aime25 \\
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
|
|
|
|
|
// ====================================================================
|
|
|
|
|
// AMD ROCm (MI300X / MI355X)
|
|
|
|
|
// --------------------------------------------------------------------
|
|
|
|
|
|
|
|
|
|
// ---------- MI300X (192GB) — Flash FP8 ----------
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi300x", variant: "flash", quant: "fp8", strategy: "low-latency", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 8192",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi300x", variant: "flash", quant: "fp8", strategy: "balanced", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--dp 8",
|
|
|
|
|
"--enable-dp-attention",
|
|
|
|
|
"--enable-prefill-delayer",
|
|
|
|
|
"--prefill-delayer-max-delay-ms 5000",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 65536",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi300x", variant: "flash", quant: "fp8", strategy: "high-throughput", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--dp 8",
|
|
|
|
|
"--enable-dp-attention",
|
|
|
|
|
"--enable-prefill-delayer",
|
|
|
|
|
"--prefill-delayer-max-delay-ms 5000",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 65536",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
|
|
|
|
|
// ---------- MI355X (288GB) — Flash FP4 ----------
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi355x", variant: "flash", quant: "fp4", strategy: "low-latency", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 8192",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi355x", variant: "flash", quant: "fp4", strategy: "balanced", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--dp 8",
|
|
|
|
|
"--enable-dp-attention",
|
|
|
|
|
"--enable-prefill-delayer",
|
|
|
|
|
"--prefill-delayer-max-delay-ms 5000",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 65536",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi355x", variant: "flash", quant: "fp4", strategy: "high-throughput", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--dp 8",
|
|
|
|
|
"--enable-dp-attention",
|
|
|
|
|
"--enable-prefill-delayer",
|
|
|
|
|
"--prefill-delayer-max-delay-ms 5000",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 65536",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
|
|
|
|
|
// ---------- MI355X (288GB) — Flash FP8 ----------
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi355x", variant: "flash", quant: "fp8", strategy: "low-latency", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 8192",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi355x", variant: "flash", quant: "fp8", strategy: "balanced", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--dp 8",
|
|
|
|
|
"--enable-dp-attention",
|
|
|
|
|
"--enable-prefill-delayer",
|
|
|
|
|
"--prefill-delayer-max-delay-ms 5000",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 65536",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi355x", variant: "flash", quant: "fp8", strategy: "high-throughput", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--dp 8",
|
|
|
|
|
"--enable-dp-attention",
|
|
|
|
|
"--enable-prefill-delayer",
|
|
|
|
|
"--prefill-delayer-max-delay-ms 5000",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 65536",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
|
|
|
|
|
// ---------- MI355X (288GB) — Pro FP4 ----------
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi355x", variant: "pro", quant: "fp4", strategy: "low-latency", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 8192",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi355x", variant: "pro", quant: "fp4", strategy: "balanced", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--dp 8",
|
|
|
|
|
"--enable-dp-attention",
|
|
|
|
|
"--enable-prefill-delayer",
|
|
|
|
|
"--prefill-delayer-max-delay-ms 5000",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 65536",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi355x", variant: "pro", quant: "fp4", strategy: "high-throughput", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--dp 8",
|
|
|
|
|
"--enable-dp-attention",
|
|
|
|
|
"--enable-prefill-delayer",
|
|
|
|
|
"--prefill-delayer-max-delay-ms 5000",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 65536",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
|
|
|
|
|
// ---------- MI355X (288GB) — Pro FP8 ----------
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi355x", variant: "pro", quant: "fp8", strategy: "low-latency", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 8192",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi355x", variant: "pro", quant: "fp8", strategy: "balanced", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--dp 8",
|
|
|
|
|
"--enable-dp-attention",
|
|
|
|
|
"--enable-prefill-delayer",
|
|
|
|
|
"--prefill-delayer-max-delay-ms 5000",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 65536",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
match: { hw: "mi355x", variant: "pro", quant: "fp8", strategy: "high-throughput", nodes: "single" },
|
|
|
|
|
verified: true,
|
|
|
|
|
env: [
|
|
|
|
|
"SGLANG_DEFAULT_THINKING=1",
|
|
|
|
|
"SGLANG_DSV4_REASONING_EFFORT=max",
|
|
|
|
|
"SGLANG_OPT_DEEPGEMM_HC_PRENORM=false",
|
|
|
|
|
"SGLANG_USE_AITER=1",
|
|
|
|
|
"SGLANG_USE_ROCM700A=0",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS=true",
|
|
|
|
|
"SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
|
|
|
|
|
"SGLANG_OPT_FP8_WO_A_GEMM=false",
|
|
|
|
|
"SGLANG_OPT_USE_JIT_INDEXER_METADATA=false",
|
|
|
|
|
"SGLANG_OPT_USE_TOPK_V2=false",
|
|
|
|
|
"SGLANG_OPT_USE_AITER_INDEXER=true",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_INDEXER=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_PRE=false",
|
|
|
|
|
"SGLANG_OPT_USE_TILELANG_MHC_POST=false",
|
|
|
|
|
"SGLANG_FP8_PAGED_MQA_LOGITS_TORCH=1",
|
|
|
|
|
"SGLANG_OPT_USE_FUSED_COMPRESS_TRITON=true",
|
|
|
|
|
"SGLANG_OPT_USE_MULTI_STREAM_OVERLAP=false",
|
|
|
|
|
"SGLANG_ROCM_USE_MULTI_STREAM=false",
|
|
|
|
|
"AITER_BF16_FP8_MOE_BOUND=0",
|
|
|
|
|
"SGLANG_EAGER_INPUT_NO_COPY=true",
|
|
|
|
|
],
|
|
|
|
|
flags: [
|
|
|
|
|
"--trust-remote-code",
|
|
|
|
|
"--model-path {{MODEL_NAME}}",
|
|
|
|
|
"--tp 8",
|
|
|
|
|
"--dp 8",
|
|
|
|
|
"--enable-dp-attention",
|
|
|
|
|
"--enable-prefill-delayer",
|
|
|
|
|
"--prefill-delayer-max-delay-ms 5000",
|
|
|
|
|
"--attention-backend dsv4",
|
|
|
|
|
"--page-size 256",
|
|
|
|
|
"--mem-fraction-static 0.90",
|
|
|
|
|
"--swa-full-tokens-ratio 0.1",
|
|
|
|
|
"--disable-shared-experts-fusion",
|
|
|
|
|
"--kv-cache-dtype fp8_e4m3",
|
|
|
|
|
"--chunked-prefill-size 65536",
|
|
|
|
|
"--speculative-algorithm EAGLE",
|
|
|
|
|
"--speculative-num-steps 3",
|
|
|
|
|
"--speculative-eagle-topk 1",
|
|
|
|
|
"--speculative-num-draft-tokens 4",
|
|
|
|
|
"--host {{HOST_IP}}",
|
|
|
|
|
"--port {{PORT}}",
|
|
|
|
|
],
|
|
|
|
|
},
|
|
|
|
|
],
|
|
|
|
|
};
|
|
|
|
|