[AMD] Update v4 amd cookbook (#28423)
This commit is contained in:
@@ -28,15 +28,15 @@ Then run the **Python** output of the command panel below in that environment.
|
|||||||
|
|
||||||
<Tab title="Docker">
|
<Tab title="Docker">
|
||||||
|
|
||||||
|
For how to launch the image, see [Install → Method 3: Using Docker](../../../docs/get-started/install#method-3-using-docker). A minimal example (substitute the inner `sglang serve ...` with whatever the command generator below produces):
|
||||||
|
|
||||||
|
**NVIDIA GPUs**
|
||||||
|
|
||||||
A single image — `lmsysorg/sglang:latest` — covers the **datacenter GPUs** in this cookbook (B200 / B300 / GB200 / GB300 / H100 / H200). For **RTX PRO 6000 (SM120)**, use the nightly `lmsysorg/sglang:dev` instead — SM120 support isn't in `:latest` yet (see the RTX PRO 6000 note below).
|
A single image — `lmsysorg/sglang:latest` — covers the **datacenter GPUs** in this cookbook (B200 / B300 / GB200 / GB300 / H100 / H200). For **RTX PRO 6000 (SM120)**, use the nightly `lmsysorg/sglang:dev` instead — SM120 support isn't in `:latest` yet (see the RTX PRO 6000 note below).
|
||||||
|
|
||||||
```bash Command
|
```bash Command
|
||||||
docker pull lmsysorg/sglang:latest
|
docker pull lmsysorg/sglang:latest
|
||||||
```
|
|
||||||
|
|
||||||
For how to launch the image, see [Install → Method 3: Using Docker](../../../docs/get-started/install#method-3-using-docker). A minimal example (substitute the inner `sglang serve ...` with whatever the command generator below produces):
|
|
||||||
|
|
||||||
```bash Command
|
|
||||||
docker run --gpus all \
|
docker run --gpus all \
|
||||||
--shm-size 32g \
|
--shm-size 32g \
|
||||||
-p 30000:30000 \
|
-p 30000:30000 \
|
||||||
@@ -47,6 +47,28 @@ docker run --gpus all \
|
|||||||
sglang serve <use args below>
|
sglang serve <use args below>
|
||||||
```
|
```
|
||||||
|
|
||||||
|
**AMD GPUs (ROCm)**
|
||||||
|
|
||||||
|
AMD uses the daily-updated `lmsysorg/sglang-rocm` images:
|
||||||
|
|
||||||
|
- **MI355X** → `lmsysorg/sglang-rocm:v0.5.13.post1-rocm720-mi35x-20260615`
|
||||||
|
- **MI300X** → `lmsysorg/sglang-rocm:v0.5.13.post1-rocm720-mi30x-20260615`
|
||||||
|
|
||||||
|
```bash Command
|
||||||
|
docker pull lmsysorg/sglang-rocm:v0.5.13.post1-rocm720-mi35x-20260615
|
||||||
|
|
||||||
|
docker run \
|
||||||
|
--device=/dev/kfd --device=/dev/dri \
|
||||||
|
--group-add video \
|
||||||
|
--cap-add=SYS_PTRACE --security-opt seccomp=unconfined \
|
||||||
|
--shm-size 32g --ipc=host \
|
||||||
|
-p 30000:30000 \
|
||||||
|
-v ~/.cache/huggingface:/root/.cache/huggingface \
|
||||||
|
--env "HF_TOKEN=<your-hf-token>" \
|
||||||
|
lmsysorg/sglang-rocm:v0.5.13.post1-rocm720-mi35x-20260615 \
|
||||||
|
sglang serve <use args below>
|
||||||
|
```
|
||||||
|
|
||||||
</Tab>
|
</Tab>
|
||||||
|
|
||||||
</Tabs>
|
</Tabs>
|
||||||
@@ -233,6 +255,14 @@ RTX PRO 6000 (96 GB) runs **Flash only** — V4-Pro doesn't fit on 8× 96 GB. It
|
|||||||
`--mem-fraction-static 0.70`; the Deploy panel greys out the other recipes for this card.
|
`--mem-fraction-static 0.70`; the Deploy panel greys out the other recipes for this card.
|
||||||
HiCache and MegaMoE are **not** supported on RTX PRO 6000. For Docker, use the nightly `lmsysorg/sglang:dev` image — SM120 support isn't in `lmsysorg/sglang:latest` yet (the Deploy panel's Docker mode already points this card at `:dev`).
|
HiCache and MegaMoE are **not** supported on RTX PRO 6000. For Docker, use the nightly `lmsysorg/sglang:dev` image — SM120 support isn't in `lmsysorg/sglang:latest` yet (the Deploy panel's Docker mode already points this card at `:dev`).
|
||||||
|
|
||||||
|
**AMD (MI300X / MI355X) note**
|
||||||
|
|
||||||
|
- **Model checkpoints** — for correct accuracy, the FP4 model uses the stock `deepseek-ai/DeepSeek-V4-{Flash,Pro}`, and the FP8 model uses the repackaged `sgl-project/DeepSeek-V4-{Flash,Pro}-FP8`.
|
||||||
|
- **Supported models** — **MI300X** supports DeepSeek-V4-Flash in FP8; **MI355X** supports DeepSeek-V4-Flash / Pro in both FP4 and FP8. All recipes run single-node.
|
||||||
|
- **TP / DP setting** — both TP=4 and TP=8 are supported. At low concurrency we recommend **TP-only**; at high concurrency use **TP + DP**, which additionally needs `--dp 8 --enable-dp-attention --enable-prefill-delayer --prefill-delayer-max-delay-ms 5000`.
|
||||||
|
- **MTP** — speculative decoding is supported; add `--speculative-algorithm EAGLE --speculative-num-steps 3 --speculative-eagle-topk 1 --speculative-num-draft-tokens 4`.
|
||||||
|
- **Kernels** — uses the Unified KV attention and the flydsl MoE.
|
||||||
|
|
||||||
**MegaMoE**
|
**MegaMoE**
|
||||||
|
|
||||||
MegaMoE fuses expert dispatch + GEMM into a single kernel for higher throughput
|
MegaMoE fuses expert dispatch + GEMM into a single kernel for higher throughput
|
||||||
|
|||||||
@@ -504,9 +504,27 @@ export const Deployment = ({ config, benchmarks }) => {
|
|||||||
const image = (config.dockerImages && config.dockerImages[sel.hw]) || "lmsysorg/sglang:dev";
|
const image = (config.dockerImages && config.dockerImages[sel.hw]) || "lmsysorg/sglang:dev";
|
||||||
const portFlag = flags.find((x) => x.split(/[\s=]/)[0] === "--port");
|
const portFlag = flags.find((x) => x.split(/[\s=]/)[0] === "--port");
|
||||||
const servePort = portFlag ? portFlag.slice("--port".length).trim() : "{{PORT}}";
|
const servePort = portFlag ? portFlag.slice("--port".length).trim() : "{{PORT}}";
|
||||||
|
const vendorOf = (hwId) => {
|
||||||
|
for (const [vendor, list] of Object.entries(HARDWARE_CATALOG)) {
|
||||||
|
if (list.some((h) => h.id === hwId)) return vendor;
|
||||||
|
}
|
||||||
|
const extra = (config.hardware || []).find((h) => h.id === hwId);
|
||||||
|
return (extra && extra.vendor) || "nvidia";
|
||||||
|
};
|
||||||
|
const gpuAccessLines = vendorOf(sel.hw) === "amd"
|
||||||
|
? [
|
||||||
|
"docker run",
|
||||||
|
" --device=/dev/kfd --device=/dev/dri",
|
||||||
|
" --group-add video",
|
||||||
|
" --cap-add=SYS_PTRACE --security-opt seccomp=unconfined",
|
||||||
|
" --shm-size 32g",
|
||||||
|
]
|
||||||
|
: [
|
||||||
|
"docker run --gpus all",
|
||||||
|
" --shm-size 32g",
|
||||||
|
];
|
||||||
const dockerLines = [
|
const dockerLines = [
|
||||||
"docker run --gpus all",
|
...gpuAccessLines,
|
||||||
" --shm-size 32g",
|
|
||||||
// Multi-node needs host networking so the cross-node rendezvous port
|
// Multi-node needs host networking so the cross-node rendezvous port
|
||||||
// (--dist-init-addr) and NCCL/GLOO traffic are reachable; single-node
|
// (--dist-init-addr) and NCCL/GLOO traffic are reachable; single-node
|
||||||
// just maps the serve port.
|
// just maps the serve port.
|
||||||
|
|||||||
@@ -261,4 +261,25 @@ export const benchmarks = [
|
|||||||
{
|
{
|
||||||
match: { hw: "h100", variant: "pro", quant: "fp4", strategy: "high-throughput", nodes: "multi-2" },
|
match: { hw: "h100", variant: "pro", quant: "fp4", strategy: "high-throughput", nodes: "multi-2" },
|
||||||
},
|
},
|
||||||
|
// ====================================================================
|
||||||
|
// MI300X + FP8 (Flash)
|
||||||
|
{ match: { hw: "mi300x", variant: "flash", quant: "fp8", strategy: "low-latency", nodes: "single" } },
|
||||||
|
{ match: { hw: "mi300x", variant: "flash", quant: "fp8", strategy: "balanced", nodes: "single" } },
|
||||||
|
{ match: { hw: "mi300x", variant: "flash", quant: "fp8", strategy: "high-throughput", nodes: "single" } },
|
||||||
|
// MI355X + FP4 (Flash)
|
||||||
|
{ match: { hw: "mi355x", variant: "flash", quant: "fp4", strategy: "low-latency", nodes: "single" } },
|
||||||
|
{ match: { hw: "mi355x", variant: "flash", quant: "fp4", strategy: "balanced", nodes: "single" } },
|
||||||
|
{ match: { hw: "mi355x", variant: "flash", quant: "fp4", strategy: "high-throughput", nodes: "single" } },
|
||||||
|
// MI355X + FP8 (Flash)
|
||||||
|
{ match: { hw: "mi355x", variant: "flash", quant: "fp8", strategy: "low-latency", nodes: "single" } },
|
||||||
|
{ match: { hw: "mi355x", variant: "flash", quant: "fp8", strategy: "balanced", nodes: "single" } },
|
||||||
|
{ match: { hw: "mi355x", variant: "flash", quant: "fp8", strategy: "high-throughput", nodes: "single" } },
|
||||||
|
// MI355X + FP4 (Pro)
|
||||||
|
{ match: { hw: "mi355x", variant: "pro", quant: "fp4", strategy: "low-latency", nodes: "single" } },
|
||||||
|
{ match: { hw: "mi355x", variant: "pro", quant: "fp4", strategy: "balanced", nodes: "single" } },
|
||||||
|
{ match: { hw: "mi355x", variant: "pro", quant: "fp4", strategy: "high-throughput", nodes: "single" } },
|
||||||
|
// MI355X + FP8 (Pro)
|
||||||
|
{ match: { hw: "mi355x", variant: "pro", quant: "fp8", strategy: "low-latency", nodes: "single" } },
|
||||||
|
{ match: { hw: "mi355x", variant: "pro", quant: "fp8", strategy: "balanced", nodes: "single" } },
|
||||||
|
{ match: { hw: "mi355x", variant: "pro", quant: "fp8", strategy: "high-throughput", nodes: "single" } },
|
||||||
];
|
];
|
||||||
|
|||||||
@@ -7,6 +7,8 @@ export const config = {
|
|||||||
supportedHardware: [
|
supportedHardware: [
|
||||||
"h100", "h200", "b200", "b300", "gb200", "gb300",
|
"h100", "h200", "b200", "b300", "gb200", "gb300",
|
||||||
"rtx6000",
|
"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
|
// 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 FP8 needs the sgl-project repackaging (Hopper can't run FP4-mixed Instruct).
|
||||||
"h200|flash|fp8": "sgl-project/DeepSeek-V4-Flash-FP8",
|
"h200|flash|fp8": "sgl-project/DeepSeek-V4-Flash-FP8",
|
||||||
"h200|pro|fp8": "sgl-project/DeepSeek-V4-Pro-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: {
|
placeholders: {
|
||||||
@@ -147,6 +153,10 @@ sgl-eval run aime25 \\
|
|||||||
b300: "lmsysorg/sglang:latest",
|
b300: "lmsysorg/sglang:latest",
|
||||||
gb200: "lmsysorg/sglang:latest",
|
gb200: "lmsysorg/sglang:latest",
|
||||||
gb300: "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".
|
// Pre-selects the issue template's `model` dropdown on "Submit verified cell".
|
||||||
@@ -1200,5 +1210,719 @@ sgl-eval run aime25 \\
|
|||||||
"--port {{PORT}}",
|
"--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}}",
|
||||||
|
],
|
||||||
|
},
|
||||||
],
|
],
|
||||||
};
|
};
|
||||||
|
|||||||
Reference in New Issue
Block a user