GitLab

Co-authored-by: mgoin <michael@neuralmagic.com>
Co-authored-by: Dipika Sikka <dipikasikka1@gmail.com>

Co-authored-by: Michael Goin <michael@neuralmagic.com>
Co-authored-by: youkaichao <youkaichao@gmail.com>
Co-authored-by: zifeitong <zifei.tong@parasail.io>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-neuralmagic@users.noreply.github.com>

Tune Qwen2-57B-A14B configs based on #4921

Throughput Performance
command: python benchmarks/benchmark_throughput.py --model=Qwen/Qwen2-57B-A14B-Instruct --input-len 1000 --output-len 50 -tp 2

A100 GPU

benchmark	no config	w/ PR
tp=2	10.53 requests/s, 11058.17 tokens/s	12.47 requests/s, 13088.57 tokens/s
tp=4	17.77 requests/s, 18662.95 tokens/s	20.20 requests/s, 21212.32 tokens/s

Co-authored-by: Robert Shaw <114415538+robertgshaw2-neuralmagic@users.noreply.github.com>

Co-authored-by: Michael Goin <michael@neuralmagic.com>

Bug description:
With torch 2.4.0.dev20240603+cu121,
cutlass_fp8_supported outputs False, and the (capability, version) before the comparison is (90, 11111111112)

This PR fixes the support check for FP8 CUTLASS ( cutlass_fp8_supported) which was introduced in https://github.com/vllm-project/vllm/pull/5183.

Co-authored-by: Varun Sundar Rabindranath <varunsundar08@gmail.com>
Co-authored-by: Varun Sundar Rabindranath <varun@neuralmagic.com>

Switching from torch._scaled_mm to vLLM's cutlass fp8 kernels when supported as we are seeing 5-15% improvement in e2e performance on neuralmagic/Meta-Llama-3-8B-Instruct-FP8

see https://docs.google.com/spreadsheets/d/1GiAnmzyGHgZ6zL_LDSTm35Bdrt4A8AaFEurDlISYYA4/ for some quick e2e benchmarks and #5144 for comparisons across different GEMM sizes.

This PR enables the fused topk_softmax kernel used in moe layer for HIP

This PR adds Triton kernel configs for the MoE kernel for MI300X

Co-authored-by: Elisei Smirnov <el.smirnov@innopolis.university>

Co-authored-by: Varun Sundar Rabindranath <varunsundar08@gmail.com>
Co-authored-by: Varun Sundar Rabindranath <varun@neuralmagic.com>

The 2nd PR for #4532.

This PR supports loading FP8 kv-cache scaling factors from a FP8 checkpoint (with .kv_scale parameter).

Currently we need to call rotary embedding kernel for each LoRA, which makes it hard to serve multiple long context length LoRA. Add batched rotary embedding kernel and pipe it through.

It replaces the rotary embedding layer to the one that is aware of multiple cos-sin-cache per scaling factors.

Follow up of https://github.com/vllm-project/vllm/pull/3095/files

Co-authored-by: Robert Shaw <rshaw@neuralmagic.com>