GitLab

 [Speculative Decoding] Support draft model on different tensor-parallel size than target model (#5414)

This PR adds basic support for OpenTelemetry distributed tracing.
It includes changes to enable tracing functionality and improve monitoring capabilities.

I've also added a markdown with print-screens to guide users how to use this feature. You can find it here

[CI] Improve the readability of performance benchmarking results and prepare for upcoming performance dashboard (#5571)

Co-authored-by: Jiang Li <jiang1.li@intel.com>
Co-authored-by: Abhilash Majumder <abhilash.majumder@intel.com>
Co-authored-by: Abhilash Majumder <30946547+abhilash1910@users.noreply.github.com>

[CI/Build][Misc] Add CI that benchmarks vllm performance on those PRs with `perf-benchmarks` label (#5073)
Co-authored-by: simon-mo <simon.mo@hey.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>

Signed-off-by: Wang, Yi A <yi.a.wang@intel.com>
Co-authored-by: Roger Wang <ywang@roblox.com>

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

Co-authored-by: Zhuohan Li <zhuohan123@gmail.com>

Signed-off-by: pandyamarut <pandyamarut@gmail.com>

Co-authored-by: Roger Wang <ywang@roblox.com>

The 2nd PR for #4532.

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

Co-authored-by: Cade Daniel <edacih@gmail.com>
Co-authored-by: Cade Daniel <cade@anyscale.com>

[Core][Hash][Automatic Prefix caching] Accelerating the hashing function by avoiding deep copies (#4696)

This PR updates the tuning script for the fused_moe kernel to support FP8 and also adds configurations for TP4. Note that for the configuration I removed num_warps and num_stages for small batch sizes since that improved performance and brought the benchmarks on par with the numbers before in that regime to make sure this is a strict improvement over the status quo.

All the numbers below are for mistralai/Mixtral-8x7B-Instruct-v0.1, 1000 input and 50 output tokens.

Before this PR (with static activation scaling):

qps = 1: 9.8 ms ITL, 0.49s e2e latency
qps = 2: 9.7 ms ITL, 0.49s e2e latency 
qps = 4: 10.1 ms ITL, 0.52s e2e latency
qps = 6: 11.9 ms ITL, 0.59s e2e latency
qps = 8: 14.0 ms ITL, 0.70s e2e latency
qps = 10: 15.7 ms ITL, 0.79s e2e latency

After this PR (with static activation scaling):

qps = 1: 9.8 ms ITL, 0.49s e2e latency
qps = 2: 9.7 ms ITL, 0.49s e2e latency
qps = 4: 10.2 ms ITL, 0.53s e2e latency
qps = 6: 11.9 ms ITL, 0.59s e2e latency
qps = 8: 11.9 ms ITL, 0.59s e2e latency
qps = 10: 12.1 ms ITL, 0.61s e2e latency

Co-authored-by: Lei Wen <wenlei03@qiyi.com>
Co-authored-by: Sage Moore <sagemoore@utexas.edu>