Upstream Contributions
| Status | Contribution | Upstream |
|---|---|---|
| MERGED | Spectrum-based diffusion acceleration for Flux, WanVideo, HunyuanVideo, and StableDiffusion3 paths. | SGLang PR #31491 → |
| MERGED | Make explicit zero-level host and Python tracing settings work in the SGLang JAX profiler. | SGLang JAX PR #1511 → |
| MERGED | Fused sparse-MLA prefill for DeepSeek Sparse Attention on SGLang JAX. | SGLang JAX PR #1508 → |
| IN REVIEW | Packed-ragged sparse-MLA prefill with batching, radix, and chunked prefill. | SGLang JAX PR #1514 → |
| MERGED | Reshard block-quant scales for the JAX 0.10 explicit-mesh shard-map change. | SGLang JAX PR #1513 → |
| IN REVIEW | Fix sparse prefill so full skip-offset layers stay sparse. | SGLang JAX PR #1512 → |
| IN REVIEW | Correct DSA indexer-key cache budgeting and memory reporting for sparse MLA. | SGLang JAX PR #1532 → |
| IN REVIEW | Serve Kimi-K3 on AMD CDNA3 with NullStack MLA multi-query verification, MXFP4 MoE, and chunked prefix KV. | SGLang PR #34007 → |
| IN REVIEW | NullStack MLA, an MI300X/CDNA3 MLA attention backend for bf16-query / fp8-KV decode. | SGLang PR #32767 → |
| IN REVIEW | LiteAttention ROCm, an opt-in AMD MI300X attention backend for diffusion. | SGLang PR #28049 → |
Contribution Deep Dives
Spectrum-based diffusion acceleration for Flux, WanVideo, HunyuanVideo, and StableDiffusion3 paths.
MERGEDThis contribution adds Spectrum-based adaptive step-skipping and feature forecasting to SGLang's diffusion runtime. It reduces denoising latency while preserving output quality.
Make explicit zero-level host and Python tracing settings work in the SGLang JAX profiler.
MERGEDThis contribution makes explicit host_tracer_level=0 and python_tracer_level=0 settings work as intended in the SGLang JAX profiler.
Fused sparse-MLA prefill for DeepSeek Sparse Attention on SGLang JAX.
MERGEDThis opt-in sparse-prefill path makes DeepSeek Sparse Attention useful during prefill, not only decode.
Packed-ragged sparse-MLA prefill with batching, radix, and chunked prefill.
IN REVIEWThis builds on the sparse-MLA work with batching, radix, and chunked prefill support.
Reshard block-quant scales for the JAX 0.10 explicit-mesh shard-map change.
MERGEDThis fixes block-quant scale resharding for JAX 0.10's explicit-mesh shard-map behavior.
Fix sparse prefill so full skip-offset layers stay sparse.
IN REVIEWThis fixes the sparse-prefill issue we found in Google's code, so full skip-offset layers remain sparse.
Correct DSA indexer-key cache budgeting and memory reporting for sparse MLA.
IN REVIEWThis corrects sparse-MLA KV-pool sizing and memory reporting so DSA indexer-key buffers are budgeted alongside the latent cache.
Serve Kimi-K3 on AMD CDNA3 with NullStack MLA multi-query verification, MXFP4 MoE, and chunked prefix KV.
IN REVIEWThis makes Kimi-K3 serveable on MI300X and MI325X with NullStack MLA multi-query verification, gfx942 MXFP4 MoE support, and chunked prefix KV for long contexts.
NullStack MLA, an MI300X/CDNA3 MLA attention backend for bf16-query / fp8-KV decode.
IN REVIEWnullstack_mla routes the MI300X shape we run to a device-driven paged A16W8 kernel, while all other shapes fall back to aiter.
LiteAttention ROCm, an opt-in AMD MI300X attention backend for diffusion.
IN REVIEWThis opt-in backend brings NullStack's bf16 forward-attention kernel to SGLang's diffusion runtime on AMD MI300X.
NullStack Research
@nullstack_in
Low-level algorithms and hardware-aware systems power private, high-throughput AI inference for coding agents. Stay tuned as we release further benchmarks.