Fetching from the wire…
01
02
03
04
05
06
07
08
Infra2026-09-01 · source-backed
Build b10730 rewrites the qwen4exp indexer head reduction, which ran a transpose then a sum_rows over ne[1] with ne0 = 4, one GPU block per row for a four-element reduction, while the transpose copied the whole block-by-token surface twice on the way in (GitHub). Heads are adjacent on ne[1], so each is now a strided view and the sum is a short chain of adds. Prompt processing goes from 2,170 to 2,366 t/s at 55k context with generation unaffected, and the gain grows with context and ubatch size.
Each link below shares sources, entities, or timing with this story.
Build b10687 defaults the xmem F16xF32 GEMM on for Adreno X2E after measuring kernel_mul_mm_f16_f32_l4_lm running gpt-oss-20b attention projections at about a quarter the speed of the tuned dense q4_0 GEMM. Worth 25% prefill. The route already existed but was opt-in, so nobody...
Build b10691 fixes a null-pipeline crash for F16 src1 mul_mat and mul_mat_id by declining unsupported shapes in supports_op and aborting on a nil pipeline in encoder_set_pipeline, sharing the mm dispatch between both paths so support checks and execution can't disagree. Build...
440 stars, created August 11, covering 12 sections from model APIs through structured output, RAG, evals, agent loops, LoRA versus fine-tuning, security, LLMOps and serving, plus three case studies and a capstone (GitHub). The stance is that you write the agent loop, RAG and e...
Warp released its client codebase under AGPL-3.0, surged to 56,000 GitHub stars and #2 on GitHub Trending. But the real story isn't the open-sourcing. It's the repositioning. Warp isn't calling itself a terminal anymore. It's an "agentic development environment." The product n...
Eight thousand stars in a single day. That's what happened when Warp open-sourced its Rust-based, GPU-accelerated terminal on April 28. The repo shot to 47.9K total stars, making it the highest-velocity project on GitHub this week by a wide margin. But the interesting part isn...
PR #19378 landed in llama.cpp this week, and I think most people are underselling what it means. Backend-agnostic tensor parallelism via --split-mode tensor makes multi-GPU inference work across AMD, Intel, and Apple Silicon. Not just CUDA. Everything. For context: llama.cpp h...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.