From 1b5920ca10aa2b9bd76d90bd9d90b5dd8eb5a7c5 Mon Sep 17 00:00:00 2001 From: tannevaled Date: Wed, 7 Oct 2026 17:39:23 +0200 Subject: [PATCH] docs(README): mark the Dot vs threaded-BLAS figure as pre-v0.6.0 Co-Authored-By: Claude Opus 5.5 --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index fb563a3..92792ae 100644 --- a/README.md +++ b/README.md @@ -88,7 +88,7 @@ is wrong on subnormals one core. - **Slower:** `MatMul` against 16-thread OpenBLAS at 512² and up (0.42× at 512², 0.51× at 1024², measured in one run with OpenBLAS; at 256² ours is - 1.44× faster); `Dot`/mat·vec against NumPy's threaded BLAS (0.16×); `Exp` + 1.44× faster); `Dot`/mat·vec against NumPy's threaded BLAS (0.16×, measured before v0.6.0's helper pool, which took the 8-core dot from 625 to 58 µs on POWER9; not yet re-measured on the Zen 3); `Exp` below 256 Ki elements (0.82×); the *allocating* forms of elementwise ops up to 256 Ki elements, `Concatenate`/`Stack` and slice copies outside a `Workspace` (0.2–0.5×: Go's allocator and garbage collector, which