From 560ab75b0f9bb7f2ca18271c759687c8c64db7fb Mon Sep 17 00:00:00 2001 From: Oscar Smith Date: Fri, 30 May 2025 12:53:39 -0400 Subject: [PATCH 1/2] remove outdated performance warning from readme As of https://github.com/JuliaMath/BFloat16s.jl/pull/51 this code should be generating fast code on supported hardware. --- README.md | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/README.md b/README.md index 48011eb..3a3f98f 100644 --- a/README.md +++ b/README.md @@ -16,9 +16,7 @@ for more information. This package is suitable to evaluate whether using BFloat16 would cause precision problems for any particular algorithm, even without access to supporting -hardware. Note that this package is designed for functionality, not performance, -so this package should be used for precision experiments only, not performance -experiments. +hardware. ## Usage From 670aa9e2294df78469b90ebe8997c8ebc07985eb Mon Sep 17 00:00:00 2001 From: Oscar Smith Date: Fri, 30 May 2025 23:48:24 -0400 Subject: [PATCH 2/2] Update README.md Co-authored-by: Mark Kittisopikul --- README.md | 1 + 1 file changed, 1 insertion(+) diff --git a/README.md b/README.md index 3a3f98f..d92bfe0 100644 --- a/README.md +++ b/README.md @@ -18,6 +18,7 @@ This package is suitable to evaluate whether using BFloat16 would cause precision problems for any particular algorithm, even without access to supporting hardware. +Native hardware support, implemented in this package, began with Julia 1.11 via `Core.BFloat16` with support for Intel processors (x86_64, i686) via LLVM 16. Support for 64-bit ARM processors (aarch64) is is available with Julia 1.12 which uses LLVM 19. ## Usage This package exports the `BFloat16` data type. This datatype behaves