-
Notifications
You must be signed in to change notification settings - Fork 1.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Enable emission of float16/32 casts on x86 #7837
Conversation
@@ -638,14 +644,17 @@ int main(int argc, char **argv) { | |||
{ | |||
Target("x86-32-linux"), | |||
Target("x86-32-linux-sse41"), | |||
Target("x86-64-linux-sse41-avx"), | |||
Target("x86-64-linux-sse41-avx-avx2"), | |||
// Always turn on f16c when using avx. Haswell had avx without f16c, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Actually, Haswell had F16C (together with newly-added AVX2+FMA). Even IvyBridge supported F16C (without AVX2 or FMA). Only SandyBridge was pure AVX.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oops, I had haswell and ivy bridge in the wrong order in my head. I'll fix the comment.
Failures unrelated. |
* Enable emission of float16/32 casts on x86 Fixes halide#7836 Fixes halide#4166 * Fix comment * Don't catch bfloat casts * Fix missing word in comment
Fixes #7836
Fixes #4166
These weren't enabled yet because the last time this code was touched, float16 wasn't a legal type in llvm on x86 (See https://bugs.llvm.org/show_bug.cgi?id=43065)