Skip to content
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

Cast fragment #311

Merged
merged 9 commits into from
Nov 27, 2024
Merged

Cast fragment #311

merged 9 commits into from
Nov 27, 2024

Conversation

nathanielsimard
Copy link
Member

No description provided.

@nathanielsimard nathanielsimard marked this pull request as draft November 26, 2024 20:55
@nathanielsimard nathanielsimard marked this pull request as ready for review November 27, 2024 18:13
pub fn test_simple_tf32<R: Runtime>(
client: ComputeClient<R::Server, R::Channel>,
cube_dimensions: CubeDim,
) {
if !client.properties().feature_enabled(Feature::Cmma {
a: Elem::Float(FloatKind::TF32),
b: Elem::Float(FloatKind::TF32),
a: Elem::Float(FloatKind::F16),
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Isn't this a test specifically for TF32?

@@ -59,8 +59,7 @@ impl WmmaCompiler<CudaDialect<Self>> for CudaWmmaCompiler {
fn supported_wmma_combinations(arch: &Self::Architecture) -> SupportedWmmaCombinations {
let mut result: SupportedWmmaCombinations = vec![];
if arch.version >= WMMA_MINIMUM_VERSION {
// m n k
let tdims = vec![(16, 16, 16), (32, 16, 8), (8, 16, 32)];
let tdims = vec![(16, 16, 16), (32, 16, 8), (8, 16, 32), (32, 8, 16)];
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think it should be [(16, 16, 16), (32, 8, 16), (8, 32, 16)]

@nathanielsimard nathanielsimard merged commit 6de4938 into main Nov 27, 2024
5 checks passed
@nathanielsimard nathanielsimard deleted the matmul/cast-fragment branch November 27, 2024 21:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants