-
Notifications
You must be signed in to change notification settings - Fork 21.4k
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
Check SM version before calling flash attention with BFloat16 #86600
Conversation
The flash attention code path requires sm80 or newer to run on BFloat16, so any OpInfo tests running with BFloat16 would fail with the error: ``` RuntimeError: Expected q_dtype == at::kHalf || (is_sm8x && q_dtype == at::kBFloat16) to be true, but got false. ``` [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/86600
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 Failures, 6 PendingAs of commit eda71c5: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
The flash attention code path requires sm80 or newer to run on BFloat16, so any OpInfo tests running with BFloat16 would fail with the error: ``` RuntimeError: Expected q_dtype == at::kHalf || (is_sm8x && q_dtype == at::kBFloat16) to be true, but got false. ``` ghstack-source-id: 42122448b1213f77ebb3b4503c6f2c594e859ccc Pull Request resolved: #86600
…t16" The flash attention code path requires sm80 or newer to run on BFloat16, so any OpInfo tests running with BFloat16 would fail with the error: ``` RuntimeError: Expected q_dtype == at::kHalf || (is_sm8x && q_dtype == at::kBFloat16) to be true, but got false. ``` [ghstack-poisoned]
The flash attention code path requires sm80 or newer to run on BFloat16, so any OpInfo tests running with BFloat16 would fail with the error: ``` RuntimeError: Expected q_dtype == at::kHalf || (is_sm8x && q_dtype == at::kBFloat16) to be true, but got false. ``` ghstack-source-id: 95e4d520b89b76329d3630cdc4d63a2cb669acc9 Pull Request resolved: #86600
The flash attention code path requires sm80 or newer to run on BFloat16, so any OpInfo tests running with BFloat16 would fail with the error: ``` RuntimeError: Expected q_dtype == at::kHalf || (is_sm8x && q_dtype == at::kBFloat16) to be true, but got false. ``` ghstack-source-id: 95e4d520b89b76329d3630cdc4d63a2cb669acc9 Pull Request resolved: pytorch#86600
…t16" The flash attention code path requires sm80 or newer to run on BFloat16, so any OpInfo tests running with BFloat16 would fail with the error: ``` RuntimeError: Expected q_dtype == at::kHalf || (is_sm8x && q_dtype == at::kBFloat16) to be true, but got false. ``` [ghstack-poisoned]
The flash attention code path requires sm80 or newer to run on BFloat16, so any OpInfo tests running with BFloat16 would fail with the error: ``` RuntimeError: Expected q_dtype == at::kHalf || (is_sm8x && q_dtype == at::kBFloat16) to be true, but got false. ``` ghstack-source-id: b86ec8a34a8feebbfc7fcaf74ca247e67babdf0c Pull Request resolved: #86600
…t16" The flash attention code path requires sm80 or newer to run on BFloat16, so any OpInfo tests running with BFloat16 would fail with the error: ``` RuntimeError: Expected q_dtype == at::kHalf || (is_sm8x && q_dtype == at::kBFloat16) to be true, but got false. ``` [ghstack-poisoned]
The flash attention code path requires sm80 or newer to run on BFloat16, so any OpInfo tests running with BFloat16 would fail with the error: ``` RuntimeError: Expected q_dtype == at::kHalf || (is_sm8x && q_dtype == at::kBFloat16) to be true, but got false. ``` ghstack-source-id: cb731ee9f0c81823f4f08ec395547a6142449904 Pull Request resolved: #86600
@pytorchbot merge |
@pytorchbot merge -f "Dynamo failure is unrelated and present on master" |
Merge startedYour change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
…h#86600) The flash attention code path requires sm80 or newer to run on BFloat16, so any OpInfo tests running with BFloat16 would fail with the error: ``` RuntimeError: Expected q_dtype == at::kHalf || (is_sm8x && q_dtype == at::kBFloat16) to be true, but got false. ``` Pull Request resolved: pytorch#86600 Approved by: https://github.com/ngimel
…h#86600) The flash attention code path requires sm80 or newer to run on BFloat16, so any OpInfo tests running with BFloat16 would fail with the error: ``` RuntimeError: Expected q_dtype == at::kHalf || (is_sm8x && q_dtype == at::kBFloat16) to be true, but got false. ``` Pull Request resolved: pytorch#86600 Approved by: https://github.com/ngimel
Stack from ghstack (oldest at bottom):
The flash attention code path requires sm80 or newer to run on
BFloat16, so any OpInfo tests running with BFloat16 would fail with
the error: