-
Notifications
You must be signed in to change notification settings - Fork 528
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
[Feature Request]: atomicAdd() to support half2 #3573
Comments
For half, we have |
is there any risk concern for |
Its unsafe because it causes the fast HW instruction to be generated, but those instructions don't work if they act on memory that is not cached, e.g. across a PCIe bus. The developer needs to assert that they are willing to take that risk. |
Does ROCm 6.2 support it ? /opt/rocm-6.2.0/lib/llvm/bin/../../../include/hip/amd_detail/amd_hip_fp16.h does not contain the function. |
Do you think it is better to have two types of atomic add functions than a single function in CUDA ? |
Suggestion Description
hi, hip team,
here is cuda version,
looks there's non hip alternative yet, if built with hipcc, it gives:
Operating System
Ubuntu 22.04
GPU
mi300
ROCm Component
6.1.3 + rocblas + rocwmma
The text was updated successfully, but these errors were encountered: