-
Notifications
You must be signed in to change notification settings - Fork 55
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
[Issue]: __hadd(int x, int y) and __rhadd(int x, int y) is wrong #95
Labels
Comments
Hmm strange, thanks for reporting. |
|
I found a few bugs. Raised a patch internally to fix it. |
I think we can close this: 5122b8c |
@darren-amd |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Problem Description
The calculation result of __hadd(int x, int y) and __rhadd(int x, int y) is inconsistent with NV.
For example, to calculate __hadd(123, -789),the result on Rocm is 1, but the result on NV is -666.
The following are the descriptions of the __hadd function by Rocm and NV respectively.
Rocm
NV
The functions described are the same, but the calculation results are very different
Operating System
Centos7
CPU
AMD Ryzen 9 5900HX with Radeon Graphics
GPU
AMD Instinct MI100
ROCm Version
ROCm 5.7.0
ROCm Component
clr
Steps to Reproduce
kernel function as follows
(Optional for Linux users) Output of /opt/rocm/bin/rocminfo --support
No response
Additional Information
Used cuda 11.8 to compare
The text was updated successfully, but these errors were encountered: