-
Notifications
You must be signed in to change notification settings - Fork 360
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
New YAKL 2022-02-18 #4804
New YAKL 2022-02-18 #4804
Conversation
… a noinline clause to the FFT init to fix it.
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.
looks good to me.
@mrnorman this one fails:
|
Upgrading to the latest YAKL. Includes performance improvements for CUDA and HIP as well as a bug (memory leak) fix for the pool allocator. The new YAKL also has a fully functioning SYCL backend for Intel GPUs. CUDA has what appears to be a compiler bug in the FFT routines. These have been worked around in the new YAKL. The FFT API changed by no longer requiring a parameter to init(). This has been reflected in pressure.cpp and crm_variance_transport.cpp An issue with the standalone samxx has been fixed, and new machine files have been added. This PR passes ./create_test SMS_Ld10_P6x7.ne4pg2_ne4pg2.F-MMFXX.summit_gnugpu.eam-rrtmgpxx [BFB]
Merged to next. |
There are some IBM compiler build (link) errors on next that appear to be coming from this PR:
https://my.cdash.org/viewTest.php?onlydelta&buildid=2136782 Do these (8) tests work on Summit? |
I'll look at these on summit when I get the chance. |
…nd PGI compilers, but more testing is needed before merging to next again
…oves the need for usin gYAKL_DEVICE_LAMBDA
Upgrading to the latest YAKL. Includes performance improvements for CUDA and HIP as well as a bug (memory leak) fix for the pool allocator. The new YAKL also has a fully functioning SYCL backend for Intel GPUs. CUDA has what appears to be a compiler bug in the FFT routines. These have been worked around in the new YAKL. The FFT API changed by no longer requiring a parameter to init(). This has been reflected in pressure.cpp and crm_variance_transport.cpp An issue with the standalone samxx has been fixed, and new machine files have been added. This PR passes ./create_test SMS_Ld10_P6x7.ne4pg2_ne4pg2.F-MMFXX.summit_gnugpu.eam-rrtmgpxx [BFB]
Merged to master. |
Upgrading to the latest YAKL. Includes performance improvements for CUDA and HIP as well as a bug (memory leak) fix for the pool allocator. The new YAKL also has a fully functioning SYCL backend for Intel GPUs.
CUDA has what appears to be a compiler bug in the FFT routines. These have been worked around in the new YAKL. The FFT API changed by no longer requiring a parameter to init(). This has been reflected in pressure.cpp and crm_variance_transport.cpp
An issue with the standalone samxx has been fixed, and new machine files have been added.
This PR passes
./create_test SMS_Ld10_P6x7.ne4pg2_ne4pg2.F-MMFXX.summit_gnugpu.eam-rrtmgpxx
[BFB]