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.

Sign up for GitHub

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

Jump to bottom

Use the RoCM/HIP device to accelerate certain DPLASMA kernels #57

Merged
merged 50 commits into from
Aug 9, 2024
Merged

Use the RoCM/HIP device to accelerate certain DPLASMA kernels #57

merged 50 commits into from
Aug 9, 2024

Conversation

abouteiller
Copy link
Contributor

This PR adds RoCM enabled kernels to the GEMM, PORTF and memory-aware GEMM operations.

@bosilca
Copy link
Contributor

bosilca commented Mar 31, 2023

As discussed on 03/31/23 we need to rebase and check the result. This will be tested next week on Frontier, we need it to be ready.

@abouteiller abouteiller force-pushed the feature/hip branch 2 times, most recently from 5deec12 to fae375c Compare June 5, 2023 19:13
@QingleiCao QingleiCao mentioned this pull request Jun 6, 2023
Merged
@abouteiller
configure: Add the --with-hip option
2b7c65f 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
hip: Configury
d9d8a18 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
hip: kernel typedefs
828011f 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
hip: update for hip-enabled parsec
26635ca 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
hip: detect hipblas and rocsolvers
ffd7bdd 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>

Conflicts:
	src/CMakeLists.txt
@abouteiller
hip: precision generator rules
7db5636 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
hip: cleanup unused dyld hipblas functions
d2044bc 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
hip: Update lapack stagein
d283bde 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
Update for feature/common_gpu parsec branch changes
a7de750 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
Some conflicting updates between hip and common_gpu need more resolution
92e7fe8 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
hip: stream info registration
c02ebcd
@abouteiller
hip: potrf on AMD
0d2367a 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
hip:po: Some errors introduced when merging
1fc2d14 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
Add HIP to the lookahead gpu gemm
0cbeece 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
Add HIP to zgemm_summa
b27190b 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
hip: rework of PO and workspaces
8d88461 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
hip: remove unecessary hiblas init calls
3afda75 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
hip:po:errors in ldam asserts
9b92289 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
hip:po: some of the changes had broken cusolver
d41c3b6 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
fix printlogcuda/hip
0db93e7 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
Auto-generate hip stage-in/out functions
ecc4e77 
Use proper error checks instead of asserts
@abouteiller
hip:zgemm_gpu: don't use hipComplex
b546f67
@abouteiller
Return the proper PARSEC_HOOK_RETURN_ERROR in GPU error cases
a6c6053 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
Update for the new device mask for incarnations
ffe59a8 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller
So far only NN gemm can run with HIP
ab2668a 
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
@abouteiller abouteiller mentioned this pull request Jan 31, 2024
Merged
@abouteiller
Copy link
Contributor Author

This is in ready to merge state beside the 'squash to less commits'.

@abouteiller abouteiller enabled auto-merge (squash) May 21, 2024 03:57
@abouteiller
Copy link
Contributor Author

I decided to go for a squash merge, this is ready for final review @bosilca @devreal

Error in ctest is #115, preexisting and unrelated to hip.

bosilca
bosilca reviewed Jun 14, 2024
View reviewed changes
Copy link
Contributor

@bosilca bosilca left a comment

Choose a reason for hiding this comment

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

Why only certain kernels are HIP-enabled instead of having all CUDA-enabled kernel also be HIP-enabled ?

src/dplasmaaux.h Show resolved Hide resolved
src/zgemm_TN.jdf Show resolved Hide resolved
devreal
devreal reviewed Jul 11, 2024
View reviewed changes
share/help-dplasma.txt Show resolved Hide resolved
src/dplasmaaux_hip.h Show resolved Hide resolved
@abouteiller
Cublas: sometimes we would include cublas.h before including cusolver.h
a985fa6 
use cublasv2 by default (req. for cusolver anyway) except if cublas(v1)
is included before the dplasmajdf/aux.h include
This reverts the logic for cublas.h determination, and is a first steps
toward deprecating cublas(v1) usage entirely
@abouteiller abouteiller force-pushed the feature/hip branch 5 times, most recently from e5ed5f0 to a985fa6 Compare August 9, 2024 20:07
@abouteiller abouteiller merged commit edf5be3 into ICLDisco:master Aug 9, 2024
2 of 3 checks passed
@abouteiller abouteiller deleted the feature/hip branch October 24, 2024 14:15
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@bosilca bosilca bosilca left review comments

@devreal devreal devreal approved these changes

Assignees

@abouteiller abouteiller

Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@abouteiller @bosilca @devreal