Shift nightly builds from gfx103X-dgpu to gfx103X-all by harkgill-amd · Pull Request #3763 · ROCm/TheRock

harkgill-amd · 2026-03-04T22:32:47Z

Motivation

Addresses #3404

Our current gfx103X-dgpu family misses out on gfx1033 , gfx1035 and gfx1036. Shifting the nightly builds to gfx103X-all which will cover the aforementioned architectures.

Test Plan

Build TheRock with -DTHEROCK_AMDGPU_FAMILIES=gfx103X-all on Linux and Windows

Test Result

Windows = Build Successful
Linux = Build Successful

Submission Checklist

Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

LuXuxue · 2026-03-27T05:57:04Z

Is there any progress recently? ROCm/rocm-libraries#5141 has been approved but has not been merged.

harkgill-amd · 2026-03-31T19:01:15Z

We'd need ROCm/rocm-libraries#5141 to be merged before we can continue forward with this change. The former is being blocked by unrelated CI failures which the CK team is working hard to resolve. Hopefully we can get it in by the end of this week.

## Motivation Resolving PyTorch build failures when enabling builds for gfx103X-all family in TheRock. ROCm/TheRock#3763. `gfx1033` is the only failing architecture in the family and the failures point to missing support in CK. ## Technical Details PyTorch build fails with repeated error message ``` /__w/TheRock/TheRock/external-builds/pytorch/pytorch/aten/src/ATen/../../../third_party/composable_kernel/include/ck/utility/amd_buffer_addressing_builtins.hpp:33:48: error: use of undeclared identifier 'CK_BUFFER_RESOURCE_3RD_DWORD' 33 | wave_buffer_resource.config(Number<3>{}) = CK_BUFFER_RESOURCE_3RD_DWORD; | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~ ``` `gfx1033` is missing from the `__gfx103__` group which results in `CK_BUFFER_RESOURCE_3RD_DWORD` never being defined for it. Adding in `gfx1033` to the missing files which should be the minimum fix to allow torch builds to pass. ## Test Plan Compile sample test file and target gfx1033 ``` ... #ifdef __HIP_DEVICE_COMPILE__ static_assert(CK_BUFFER_RESOURCE_3RD_DWORD == 0x31014000, "wrong device value"); #else static_assert(CK_BUFFER_RESOURCE_3RD_DWORD == -1, "wrong host value"); #endif ``` ## Test Result Prior to the applying patch, compilation fails with `error: use of undeclared identifier 'CK_BUFFER_RESOURCE_3RD_DWORD'` After applying patch, test file compiles successfully. ## Submission Checklist - [X] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests. --------- Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>

Add missing gfx1033 to gfx103 group definition in ck ## Motivation Resolving PyTorch build failures when enabling builds for gfx103X-all family in TheRock. ROCm/TheRock#3763. `gfx1033` is the only failing architecture in the family and the failures point to missing support in CK. ## Technical Details PyTorch build fails with repeated error message ``` /__w/TheRock/TheRock/external-builds/pytorch/pytorch/aten/src/ATen/../../../third_party/composable_kernel/include/ck/utility/amd_buffer_addressing_builtins.hpp:33:48: error: use of undeclared identifier 'CK_BUFFER_RESOURCE_3RD_DWORD' 33 | wave_buffer_resource.config(Number<3>{}) = CK_BUFFER_RESOURCE_3RD_DWORD; | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~ ``` `gfx1033` is missing from the `__gfx103__` group which results in `CK_BUFFER_RESOURCE_3RD_DWORD` never being defined for it. Adding in `gfx1033` to the missing files which should be the minimum fix to allow torch builds to pass. ## Test Plan Compile sample test file and target gfx1033 ``` ... #ifdef __HIP_DEVICE_COMPILE__ static_assert(CK_BUFFER_RESOURCE_3RD_DWORD == 0x31014000, "wrong device value"); #else static_assert(CK_BUFFER_RESOURCE_3RD_DWORD == -1, "wrong host value"); #endif ``` ## Test Result Prior to the applying patch, compilation fails with `error: use of undeclared identifier 'CK_BUFFER_RESOURCE_3RD_DWORD'` After applying patch, test file compiles successfully. ## Submission Checklist - [X] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

CarlGao4 · 2026-04-04T12:47:06Z

Now it is merged!

geomin12

lgtm but let's wait on CI checks

LuXuxue · 2026-04-09T15:50:03Z

Now it is merged!

Maybe we have to waiting for a rocm-libraries bump in this repositories before this pr, or will get endless error

harkgill-amd · 2026-04-09T15:54:55Z

Maybe we have to waiting for a rocm-libraries bump in this repositories before this pr, or will get endless error

The below PRs got the changes into the release/2.X> branches which TheRock uses to build torch from.

[release/2.8] Update composable_kernel submodule with gfx1033 support pytorch#3137
[release/2.9] Update composable_kernel submodule with gfx1033 support pytorch#3138
[release/2.10] Update composable_kernel submodule with gfx1033 support pytorch#3139

Just waiting on ROCm/pytorch#3144 for release/2.11 as that was recently enabled as well. Once this is in, we shouldn't see any of the gfx1033/CK failures.

## Motivation Resolving PyTorch build failures when enabling builds for gfx103X-all family in TheRock. ROCm/TheRock#3763. `gfx1033` is the only failing architecture in the family and the failures point to missing support in CK. ## Technical Details PyTorch build fails with repeated error message ``` /__w/TheRock/TheRock/external-builds/pytorch/pytorch/aten/src/ATen/../../../third_party/composable_kernel/include/ck/utility/amd_buffer_addressing_builtins.hpp:33:48: error: use of undeclared identifier 'CK_BUFFER_RESOURCE_3RD_DWORD' 33 | wave_buffer_resource.config(Number<3>{}) = CK_BUFFER_RESOURCE_3RD_DWORD; | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~ ``` `gfx1033` is missing from the `__gfx103__` group which results in `CK_BUFFER_RESOURCE_3RD_DWORD` never being defined for it. Adding in `gfx1033` to the missing files which should be the minimum fix to allow torch builds to pass. ## Test Plan Compile sample test file and target gfx1033 ``` ... #ifdef __HIP_DEVICE_COMPILE__ static_assert(CK_BUFFER_RESOURCE_3RD_DWORD == 0x31014000, "wrong device value"); #else static_assert(CK_BUFFER_RESOURCE_3RD_DWORD == -1, "wrong host value"); #endif ``` ## Test Result Prior to the applying patch, compilation fails with `error: use of undeclared identifier 'CK_BUFFER_RESOURCE_3RD_DWORD'` After applying patch, test file compiles successfully. ## Submission Checklist - [X] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests. --------- Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>

## Motivation Update `ROADMAP.md` to reflect recently added support. ## Technical Details `gfx103X-all` builds passing for Linux/Windows: #3763 (Pytorch failing until ROCm/rocm-libraries#5141 lands) `gfx900` builds passing: #3564 `gfx90c` builds awaiting ROCm/rocm-libraries#5282 to go green ## Test Plan `gfx90c` builds to be tested (#3818) ## Test Result N/A ## Submission Checklist - [x] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

## Motivation Resolving PyTorch build failures when enabling builds for gfx103X-all family in TheRock. ROCm/TheRock#3763. `gfx1033` is the only failing architecture in the family and the failures point to missing support in CK. ## Technical Details PyTorch build fails with repeated error message ``` /__w/TheRock/TheRock/external-builds/pytorch/pytorch/aten/src/ATen/../../../third_party/composable_kernel/include/ck/utility/amd_buffer_addressing_builtins.hpp:33:48: error: use of undeclared identifier 'CK_BUFFER_RESOURCE_3RD_DWORD' 33 | wave_buffer_resource.config(Number<3>{}) = CK_BUFFER_RESOURCE_3RD_DWORD; | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~ ``` `gfx1033` is missing from the `__gfx103__` group which results in `CK_BUFFER_RESOURCE_3RD_DWORD` never being defined for it. Adding in `gfx1033` to the missing files which should be the minimum fix to allow torch builds to pass. ## Test Plan Compile sample test file and target gfx1033 ``` ... #ifdef __HIP_DEVICE_COMPILE__ static_assert(CK_BUFFER_RESOURCE_3RD_DWORD == 0x31014000, "wrong device value"); #else static_assert(CK_BUFFER_RESOURCE_3RD_DWORD == -1, "wrong host value"); #endif ``` ## Test Result Prior to the applying patch, compilation fails with `error: use of undeclared identifier 'CK_BUFFER_RESOURCE_3RD_DWORD'` After applying patch, test file compiles successfully. ## Submission Checklist - [X] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests. --------- Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>

geomin12

lgtm, builds pass for both linux + windows

Shift nightly builds from gfx103X-dgpu to gfx103X-all

9d979a2

github-project-automation Bot added this to TheRock Triage Mar 4, 2026

github-project-automation Bot moved this to TODO in TheRock Triage Mar 4, 2026

harkgill-amd added the gfx103X label Mar 4, 2026

harkgill-amd mentioned this pull request Mar 5, 2026

Add missing gfx1033 to gfx103 group definition in ck ROCm/rocm-libraries#5141

Merged

1 task

lucbruni-amd mentioned this pull request Mar 12, 2026

ROADMAP.md: gfx103X builds passing; add gfx900/gfx90c rows #3933

Merged

1 task

Merge branch 'main' into users/harkgill/enable-gfx103X-all

a486cc2

harkgill-amd marked this pull request as ready for review April 6, 2026 14:05

harkgill-amd requested a review from geomin12 April 6, 2026 14:05

Merge branch 'main' into users/harkgill/enable-gfx103X-all

ee51c8b

geomin12 reviewed Apr 7, 2026

View reviewed changes

Merge branch 'main' into users/harkgill/enable-gfx103X-all

2bcd8ee

Merge branch 'main' into users/harkgill/enable-gfx103X-all

cbc8adf

Merge branch 'main' into users/harkgill/enable-gfx103X-all

d0ab1d9

geomin12 approved these changes Apr 21, 2026

View reviewed changes

harkgill-amd merged commit 2722ebf into main Apr 21, 2026
130 of 138 checks passed

harkgill-amd deleted the users/harkgill/enable-gfx103X-all branch April 21, 2026 15:22

github-project-automation Bot moved this from TODO to Done in TheRock Triage Apr 21, 2026

harkgill-amd mentioned this pull request Apr 21, 2026

[Feature]: Change nightly build from gfx103x-dgpu to gfx103x-all #3404

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Shift nightly builds from gfx103X-dgpu to gfx103X-all#3763

Shift nightly builds from gfx103X-dgpu to gfx103X-all#3763
harkgill-amd merged 6 commits intomainfrom
users/harkgill/enable-gfx103X-all

harkgill-amd commented Mar 4, 2026 •

edited

Loading

Uh oh!

LuXuxue commented Mar 27, 2026

Uh oh!

harkgill-amd commented Mar 31, 2026

Uh oh!

CarlGao4 commented Apr 4, 2026

Uh oh!

geomin12 left a comment

Uh oh!

LuXuxue commented Apr 9, 2026

Uh oh!

harkgill-amd commented Apr 9, 2026

Uh oh!

geomin12 left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

harkgill-amd commented Mar 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Motivation

Test Plan

Test Result

Submission Checklist

Uh oh!

LuXuxue commented Mar 27, 2026

Uh oh!

harkgill-amd commented Mar 31, 2026

Uh oh!

CarlGao4 commented Apr 4, 2026

Uh oh!

geomin12 left a comment

Choose a reason for hiding this comment

Uh oh!

LuXuxue commented Apr 9, 2026

Uh oh!

harkgill-amd commented Apr 9, 2026

Uh oh!

geomin12 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

harkgill-amd commented Mar 4, 2026 •

edited

Loading