refactor(expr): move buffer_ops and split gfx1250 mcast policy to kernels.common#880
Merged
Conversation
Phil-amd
commented
Jul 22, 2026
Phil-amd
commented
Member
Phil-amd
force-pushed
the
lhc/refactor-expr
branch
2 times, most recently
from
July 23, 2026 05:35
7273ad8 to
bfe30ce
Compare
Phil-amd
changed the title
Phil-amd
changed the title
Phil-amd
force-pushed
the
lhc/refactor-expr
branch
from
July 23, 2026 07:00
bfe30ce to
faac4f1
Compare
… remove expr/vector.py Reconcile ROCm/FlyDSL PR ROCm#853 onto current main: relocate buffer_ops to kernels.common, delete expr/vector.py with a full vector migration to raw _mlir.dialects.vector + as_ir_value, adopt the tdm_ops raw-vector cleanup, and update __init__/docs/tests. Also covers upstream-added/relocated files ROCm#853's stale base missed (flash_attn_utils, rmsnorm_bwd_kernel, common/mma helpers). Co-Authored-By: Feng Shijie <Shijie.Feng@amd.com>
…250_cluster Move the gfx1250 MCAST workgroup-mask policy out of expr/rocdl/cluster.py into kernels/common/gfx1250_cluster.py (it is kernel policy, not a ROCDL primitive) and point the gfx1250 gemm/moe call sites and tdm tests at the new module.
Phil-amd
force-pushed
the
lhc/refactor-expr
branch
2 times, most recently
from
July 23, 2026 08:45
ab5aa59 to
be5a37b
Compare
coderfeli
approved these changes
Jul 23, 2026
coderfeli
pushed a commit
that referenced
this pull request
Jul 24, 2026
#880 moved buffer_ops out of flydsl.expr into kernels.common; the fused mxfp_moe kernels (added after #880) still imported it from flydsl.expr, breaking a fresh-build import (CI). Point mxfp4_gemm_common at kernels.common.buffer_ops (same API: get_element_ptr / create_buffer_resource_from_addr). Local venv is pre-#880 so this was masked locally. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
xudoyuan
added a commit
that referenced
this pull request
Jul 24, 2026
PR #880 moved buffer_ops to kernels/common and dropped the flydsl.expr alias, so neither fx.buffer_ops nor 'from flydsl.expr import buffer_ops' resolves on main. Write outputs with plain tensor subscripting (Out[i] = val), matching the other system tests (e.g. test_while_e2e) and dropping the buffer_ops dependency. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
loctxmoreh
added a commit
to loctxmoreh/FlyDSL
that referenced
this pull request
Jul 24, 2026
Upstream 0.3.0 (19 commits, ROCm#893 version bump). Conflicts, all resolved keeping the gfx90a logic: - flash_attn_generic.py / flash_attn_utils.py: upstream added a gfx942 async-DMA opt-in and re-widened `has_lds_load_b128` to `not gfx942`. Kept the gfx90a-correct `startswith("gfx950")` test (b128 buffer_load_dwordx4_lds is CDNA4-only; on gfx90a it aborts the AMDGPU backend) and layered upstream's gfx942 opt-in on top, so gfx90a takes neither path. - hgemm_splitk.py / splitk_hgemm.py: import-only conflict; kept `is_cdna4` alongside upstream's relocated `kernels.common.buffer_ops`. One non-conflict fixup the merge required: ROCm#880 moved `expr/buffer_ops.py` to `kernels.common`, which `expr/rocdl/atomic.py` imported for `extract_base_index`. `expr/` must not import from `kernels/`, so the base-address extraction is now inlined in `atomic.py` using the same fly/memref primitives `expr/rocdl/tdm_ops.py` already uses. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.