Page MenuHomePhabricator

jrbyrnes (Jeffrey Byrnes)
User

Projects

User does not belong to any projects.

User Details

User Since
Feb 27 2022, 6:44 PM (23 w, 17 h)

Recent Activity

Fri, Aug 5

jrbyrnes updated the diff for D130797: [AMDGPU] Implement pipeline solver for non-trivial pipelines.

Address Review comments.

Fri, Aug 5, 2:15 PM · Restricted Project, Restricted Project
jrbyrnes added inline comments to D130797: [AMDGPU] Implement pipeline solver for non-trivial pipelines.
Fri, Aug 5, 8:35 AM · Restricted Project, Restricted Project

Thu, Aug 4

jrbyrnes added inline comments to D130729: [InferAddressSpaces] [AMDGPU] Add inference for flat_atomic intrinsics.
Thu, Aug 4, 1:38 PM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D130729: [InferAddressSpaces] [AMDGPU] Add inference for flat_atomic intrinsics.

Move codegen tests to CodeGen, add IR test for InferAddressSpace flat_atomic.

Thu, Aug 4, 1:21 PM · Restricted Project, Restricted Project
jrbyrnes retitled D130729: [InferAddressSpaces] [AMDGPU] Add inference for flat_atomic intrinsics from [SeparateConstOffsetFromGEP] [AMDGPU] Check legality for all uses of transformed GEP to [InferAddressSpaces] [AMDGPU] Add inference for flat_atomic intrinsics.
Thu, Aug 4, 12:19 PM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D130729: [InferAddressSpaces] [AMDGPU] Add inference for flat_atomic intrinsics.

Rework approach of fix.

Thu, Aug 4, 12:17 PM · Restricted Project, Restricted Project

Wed, Aug 3

jrbyrnes added a comment to D130729: [InferAddressSpaces] [AMDGPU] Add inference for flat_atomic intrinsics.

Hey Matt, Jay,

Wed, Aug 3, 3:35 PM · Restricted Project, Restricted Project
jrbyrnes committed rGe0b16aaaf997: [AMDGPU] Precommit test case for D130729 (authored by jrbyrnes).
[AMDGPU] Precommit test case for D130729
Wed, Aug 3, 3:23 PM · Restricted Project, Restricted Project

Tue, Aug 2

jrbyrnes added a reviewer for D130729: [InferAddressSpaces] [AMDGPU] Add inference for flat_atomic intrinsics: foad.
Tue, Aug 2, 2:27 PM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D130797: [AMDGPU] Implement pipeline solver for non-trivial pipelines.

Address Review Comments.

Tue, Aug 2, 2:21 PM · Restricted Project, Restricted Project

Mon, Aug 1

jrbyrnes added a comment to D130797: [AMDGPU] Implement pipeline solver for non-trivial pipelines.

Thanks! I like the idea behind the greedy solver. Not sure about SchedGroupSU. Maybe just a map between SUs and lists of schedgroups? I think trying to track sched_group_barriers by their order and assigning that an index is a bit confusing.

Mon, Aug 1, 1:31 PM · Restricted Project, Restricted Project

Fri, Jul 29

jrbyrnes requested review of D130797: [AMDGPU] Implement pipeline solver for non-trivial pipelines.
Fri, Jul 29, 1:07 PM · Restricted Project, Restricted Project

Thu, Jul 28

jrbyrnes updated the diff for D130729: [InferAddressSpaces] [AMDGPU] Add inference for flat_atomic intrinsics.

Remove unnecessary local var

Thu, Jul 28, 1:53 PM · Restricted Project, Restricted Project
jrbyrnes requested review of D130729: [InferAddressSpaces] [AMDGPU] Add inference for flat_atomic intrinsics.
Thu, Jul 28, 1:52 PM · Restricted Project, Restricted Project
jrbyrnes accepted D128158: [AMDGPU] Add amdgcn_sched_group_barrier builtin.
Thu, Jul 28, 9:37 AM · Restricted Project, Restricted Project, Restricted Project

Tue, Jul 26

jrbyrnes added a comment to D128158: [AMDGPU] Add amdgcn_sched_group_barrier builtin.

LGTM

Tue, Jul 26, 2:30 PM · Restricted Project, Restricted Project, Restricted Project

Thu, Jul 14

jrbyrnes committed rG8a12f20ef773: [AMDGPU] Update the mechanism used to check for cycles and add eges in power… (authored by jrbyrnes).
[AMDGPU] Update the mechanism used to check for cycles and add eges in power…
Thu, Jul 14, 4:25 PM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D129667: [AMDGPU] Update the mechanism used to check for cycles and add eges in power-sched mutation.

Address review comments.

Thu, Jul 14, 10:14 AM · Restricted Project, Restricted Project

Wed, Jul 13

jrbyrnes updated the diff for D129667: [AMDGPU] Update the mechanism used to check for cycles and add eges in power-sched mutation.

Include test which which minimally reproduces the SmallVector error reported.

Wed, Jul 13, 4:41 PM · Restricted Project, Restricted Project
jrbyrnes requested review of D129667: [AMDGPU] Update the mechanism used to check for cycles and add eges in power-sched mutation.
Wed, Jul 13, 10:45 AM · Restricted Project, Restricted Project

Jul 6 2022

jrbyrnes updated the diff for D127994: [AMDGPU] Expose CLI controls for IGroup ordering.

Remove unnecessary debug code.

Jul 6 2022, 11:09 AM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D127994: [AMDGPU] Expose CLI controls for IGroup ordering.

Addressed review comments.

Jul 6 2022, 11:03 AM · Restricted Project, Restricted Project

Jul 1 2022

jrbyrnes added a comment to D128158: [AMDGPU] Add amdgcn_sched_group_barrier builtin.

Hey Austin -- I like the removal of canAddMIs. In the original design, I was leaving open the possibility for users to pass in canAddMIs rather than a mask / SchedGroup name, but it looks like this isn't the direction we're going, and the classification functions defined in a general canAddMI makes things easier.

Jul 1 2022, 11:45 AM · Restricted Project, Restricted Project, Restricted Project

Jun 30 2022

jrbyrnes committed rG09424f802c8f: [AMDGPU] Check for CopyToReg PhysReg clobbers in pre-RA-sched (authored by jrbyrnes).
[AMDGPU] Check for CopyToReg PhysReg clobbers in pre-RA-sched
Jun 30 2022, 9:18 AM · Restricted Project, Restricted Project
jrbyrnes closed D128681: [AMDGPU][ScheduleDAG] Check for CopyToReg PhysReg clobbers in pre-RA-sched.
Jun 30 2022, 9:18 AM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D128681: [AMDGPU][ScheduleDAG] Check for CopyToReg PhysReg clobbers in pre-RA-sched.

Remove accidental files

Jun 30 2022, 7:43 AM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D128681: [AMDGPU][ScheduleDAG] Check for CopyToReg PhysReg clobbers in pre-RA-sched.

Run instnamer on testfile, explicitly use "source" (RRList) scheduler for InstSelection Scheduler in test.

Jun 30 2022, 7:41 AM · Restricted Project, Restricted Project

Jun 28 2022

jrbyrnes added inline comments to D128681: [AMDGPU][ScheduleDAG] Check for CopyToReg PhysReg clobbers in pre-RA-sched.
Jun 28 2022, 4:13 PM · Restricted Project, Restricted Project

Jun 27 2022

jrbyrnes updated the diff for D128681: [AMDGPU][ScheduleDAG] Check for CopyToReg PhysReg clobbers in pre-RA-sched.

Broke up logic in ScheduleDAGFast CheckForLiveRegDef to remove redundancy.

Jun 27 2022, 4:44 PM · Restricted Project, Restricted Project
jrbyrnes retitled D128681: [AMDGPU][ScheduleDAG] Check for CopyToReg PhysReg clobbers in pre-RA-sched from [AMDGPU] Check for CopyToReg PhysReg clobbers in pre-RA-sched to [AMDGPU][ScheduleDAG] Check for CopyToReg PhysReg clobbers in pre-RA-sched.
Jun 27 2022, 2:05 PM · Restricted Project, Restricted Project
jrbyrnes added reviewers for D128681: [AMDGPU][ScheduleDAG] Check for CopyToReg PhysReg clobbers in pre-RA-sched: arsenm, rampitec, kerbowa, foad, Hipony.
Jun 27 2022, 2:04 PM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D128681: [AMDGPU][ScheduleDAG] Check for CopyToReg PhysReg clobbers in pre-RA-sched.

Ported over to phab review to land in Trunk. Addressed the requests in initial review, renamed test file to better align with naming of previous test.

Jun 27 2022, 2:02 PM · Restricted Project, Restricted Project
jrbyrnes updated the summary of D128681: [AMDGPU][ScheduleDAG] Check for CopyToReg PhysReg clobbers in pre-RA-sched.
Jun 27 2022, 1:58 PM · Restricted Project, Restricted Project
jrbyrnes requested review of D128681: [AMDGPU][ScheduleDAG] Check for CopyToReg PhysReg clobbers in pre-RA-sched.
Jun 27 2022, 1:57 PM · Restricted Project, Restricted Project

Jun 16 2022

jrbyrnes added inline comments to D127994: [AMDGPU] Expose CLI controls for IGroup ordering.
Jun 16 2022, 12:46 PM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D127994: [AMDGPU] Expose CLI controls for IGroup ordering.

Fix naming of the parser.

Jun 16 2022, 11:37 AM · Restricted Project, Restricted Project
jrbyrnes requested review of D127994: [AMDGPU] Expose CLI controls for IGroup ordering.
Jun 16 2022, 11:24 AM · Restricted Project, Restricted Project

Jun 13 2022

jrbyrnes added a comment to D127123: [AMDGPU] Add more expressive sched_barrier controls.

Thanks for addressing the post-commit requests in D125997. LGTM

Jun 13 2022, 1:13 PM · Restricted Project, Restricted Project

May 26 2022

jrbyrnes added a comment to D125997: [AMDGPU] Instruction Type Pipeline.

I think your last diff overwrote the changes before last. I.e. the added comments and renamed cl opts are gone.

May 26 2022, 5:20 PM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D125997: [AMDGPU] Instruction Type Pipeline.

Write back the changes that were accidentally overwritten

May 26 2022, 5:19 PM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D125997: [AMDGPU] Instruction Type Pipeline.

Testing on CK uncovered an edge case error w/ the way bundle iteration was done.

May 26 2022, 4:59 PM · Restricted Project, Restricted Project

May 25 2022

jrbyrnes added inline comments to D125997: [AMDGPU] Instruction Type Pipeline.
May 25 2022, 9:58 AM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D125997: [AMDGPU] Instruction Type Pipeline.

Addressed Review comments:

  1. Command line option naming.
  2. Included comments where code was potentially confusing
May 25 2022, 9:52 AM · Restricted Project, Restricted Project

May 23 2022

jrbyrnes updated the diff for D125997: [AMDGPU] Instruction Type Pipeline.

Addressed review comments:

  1. Added options for sizing of specific instruction groups
  2. Renamed the DAG Mutation -- renaming suggestions welcomed.
  3. Minor details
May 23 2022, 12:36 PM · Restricted Project, Restricted Project

May 19 2022

jrbyrnes retitled D125997: [AMDGPU] Instruction Type Pipeline from Instruction Type Pipeline to [AMDGPU] Instruction Type Pipeline.
May 19 2022, 11:51 AM · Restricted Project, Restricted Project
jrbyrnes requested review of D125997: [AMDGPU] Instruction Type Pipeline.
May 19 2022, 11:31 AM · Restricted Project, Restricted Project

May 10 2022

jrbyrnes committed rGf822db7670d4: [AMDGPU] Allow for MFMA Inst Clustering (authored by jrbyrnes).
[AMDGPU] Allow for MFMA Inst Clustering
May 10 2022, 12:59 PM · Restricted Project, Restricted Project
jrbyrnes closed D124678: [AMDGPU] Allow for MFMA Inst Clustering.
May 10 2022, 12:59 PM · Restricted Project, Restricted Project
jrbyrnes committed rG3ff8ee2447ec: [NFC] Fix typo (authored by jrbyrnes).
[NFC] Fix typo
May 10 2022, 12:12 PM · Restricted Project, Restricted Project
jrbyrnes closed D124647: [NFC] Fix typo.
May 10 2022, 12:12 PM · Restricted Project, Restricted Project

May 6 2022

jrbyrnes updated the diff for D124678: [AMDGPU] Allow for MFMA Inst Clustering.

Resolve NITs

May 6 2022, 4:59 PM · Restricted Project, Restricted Project

May 4 2022

jrbyrnes updated the diff for D124678: [AMDGPU] Allow for MFMA Inst Clustering.

Fix algorithmic flaws:

  1. Use chain as cluster shape (A->B->C->D) instead of fanout (A->{B,C,D}). With a chain, the scheduler will not miss cluster edges due to multiple cluster succs.
  2. Create artificial edges in the cluster. This will coerce the scheduler to start from either the root or leaf of the cluster rather than potentially selecting the middle. In post RA scheduling, if the scheduler selects the middle, it will lose the cluster prefix.
May 4 2022, 10:15 AM · Restricted Project, Restricted Project
jrbyrnes added inline comments to D124678: [AMDGPU] Allow for MFMA Inst Clustering.
May 4 2022, 9:23 AM · Restricted Project, Restricted Project

May 3 2022

jrbyrnes added inline comments to D124678: [AMDGPU] Allow for MFMA Inst Clustering.
May 3 2022, 12:43 PM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D124678: [AMDGPU] Allow for MFMA Inst Clustering.

Remove extraneous file

May 3 2022, 11:41 AM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D124678: [AMDGPU] Allow for MFMA Inst Clustering.

Add test to MIR after both scheduling passes in sequence to confirm clustering results hold.

May 3 2022, 11:40 AM · Restricted Project, Restricted Project
jrbyrnes updated the diff for D124678: [AMDGPU] Allow for MFMA Inst Clustering.

Resolve silly arcconfig issue.

May 3 2022, 11:13 AM · Restricted Project, Restricted Project

May 2 2022

jrbyrnes updated the diff for D124678: [AMDGPU] Allow for MFMA Inst Clustering.

Updating D124678: [AMDGPU] Allow for MFMA Inst Clustering

May 2 2022, 6:22 PM · Restricted Project, Restricted Project

Apr 29 2022

jrbyrnes added a comment to D124678: [AMDGPU] Allow for MFMA Inst Clustering.

One thing to mention: transferring all successors and predecessors technically does not guarantee that an independent instruction will not be scheduled in between. I see that you are adding SDep::Cluster, but I remember it didn't use to work in the very similar scenario before. Not sure it is still so. It can happen that in a bigger program the cluster may be broken.

Also we probably need to consider dropping MFMA clustering along with load clustering during GCNScheduleDAGMILive::UnclusteredReschedule stage.

Apr 29 2022, 4:19 PM · Restricted Project, Restricted Project
jrbyrnes added inline comments to D124678: [AMDGPU] Allow for MFMA Inst Clustering.
Apr 29 2022, 10:05 AM · Restricted Project, Restricted Project
jrbyrnes requested review of D124678: [AMDGPU] Allow for MFMA Inst Clustering.
Apr 29 2022, 8:35 AM · Restricted Project, Restricted Project

Apr 28 2022

jrbyrnes requested review of D124647: [NFC] Fix typo.
Apr 28 2022, 5:09 PM · Restricted Project, Restricted Project