Page MenuHomePhabricator

ftynse (Alex Zinenko)
User

Projects

User does not belong to any projects.

User Details

User Since
Jun 28 2018, 2:24 AM (133 w, 5 d)

Recent Activity

Today

ftynse added a comment to D94923: [mlir:Async] Add intermediate async.coro and async.runtime operations to simplify Async to LLVM lowering.

https://github.com/llvm/llvm-project/blob/main/mlir/include/mlir/Dialect/LLVMIR/NVVMOps.td as an example. We also have support in mlir-tblgen to generate the boilerplate for ops given LLVM IR intrinsic definition file, example of usage: https://github.com/llvm/llvm-project/blob/main/mlir/test/mlir-tblgen/llvm-intrinsics.td

Tue, Jan 19, 5:04 AM · Restricted Project
ftynse committed rG9a60ad216d2f: [mlir] Clarify docs around LLVM dialect-compatible types (authored by ftynse).
[mlir] Clarify docs around LLVM dialect-compatible types
Tue, Jan 19, 4:43 AM
ftynse added a comment to D94923: [mlir:Async] Add intermediate async.coro and async.runtime operations to simplify Async to LLVM lowering.

Intermediate coro/runtime operations will allow to run transformations on a higher level IR and do not try to match IR based on the LLVM::CallOp properties.

Although async.coro is very close to LLVM coroutines, it is not exactly the same API, instead it is optimized for usability in async lowering, and misses a lot of details that are present in @llvm.coro intrinsic.

Tue, Jan 19, 4:26 AM · Restricted Project
ftynse accepted D94752: [mlir][python] Add facility for extending generated python ODS..
Tue, Jan 19, 1:50 AM · Restricted Project
ftynse accepted D94738: [mlir][python] Factor out standalone OpView._ods_build_default class method..
Tue, Jan 19, 1:29 AM · Restricted Project
ftynse accepted D94920: [mlir][Affine] Revisit and simplify composeAffineMapAndOperands..
Tue, Jan 19, 1:00 AM · Restricted Project
ftynse accepted D94939: Implement constant folding for DivFOp.
Tue, Jan 19, 12:15 AM · Restricted Project

Yesterday

ftynse added inline comments to D94738: [mlir][python] Factor out standalone OpView._ods_build_default class method..
Mon, Jan 18, 10:29 AM · Restricted Project
ftynse accepted D94918: [MLIR] NFC: simplify PresburgerSet::isEqual.

We generally use code review for everything except build unbreaks, reverts and minor syntactic changes such as typos.

Mon, Jan 18, 8:50 AM · Restricted Project
ftynse accepted D94915: [MLIR] Support checking if two FlatAffineConstraints are equal.
Mon, Jan 18, 7:24 AM · Restricted Project
ftynse added a comment to D94810: [mlir][python] Swap operands/results order on default builder..

Not sure about this one... I can understand the rationale, but this starts to diverge from C++/ODS, which already has suboptimal documentation and I'm apprehensive of the complexity it might add. If we agree we want to put result types after operands, we should be doing so consistently and that could be a problem for generating builders automatically if we decide we want that.

Mon, Jan 18, 6:23 AM · Restricted Project
ftynse added inline comments to D94911: [mlir] Add `complex.abs`, `complex.div` and `complex.mul` to ComplexOps..
Mon, Jan 18, 6:17 AM · Restricted Project
ftynse committed rGaca240b4f69e: [mlir] Fix cross-compilation (Linalg ODS gen) (authored by vinograd47).
[mlir] Fix cross-compilation (Linalg ODS gen)
Mon, Jan 18, 2:58 AM
ftynse closed D94598: [mlir] Fix cross-compilation (Linalg ODS gen).
Mon, Jan 18, 2:58 AM · Restricted Project
ftynse accepted D94812: [mlir][python] Swap shape and element_type order for MemRefType..

Hmm, this change actually makes Python API align on C++ API. I don't remember if there was a good reason to swap the order in C.

Mon, Jan 18, 2:57 AM · Restricted Project
ftynse added a comment to D94812: [mlir][python] Swap shape and element_type order for MemRefType..

Matches how all of the other shaped types are declared.

Mon, Jan 18, 2:55 AM · Restricted Project

Fri, Jan 15

ftynse accepted D94737: [mlir] Add better support for f80 and f128.
Fri, Jan 15, 12:27 AM · Restricted Project

Thu, Jan 14

ftynse committed rG6ebeba88f519: Support emptiness checks for unbounded FlatAffineConstraints. (authored by arjunp).
Support emptiness checks for unbounded FlatAffineConstraints.
Thu, Jan 14, 10:34 AM
ftynse closed D94272: Support emptiness checks for unbounded FlatAffineConstraints..
Thu, Jan 14, 10:34 AM · Restricted Project
ftynse accepted D94272: Support emptiness checks for unbounded FlatAffineConstraints..

Thank you!

Thu, Jan 14, 10:23 AM · Restricted Project
ftynse added a comment to D94272: Support emptiness checks for unbounded FlatAffineConstraints..

We can't test for this because the Euclidean GCD algorithm is still correct and terminates with machine division (about efficiency, I am not sure).

Thu, Jan 14, 3:54 AM · Restricted Project
ftynse added a comment to D94651: [mlir] Update doc to omit the usage of LLVMIntegerType.

Please consider avoiding "advanced" markdown features in the commit messages, a lot of people look at them in the terminal and it may be hard to parse

Thu, Jan 14, 12:31 AM · Restricted Project
ftynse committed rGed205f63b4a2: [mlir] Update doc to omit the usage of LLVMIntegerType (authored by Lewuathe).
[mlir] Update doc to omit the usage of LLVMIntegerType
Thu, Jan 14, 12:29 AM
ftynse closed D94651: [mlir] Update doc to omit the usage of LLVMIntegerType.
Thu, Jan 14, 12:29 AM · Restricted Project

Wed, Jan 13

ftynse added a comment to D94551: [mlir][sparse] add vectorization strategies to sparse compiler.

On one hand, it's great to see minimal additional patterns translate to new capabilities, very nice!

OTOH, this has grown far beyond a few reasonable abstraction gap jumps, even for a prototype.
As time passes and the abstraction gap grows between the input and the output, the more difficult it will be to break down and refactor into composable pieces.

I am curious what the expected integration will look like and what other features you want to prototype before this integration can start?
For instance, it seems unlikely to me that the tests, as they are written, will be gradually ported to new abstractions.
One outcome that I would find unfortunate is whether people actually start to rely on all the functionality needing to be kept green at all times..

Thoughts?

Wed, Jan 13, 4:05 AM · Restricted Project
ftynse added inline comments to D94551: [mlir][sparse] add vectorization strategies to sparse compiler.
Wed, Jan 13, 3:59 AM · Restricted Project
ftynse accepted D94575: [mlir][OpFormatGen] Format enum attribute cases as keywords when possible.

Yay!

Wed, Jan 13, 1:14 AM · Restricted Project

Tue, Jan 12

ftynse committed rG7fd185081341: [mlir] Update LLVM dialect type documentation (authored by ftynse).
[mlir] Update LLVM dialect type documentation
Tue, Jan 12, 1:38 PM
ftynse closed D94485: [mlir] Update LLVM dialect type documentation.
Tue, Jan 12, 1:38 PM · Restricted Project
ftynse committed rG9667d15e7496: [mlir] Fix for LIT tests (authored by vinograd47).
[mlir] Fix for LIT tests
Tue, Jan 12, 8:07 AM
ftynse closed D94407: [mlir] Fix for LIT tests.
Tue, Jan 12, 8:07 AM · Restricted Project
ftynse added a comment to D94407: [mlir] Fix for LIT tests.

I took your name and email from Github commits. Please consider sending patches with Arcanist to preserve author (and base revision) info.

Tue, Jan 12, 8:07 AM · Restricted Project
ftynse committed rG4fa01f72de6c: [mlir][CAPI] Fix inline function declaration (authored by vinograd47).
[mlir][CAPI] Fix inline function declaration
Tue, Jan 12, 8:05 AM
ftynse closed D94496: [mlir][CAPI] Fix inline function declaration.
Tue, Jan 12, 8:05 AM · Restricted Project
ftynse added a comment to D94497: [mlir] Define inline functions in headers as static.

Looks like a duplicate of D94496

Tue, Jan 12, 7:45 AM · Restricted Project
ftynse accepted D94496: [mlir][CAPI] Fix inline function declaration.
Tue, Jan 12, 7:44 AM · Restricted Project
ftynse requested review of D94485: [mlir] Update LLVM dialect type documentation.
Tue, Jan 12, 3:08 AM · Restricted Project
ftynse committed rGbd30a796fc4b: [mlir] use built-in vector types instead of LLVM dialect types when possible (authored by ftynse).
[mlir] use built-in vector types instead of LLVM dialect types when possible
Tue, Jan 12, 1:04 AM
ftynse closed D94405: [mlir] use built-in vector types instead of LLVM dialect types when possible.
Tue, Jan 12, 1:04 AM · Restricted Project
ftynse added inline comments to D94405: [mlir] use built-in vector types instead of LLVM dialect types when possible.
Tue, Jan 12, 1:00 AM · Restricted Project

Mon, Jan 11

ftynse accepted D94441: [mlir][vector] verify memref of vector memory ops.
Mon, Jan 11, 12:10 PM · Restricted Project
ftynse committed rG547e3eef14a8: [mlir] Expose MemRef layout in Python bindings (authored by ftynse).
[mlir] Expose MemRef layout in Python bindings
Mon, Jan 11, 10:58 AM
ftynse closed D94302: [mlir] Expose MemRef layout in Python bindings.
Mon, Jan 11, 10:57 AM · Restricted Project
ftynse committed rGe79bd0b4f25e: [mlir] More Python bindings for AffineMap (authored by ftynse).
[mlir] More Python bindings for AffineMap
Mon, Jan 11, 10:57 AM
ftynse committed rG74628c43053b: [mlir] Add Python bindings for AffineExpr (authored by ftynse).
[mlir] Add Python bindings for AffineExpr
Mon, Jan 11, 10:57 AM
ftynse closed D94297: [mlir] More Python bindings for AffineMap.
Mon, Jan 11, 10:57 AM · Restricted Project
ftynse closed D94225: [mlir] Add Python bindings for AffineExpr.
Mon, Jan 11, 10:57 AM · Restricted Project
ftynse requested changes to D94272: Support emptiness checks for unbounded FlatAffineConstraints..
Mon, Jan 11, 9:15 AM · Restricted Project
ftynse updated the diff for D94405: [mlir] use built-in vector types instead of LLVM dialect types when possible.

Rebase

Mon, Jan 11, 8:32 AM · Restricted Project
ftynse accepted D94368: Enable python bindings for tensor, shape and linalg dialects..
Mon, Jan 11, 6:03 AM · Restricted Project
ftynse accepted D94370: [mlir][CAPI] Introduce standard source layout for mlir-c dialect registration..
Mon, Jan 11, 6:01 AM · Restricted Project
ftynse updated the diff for D94297: [mlir] More Python bindings for AffineMap.

Address review

Mon, Jan 11, 5:54 AM · Restricted Project
ftynse requested review of D94405: [mlir] use built-in vector types instead of LLVM dialect types when possible.
Mon, Jan 11, 5:13 AM · Restricted Project
ftynse accepted D94400: [mlir] Remove unnecessary llvm.mlir.cast in AsyncToLLVM lowering..
Mon, Jan 11, 5:12 AM · Restricted Project
ftynse accepted D94397: [mlir] Make GpuAsyncRegion pass depend on async dialect..
Mon, Jan 11, 4:56 AM · Restricted Project
ftynse accepted D94402: [mlir] Fix gpu-to-llvm lowering for gpu.alloc with dynamic sizes..
Mon, Jan 11, 4:44 AM · Restricted Project
ftynse added a comment to D94400: [mlir] Remove unnecessary llvm.mlir.cast in AsyncToLLVM lowering..

Is it always unnecessary or is there a possibility that it still might be (e.g., async pointer)?

Mon, Jan 11, 4:28 AM · Restricted Project

Fri, Jan 8

ftynse committed rGdd5165a920f6: [mlir] replace LLVM dialect float types with built-ins (authored by ftynse).
[mlir] replace LLVM dialect float types with built-ins
Fri, Jan 8, 8:38 AM
ftynse closed D94179: [mlir] replace LLVM dialect float types with built-ins.
Fri, Jan 8, 8:38 AM · Restricted Project
ftynse updated the diff for D94179: [mlir] replace LLVM dialect float types with built-ins.

Rebase.

Fri, Jan 8, 8:36 AM · Restricted Project
ftynse requested review of D94302: [mlir] Expose MemRef layout in Python bindings.
Fri, Jan 8, 6:25 AM · Restricted Project
ftynse committed rG195ffcd890f6: [MLIR][NFC] remove unnecessary includes form tablegen command (authored by xgupta).
[MLIR][NFC] remove unnecessary includes form tablegen command
Fri, Jan 8, 5:26 AM
ftynse closed D94216: [MLIR][NFC] remove unnecessary includes form tablegen command.
Fri, Jan 8, 5:26 AM · Restricted Project
ftynse added inline comments to D94297: [mlir] More Python bindings for AffineMap.
Fri, Jan 8, 5:24 AM · Restricted Project
ftynse requested review of D94297: [mlir] More Python bindings for AffineMap.
Fri, Jan 8, 5:21 AM · Restricted Project
ftynse added a comment to D94216: [MLIR][NFC] remove unnecessary includes form tablegen command.

@ftynse please commit the revision, I don't have commit access.

Fri, Jan 8, 4:26 AM · Restricted Project
ftynse requested changes to D94272: Support emptiness checks for unbounded FlatAffineConstraints..
Fri, Jan 8, 1:13 AM · Restricted Project
ftynse added a comment to D94272: Support emptiness checks for unbounded FlatAffineConstraints..

It looks like something went wrong when updating the second revision and the diff now only contains the comment change. Note that if you use arc diff on a branch of multiple commits, you must specify the commit that the branch is based on rather than HEAD^, which is merely the previous-to-head commit.

Fri, Jan 8, 1:13 AM · Restricted Project
ftynse accepted D94216: [MLIR][NFC] remove unnecessary includes form tablegen command.
Fri, Jan 8, 1:10 AM · Restricted Project

Thu, Jan 7

ftynse committed rG906efeec0a47: [mlir] don't match the text produced only in debug mode in Python tests (authored by ftynse).
[mlir] don't match the text produced only in debug mode in Python tests
Thu, Jan 7, 10:58 AM
ftynse closed D94221: [mlir] don't match the text produced only in debug mode in Python tests.
Thu, Jan 7, 10:57 AM · Restricted Project
ftynse committed rG2230bf99c712: [mlir] replace LLVMIntegerType with built-in integer type (authored by ftynse).
[mlir] replace LLVMIntegerType with built-in integer type
Thu, Jan 7, 10:49 AM
ftynse closed D94178: [mlir] replace LLVMIntegerType with built-in integer type.
Thu, Jan 7, 10:49 AM · Restricted Project
ftynse accepted D94235: [MLIR][OpenMP] Attribute to include WsLoop upperbound.
Thu, Jan 7, 8:55 AM · Restricted Project
ftynse added inline comments to D94179: [mlir] replace LLVM dialect float types with built-ins.
Thu, Jan 7, 6:20 AM · Restricted Project
ftynse updated the diff for D94179: [mlir] replace LLVM dialect float types with built-ins.

Address review

Thu, Jan 7, 6:10 AM · Restricted Project
ftynse committed rGa7cbc32a916a: [mlir] remove a use of deprecated OpState::setAttr (authored by ftynse).
[mlir] remove a use of deprecated OpState::setAttr
Thu, Jan 7, 5:21 AM
ftynse added a comment to D92485: [mlir] Add fastmath flags support to some LLVM dialect ops.

FYI, if you send commits with arcanist, the committer credentials are preserved.

Thu, Jan 7, 5:00 AM · Restricted Project
ftynse committed rGc1d58c2b0023: [mlir] Add fastmath flags support to some LLVM dialect ops (authored by Hardcode84).
[mlir] Add fastmath flags support to some LLVM dialect ops
Thu, Jan 7, 5:00 AM
ftynse closed D92485: [mlir] Add fastmath flags support to some LLVM dialect ops.
Thu, Jan 7, 5:00 AM · Restricted Project
ftynse committed rG10164a2e50b4: [mlir] Refactor translation of OpenMP dialect ops to LLVM IR (authored by ftynse).
[mlir] Refactor translation of OpenMP dialect ops to LLVM IR
Thu, Jan 7, 4:34 AM
ftynse closed D94086: [mlir] Refactor translation of OpenMP dialect ops to LLVM IR.
Thu, Jan 7, 4:34 AM · Restricted Project
ftynse added inline comments to D94086: [mlir] Refactor translation of OpenMP dialect ops to LLVM IR.
Thu, Jan 7, 4:33 AM · Restricted Project
ftynse added a comment to D92485: [mlir] Add fastmath flags support to some LLVM dialect ops.

Can someone merge this, please?

Thu, Jan 7, 4:27 AM · Restricted Project
ftynse requested review of D94225: [mlir] Add Python bindings for AffineExpr.
Thu, Jan 7, 4:22 AM · Restricted Project
ftynse requested review of D94221: [mlir] don't match the text produced only in debug mode in Python tests.
Thu, Jan 7, 2:31 AM · Restricted Project

Wed, Jan 6

ftynse requested review of D94179: [mlir] replace LLVM dialect float types with built-ins.
Wed, Jan 6, 7:21 AM · Restricted Project
ftynse requested review of D94178: [mlir] replace LLVMIntegerType with built-in integer type.
Wed, Jan 6, 7:19 AM · Restricted Project
ftynse abandoned D88012: WIP: DSL-style builder generator.
Wed, Jan 6, 4:10 AM · Restricted Project
ftynse abandoned D89474: [DO NOT COMMIT][mlir] modernizer for ODS OpBuilder syntax.
Wed, Jan 6, 4:09 AM · Restricted Project, Restricted Project
ftynse abandoned D88960: [mlir] use enum for C MlirLogicalResult.
Wed, Jan 6, 4:09 AM · Restricted Project
ftynse committed rG74438eff511e: [mlir] Use thread_local stack in LLVM dialect type parsing and printing (authored by ftynse).
[mlir] Use thread_local stack in LLVM dialect type parsing and printing
Wed, Jan 6, 3:05 AM
ftynse closed D93714: [mlir] Use thread_local stack in LLVM dialect type parsing and printing.
Wed, Jan 6, 3:05 AM · Restricted Project
ftynse accepted D92485: [mlir] Add fastmath flags support to some LLVM dialect ops.

Thanks for iterating on this!

Wed, Jan 6, 1:52 AM · Restricted Project
ftynse accepted D94158: [mlir] NFC: fix trivial typos fix typos under docs, test, and tools directories.
Wed, Jan 6, 1:47 AM · Restricted Project

Tue, Jan 5

ftynse accepted D94052: [mlir] Add sret and byval argument attribute support to LLVM-IR dialect.
Tue, Jan 5, 9:00 AM · Restricted Project
ftynse committed rGc69c9e0f0fd2: [mlir] Remove LLVMType, LLVM dialect types now derive Type directly (authored by ftynse).
[mlir] Remove LLVMType, LLVM dialect types now derive Type directly
Tue, Jan 5, 8:37 AM
ftynse closed D93713: [mlir] Remove LLVMType, LLVM dialect types now derive Type directly.
Tue, Jan 5, 8:37 AM · Restricted Project
ftynse added inline comments to D93714: [mlir] Use thread_local stack in LLVM dialect type parsing and printing.
Tue, Jan 5, 8:04 AM · Restricted Project
ftynse updated the diff for D93714: [mlir] Use thread_local stack in LLVM dialect type parsing and printing.

Address review

Tue, Jan 5, 8:03 AM · Restricted Project