Page MenuHomePhabricator

fhahn (Florian Hahn)
User

Projects

User does not belong to any projects.

User Details

User Since
Aug 18 2016, 4:39 AM (189 w, 1 d)

Recent Activity

Today

fhahn updated the diff for D76612: [Matrix] Add draft specification for matrix support in Clang..

Specify that standard conversion rules do not apply to assignments for matrix types.

Fri, Apr 3, 11:20 AM · Restricted Project
fhahn added a comment to D70456: [Matrix] Add first set of matrix intrinsics and initial lowering pass..

I have the similar question on how to lower matrix intrinsics to some HW specific intrinsics/instruction. For example, X86 have the AVX512_VNNI feature (https://software.intel.com/sites/landingpage/IntrinsicsGuide/#expand=39,5370,5361,364,142,139,2210&text=vnni). It can perform dot product computation. But after matrix intrinsic is lowered to vector, it seems difficult to transform the vector operation to AVX512_VNNI intrinsic/instruction.

Fri, Apr 3, 10:47 AM · Restricted Project
fhahn added inline comments to D76149: [AssumeBundles] Use assume bundles in isKnownNonZero.
Fri, Apr 3, 10:15 AM · Restricted Project
fhahn added a comment to D70456: [Matrix] Add first set of matrix intrinsics and initial lowering pass..

Hi Renato,

Fri, Apr 3, 5:52 AM · Restricted Project
fhahn added inline comments to D76838: [LV][LoopInfo] Transform counting-down loops to counting-up loop.
Fri, Apr 3, 4:48 AM
fhahn updated the diff for D72281: [Matrix] Add matrix type to Clang..

Use 20 bits in MatrixTypeBitfields for both number of rows and number of columns. This leaves 24 bits for NumTypeBits, while providing a large ranges for number of rows/columns.

Fri, Apr 3, 4:15 AM · Restricted Project
fhahn updated the diff for D77378: [Clang] Include size and maximum in vector size error message..

Fix wrong formatting of check lines..

Fri, Apr 3, 3:44 AM · Restricted Project
fhahn created D77378: [Clang] Include size and maximum in vector size error message..
Fri, Apr 3, 3:43 AM · Restricted Project
fhahn added a comment to D77363: Use git-clang-format as Arcanist linter.

This does not work for me

Fri, Apr 3, 3:43 AM · Restricted Project
fhahn added inline comments to D77335: [AST] clang::VectorType supports any size (that fits in unsigned).
Fri, Apr 3, 3:11 AM · Restricted Project

Yesterday

fhahn added a comment to D73763: [DSE] Lift post-dominance restriction..

I think there are 2 cases to distinguish:

  1. For accesses to non-alloca objects, this matches what the intrinsic achieves I think. We can only eliminate stores to objects that are visible after the function returns, if they are overwritten along all paths to the exit. So if we have determined a set of blocks that overwrite A (and there are no reads in between), we could check if all paths to the exit from A must go through one of the overwriting blocks. I think that matches your suggestion.

Sounds like what I was thinking, yeah.

For objects that are not visible to the caller (e.g. alloca), the intrinsic achieves something slightly different: Given A and B as above, we can eliminate A, if there are no reads of A between A and any of the exits from A. We can stop walking any paths that contain overwrites of A.

I'm confused about why allocas need special treatment. This new set that we're building is essentially a fast way to answer the question "are there potentially unknowable reads after this Def/Phi?" If we know memory (e.g., from an alloca) is unreadable outside of the function, I'd hope we can ignore the set entirely, like this patch is doing now? Baked into my suggestion was the assumption that we were already:

Thu, Apr 2, 2:06 PM · Restricted Project
fhahn added inline comments to D70456: [Matrix] Add first set of matrix intrinsics and initial lowering pass..
Thu, Apr 2, 1:33 PM · Restricted Project
fhahn updated the diff for D76794: [Matrix] Implement * binary operator for MatrixType..

clang-format

Thu, Apr 2, 1:00 PM · Restricted Project
fhahn updated the diff for D76794: [Matrix] Implement * binary operator for MatrixType..

Implement conversion for matrix/scalar variants.

Thu, Apr 2, 1:00 PM · Restricted Project
fhahn updated the diff for D76793: [Matrix] Implement + and - operators for MatrixType..

Implement conversions for matrix/scalar variants.

Thu, Apr 2, 1:00 PM · Restricted Project
fhahn updated the diff for D76612: [Matrix] Add draft specification for matrix support in Clang..

Update arithmetic conversion rules after recent discussion on cfe-dev.

Thu, Apr 2, 7:01 AM · Restricted Project
fhahn committed rGa63b5c9e5319: [CallSiteSplitting] Simplify isPredicateOnPHI & continue checking PHIs. (authored by fhahn).
[CallSiteSplitting] Simplify isPredicateOnPHI & continue checking PHIs.
Thu, Apr 2, 2:41 AM
fhahn closed D77089: [CallSiteSplitting] Simplify isPredicateOnPHI & continue checking PHIs..
Thu, Apr 2, 2:41 AM · Restricted Project

Wed, Apr 1

fhahn added a comment to D76611: [SCCP] Use ranges for predicate info conditions..

Code looks fine.

I'm a little concerned that the recent clarification in LangRef doesn't actually match the code at the moment, so this could end up exposing an issue with some other transform. (grep for Attribute::SanitizeMemory.)

I assume you are referring to code in other passes, right? I think we probably have code assuming both the new behaviour and the old one and we somehow have to consolidate the code base. Probably best to discuss that at the langref change/llvm-Dev?

Wed, Apr 1, 2:13 PM · Restricted Project
fhahn added inline comments to D73763: [DSE] Lift post-dominance restriction..
Wed, Apr 1, 12:22 PM · Restricted Project
fhahn updated the diff for D73763: [DSE] Lift post-dominance restriction..

Avoid inserting calls by pre-computing set of the last memorydefs or phis before a function exit.

Wed, Apr 1, 12:04 PM · Restricted Project
fhahn updated the diff for D76791: [Matrix] Implement matrix index expressions ([][])..

Use placeholder type for incomplete matrix index expressions, as suggested by @rjmccall

Wed, Apr 1, 9:53 AM · Restricted Project
fhahn updated the diff for D76611: [SCCP] Use ranges for predicate info conditions..

Rebased after adding constantrange_including_undef to ValueLattice.

Wed, Apr 1, 7:07 AM · Restricted Project
fhahn added a comment to D77211: [LSR] do not replace PHI nodes if its scev is SCEVUnknown.

I think it would be helpful to describe what goes wrong in the patch description.

Wed, Apr 1, 4:58 AM · Restricted Project
fhahn added a comment to D76193: [ValueTracking] Use assumptions in computeConstantRange..

That's a neat site. It looks very useful. Do you know if there's any more information about the site? Is there a way to also check the patches together with D76228?

With respect to the regression, I am a bit surprised that there are seem to be quite a few assumptions encoded using assume() from clang. I'll check.

If you push a branch that starts with perf/ to your fork, it should get picked up. I've also done a run with the additional patch: Individual commits and all three combined.

Wed, Apr 1, 4:24 AM · Restricted Project
fhahn updated the diff for D76228: [ValueTracking] Use Inst::comesBefore in isValidAssumeForCtx (NFC)..

Generally, I think this code is a bit convoluted, and I think it would be clearer to rewrite along these lines.

Wed, Apr 1, 4:24 AM · Restricted Project
fhahn added inline comments to D76149: [AssumeBundles] Use assume bundles in isKnownNonZero.
Wed, Apr 1, 3:18 AM · Restricted Project
fhahn planned changes to D76992: [VPlan] Add & use VPValue operands for VPWidenRecipe (NFC)..
In D76992#1954253, @Ayal wrote:>

Yes, let's go for this cleaner solution. The first phase of VPlan integration using recipes is focused on refactoring ILV logic so that decisions are made and recorded in VPlan rather than during codegen. The latter, taken care of by recipe::execute(), should be as straightforward as possible.

Wed, Apr 1, 2:50 AM · Restricted Project
fhahn added inline comments to D77089: [CallSiteSplitting] Simplify isPredicateOnPHI & continue checking PHIs..
Wed, Apr 1, 2:43 AM · Restricted Project
fhahn committed rGd307174e1d9e: [ConstantRange] Use APInt::or/APInt::and for single elements. (authored by fhahn).
[ConstantRange] Use APInt::or/APInt::and for single elements.
Wed, Apr 1, 2:11 AM
fhahn closed D76446: [ConstantRange] Use APInt::or/APInt::and for single elements..
Wed, Apr 1, 2:10 AM · Restricted Project
fhahn committed rGe20cac365025: [Matrix] Add new test case with getelementptr constant exprs. (authored by fhahn).
[Matrix] Add new test case with getelementptr constant exprs.
Wed, Apr 1, 1:44 AM
fhahn committed rG862766e01e90: [Verifier] Verify matrix dimensions operands match vector size. (authored by fhahn).
[Verifier] Verify matrix dimensions operands match vector size.
Wed, Apr 1, 1:43 AM
fhahn closed D77129: [Verifier] Verify matrix dimensions operands match vector size..
Wed, Apr 1, 1:42 AM · Restricted Project
fhahn added a comment to D76446: [ConstantRange] Use APInt::or/APInt::and for single elements..

I'd like to see a unit test that exhaustively tests this for all binary operators (or the ones where it holds). I don't think single element ranges are handled accurately for all the ops (e.g. probably not for urem/srem?)

Yes, ideally we would handle all single element ranges in ConstantRange::binaryOp directly, by delegating to APInt, rather than doing it separately for each one I think. Maybe we should add an APInt::binaryOp?

I don't think that's possible due to layering. APInt is in Support, so it can't depend on IR.

Or just dispatch in ConstantRange::binaryOp?

We probably don't want to make binaryOp() more powerful than the individual methods, so I think your current approach is already fine.

Wed, Apr 1, 1:36 AM · Restricted Project

Tue, Mar 31

fhahn added a comment to D73763: [DSE] Lift post-dominance restriction..

Thanks for working on this!

+1 to the general idea of "adding a read-everything intrinsic doesn't smell great," and a further +1 for containing it to only this pass if we do end up needing it.

Tue, Mar 31, 3:27 PM · Restricted Project
fhahn committed rGb0cd7b2799eb: [SCCP] Limit use of range info for binops to integers for now. (authored by fhahn).
[SCCP] Limit use of range info for binops to integers for now.
Tue, Mar 31, 9:24 AM
fhahn added a comment to D71938: [SCCP] Use constant ranges for casts..

I'm now getting an assertion failure in the LNT test suite:
http://lab.llvm.org:8011/builders/clang-s390x-linux-lnt/builds/17826/steps/test-suite/logs/stdio

clang: /home/uweigand/sandbox/buildbot/clang-s390x-linux-lnt/llvm/llvm/lib/IR/Constants.cpp:1661: static llvm::Constant* llvm::ConstantExpr::getCast(unsigned int, llvm::Constant*, llvm::Type*, bool): Assertion `CastInst::castIsValid(opc, C, Ty) && "Invalid constantexpr cast!"' failed.
PLEASE submit a bug report to https://bugs.llvm.org/ and include the crash backtrace, preprocessed source, and associated run script.
Stack dump:
0. Program arguments: /home/uweigand/sandbox/buildbot/clang-s390x-linux-lnt/stage1.install/bin/clang -DNDEBUG -O3 -DNDEBUG -w -Werror=date-time -w -o CMakeFiles/GCC-C-execute-scal-to-vec1.dir/scal-to-vec1.c.o -c /home/uweigand/sandbox/buildbot/clang-s390x-linux-lnt/test/test-suite/SingleSource/Regression/C/gcc-c-torture/execute/scal-to-vec1.c

  1. <eof> parser at end of file
  2. Per-module optimization passes
  3. Running pass 'CallGraph Pass Manager' on module '/home/uweigand/sandbox/buildbot/clang-s390x-linux-lnt/test/test-suite/SingleSource/Regression/C/gcc-c-torture/execute/scal-to-vec1.c'.
  4. Running pass 'Sparse Conditional Constant Propagation' on function '@main' [...] #9 0x000002aa1a5b6b70 llvm::ConstantExpr::getCast(unsigned int, llvm::Constant*, llvm::Type*, bool) (/home/uweigand/sandbox/buildbot/clang-s390x-linux-lnt/stage1.install/bin/clang+0x2ab6b70) #10 0x000002aa19ec233c llvm::ConstantFoldCastOperand(unsigned int, llvm::Constant*, llvm::Type*, llvm::DataLayout const&) (/home/uweigand/sandbox/buildbot/clang-s390x-linux-lnt/stage1.install/bin/clang+0x23c233c) #11 0x000002aa1ac69ab0 (anonymous namespace)::SCCPSolver::visitCastInst(llvm::CastInst&) (/home/uweigand/sandbox/buildbot/clang-s390x-linux-lnt/stage1.install/bin/clang+0x3169ab0) #12 0x000002aa1ac6ce72 llvm::InstVisitor<(anonymous namespace)::SCCPSolver, void>::visit(llvm::Instruction&) (/home/uweigand/sandbox/buildbot/clang-s390x-linux-lnt/stage1.install/bin/clang+0x316ce72) [...]

    that appears to be directly related to this visitCastInst change.
Tue, Mar 31, 8:54 AM · Restricted Project
fhahn added inline comments to D70456: [Matrix] Add first set of matrix intrinsics and initial lowering pass..
Tue, Mar 31, 8:17 AM · Restricted Project
fhahn abandoned D76428: [cmake] Disable C4129 warning for MSVC..

I've moved the expected raw string literals to variables in rG6120cb42f797. It looks like that was enough to silence the warning at least for the llvm-clang-x86_64-expensive-checks-win (http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win/builds/22891). Please let me know if there are any other configurations where this did not fix the warning.

Tue, Mar 31, 6:04 AM · Restricted Project
fhahn committed rG6120cb42f797: [VPlan] Move test strs to variables, silence spurious C4129 (NFC). (authored by fhahn).
[VPlan] Move test strs to variables, silence spurious C4129 (NFC).
Tue, Mar 31, 5:32 AM
fhahn added a comment to D76992: [VPlan] Add & use VPValue operands for VPWidenRecipe (NFC)..

Thanks for following-up quickly. The part that extends VPWidenRecipe to have a User for holding its VPValue operands looks ok, but could the patch make sure that these operands are indeed used correctly by ILV::widenInstruction(), instead of moving the latter over to its execute()? The former may involve refactoring the case that deals with Calls, to record cost-based UseVectorInstrinsic/NeedToScalarize decisions when constructing the VPlan/recipe rather than during codegen; to record InvariantCond of Selects instead of accessing getSCEV(I.getOperand(0)) during codegen, possibly cleaning up to generate ScalarCond OR Cond but not both. In any case, accesses to I’s operands should all change to access User's operands instead, possibly with their underlying ingredient if needed.

Tue, Mar 31, 5:31 AM · Restricted Project
fhahn committed rGb37543750c76: [ValueLattice] Distinguish between constant ranges with/without undef. (authored by fhahn).
[ValueLattice] Distinguish between constant ranges with/without undef.
Tue, Mar 31, 4:59 AM
fhahn closed D76931: [ValueLattice] Distinguish between constant ranges with/without undef..
Tue, Mar 31, 4:59 AM · Restricted Project
fhahn created D77129: [Verifier] Verify matrix dimensions operands match vector size..
Tue, Mar 31, 3:18 AM · Restricted Project
fhahn committed rG0c9c58ada0af: [SCCP] Use constant ranges for casts. (authored by fhahn).
[SCCP] Use constant ranges for casts.
Tue, Mar 31, 1:37 AM
fhahn closed D71938: [SCCP] Use constant ranges for casts..
Tue, Mar 31, 1:37 AM · Restricted Project

Mon, Mar 30

fhahn added inline comments to D39137: Add CallSiteSplitting pass.
Mon, Mar 30, 2:11 PM
fhahn created D77089: [CallSiteSplitting] Simplify isPredicateOnPHI & continue checking PHIs..
Mon, Mar 30, 2:11 PM · Restricted Project
fhahn updated the summary of D76931: [ValueLattice] Distinguish between constant ranges with/without undef..
Mon, Mar 30, 10:49 AM · Restricted Project
fhahn updated the diff for D76931: [ValueLattice] Distinguish between constant ranges with/without undef..

I've adjusted the defaults after rG05f0e598ab26: per default, markConstantRange & co assume that the constant range does not contain undef (existing behavior) and isConstantRange/getConstantRange allow constant ranges including undef (existing behavior).

Mon, Mar 30, 10:49 AM · Restricted Project
fhahn committed rG7899a111ea11: Revert "[Darwin] Respect -fno-unroll-loops during LTO." (authored by fhahn).
Revert "[Darwin] Respect -fno-unroll-loops during LTO."
Mon, Mar 30, 8:04 AM
fhahn added a reverting change for rG9ce198d6ed37: [Darwin] Respect -fno-unroll-loops during LTO.: rG7899a111ea11: Revert "[Darwin] Respect -fno-unroll-loops during LTO.".
Mon, Mar 30, 8:04 AM
fhahn added a comment to D76916: [Darwin] Respect -fno-unroll-loops during LTO..

Thanks for taking a look @dexonsmith!

Mon, Mar 30, 8:03 AM · Restricted Project
fhahn created D77058: [Clang] Add llvm.loop.unroll.disable to loops with -fno-unroll-loops..
Mon, Mar 30, 7:33 AM · Restricted Project
fhahn added inline comments to D76325: [Matrix] Add option to use row-major matrix layout as default..
Mon, Mar 30, 5:55 AM · Restricted Project
fhahn updated the diff for D76325: [Matrix] Add option to use row-major matrix layout as default..

Adjust documentation in a few more places, add assertions to make sure operands of binary ops agree on the layout.

Mon, Mar 30, 5:22 AM · Restricted Project
fhahn updated the diff for D76325: [Matrix] Add option to use row-major matrix layout as default..

TODO: additional tests.

Is this WIP?

Mon, Mar 30, 4:50 AM · Restricted Project
fhahn committed rG9e81249d7614: [Matrix] Rename emitChainedMatrixMultiply to emitMatrixMultiply (NFC). (authored by fhahn).
[Matrix] Rename emitChainedMatrixMultiply to emitMatrixMultiply (NFC).
Mon, Mar 30, 3:45 AM
fhahn added inline comments to D75566: [Matrix] Add initial tiling for load/multiply/store chains..
Mon, Mar 30, 3:44 AM · Restricted Project
fhahn updated the diff for D75566: [Matrix] Add initial tiling for load/multiply/store chains..

Remove unnecessary updates, use DT::dominates in stead of OrderedInstructions, after DT::dominates now uses the BB local numbering recently committed.

Mon, Mar 30, 3:44 AM · Restricted Project
fhahn committed rGc3b03f3d0c35: [AMDGPU] Drop const for value that is copied (NFC). (authored by fhahn).
[AMDGPU] Drop const for value that is copied (NFC).
Mon, Mar 30, 3:13 AM
fhahn closed D77011: [AMDGPU] Drop const for value that is copied (NFC_..
Mon, Mar 30, 3:13 AM · Restricted Project
fhahn committed rG84c1fbab5d76: [CVP] Add additional icmp for ranges with undef to test. (authored by fhahn).
[CVP] Add additional icmp for ranges with undef to test.
Mon, Mar 30, 3:13 AM

Sun, Mar 29

fhahn updated the diff for D75566: [Matrix] Add initial tiling for load/multiply/store chains..

Adjust bb numbers in tests after change to creating BBs first.

Sun, Mar 29, 1:23 PM · Restricted Project
fhahn updated the diff for D75566: [Matrix] Add initial tiling for load/multiply/store chains..

Thanks Adam.

Sun, Mar 29, 12:50 PM · Restricted Project
fhahn added a comment to D76976: [NewGVN] Make NewGVN aware of aligned_alloc.

Please prefix the commit message with something like [NewGVN]

Sun, Mar 29, 10:40 AM · Restricted Project
fhahn committed rG99913ef3d14f: [OpenMP] set_bits iterator yields unsigned elements, no reference (NFC). (authored by fhahn).
[OpenMP] set_bits iterator yields unsigned elements, no reference (NFC).
Sun, Mar 29, 9:36 AM
fhahn closed D77010: [OpenMP] set_bits iterator yields unsigned elements, no reference (NFC)..
Sun, Mar 29, 9:36 AM · Restricted Project
fhahn added inline comments to D71938: [SCCP] Use constant ranges for casts..
Sun, Mar 29, 9:36 AM · Restricted Project
fhahn updated the diff for D71938: [SCCP] Use constant ranges for casts..

Thanks Eli!

Sun, Mar 29, 9:36 AM · Restricted Project
fhahn committed rG49d00824bbbb: [VPlan] Use one VPWidenRecipe per original IR instruction. (NFC). (authored by fhahn).
[VPlan] Use one VPWidenRecipe per original IR instruction. (NFC).
Sun, Mar 29, 5:51 AM
fhahn closed D76988: [VPlan] Use on VPWidenRecipe per original IR instruction. (NFC)..
Sun, Mar 29, 5:51 AM · Restricted Project
fhahn created D77011: [AMDGPU] Drop const for value that is copied (NFC_..
Sun, Mar 29, 5:50 AM · Restricted Project
fhahn created D77010: [OpenMP] set_bits iterator yields unsigned elements, no reference (NFC)..
Sun, Mar 29, 4:14 AM · Restricted Project

Sat, Mar 28

fhahn added a comment to D76973: [LangRef] Clarify the semantics of branch on undef.

We should probably also think about what to do with all the tests that contain branches on undef. Also, bugpoint should probably stop creating branches on undef with this spelled out as is?

Sat, Mar 28, 2:00 PM · Restricted Project
fhahn added a comment to D76428: [cmake] Disable C4129 warning for MSVC..

Thanks for all the input!

Sat, Mar 28, 11:49 AM · Restricted Project
fhahn added parent revisions for D76992: [VPlan] Add & use VPValue operands for VPWidenRecipe (NFC).: D74960: [VPlan] Provide required helpers through VPTransformState (NFC)., D76988: [VPlan] Use on VPWidenRecipe per original IR instruction. (NFC)..
Sat, Mar 28, 11:49 AM · Restricted Project
fhahn added a child revision for D76988: [VPlan] Use on VPWidenRecipe per original IR instruction. (NFC).: D76992: [VPlan] Add & use VPValue operands for VPWidenRecipe (NFC)..
Sat, Mar 28, 11:49 AM · Restricted Project
fhahn added a child revision for D74960: [VPlan] Provide required helpers through VPTransformState (NFC).: D76992: [VPlan] Add & use VPValue operands for VPWidenRecipe (NFC)..
Sat, Mar 28, 11:49 AM · Restricted Project
fhahn updated the diff for D74960: [VPlan] Provide required helpers through VPTransformState (NFC)..

Rebased.

Sat, Mar 28, 11:49 AM · Restricted Project
fhahn updated the diff for D76988: [VPlan] Use on VPWidenRecipe per original IR instruction. (NFC)..

Update test checking VPBB iterator, remove newline from printing.

Sat, Mar 28, 11:49 AM · Restricted Project
fhahn created D76992: [VPlan] Add & use VPValue operands for VPWidenRecipe (NFC)..
Sat, Mar 28, 11:49 AM · Restricted Project
fhahn committed rG81f173ed0e2c: [SCCP] Remove LatticeVal alias now that transition is done (NFC). (authored by fhahn).
[SCCP] Remove LatticeVal alias now that transition is done (NFC).
Sat, Mar 28, 9:08 AM
fhahn committed rGa44bf59c93b4: [SCCP] Remove unused toLatticeValue helper (NFC). (authored by fhahn).
[SCCP] Remove unused toLatticeValue helper (NFC).
Sat, Mar 28, 9:08 AM
fhahn created D76988: [VPlan] Use on VPWidenRecipe per original IR instruction. (NFC)..
Sat, Mar 28, 7:32 AM · Restricted Project

Fri, Mar 27

fhahn committed rG9ce198d6ed37: [Darwin] Respect -fno-unroll-loops during LTO. (authored by fhahn).
[Darwin] Respect -fno-unroll-loops during LTO.
Fri, Mar 27, 3:27 PM
fhahn closed D76916: [Darwin] Respect -fno-unroll-loops during LTO..
Fri, Mar 27, 3:27 PM · Restricted Project
fhahn added inline comments to D76941: [Module] Track intrinsic IDs in module to skip unnecessary lowering..
Fri, Mar 27, 3:27 PM · Restricted Project
fhahn added a comment to D71938: [SCCP] Use constant ranges for casts..

ping :)

Fri, Mar 27, 3:27 PM · Restricted Project
fhahn added a comment to D76327: [Matrix] Also run LowerMatrixIntrinsics for -O0..

We could make the Module automatically compute a map from intrinsics to overloads of that intrinsic in the module, I guess. Probably not too expensive to maintain.

Fri, Mar 27, 11:28 AM · Restricted Project
fhahn created D76941: [Module] Track intrinsic IDs in module to skip unnecessary lowering..
Fri, Mar 27, 11:28 AM · Restricted Project
fhahn added a comment to D76931: [ValueLattice] Distinguish between constant ranges with/without undef..

Hi,
The semantics of whether br undef is nondet. choice or UB really has been a topic between me and Nuno too.
As @nikic said, we can define it as UB if which branch to take depends on an undef value.

Fri, Mar 27, 11:28 AM · Restricted Project
fhahn added a comment to D75566: [Matrix] Add initial tiling for load/multiply/store chains..

ping

Fri, Mar 27, 9:07 AM · Restricted Project
fhahn added a comment to D76611: [SCCP] Use ranges for predicate info conditions..

If I'm understanding correctly, you're saying that SCCP doesn't care about the difference between constantrange and a hypothetical constantrangefromundef, therefore it's fine if the value is actually undef? I guess that's reasonable. But it would be more clear if we represented it explicitly in the lattice.

Fri, Mar 27, 9:05 AM · Restricted Project
fhahn created D76931: [ValueLattice] Distinguish between constant ranges with/without undef..
Fri, Mar 27, 9:05 AM · Restricted Project
fhahn created D76916: [Darwin] Respect -fno-unroll-loops during LTO..
Fri, Mar 27, 4:17 AM · Restricted Project

Thu, Mar 26

fhahn updated the diff for D76612: [Matrix] Add draft specification for matrix support in Clang..

Update according to comments on cfe-dev.

Thu, Mar 26, 3:47 PM · Restricted Project
fhahn added inline comments to D73763: [DSE] Lift post-dominance restriction..
Thu, Mar 26, 2:08 PM · Restricted Project
fhahn updated the diff for D73763: [DSE] Lift post-dominance restriction..

Thanks for taking a look!

Thu, Mar 26, 2:08 PM · Restricted Project