[PATCH] D80416: [RFC][OpenCL] Set fp contract flag on -cl-mad-enable

arsenm requested changes to this revision. arsenm added a comment. This revision now requires changes to proceed. Please rebase if still relevant CHANGES SINCE LAST ACTION https://reviews.llvm.org/D80416/new/ https://reviews.llvm.org/D80416 ___

[PATCH] D80416: [RFC][OpenCL] Set fp contract flag on -cl-mad-enable

arsenm added a comment. Herald added a subscriber: Naghasan. Herald added a project: All. Was this addressed already? Today it looks like we have: if (Opts.FastRelaxedMath || Opts.CLUnsafeMath) Opts.setDefaultFPContractMode(LangOptions::FPM_Fast); CHANGES SINCE LAST ACTION

[PATCH] D80416: [RFC][OpenCL] Set fp contract flag on -cl-mad-enable

arsenm added a comment. Is this still necessary? CHANGES SINCE LAST ACTION https://reviews.llvm.org/D80416/new/ https://reviews.llvm.org/D80416 ___ cfe-commits mailing list cfe-commits@lists.llvm.org

[PATCH] D80416: [RFC][OpenCL] Set fp contract flag on -cl-mad-enable

Anastasia marked an inline comment as done. Anastasia added inline comments. Comment at: clang/test/CodeGenOpenCL/relaxed-fpmath.cl:21 +float fused_mad(float a, float b, float c) { + // NORMAL: @llvm.fmuladd.f32 + // FAST: fmul fast float Anastasia wrote: >

[PATCH] D80416: [RFC][OpenCL] Set fp contract flag on -cl-mad-enable

arsenm added a comment. I think the contract flag needs clarification. I would interpret an instruction with only a contract flag as meaning allow precision increasing FMA formation, and contract+afn to mean combining while reducing precision CHANGES SINCE LAST ACTION

[PATCH] D80416: [RFC][OpenCL] Set fp contract flag on -cl-mad-enable

Anastasia marked an inline comment as done. Anastasia added inline comments. Comment at: clang/test/CodeGenOpenCL/relaxed-fpmath.cl:21 +float fused_mad(float a, float b, float c) { + // NORMAL: @llvm.fmuladd.f32 + // FAST: fmul fast float Anastasia wrote: >

[PATCH] D80416: [RFC][OpenCL] Set fp contract flag on -cl-mad-enable

Anastasia marked an inline comment as done. Anastasia added inline comments. Comment at: clang/test/CodeGenOpenCL/relaxed-fpmath.cl:21 +float fused_mad(float a, float b, float c) { + // NORMAL: @llvm.fmuladd.f32 + // FAST: fmul fast float Anastasia wrote: > I

[PATCH] D80416: [RFC][OpenCL] Set fp contract flag on -cl-mad-enable

Anastasia added a comment. In D80416#2050250 , @arsenm wrote: > The langref wording makes me think this isn't quite right. This depends on > your definition of floating point contraction. I've always assumed it meant > allow FMA, potentially increasing

[PATCH] D80416: [RFC][OpenCL] Set fp contract flag on -cl-mad-enable

arsenm added a comment. The langref wording makes me think this isn't quite right. This depends on your definition of floating point contraction. I've always assumed it meant allow FMA, potentially increasing precision. Is contracting into something less precise allowed? If not, that's

[PATCH] D80416: [RFC][OpenCL] Set fp contract flag on -cl-mad-enable

Anastasia marked an inline comment as done. Anastasia added inline comments. Comment at: clang/test/CodeGenOpenCL/relaxed-fpmath.cl:21 +float fused_mad(float a, float b, float c) { + // NORMAL: @llvm.fmuladd.f32 + // FAST: fmul fast float I don't find this

[PATCH] D80416: [RFC][OpenCL] Set fp contract flag on -cl-mad-enable

Anastasia created this revision. Anastasia added reviewers: rjmccall, arsenm. Herald added subscribers: ebevhan, yaxunl, wdng. Anastasia marked an inline comment as done. Anastasia added inline comments. Anastasia retitled this revision from "[RCF][OpenCL] Set fp contract flag on -cl-mad-enable"