forked from OSchip/llvm-project
75cf30918f
This will likely introduce catastrophic performance regressions on older subtargets, but should be correct. A follow up change will remove the old fp32-denormals subtarget features, and switch to using the new denormal-fp-math/denormal-fp-math-f32 attributes. Frontends should be making sure to add the denormal-fp-math-f32 attribute when appropriate to avoid performance regressions. |
||
|---|---|---|
| .. | ||
| expected-target-index-name.mir | ||
| intrinsics.mir | ||
| invalid-target-index-operand.mir | ||
| lit.local.cfg | ||
| llc-target-cpu-attr-from-cmdline-ir.mir | ||
| llc-target-cpu-attr-from-cmdline.mir | ||
| load-store-opt-dlc.mir | ||
| machine-function-info-no-ir.mir | ||
| machine-function-info-register-parse-error1.mir | ||
| machine-function-info-register-parse-error2.mir | ||
| machine-function-info.ll | ||
| mfi-frame-offset-reg-class.mir | ||
| mfi-parse-error-frame-offset-reg.mir | ||
| mfi-parse-error-scratch-rsrc-reg.mir | ||
| mfi-parse-error-stack-ptr-offset-reg.mir | ||
| mfi-scratch-rsrc-reg-reg-class.mir | ||
| mfi-stack-ptr-offset-reg-class.mir | ||
| mir-canon-multi.mir | ||
| mircanon-memoperands.mir | ||
| parse-order-reserved-regs.mir | ||
| stack-id.mir | ||
| syncscopes.mir | ||
| target-flags.mir | ||
| target-index-operands.mir | ||