Sanjoy Das 7d237893d7 [BypassSlowDivision] Improve our handling of divisions by constants ...

(This reapplies r314253.  r314253 was reverted on r314482 because of a
correctness regression on P100, but that regression was identified to be
something else.)

Summary:
Don't bail out on constant divisors for divisions that can be narrowed without
introducing control flow .  This gives us a 32 bit multiply instead of an
emulated 64 bit multiply in the generated PTX assembly.

Reviewers: jlebar

Subscribers: jholewinski, mcrosier, llvm-commits

Differential Revision: https://reviews.llvm.org/D38265

llvm-svn: 319677

2017-12-04 19:21:58 +00:00

..

bypass-slow-div-constant-numerator.ll

[BypassSlowDivision] Handle division by constant numerators better.

2016-11-16 00:44:47 +00:00

bypass-slow-div-not-exact.ll

…

bypass-slow-div-special-cases.ll

[BypassSlowDivision] Do not bypass division of hash-like values

2017-04-02 13:14:30 +00:00

bypass-slow-div.ll

[BypassSlowDivision] Improve our handling of divisions by constants

2017-12-04 19:21:58 +00:00

dont-sink-nop-addrspacecast.ll

[CodeGenPrepare] Don't sink non-cheap addrspacecasts.

2016-11-21 22:49:15 +00:00

lit.local.cfg

Add missing lit.local.cfg to llvm/test/Transforms/CodeGenPrepare/NVPTX.

2016-10-28 21:56:07 +00:00