search for: icmpzero

...k at apple.com] >>> Sent: 18 January 2013 22:23 >>> To: Manish Verma >>> Cc: 'Dmitri Gribenko'; Renato Golin; llvm-commits at cs.uiuc.edu >>> Subject: Re: [llvm-commits] [llvm] r172534 - >>> /llvm/trunk/test/Transforms/LoopStrengthReduce/post-inc-icmpzero.ll >>> >>> Manish, >>> >>> For some reason, the arm-hosted LSR does not seem to recognize -1 as >>> a legal address here. >>> >>> I'm baffled at how a fundamental problem would only show up in one >>> test, but as a quic...

...Trick [mailto:atrick at apple.com] >> Sent: 18 January 2013 22:23 >> To: Manish Verma >> Cc: 'Dmitri Gribenko'; Renato Golin; llvm-commits at cs.uiuc.edu >> Subject: Re: [llvm-commits] [llvm] r172534 - >> /llvm/trunk/test/Transforms/LoopStrengthReduce/post-inc-icmpzero.ll >> >> Manish, >> >> For some reason, the arm-hosted LSR does not seem to recognize -1 as >> a legal address here. >> >> I'm baffled at how a fundamental problem would only show up in one >> test, but as a quick sanity check you can add an un...

...p eq i32 %i.010, 0 Cost: 0 LSR has identified the following interesting factors and types: *8 LSR is examining the following fixup sites: UserInst=%cmp11, OperandValToReplace=%i.010 UserInst=%0, OperandValToReplace=%arrayidx LSR found 2 uses: LSR is examining the following uses: LSR Use: Kind=ICmpZero, Offsets={0}, widest fixup type: i32 reg({0,+,-1}<nw><%while.body>) LSR Use: Kind=Address of i32 in addrspace(0), Offsets={0}, widest fixup type: i32* reg({@buffer,+,8}<nsw><%while.body>) After generating reuse formulae: LSR is examining the following uses: LSR...

Hi, Transform/LoopStrengthReduction/post-inc-icmpzero.ll was failing for clang-native-arm-cortex-a9 build-bot configuration. The reason for the failure was the test was using hardcoded names. The attached patch fixes this failure by replacing the hard-coded variables names with pattern-matched variable names. I have checked that the test pass after a...

...ks for the patch, I'll check and commit. cheers, --renato PS: Patches are normally sent to llvm-commit list, unless they require a lot of discussion. On 15 January 2013 14:52, Manish Verma <manish.avtaar at gmail.com> wrote: > Hi, > > Transform/LoopStrengthReduction/post-inc-icmpzero.ll was failing for > clang-native-arm-cortex-a9 build-bot configuration. The reason for the > failure was the test was using hardcoded names. The attached patch fixes > this failure by replacing the hard-coded variables names with > pattern-matched variable names. > > I have check...

...eers, > --renato > > PS: Patches are normally sent to llvm-commit list, unless they require a > lot of discussion. > > > On 15 January 2013 14:52, Manish Verma <manish.avtaar at gmail.com> wrote: > >> Hi, >> >> Transform/LoopStrengthReduction/post-inc-icmpzero.ll was failing for >> clang-native-arm-cortex-a9 build-bot configuration. The reason for the >> failure was the test was using hardcoded names. The attached patch fixes >> this failure by replacing the hard-coded variables names with >> pattern-matched variable names. >&g...

...ollowing interesting factors and types: *8 >> LSR is examining the following fixup sites: >> UserInst=%cmp11, OperandValToReplace=%i.010 >> UserInst=%0, OperandValToReplace=%arrayidx >> LSR found 2 uses: >> LSR is examining the following uses: >> LSR Use: Kind=ICmpZero, Offsets={0}, widest fixup type: i32 >> reg({0,+,-1}<nw><%while.body>) >> LSR Use: Kind=Address of i32 in addrspace(0), Offsets={0}, widest fixup type: >> i32* >> reg({@buffer,+,8}<nsw><%while.body>) >> >> After generating reuse fo...

...t;>>> LSR is examining the following fixup sites: >>>> UserInst=%cmp11, OperandValToReplace=%i.010 >>>> UserInst=%0, OperandValToReplace=%arrayidx >>>> LSR found 2 uses: >>>> LSR is examining the following uses: >>>> LSR Use: Kind=ICmpZero, Offsets={0}, widest fixup type: i32 >>>> reg({0,+,-1}<nw><%while.body>) >>>> LSR Use: Kind=Address of i32 in addrspace(0), Offsets={0}, widest fixup >> type: >>>> i32* >>>> reg({@buffer,+,8}<nsw><%while.body>) >&g...

On Mon, Jan 21, 2013 at 11:49 AM, Andrew Trick <atrick at apple.com> wrote: > Moving to llvmdev... > > On Jan 21, 2013, at 11:37 AM, Nadav Rotem <nrotem at apple.com> wrote: > > > Hi Manish, > > > > Thank you for looking at this. Recently opt started using the "-mtriple" > command line flag to initialize the TargetMachine for IR-level >

On Jan 21, 2013, at 12:37 PM, Chandler Carruth <chandlerc at google.com> wrote: > On Mon, Jan 21, 2013 at 11:49 AM, Andrew Trick <atrick at apple.com> wrote: > Moving to llvmdev... > > On Jan 21, 2013, at 11:37 AM, Nadav Rotem <nrotem at apple.com> wrote: > > > Hi Manish, > > > > Thank you for looking at this. Recently opt started using the

On Jan 21, 2013, at 3:12 PM, Chandler Carruth <chandlerc at google.com> wrote: > On Mon, Jan 21, 2013 at 2:59 PM, Andrew Trick <atrick at apple.com> wrote: > > On Jan 21, 2013, at 12:37 PM, Chandler Carruth <chandlerc at google.com> wrote: > >> On Mon, Jan 21, 2013 at 11:49 AM, Andrew Trick <atrick at apple.com> wrote: >> Moving to llvmdev...

...(and my local box). The Clang one I think it's assuming an Intel box, the other two look like the FileCheck are not good enough. http://lab.llvm.org:8011/builders/clang-native-arm-cortex-a9/builds/4305 Clang :: CodeGen/compound-assign-overflow.c LLVM :: Transforms/LoopStrengthReduce/post-inc-icmpzero.ll LLVM :: Transforms/LoopStrengthReduce/2012-07-18-LimitReassociate.ll cheers, --renato -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20130108/f0cea63a/attachment.html>

On 8 January 2013 17:01, Dmitri Gribenko <gribozavr at gmail.com> wrote: > r171853 for this one. Build not finished yet. > http://lab.llvm.org:8011/builders/clang-native-arm-cortex-a9/builds/4312 Thanks!! > LLVM :: Transforms/LoopStrengthReduce/post-inc-icmpzero.ll > > Looks like a better regex can fix this. > I think both of them are just bad FileChecks... This is the first build that fails: http://lab.llvm.org:8011/builders/clang-native-arm-cortex-a9/builds/4294 And they haven't introduced the tests... Most of them are debug, there's...

...ttp://lab.llvm.org:8011/builders/clang-native-arm-cortex-a9/builds/4305 > > Clang :: CodeGen/compound-assign-overflow.c r171853 for this one. Build not finished yet. http://lab.llvm.org:8011/builders/clang-native-arm-cortex-a9/builds/4312 > LLVM :: Transforms/LoopStrengthReduce/post-inc-icmpzero.ll Looks like a better regex can fix this. Dmitri -- main(i,j){for(i=2;;i++){for(j=2;j<i;j++){if(!(i%j)){j=0;break;}}if (j){printf("%d\n",i);}}} /*Dmitri Gribenko <gribozavr at gmail.com>*/

On Tue, Jan 8, 2013 at 3:04 PM, Renato Golin <renato.golin at linaro.org> wrote: > The following failures are consistent on buildbot (and my local box). [...] > LLVM :: Transforms/LoopStrengthReduce/post-inc-icmpzero.ll > LLVM :: Transforms/LoopStrengthReduce/2012-07-18-LimitReassociate.ll It is interesting that I don't see this on my ARM box. Instead I see these: Failing Tests (5): LLVM :: ExecutionEngine/MCJIT/2003-01-04-ArgumentBug.ll LLVM :: ExecutionEngine/MCJIT/pr13727.ll LLVM :: Exe...

...ng one I think it's assuming an Intel box, the other two look like > the FileCheck are not good enough. > > http://lab.llvm.org:8011/builders/clang-native-arm-cortex-a9/builds/4305 > > Clang :: CodeGen/compound-assign-overflow.c > LLVM :: Transforms/LoopStrengthReduce/post-inc-icmpzero.ll > LLVM :: Transforms/LoopStrengthReduce/2012-07-18-LimitReassociate.ll Usually the best way to get traction on such things is to reply to the commit that caused the regression. Whoever broke things is usually more invested in making sure the change is solid (& doesn't get reverted)....

I have a few questions regarding cost calculation in loop strength reduction. 1. Ignoring LSR uses of kind ICmpZero. These usually come from the loop latch conditions. On Hexagon most loops that would contain such a LSR use are going to be converted to hardware loops and the entire check will disappear. There is no need to analyze these cases in LSR. In fact, completely ignoring them improves performance. Wo...

...to Golin Cc: Clang Dev; LLVM Dev Subject: Re: [cfe-dev] [LLVMdev] ARM failures On Tue, Jan 8, 2013 at 3:04 PM, Renato Golin <renato.golin at linaro.org> wrote: > The following failures are consistent on buildbot (and my local box). [...] > LLVM :: Transforms/LoopStrengthReduce/post-inc-icmpzero.ll > LLVM :: Transforms/LoopStrengthReduce/2012-07-18-LimitReassociate.ll It is interesting that I don't see this on my ARM box. Instead I see these: Failing Tests (5): LLVM :: ExecutionEngine/MCJIT/2003-01-04-ArgumentBug.ll LLVM :: ExecutionEngine/MCJIT/pr13727.ll LLVM :: Exe...