Page MenuHomePhabricator

chrisdiamand_arm (Chris Diamand)
User

Projects

User does not belong to any projects.

User Details

User Since
Jan 14 2016, 4:01 AM (195 w, 5 d)

Recent Activity

Jun 29 2016

chrisdiamand_arm added a comment to D17203: [LICM] Sink entire inner loops..

Hi - does anyone have any thoughts on this?

Jun 29 2016, 3:48 AM

Jun 23 2016

chrisdiamand_arm updated the diff for D17203: [LICM] Sink entire inner loops..

Rebase and fix conflicts with renaming 'LICMSafetyInfo' to 'LoopSafetyInfo'. Also ping :)

Jun 23 2016, 2:54 AM

Jun 1 2016

chrisdiamand_arm updated the diff for D17203: [LICM] Sink entire inner loops..

Add messages to assertions.

Jun 1 2016, 9:10 AM
chrisdiamand_arm added a comment to D17203: [LICM] Sink entire inner loops..

Replies inline - cheers!

Jun 1 2016, 7:37 AM
chrisdiamand_arm added a comment to D17203: [LICM] Sink entire inner loops..

Ping.

Jun 1 2016, 5:23 AM

May 25 2016

chrisdiamand_arm added a comment to D17203: [LICM] Sink entire inner loops..

Ping. Anyone?

May 25 2016, 9:20 AM

May 9 2016

chrisdiamand_arm updated the diff for D17203: [LICM] Sink entire inner loops..

Updating the diff after rebasing on ToT (there was a conflict with some reformatting).

May 9 2016, 6:26 AM

Apr 11 2016

chrisdiamand_arm added a comment to D17203: [LICM] Sink entire inner loops..

Ping.

Apr 11 2016, 5:50 AM

Mar 24 2016

chrisdiamand_arm added a comment to D17203: [LICM] Sink entire inner loops..

Ping. Also (I should have mentioned this earlier), I've tested this with the regression tests, the LLVM test suite, and several proprietary benchmarks.

Mar 24 2016, 9:16 AM

Mar 11 2016

chrisdiamand_arm added a comment to D17203: [LICM] Sink entire inner loops..

Has anyone had a chance to look at this yet?

Mar 11 2016, 2:02 AM

Mar 4 2016

chrisdiamand_arm updated the diff for D17203: [LICM] Sink entire inner loops..

This now updates the DominatorTree incrementally, rather than recalculating the whole thing at the end of runOnLoop().

Mar 4 2016, 9:12 AM

Mar 1 2016

chrisdiamand_arm added inline comments to D17203: [LICM] Sink entire inner loops..
Mar 1 2016, 11:32 AM
chrisdiamand_arm updated the diff for D17203: [LICM] Sink entire inner loops..

This update removes the hoisting code, in a bid to make the diff a bit more manageable.

Mar 1 2016, 11:31 AM

Feb 24 2016

chrisdiamand_arm added a comment to D17203: [LICM] Sink entire inner loops..

Hi, thanks for taking a look at this! Comments inline.

Feb 24 2016, 8:05 AM

Feb 18 2016

chrisdiamand_arm updated subscribers of D12116: [AArch64] Improve load/store optimizer to handle LDUR + LDR..
Feb 18 2016, 5:30 AM

Feb 17 2016

chrisdiamand_arm updated the diff for D17203: [LICM] Sink entire inner loops..

This adds one '; CHECK-LABEL: @main(' line to each new test.

Feb 17 2016, 8:05 AM
chrisdiamand_arm added inline comments to D17203: [LICM] Sink entire inner loops..
Feb 17 2016, 7:29 AM
chrisdiamand_arm added a comment to D17203: [LICM] Sink entire inner loops..

Taking a step back, can you give a motivating example on why we might want to do this? Your tests look like they'd be caught by loop-unswitch and LICM together, but I suspect that's just because the tests are (correctly) simple.

Feb 17 2016, 5:33 AM
chrisdiamand_arm updated D17203: [LICM] Sink entire inner loops..
Feb 17 2016, 3:56 AM
chrisdiamand_arm updated the diff for D17203: [LICM] Sink entire inner loops..

Hopefully this address all your feedback, James.

Feb 17 2016, 3:54 AM

Feb 16 2016

chrisdiamand_arm added inline comments to D17203: [LICM] Sink entire inner loops..
Feb 16 2016, 4:02 AM

Feb 12 2016

chrisdiamand_arm retitled D17203: [LICM] Sink entire inner loops. from to [LICM] Hoist and sink entire inner loops..
Feb 12 2016, 9:36 AM

Feb 9 2016

chrisdiamand_arm updated subscribers of D17000: [AArch64] Reduce number of callee-save save/restores..
Feb 9 2016, 8:08 AM

Feb 4 2016

chrisdiamand_arm closed D16195: [AArch64] Multiply extended 32-bit ints with `[U|S]MADDL'..

r259800

Feb 4 2016, 8:55 AM
chrisdiamand_arm added a comment to D16195: [AArch64] Multiply extended 32-bit ints with `[U|S]MADDL'..

Thanks James. Tim - do you think this is ready to be merged?

Feb 4 2016, 7:00 AM

Feb 1 2016

chrisdiamand_arm updated subscribers of D16734: [AArch64] Simplify prolog/epilog callee save/restore. NFC..
Feb 1 2016, 1:58 AM

Jan 28 2016

chrisdiamand_arm added a comment to D16590: Refactor backend diagnostics for unsupported features.

The shared library build is still broken - as mentioned before, there's a circular dependency between CodeGen and SelectionDAG because of the Value->print(OS); line.

Jan 28 2016, 3:58 AM
chrisdiamand_arm updated subscribers of D16590: Refactor backend diagnostics for unsupported features.
Jan 28 2016, 3:44 AM

Jan 26 2016

chrisdiamand_arm added a comment to D16195: [AArch64] Multiply extended 32-bit ints with `[U|S]MADDL'..

Did anyone get a chance to look at the updated patch?

Jan 26 2016, 7:47 AM

Jan 18 2016

chrisdiamand_arm updated the diff for D16195: [AArch64] Multiply extended 32-bit ints with `[U|S]MADDL'..

Good point - I've updated the patch to handle those cases.

Jan 18 2016, 7:04 AM

Jan 14 2016

chrisdiamand_arm updated D16195: [AArch64] Multiply extended 32-bit ints with `[U|S]MADDL'..
Jan 14 2016, 9:15 AM
chrisdiamand_arm retitled D16195: [AArch64] Multiply extended 32-bit ints with `[U|S]MADDL'. from to [AArch64] Multiply extended 32-bit ints with `[U|S]MADDL'..
Jan 14 2016, 9:10 AM