Page MenuHomePhabricator
Feed Advanced Search

Jul 13 2017

steleman updated the diff for D35319: LSE Atomics reorg - Part I.

Updated and corrected AArch64DeadRegisterDefinitions::ShouldSkip.

Jul 13 2017, 12:01 PM
steleman added inline comments to D35319: LSE Atomics reorg - Part I.
Jul 13 2017, 11:13 AM
steleman added inline comments to D35319: LSE Atomics reorg - Part I.
Jul 13 2017, 10:19 AM
steleman added inline comments to D35319: LSE Atomics reorg - Part I.
Jul 13 2017, 8:59 AM
steleman added inline comments to D35319: LSE Atomics reorg - Part I.
Jul 13 2017, 8:44 AM
steleman added a comment to D35319: LSE Atomics reorg - Part I.

Hi Stefan,

I went ahead and committed D35309, so you will likely need to rebase the changes to AArch64InstrAtomics.td and AArch64DeadRegisterDefinitionsPass.cpp. I would also recommend separating out the ThunderX scheduling changes into their own patch, at least when committing.

Jul 13 2017, 8:38 AM
steleman added inline comments to D35319: LSE Atomics reorg - Part I.
Jul 13 2017, 8:38 AM
steleman added inline comments to D35319: LSE Atomics reorg - Part I.
Jul 13 2017, 8:29 AM
steleman added inline comments to D35319: LSE Atomics reorg - Part I.
Jul 13 2017, 7:52 AM
steleman added a comment to D35319: LSE Atomics reorg - Part I.

This diff covers lots of different areas:

Jul 13 2017, 7:30 AM
steleman added inline comments to D35319: LSE Atomics reorg - Part I.
Jul 13 2017, 7:25 AM
steleman added a comment to D35319: LSE Atomics reorg - Part I.

If at all possible, I would like to see D35309 committed before/with this patch.

Jul 13 2017, 7:01 AM

Jul 12 2017

steleman created D35319: LSE Atomics reorg - Part I.
Jul 12 2017, 12:16 PM
steleman added a comment to D34958: [AArch64] Add AArch64Subtarget::isFusion function..

Ping. @joelkevinjones @steleman are you happy for this to go in?

I am running SPECint2K6 with and without your patch for comparison and
I will let you know as soon as I have all the numbers.

Jul 12 2017, 11:04 AM
steleman added a comment to D34958: [AArch64] Add AArch64Subtarget::isFusion function..

Ping. @joelkevinjones @steleman are you happy for this to go in?

Jul 12 2017, 9:10 AM

May 9 2017

steleman added a comment to D31801: Performance enhancements for Cavium ThunderX2 T99.

Ping2!! :-)

May 9 2017, 12:09 PM

May 3 2017

steleman added a comment to D31801: Performance enhancements for Cavium ThunderX2 T99.

Ping! :-)

May 3 2017, 9:55 PM

Apr 28 2017

steleman added a comment to D31801: Performance enhancements for Cavium ThunderX2 T99.

Right, I clicked the link and it took me to the old code. :)

I'm not sure how f16/f128 would fare there. Probably not well.

Apr 28 2017, 7:00 AM
steleman added inline comments to D31801: Performance enhancements for Cavium ThunderX2 T99.
Apr 28 2017, 6:45 AM

Apr 27 2017

steleman updated the diff for D31801: Performance enhancements for Cavium ThunderX2 T99.

Updated to the most recent LLVM git mirror from 2017/04/27.
Several regex and instruction improvements.

Apr 27 2017, 8:10 PM

Apr 25 2017

steleman updated the diff for D31801: Performance enhancements for Cavium ThunderX2 T99.

Updates to the T99 Scheduler.

Apr 25 2017, 11:52 PM

Apr 11 2017

steleman added inline comments to D31801: Performance enhancements for Cavium ThunderX2 T99.
Apr 11 2017, 2:55 PM
steleman added inline comments to D31801: Performance enhancements for Cavium ThunderX2 T99.
Apr 11 2017, 9:11 AM
steleman added a comment to D31801: Performance enhancements for Cavium ThunderX2 T99.

I will re-submit with changes shortly - I also have a few more additions to the *.td
file as well.

Apr 11 2017, 7:49 AM

Apr 6 2017

steleman created D31801: Performance enhancements for Cavium ThunderX2 T99.
Apr 6 2017, 8:25 PM

Mar 5 2017

steleman updated the diff for D30510: Vulcan is now ThunderX2T99.

Updated diff based on Renato's comments.

Mar 5 2017, 9:06 PM

Mar 2 2017

steleman added a comment to D30510: Vulcan is now ThunderX2T99.

Hi Renato,

Mar 2 2017, 8:40 AM

Mar 1 2017

steleman updated the diff for D30510: Vulcan is now ThunderX2T99.
Mar 1 2017, 10:42 PM
steleman added a comment to D30510: Vulcan is now ThunderX2T99.

Diff is based on LLVM trunk from Git mirror on 03/01/2017.

Mar 1 2017, 2:13 PM
steleman created D30510: Vulcan is now ThunderX2T99.
Mar 1 2017, 2:05 PM

Feb 13 2017

steleman updated the diff for D28891: add support for Cavium ThunderX ARM64 processors.

Sorted the ThunderX CPU names alphabetically in AArch64Subtarget.h.

Feb 13 2017, 9:47 AM

Feb 9 2017

steleman updated the diff for D28891: add support for Cavium ThunderX ARM64 processors.
  • Added tests for ARMV8.1-A LSE Extensions.
  • Added tests in unittests/Support/TargetParserTest.cpp
Feb 9 2017, 1:15 PM
steleman added a comment to D28891: add support for Cavium ThunderX ARM64 processors.

Hi, can you also please add the relevant tests to unittests/Support/TargetParserTest.cpp? It should be pretty obvious from the context.

Feb 9 2017, 7:16 AM

Feb 2 2017

steleman updated the diff for D28891: add support for Cavium ThunderX ARM64 processors.
Feb 2 2017, 8:45 PM
steleman added a comment to D28891: add support for Cavium ThunderX ARM64 processors.

This latest changeset is based on ToT from 02/02/2017.

Feb 2 2017, 3:39 PM
steleman updated the diff for D28891: add support for Cavium ThunderX ARM64 processors.

Fixes and corrections as per review comments.
Added test case for ARMV8.1-A and LSE (test/MC/AArch64/armv8.1a-lse.s).

Feb 2 2017, 3:30 PM

Jan 19 2017

steleman added a comment to D28891: add support for Cavium ThunderX ARM64 processors.

Hi:
Some comments specifically on the scheduler. Overall the sched-model looks good, although I don't have the detailed latency and micro-architecture information of ThunderX.

A few points you may want to consider :

  1. You could remove the lines with 'let ResourceCycles = [1];' as that's by-default. That would help you reduce size of descriptions in a number of places.
  2. Same for 'let Latency = 1;'
  3. I don't know the micro-arch of ThunderX, but is there a typo for the latency of THXT8XWriteVST3 as it exceeds corresponding resource-cycle for the same sched-class. You may want to just double-check that.
  4. ReadAdvance seems to exceed write latency in some cases, and so you may want to double check that - e.g. ReadI (2) verus WriteI (1).

    Best Regards Javed
Jan 19 2017, 9:37 AM

Jan 18 2017

steleman updated the diff for D28891: add support for Cavium ThunderX ARM64 processors.

Support for Cavium ThunderX T8X ARM64 processors.

Jan 18 2017, 10:12 PM
steleman updated the diff for D28891: add support for Cavium ThunderX ARM64 processors.
Jan 18 2017, 10:08 PM
steleman updated the diff for D28891: add support for Cavium ThunderX ARM64 processors.
Jan 18 2017, 10:07 PM
steleman updated the diff for D28891: add support for Cavium ThunderX ARM64 processors.
Jan 18 2017, 10:07 PM
steleman updated the diff for D28891: add support for Cavium ThunderX ARM64 processors.
Jan 18 2017, 10:06 PM
steleman created D28891: add support for Cavium ThunderX ARM64 processors.
Jan 18 2017, 10:05 PM

Sep 3 2015

steleman added a comment to D12598: clang should emulate GCC's -fabi-version=<X>.

The diff is based on 3.6.2.

Sep 3 2015, 6:11 AM
steleman retitled D12598: clang should emulate GCC's -fabi-version=<X> from to clang should emulate GCC's -fabi-version=<X>.
Sep 3 2015, 6:04 AM