Page MenuHomePhabricator

jakev (Jake VanAdrighem)
User

Projects

User does not belong to any projects.

User Details

User Since
May 24 2016, 7:08 PM (219 w, 3 d)

Recent Activity

Jul 22 2016

jakev added a comment to D22607: [Profile] Don't use the override API in all cases..

Sean or David, do you think you could commit the pair of patches for me? I don't yet have commit access so I would appreciate that very much.

Jul 22 2016, 7:57 PM

Jul 20 2016

jakev updated D22607: [Profile] Don't use the override API in all cases..
Jul 20 2016, 8:33 PM
jakev retitled D22608: [Profile] Use a flag to enable PGO rather than the profraw filename. from to [Profile] Use a flag to enable PGO rather than the profraw filename..
Jul 20 2016, 8:32 PM
jakev retitled D22607: [Profile] Don't use the override API in all cases. from to [Profile] Don't use the override API in all cases..
Jul 20 2016, 8:31 PM
jakev added a comment to D22600: [PGO] Fix profile mismatch in Comdat function with pre-inliner.

Just a couple nits from me.

Jul 20 2016, 5:38 PM

Jul 18 2016

jakev abandoned D22030: [PGO] Change the ProfileSummary metadata merge behavior.
Jul 18 2016, 8:09 PM

Jul 15 2016

jakev added a comment to D21823: [Driver] Add flags for enabling both types of PGO Instrumentation.

Fantastic. Sean or David, could one of you commit this for me? It would be much appreciated.

Jul 15 2016, 6:19 PM

Jul 14 2016

jakev added inline comments to D21823: [Driver] Add flags for enabling both types of PGO Instrumentation.
Jul 14 2016, 3:46 PM
jakev updated the diff for D21823: [Driver] Add flags for enabling both types of PGO Instrumentation.

Add a couple notes to the docs and a fixme to a test. We can more thoroughly fix up the docs in a separate patch.

Jul 14 2016, 3:46 PM

Jul 11 2016

jakev updated the diff for D21823: [Driver] Add flags for enabling both types of PGO Instrumentation.

Change patch to use -fprofile-generate to enable IRPGO.

Jul 11 2016, 7:55 PM
jakev added a comment to D22030: [PGO] Change the ProfileSummary metadata merge behavior.

I am trying to understand the use case here. Do you need to support a case where you compile different modules with different profiles and LTO them together? Isn't merging the profiles and using the merged profile an option there?

Apologies for the delay.

Jul 11 2016, 7:50 PM
jakev added a comment to D22028: [PGO] Don't include full file path in static function profile counter names.

Sean or David, could one of you please commit this for me, I don't have commit access. Thanks!

Jul 11 2016, 7:35 PM

Jul 8 2016

jakev updated the diff for D22028: [PGO] Don't include full file path in static function profile counter names.

Added a flag for controlling the behavior of this. Keeping the name short but informative makes the flag name somewhat weird but just let me know if you have some preference on the naming.

Jul 8 2016, 6:58 PM
jakev updated the diff for D21823: [Driver] Add flags for enabling both types of PGO Instrumentation.

Reduce instrumentation flags to fpgo-train and change profile output file flag to fpgo-train-output.

Jul 8 2016, 6:27 PM

Jul 6 2016

jakev added a comment to D22030: [PGO] Change the ProfileSummary metadata merge behavior.

All the modules are expected to have the same profile summary and changing the behavior to a warning is not the right thing to do. If we want to support a use case where we LTO files with different profile data, perhaps we should have a way to disable profile summary for those builds? That seems better than picking the first summary we see (as would be the case with a warning)

Jul 6 2016, 8:16 PM

Jul 5 2016

jakev updated subscribers of D22028: [PGO] Don't include full file path in static function profile counter names.
Jul 5 2016, 8:19 PM
jakev retitled D22030: [PGO] Change the ProfileSummary metadata merge behavior from to [PGO] Change the ProfileSummary metadata merge behavior.
Jul 5 2016, 8:17 PM
jakev retitled D22028: [PGO] Don't include full file path in static function profile counter names from to [PGO] Don't include full file path in static function profile counter names.
Jul 5 2016, 6:35 PM

Jul 1 2016

jakev updated D21823: [Driver] Add flags for enabling both types of PGO Instrumentation.
Jul 1 2016, 1:42 PM
jakev added a comment to D21823: [Driver] Add flags for enabling both types of PGO Instrumentation.
In D21823#470516, @xur wrote:

(1) It seems the original options of -fprofile-instr-generate and -fprofile-instr-use will be kept. Is so, what is the guideline to use these two sets of options.

Jul 1 2016, 1:41 PM
jakev updated the diff for D21823: [Driver] Add flags for enabling both types of PGO Instrumentation.

Don't allow -fpgo-train and -fpgo-apply together. Add -fpgo-train-default-output=* to set the default profile output file.

Jul 1 2016, 1:39 PM

Jun 29 2016

jakev updated the diff for D21823: [Driver] Add flags for enabling both types of PGO Instrumentation.

Address comments.

Jun 29 2016, 1:47 PM

Jun 28 2016

jakev retitled D21823: [Driver] Add flags for enabling both types of PGO Instrumentation from to [Driver] Add flags for enabling both types of PGO Instrumentation.
Jun 28 2016, 5:20 PM

Jun 20 2016

jakev added a comment to D21546: Fix PR28219: Use profile summary from reader and not compute it.

Scratch that, please ignore my previous comment. My own naivety and an IR file with stale ProfileSummary info in the build was causing the error. This patch does in fact fix the problem when not using IR/bitcode files in that way.

Jun 20 2016, 11:07 PM
jakev added a comment to D21546: Fix PR28219: Use profile summary from reader and not compute it.

So even after applying this patch, IRMover is still emitting an error. I
believe the root cause of this is that we do a pointer based comparison of
the metadata values. This has worked until now when we have constant values
because we unique them in the IR, but doesn't for the ProfileSummary value
which references another metadata entry. If you could, please take a look
and let me know if I'm seeing things correctly.

Jun 20 2016, 8:14 PM

May 26 2016

jakev updated the diff for D20610: [TRE] Slight refactor and cleanup..

Moved the markTails stuff out of runOnFunction so we don't have to duplicate this when we port to the new PM.

May 26 2016, 7:07 PM
jakev updated the diff for D20610: [TRE] Slight refactor and cleanup..

Addressed dcci's comments.

May 26 2016, 1:57 PM

May 24 2016

jakev retitled D20610: [TRE] Slight refactor and cleanup. from to [TRE] Slight refactor and cleanup..
May 24 2016, 7:21 PM