In D71617#1921143, @uenoku wrote:I think the title should be more descriptive, like "improve noalias preservation using reachability" or something.
- Queries
- All Stories
- Search
- Advanced Search
- Transactions
- Transaction Logs
Feed Advanced Search
Advanced Search
Advanced Search
Mar 13 2020
Mar 13 2020
Mar 13 2020, 3:41 AM · Restricted Project
pgode retitled D71617: [Attributor] Improve noalias preservation using reachability from [Attributor] noalias attribute deduction fixme to [Attributor] Improve noalias preservation using reachability.
Mar 13 2020, 3:41 AM · Restricted Project
pgode retitled D71617: [Attributor] Improve noalias preservation using reachability from [WIP][NFC][Attributor] noalias attribute deduction fixme to [Attributor] noalias attribute deduction fixme.
Mar 13 2020, 2:15 AM · Restricted Project
In D71617#1921067, @jdoerfert wrote:@pgode Can you modify the title (this is not a WIP and not a NFC patch).
Mar 13 2020, 2:15 AM · Restricted Project
Mar 12 2020
Mar 12 2020
Updated the patch on top on Hideto's changes.
Added description to the test in noalias.ll.
Needed to increase the max-iterations for nonnull.ll
Mar 12 2020, 11:23 AM · Restricted Project
In D71617#1919565, @uenoku wrote:It was caused by a bug that the dependence track was missing for NoAliasAA so I have pushed the fix. (https://github.com/llvm/llvm-project/commit/d9bf79f4e9952acca1fa353e39bcee89cd69550f)
Mar 12 2020, 10:51 AM · Restricted Project
Mar 11 2020
Mar 11 2020
Additional conditions are included. Added new test.
Mar 11 2020, 1:35 AM · Restricted Project
Mar 9 2020
Mar 9 2020
In D71617#1911979, @jdoerfert wrote:@pgode Are you still looking into this? Do you need feedback from me?
Mar 9 2020, 7:30 AM · Restricted Project
Jan 12 2020
Jan 12 2020
Thanks for the explanation. I thought the exact opposite i.e. even though the use is potentially captured in the callee but the callsite of callee is reachable, so we propagate noalias. I apologize as I had basic incorrect understanding.
Jan 12 2020, 9:26 AM · Restricted Project
Jan 10 2020
Jan 10 2020
Thanks for reviewing further.
- Removed the Dependence flag.
- For the conditions to return 'true', in Pred: I observe that the correct condition should be below. This condition means that callsite_argument should not have nocapture and getCtxI() should be reachable from user. I think this, because for noalias propagation to happen w.r.t. (ii) condition, we should get true from checkForAllUses and Pred as well.
Jan 10 2020, 7:05 AM · Restricted Project
Jan 9 2020
Jan 9 2020
Updated patch which considers the review comments.
Jan 9 2020, 5:31 AM · Restricted Project
Jan 8 2020
Jan 8 2020
Thanks for reviewing. I understand as this was during year end.
I will make the changes and update the patch. I tried the changes locally and I think I need to further check why noalias is still propagated for other checks in pthread.ll .
Jan 8 2020, 9:33 AM · Restricted Project
Jan 7 2020
Jan 7 2020
pgode added inline comments to D71617: [Attributor] Improve noalias preservation using reachability.
Jan 7 2020, 11:38 PM · Restricted Project
Jan 1 2020
Jan 1 2020
Thanks @jdoerfert for detailed explanation and listing the corrections.
Jan 1 2020, 10:30 AM · Restricted Project
Dec 27 2019
Dec 27 2019
I have tried modifying the patch to include suggestions.
- Added condition to check if use is not marked nocapture.
- Included nullpointer check and reachability check.
Dec 27 2019, 11:03 AM · Restricted Project
Dec 26 2019
Dec 26 2019
pgode added inline comments to D71617: [Attributor] Improve noalias preservation using reachability.
Dec 26 2019, 11:11 PM · Restricted Project
Thanks for suggestions. I am learning more.
- Using AAReachability instead of AANoCapture and also removed those functions from AANoCapture.
Dec 26 2019, 6:36 AM · Restricted Project
pgode added inline comments to D71617: [Attributor] Improve noalias preservation using reachability.
Dec 26 2019, 5:17 AM · Restricted Project
Dec 25 2019
Dec 25 2019
Addressed the comments in the new diff version.
Dec 25 2019, 9:46 AM · Restricted Project
Tried addressing the comments in this patch.
- Created new methods for NoCaptureMaybeBeforeInstruction which returns true as of now. Using this method in the checkForAllUses method.
- Added additional parameter to checkForAllUses (initializing to nullptr). If this ptr is not null, then we check if use is reachable till this callsite instruction.
- Predicate checks in which case we should follow for use's user.
- checkForAllUses is called to check uses of argument. If it is valid/reachable use, then noalias deduction should happen.
Dec 25 2019, 9:45 AM · Restricted Project
Dec 17 2019
Dec 17 2019
Dec 17 2019, 9:26 AM · Restricted Project
Nov 26 2019
Nov 26 2019
Nov 26 2019, 1:38 AM · Restricted Project
Nov 22 2019
Nov 22 2019
pgode set the repository for D70233: [WIP][Attributor] AAReachability Attribute to rG LLVM Github Monorepo.
Nov 22 2019, 8:53 AM · Restricted Project
Nov 20 2019
Nov 20 2019
Updated the patch as per review comments.
Nov 20 2019, 10:55 PM · Restricted Project
Nov 20 2019, 10:45 PM · Restricted Project
Thanks for highlighting the nit. Corrected it.
Nov 20 2019, 9:26 AM · Restricted Project
In D70233#1753619, @uenoku wrote:It seems a bit weird for me to use IRPosition as a program point. Therefore, I think it is better to use Instruction instead of IRPosition. So the interface will be
bool isAssumedReachable(const Instruction &From, const Instruction &To)What do you think?
Nov 20 2019, 9:19 AM · Restricted Project
Nov 19 2019
Nov 19 2019
pgode retitled D70233: [WIP][Attributor] AAReachability Attribute from [WIP][NOT FOR COMMIT][Attributor] AAReachability Attribute to [WIP][Attributor] AAReachability Attribute.
Nov 19 2019, 9:09 PM · Restricted Project
Nov 17 2019
Nov 17 2019
Btw., do you have commit access?
Nov 17 2019, 6:26 AM · Restricted Project
Updated the function comments.
Nov 17 2019, 3:26 AM · Restricted Project
Thanks @jdoerfert for the helpful corrections. I have removed the files. (I should have removed them in the last patch.)
I will start working on the next part.
Nov 17 2019, 2:59 AM · Restricted Project
Nov 15 2019
Nov 15 2019
@jdoerfert Thank you suggesting the corrections.
Updating the corrected patch.
Nov 15 2019, 10:41 AM · Restricted Project
Thank you @jdoerfert , @sstefan1 and @uenoku for your helpful comments.
Submitting updated patch with corrections.
Nov 15 2019, 8:54 AM · Restricted Project
Nov 14 2019
Nov 14 2019
pgode retitled D70233: [WIP][Attributor] AAReachability Attribute from [WIP][NOT FOR COMMIT] AAReachability Attribute to [WIP][NOT FOR COMMIT][Attributor] AAReachability Attribute.
Nov 14 2019, 4:10 AM · Restricted Project
Nov 14 2019, 4:10 AM · Restricted Project
Oct 3 2016
Oct 3 2016
Thanks for splitting the code. LGTM.
Thanks.
Oct 2 2016
Oct 2 2016
Thanks for asking.
The first part of changes i.e. 'Fuse CMN, CMP, TST followed by Bcc', is applicable for Vulcan as well.
But, the second part i.e. 'Fuse ALU operations followed by CBZ/CBNZ' is not applicable for Vulcan, i.e. we don't fuse these. Is it possible to make this only Apple specific ?
Jul 19 2016
Jul 19 2016
[AArch64] PredictableSelectIsExpensive for Vulcan.
pgode updated the diff for D22448: [AArch64] Cache line size and PredictableSelectIsExpensive for Vulcan.
Removed the CacheLineSize (set to 64) for Vulcan, as three other parameters are required for LoopDataPrefetch to be effective.
Will include CacheLineSize & 3 other parameters in another patch, after they are finalized.
pgode added inline comments to D22448: [AArch64] Cache line size and PredictableSelectIsExpensive for Vulcan.
Jul 17 2016
Jul 17 2016
pgode updated subscribers of D22448: [AArch64] Cache line size and PredictableSelectIsExpensive for Vulcan.
pgode updated subscribers of D22448: [AArch64] Cache line size and PredictableSelectIsExpensive for Vulcan.
pgode added reviewers for D22448: [AArch64] Cache line size and PredictableSelectIsExpensive for Vulcan: t.p.northover, rengolin, MatzeB.
pgode retitled D22448: [AArch64] Cache line size and PredictableSelectIsExpensive for Vulcan from to [AArch64] Cache line size and PredictableSelectIsExpensive for Vulcan.
Jul 8 2016
Jul 8 2016
pgode committed rL274837: [AArch64] Macro fusion of simple ALU ops with branches for Broadcom's Vulcan.
[AArch64] Macro fusion of simple ALU ops with branches for Broadcom's Vulcan
pgode updated the diff for D22042: [AArch64] Macro fusion of simple ALU ops with branches for Broadcom's Vulcan.
Updated diff.
Removed fusion of additional instructions for Vulcan.
Default instructions supported by 'Macrofusion subtarget feature' will be fused.
pgode added a comment to D22042: [AArch64] Macro fusion of simple ALU ops with branches for Broadcom's Vulcan.
The approach of adding a new sub-feature for Macro-op fusion, by categorizing the instructions (my presumption) doesn't seem a good option. It will end up adding too many subfeature such as FeatureMacroOpFusionArith/FeatureMacroOpFusionLogical. Please correct me.
pgode added inline comments to D22042: [AArch64] Macro fusion of simple ALU ops with branches for Broadcom's Vulcan.
Jul 7 2016
Jul 7 2016
pgode added inline comments to D22042: [AArch64] Macro fusion of simple ALU ops with branches for Broadcom's Vulcan.
Jul 6 2016
Jul 6 2016
pgode retitled D22042: [AArch64] Macro fusion of simple ALU ops with branches for Broadcom's Vulcan from to [AArch64] Macro fusion of simple ALU ops with branches for Broadcom's Vulcan.
Jun 29 2016
Jun 29 2016
[AArch64] Add Broadcom Vulcan scheduling model.
Corrected & updated the diff.
But the code looks good to me with the extra comment removed.
Removed the extra comments (comments in 3.11 & 3.12 FP Load/store).
Modified comments part of 3.6 & 3.7 (Load/Store) to say that they are valid for 3.11 & 3.12 (FP Load/Store) as well.
[Driver][AArch64] Add support for Broadcom Vulcan core.
Out of curiosity, what are the performance improvements with this model?
Actually, this is a functional patch and we are currently validating it on the simulators. So, we will update performance numbers later.
Eric,
While submitting, I had added 'llvm-commits' as subscriber, as per the guidelines. (>> pgode added subscribers: echristo, aemerson, MatzeB, llvm-commits.)
Additionally, I also got a mail from the list llvm-commits: 'Your message entitled [PATCH] D21728: [AARCH64] Add Broadcom Vulcan scheduling modelwas successfully received by the llvm-commits mailing list.'
I can resend the patch, if it is required. Kindly suggest.
Thanks.
Jun 25 2016
Jun 25 2016
pgode retitled D21728: [AARCH64] Add Broadcom Vulcan scheduling model from to [AARCH64] Add Broadcom Vulcan scheduling model.
Jun 23 2016
Jun 23 2016
Ping ?
Jun 22 2016
Jun 22 2016
Jun 20 2016
Jun 20 2016
[AARCH64] Add support for Broadcom Vulcan
Jun 19 2016
Jun 19 2016
pgode retitled D21501: [Driver] Add support for Broadcom Vulcan core from to [Driver] Add support for Broadcom Vulcan core.
pgode retitled D21500: [AARCH64] Add support for Broadcom Vulcan from to [AARCH64] Add support for Broadcom Vulcan.
Jun 15 2016
Jun 15 2016
Test commit after access grant. Modified comment by adding a period.