Page MenuHomePhabricator

silvas (Sean Silva)
User

Projects

User does not belong to any projects.

User Details

User Since
Aug 20 2012, 11:51 AM (439 w, 6 d)

Recent Activity

Tue, Jan 19

silvas added inline comments to D94994: [mlir][splitting std] move 2 more ops to `tensor`.
Tue, Jan 19, 3:30 PM · Restricted Project
silvas added inline comments to D94994: [mlir][splitting std] move 2 more ops to `tensor`.
Tue, Jan 19, 1:59 PM · Restricted Project
silvas committed rGbe7352c00d51: [mlir][splitting std] move 2 more ops to `tensor` (authored by silvas).
[mlir][splitting std] move 2 more ops to `tensor`
Tue, Jan 19, 1:54 PM
silvas closed D94994: [mlir][splitting std] move 2 more ops to `tensor`.
Tue, Jan 19, 1:54 PM · Restricted Project
silvas added inline comments to D94994: [mlir][splitting std] move 2 more ops to `tensor`.
Tue, Jan 19, 1:47 PM · Restricted Project
silvas added inline comments to D94994: [mlir][splitting std] move 2 more ops to `tensor`.
Tue, Jan 19, 1:47 PM · Restricted Project
silvas updated the diff for D94994: [mlir][splitting std] move 2 more ops to `tensor`.

Use forward decl.

Tue, Jan 19, 1:46 PM · Restricted Project
silvas updated the diff for D94994: [mlir][splitting std] move 2 more ops to `tensor`.

Address River's comments.

Tue, Jan 19, 1:42 PM · Restricted Project
silvas requested review of D94994: [mlir][splitting std] move 2 more ops to `tensor`.
Tue, Jan 19, 12:41 PM · Restricted Project

Fri, Jan 15

silvas added inline comments to D94832: [mlir] Add a new builtin `unrealized_conversion_cast` operation.
Fri, Jan 15, 5:47 PM · Restricted Project
silvas added inline comments to D94832: [mlir] Add a new builtin `unrealized_conversion_cast` operation.
Fri, Jan 15, 5:41 PM · Restricted Project
silvas accepted D94831: [mlir] Add an interface for Cast-Like operations.

Thanks for cleaning this up!

Fri, Jan 15, 3:02 PM · Restricted Project
silvas added inline comments to D94831: [mlir] Add an interface for Cast-Like operations.
Fri, Jan 15, 2:56 PM · Restricted Project
silvas added inline comments to D94831: [mlir] Add an interface for Cast-Like operations.
Fri, Jan 15, 2:54 PM · Restricted Project
silvas accepted D94819: [mlir][Linalg] Add missing check to canonicalization of GenericOp that are identity ops..

LGTM, but ot sounds like you might want to add some more testing as well. A payload that returns an argument of the enclosing function would be enough.

Fri, Jan 15, 12:57 PM · Restricted Project

Thu, Jan 14

silvas committed rGe2d7d3cb0ead: [mlir][docs] Bring bufferization docs up to date. (authored by silvas).
[mlir][docs] Bring bufferization docs up to date.
Thu, Jan 14, 12:29 PM
silvas closed D94351: [mlir][docs] Bring bufferization docs up to date..
Thu, Jan 14, 12:29 PM · Restricted Project

Wed, Jan 13

silvas accepted D94581: [mlir][Linalg] Add canonicalization to remove no-op linalg operations..
Wed, Jan 13, 5:23 PM · Restricted Project
silvas accepted D94247: [MLIR][TOSA] First lowerings from Tosa to Linalg.
Wed, Jan 13, 4:03 PM · Restricted Project

Tue, Jan 12

silvas added inline comments to D94247: [MLIR][TOSA] First lowerings from Tosa to Linalg.
Tue, Jan 12, 5:17 PM · Restricted Project
silvas added inline comments to D94485: [mlir] Update LLVM dialect type documentation.
Tue, Jan 12, 4:29 PM · Restricted Project
silvas updated the diff for D94351: [mlir][docs] Bring bufferization docs up to date..

Address Stephan's comments.

Tue, Jan 12, 1:47 PM · Restricted Project

Mon, Jan 11

silvas added inline comments to D94247: [MLIR][TOSA] First lowerings from Tosa to Linalg.
Mon, Jan 11, 5:41 PM · Restricted Project
silvas updated the diff for D94351: [mlir][docs] Bring bufferization docs up to date..

Fix dfki-jugr@'s comments.

Mon, Jan 11, 11:19 AM · Restricted Project
silvas accepted D94405: [mlir] use built-in vector types instead of LLVM dialect types when possible.

Thanks! This looks great!

Mon, Jan 11, 11:16 AM · Restricted Project

Fri, Jan 8

silvas added inline comments to D94247: [MLIR][TOSA] First lowerings from Tosa to Linalg.
Fri, Jan 8, 4:53 PM · Restricted Project
silvas updated the diff for D94351: [mlir][docs] Bring bufferization docs up to date..

Remove old stray stuff and add a TOC

Fri, Jan 8, 4:29 PM · Restricted Project
silvas requested review of D94351: [mlir][docs] Bring bufferization docs up to date..
Fri, Jan 8, 4:28 PM · Restricted Project
silvas added inline comments to D94247: [MLIR][TOSA] First lowerings from Tosa to Linalg.
Fri, Jan 8, 1:54 PM · Restricted Project

Thu, Jan 7

silvas added inline comments to D94247: [MLIR][TOSA] First lowerings from Tosa to Linalg.
Thu, Jan 7, 1:50 PM · Restricted Project

Wed, Jan 6

silvas accepted D94179: [mlir] replace LLVM dialect float types with built-ins.

LGTM modulo Lei's comment.

Wed, Jan 6, 11:26 AM · Restricted Project
silvas accepted D94178: [mlir] replace LLVMIntegerType with built-in integer type.

Awesome! Thank you for pushing this herculean effort!

Wed, Jan 6, 11:25 AM · Restricted Project

Tue, Jan 5

silvas accepted D94133: [mlir][ODS] Rename several uses of 'description' to 'summary'.
Tue, Jan 5, 3:53 PM · Restricted Project

Dec 17 2020

silvas committed rG129d6e554e7a: [mlir] Move `std.tensor_cast` -> `tensor.cast`. (authored by silvas).
[mlir] Move `std.tensor_cast` -> `tensor.cast`.
Dec 17 2020, 4:08 PM
silvas closed D93357: [mlir] Move `std.tensor_cast` -> `tensor.cast`..
Dec 17 2020, 4:07 PM · Restricted Project
silvas added a comment to D93498: Workaround around clang 5.0 bug by including SmallVector.h in LLVM.h (PR41549).

I think you also should do the same for

Dec 17 2020, 4:02 PM · Restricted Project
silvas accepted D93492: [mlir] Add canonicalization from `tensor_cast` to `dim` op..
Dec 17 2020, 2:41 PM · Restricted Project
silvas added inline comments to D93374: [mlir][Linalg] Define a linalg.init_tensor operation..
Dec 17 2020, 2:03 PM · Restricted Project
silvas added inline comments to D93374: [mlir][Linalg] Define a linalg.init_tensor operation..
Dec 17 2020, 11:03 AM · Restricted Project
silvas accepted D93374: [mlir][Linalg] Define a linalg.init_tensor operation..
Dec 17 2020, 10:50 AM · Restricted Project

Dec 16 2020

silvas updated the diff for D93357: [mlir] Move `std.tensor_cast` -> `tensor.cast`..

Address Stephan's comments.

Dec 16 2020, 2:32 PM · Restricted Project
silvas accepted D93405: [mlir] Support index and memref types in llvm.mlir.cast.

Awesome!

Dec 16 2020, 2:26 PM · Restricted Project
silvas added inline comments to D93374: [mlir][Linalg] Define a linalg.init_tensor operation..
Dec 16 2020, 12:34 PM · Restricted Project

Dec 15 2020

silvas added inline comments to D93357: [mlir] Move `std.tensor_cast` -> `tensor.cast`..
Dec 15 2020, 6:23 PM · Restricted Project
silvas updated the diff for D93357: [mlir] Move `std.tensor_cast` -> `tensor.cast`..

Address River's comments.

Dec 15 2020, 6:23 PM · Restricted Project
silvas updated the diff for D93357: [mlir] Move `std.tensor_cast` -> `tensor.cast`..

Fix a few mentions in comments.

Dec 15 2020, 4:53 PM · Restricted Project
silvas added a reviewer for D93357: [mlir] Move `std.tensor_cast` -> `tensor.cast`.: herhut.
Dec 15 2020, 4:49 PM · Restricted Project
silvas requested review of D93357: [mlir] Move `std.tensor_cast` -> `tensor.cast`..
Dec 15 2020, 4:49 PM · Restricted Project
silvas committed rGcaf4f2e372a7: [mlir] Handle unknown ops in dynamic_tensor_from_elements bufferization (authored by silvas).
[mlir] Handle unknown ops in dynamic_tensor_from_elements bufferization
Dec 15 2020, 12:52 PM
silvas closed D93262: [mlir] Handle unknown ops in dynamic_tensor_from_elements bufferization.
Dec 15 2020, 12:51 PM · Restricted Project

Dec 14 2020

silvas added reviewers for D93262: [mlir] Handle unknown ops in dynamic_tensor_from_elements bufferization: herhut, rriddle.
Dec 14 2020, 6:33 PM · Restricted Project
silvas requested review of D93262: [mlir] Handle unknown ops in dynamic_tensor_from_elements bufferization.
Dec 14 2020, 6:33 PM · Restricted Project
silvas added a comment to D92171: [mlir][ArmNeon][RFC] Add an ArmNeon dialect.

Did we actually get consensus around the direction here? I didn't see a clear conclusion.

Dec 14 2020, 3:19 PM · Restricted Project

Dec 11 2020

silvas added a reverting change for rG0d48d265db66: Revert "[mlir] Start splitting the `tensor` dialect out of `std`.": rG444822d77a7f: Revert "Revert "[mlir] Start splitting the `tensor` dialect out of `std`."".
Dec 11 2020, 2:32 PM
silvas committed rG444822d77a7f: Revert "Revert "[mlir] Start splitting the `tensor` dialect out of `std`."" (authored by silvas).
Revert "Revert "[mlir] Start splitting the `tensor` dialect out of `std`.""
Dec 11 2020, 2:32 PM
silvas added a reverting change for rGcab8dda90f48: [mlir] Start splitting the `tensor` dialect out of `std`.: rG0d48d265db66: Revert "[mlir] Start splitting the `tensor` dialect out of `std`.".
Dec 11 2020, 2:18 PM
silvas committed rG0d48d265db66: Revert "[mlir] Start splitting the `tensor` dialect out of `std`." (authored by silvas).
Revert "[mlir] Start splitting the `tensor` dialect out of `std`."
Dec 11 2020, 2:18 PM
silvas added a reverting change for D92991: [mlir] Start splitting the `tensor` dialect out of `std`.: rG0d48d265db66: Revert "[mlir] Start splitting the `tensor` dialect out of `std`.".
Dec 11 2020, 2:18 PM · Restricted Project
silvas committed rGcab8dda90f48: [mlir] Start splitting the `tensor` dialect out of `std`. (authored by silvas).
[mlir] Start splitting the `tensor` dialect out of `std`.
Dec 11 2020, 1:58 PM
silvas closed D92991: [mlir] Start splitting the `tensor` dialect out of `std`..
Dec 11 2020, 1:58 PM · Restricted Project
silvas updated the diff for D92991: [mlir] Start splitting the `tensor` dialect out of `std`..

One last nit.

Dec 11 2020, 1:45 PM · Restricted Project
silvas updated the diff for D92991: [mlir] Start splitting the `tensor` dialect out of `std`..

Touch up a few more places that mention extract_element.

Dec 11 2020, 1:44 PM · Restricted Project
silvas updated the diff for D92991: [mlir] Start splitting the `tensor` dialect out of `std`..

Partially revert https://reviews.llvm.org/D92980

Dec 11 2020, 1:04 PM · Restricted Project

Dec 10 2020

silvas added a comment to D93069: [SmallVector] Copy new docs into Doxygen comment.

You might want to link to the ProgrammersManual directly from here as well.

Dec 10 2020, 2:32 PM · Restricted Project
silvas updated the diff for D92991: [mlir] Start splitting the `tensor` dialect out of `std`..

Remove some more stray copypasta.

Dec 10 2020, 1:29 PM · Restricted Project
silvas added inline comments to D92991: [mlir] Start splitting the `tensor` dialect out of `std`..
Dec 10 2020, 12:46 PM · Restricted Project
silvas updated the diff for D92991: [mlir] Start splitting the `tensor` dialect out of `std`..

Address River's comments.

Dec 10 2020, 12:46 PM · Restricted Project
silvas added a comment to D92562: [MLIR] Refactored BufferDeallocation pass to enable support for individual alloc/clone/free operations..

Having this customizable with a callback will allow users to potentially start doing all sorts of random stuff inside. What if a user starts depending on that? For example, what if the callback references some side data structure and generates different clone/free ops depending on that, possibly in a less structured way than we would like to support upstream. That suddenly becomes much more difficult to unwind. Of course, implementing that callback is likely to be done in a downstream project, where doing something like this as a shortcut might seem more acceptable. But that shortcut then becomes a cost that the MLIR community needs to pay to unwind!

We could, for example, explicitly state that this is not allowed and that the call back should always create the same operations. Then we have a contract of what we want to support and a clear mandate to break downstream that does not follow the contract. Would that approach work for you?

Dec 10 2020, 10:02 AM · Restricted Project
silvas added a comment to D92991: [mlir] Start splitting the `tensor` dialect out of `std`..

I generally agree with @bondhugula on tensor.get_element being somewhat better. However, I think tensor.extract is the right thing for this patch because:

Dec 10 2020, 9:39 AM · Restricted Project
silvas accepted D93037: [YAML] Use correct source location for unknown key errors..

Would be nice to find a way to test this (Maybe a FileCheck test? not sure how to set that up though; don't worry if it is too much work.)

Dec 10 2020, 9:34 AM · Restricted Project

Dec 9 2020

silvas requested review of D92991: [mlir] Start splitting the `tensor` dialect out of `std`..
Dec 9 2020, 7:05 PM · Restricted Project
silvas accepted D92980: [mlir] Remove the dependency on StandardOps from FoldUtils.
Dec 9 2020, 4:05 PM · Restricted Project
silvas added a comment to D92562: [MLIR] Refactored BufferDeallocation pass to enable support for individual alloc/clone/free operations..

Totally agreed that MLIR is in a youthful stage. I think it is a valuable discussion to be had explicitly with the community about how we approach the tradeoffs in adopting more incremental code vs more vetted designs. Getting clarity there would greatly simplify this discussion!

Dec 9 2020, 1:13 PM · Restricted Project
silvas added inline comments to D92927: [mlir][Shape] Make sure tensor_cast(constant_shape) folding uses the correct type.
Dec 9 2020, 11:25 AM · Restricted Project
silvas accepted D92927: [mlir][Shape] Make sure tensor_cast(constant_shape) folding uses the correct type.
Dec 9 2020, 11:19 AM · Restricted Project

Dec 8 2020

silvas accepted D92901: [MLIR] Minor cleanup for Shape dialect..
Dec 8 2020, 7:35 PM · Restricted Project
silvas accepted D92878: [mlir] Use mlir::OpState::operator->() to get to methods of mlir::Operation. This is a preparation step to remove the corresponding methods from OpState..
Dec 8 2020, 12:45 PM · Restricted Project

Dec 7 2020

silvas added inline comments to D92755: [YAML] Support extended spellings when parsing bools..
Dec 7 2020, 12:52 PM · Restricted Project
silvas added inline comments to D92755: [YAML] Support extended spellings when parsing bools..
Dec 7 2020, 12:51 PM · Restricted Project
silvas accepted D92755: [YAML] Support extended spellings when parsing bools..

Thanks. This LGTM.

Dec 7 2020, 12:26 PM · Restricted Project
silvas added a comment to D92562: [MLIR] Refactored BufferDeallocation pass to enable support for individual alloc/clone/free operations..

@silvas Our intention was not to "bypass anything", sorry for the confusion. As you have mentioned, the discussion went a bit "idle" and we had to more forwards to simplify the integration of the BufferDeallocation pass into other third-party sub projects. Therefore, we proposed this CL that realizes an extremely light-weight version of the Pass Interface being discussed in the forums. However, the proposed solution in the scope of this CL is meant to unblock others on the one hand and being minimally invasive on the other hand. In other words: as soon as we can come up with a better solution based on the Discourse discussion, we can easily "revert" or "update" this pass interface. Please note further that existing code will not break or require further updates after merging this CL.

Dec 7 2020, 10:54 AM · Restricted Project
silvas added a comment to D92755: [YAML] Support extended spellings when parsing bools..

The spec link doesn't seem to support AllowNumeric. Why do we have that feature?

Dec 7 2020, 10:39 AM · Restricted Project

Dec 4 2020

silvas added a comment to D92562: [MLIR] Refactored BufferDeallocation pass to enable support for individual alloc/clone/free operations..

As a general matter of policy in LLVM, we do not bypass ongoing design discussions to unblock downstream users.

Just a note. It's fair to say that they should have pinged the thread one more time, but that thread had a discussion, where dfki and herhut responded to the debate and then received no response for 17 days. That's not really an ongoing discussion at that point.

Dec 4 2020, 7:18 PM · Restricted Project
silvas added a comment to D92562: [MLIR] Refactored BufferDeallocation pass to enable support for individual alloc/clone/free operations..

I disagree with @silvas on this. This particular change does not take any opinion on the discussion you reference. It merely makes the hard-wired current behavior configurable by choosing different ops. It does not mandate how the decision which free operation to use is done.

Dec 4 2020, 1:12 PM · Restricted Project

Dec 3 2020

silvas abandoned D90884: [SmallVector] Add `SVec<T>` / `Vec<T>` convenience wrappers..
Dec 3 2020, 6:35 PM · Restricted Project
silvas added a comment to D90884: [SmallVector] Add `SVec<T>` / `Vec<T>` convenience wrappers..

Abandoned in favor of https://reviews.llvm.org/D92522

Dec 3 2020, 6:35 PM · Restricted Project
silvas added a comment to D92269: [TableGen] Eliminate the 'code' type.

Wow, awesome work!

Dec 3 2020, 5:47 PM · Restricted Project, Restricted Project, Restricted Project
silvas committed rGae9fd5578e8a: [SmallVector] Allow SmallVector<T> (authored by silvas).
[SmallVector] Allow SmallVector<T>
Dec 3 2020, 5:22 PM
silvas closed D92522: [SmallVector] Allow SmallVector<T>.
Dec 3 2020, 5:22 PM · Restricted Project
silvas updated the diff for D92522: [SmallVector] Allow SmallVector<T>.

Replace constexpr function with old-school class with static members.

Dec 3 2020, 5:13 PM · Restricted Project
silvas added a comment to D92562: [MLIR] Refactored BufferDeallocation pass to enable support for individual alloc/clone/free operations..

Can we please finish the discussion in https://llvm.discourse.group/t/remove-tight-coupling-of-the-bufferdeallocation-pass-to-std-and-linalg-operations/2162/12 before moving forward with this? It's not clear to me that we all agreed that this was the right direction.

Dec 3 2020, 12:46 PM · Restricted Project
silvas added inline comments to D92540: [mlir][linalg] Add vectorization for element-wise linalg ops.
Dec 3 2020, 12:19 PM · Restricted Project
silvas added inline comments to D92540: [mlir][linalg] Add vectorization for element-wise linalg ops.
Dec 3 2020, 12:18 PM · Restricted Project
silvas added inline comments to D92343: [mlir] Add initial support for an alias analysis framework in MLIR.
Dec 3 2020, 12:15 PM · Restricted Project
silvas added inline comments to D92522: [SmallVector] Allow SmallVector<T>.
Dec 3 2020, 10:42 AM · Restricted Project
silvas updated the diff for D92522: [SmallVector] Allow SmallVector<T>.

Address Chris's comments about the docs.

Dec 3 2020, 10:41 AM · Restricted Project

Dec 2 2020

silvas added a comment to D92522: [SmallVector] Allow SmallVector<T>.

I'm generally good with this in terms of incremental forward progress. Still don't feel super great about the "at least 1" thing and the disparate storage limits (preferred size versus break-because-too-big size), but given the varied opinions, seems like the incremental progress is probably the best path forward for now. (perfect being the enemy of good and all that)

Dec 2 2020, 5:04 PM · Restricted Project
silvas added inline comments to D92522: [SmallVector] Allow SmallVector<T>.
Dec 2 2020, 4:54 PM · Restricted Project
silvas updated the diff for D92522: [SmallVector] Allow SmallVector<T>.

Use SmallVector<T, 0> as the header size.

Dec 2 2020, 4:50 PM · Restricted Project
silvas abandoned D92526: [SmallVector] Allow SmallVector<T>.

Created mistakenly (bad script). Abandoning.

Dec 2 2020, 4:46 PM · Restricted Project