[opengl] Optimize range_for for ndarrays #3884
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Note this PR reduces number of generated shaders for mpm88 from 10 to 6. There's still one gtmp related shader remaining for temporary values used across multiple shaders. Whether to further remove that shader need more benchmark so let's just get rid of addtional shaders introduced by ndarray range_for for now. [ghstack-poisoned]
ailzhang
pushed a commit
that referenced
this pull request
Dec 27, 2021
Note this PR reduces number of generated shaders for mpm88 from 10 to 6. There's still one gtmp related shader remaining for temporary values used across multiple shaders. Whether to further remove that shader need more benchmark so let's just get rid of addtional shaders introduced by ndarray range_for for now. ghstack-source-id: ae80e9c8d92ee131a37f54bff10a710dd21e79de Pull Request resolved: #3884
Note this PR reduces number of generated shaders for mpm88 from 10 to 6. There's still one gtmp related shader remaining for temporary values used across multiple shaders. Whether to further remove that shader need more benchmark so let's just get rid of addtional shaders introduced by ndarray range_for for now. [ghstack-poisoned]
ailzhang
pushed a commit
that referenced
this pull request
Dec 27, 2021
Note this PR reduces number of generated shaders for mpm88 from 10 to 6. There's still one gtmp related shader remaining for temporary values used across multiple shaders. Whether to further remove that shader need more benchmark so let's just get rid of addtional shaders introduced by ndarray range_for for now. ghstack-source-id: bdb9feba35fa8cd6cc99b826679dde353d706e07 Pull Request resolved: #3884
ailzhang
changed the title
bobcao3
reviewed
Dec 28, 2021
taichi/ir/statements.h
Outdated
| @@ -756,7 +762,8 @@ class RangeForStmt : public Stmt { | |||
| bit_vectorize, | |||
| num_cpu_threads, | |||
| block_dim, | |||
| strictly_serialized); | |||
| strictly_serialized, | |||
| range_of_array); | |||
There was a problem hiding this comment.
This parameter name seems slightly confusing?
bobcao3
reviewed
Dec 28, 2021
| @@ -85,19 +85,35 @@ class Offloader { | |||
| } else { | |||
| offloaded->block_dim = s->block_dim; | |||
| } | |||
| if (auto val = s->begin->cast<ConstStmt>()) { | |||
|
|
|||
| // TODO: We need to update codegen for each backend gradually so let's | |||
There was a problem hiding this comment.
When we merge this in let's also notify people working on the vulkan ndarray this change
bobcao3
reviewed
Dec 28, 2021
| @@ -202,7 +202,7 @@ def init(d: ti.i32, density1: ti.any_arr(), density2: ti.any_arr(), | |||
|
|
|||
| @ti.test(arch=ti.opengl) | |||
| def test_opengl_exceed_max_ssbo(): | |||
| # 7 ndarrays + gtmp + args > 8 (maximum allowed) | |||
| # 8 ndarrays + args > 8 (maximum allowed) | |||
There was a problem hiding this comment.
Do we really want this test? This seems a bit arch specific & a lot of devices support more
There was a problem hiding this comment.
Yea this test was mainly a self reminder to double check how many ssbos we create in normal cases - we can remove it later :D
bobcao3
approved these changes
Dec 28, 2021
Note this PR reduces number of generated shaders for mpm88 from 10 to 6. There's still one gtmp related shader remaining for temporary values used across multiple shaders. Whether to further remove that shader need more benchmark so let's just get rid of addtional shaders introduced by ndarray range_for for now. [ghstack-poisoned]
This was referenced Dec 29, 2021
Note this PR reduces number of generated shaders for mpm88 from 10 to 6. There's still one gtmp related shader remaining for temporary values used across multiple shaders. Whether to further remove that shader need more benchmark so let's just get rid of addtional shaders introduced by ndarray range_for for now. [ghstack-poisoned]
ailzhang
changed the title
ailzhang
pushed a commit
that referenced
this pull request
Dec 29, 2021
Note this PR reduces number of generated shaders for mpm88 from 10 to 6. There's still one gtmp related shader remaining for temporary values used across multiple shaders. Whether to further remove that shader need more benchmark so let's just get rid of addtional shaders introduced by ndarray range_for for now. ghstack-source-id: 3e2c2827bf752d6397cfc7780f2a79f9f2853a3d Pull Request resolved: #3884
quadpixels
pushed a commit
to quadpixels/taichi
that referenced
this pull request
Jan 5, 2022
Note this PR reduces number of generated shaders for mpm88 from 10 to 6. There's still one gtmp related shader remaining for temporary values used across multiple shaders. Whether to further remove that shader need more benchmark so let's just get rid of addtional shaders introduced by ndarray range_for for now. ghstack-source-id: 3e2c2827bf752d6397cfc7780f2a79f9f2853a3d Pull Request resolved: taichi-dev#3884
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Stack from ghstack:
Note this PR reduces number of generated shaders for mpm88 from 10 to 6.
There's still one gtmp related shader remaining for temporary values
used across multiple shaders. Whether to further remove that shader need
more benchmark so let's just get rid of addtional shaders introduced by
ndarray range_for for now.