Open
Conversation
….md to use items and added the corresponding code to solution.cpp
| ``` | ||
| auto rowMajorLinearId = (idx[1] * width) + idx[0]; // row-major | ||
| auto columnMajorLinearId = (idx[0] * height) + idx[1]; // column-major | ||
| auto rowMajorLinearId = sycl::id(globalId[0], globalId[1]); |
Contributor
There was a problem hiding this comment.
I don't think this is the original intent, here, since it talks about the linear id presumably the intent is to linearise it manually (to demonstrate the difference in striding). Can we change the other side instead to better match this?
Comment on lines
+82
to
+83
| auto rowMajorLinearId = sycl::id(globalId[0], globalId[1]); | ||
| auto columnMajorLinearId = sycl::id(globalId[1], globalId[0]); |
Contributor
There was a problem hiding this comment.
Similarly these are 2d ids, not linear ones
Contributor
Author
|
Hi,
To achieve a better match with as few changes to the code as possible, I think this can be a good solution:
```cpp
cgh.parallel_for<image_convolution>(
ndRange, [=](sycl::id<2> idx) {
auto rowMajorLinearId = idx[1] * inputImgWidth + idx[0];
auto columnMajorLinearId = idx[0] * inputImgWidth + idx[1];
auto globalId(rowMajorLinearId % inputImgWidth, rowMajorLinearId / inputImgWidth);
what do you think ?
________________________________
From: Duncan McBain ***@***.***>
Sent: 31 January 2024 18:46
To: codeplaysoftware/syclacademy ***@***.***>
Cc: Pablo ***@***.***>; Author ***@***.***>
Subject: Re: [codeplaysoftware/syclacademy] Problem with exercise 16 (PR #326)
@DuncanMcBain commented on this pull request.
________________________________
In Code_Exercises/Exercise_16_Coalesced_Global_Memory/README.md<#326 (comment)>:
@@ -17,8 +17,8 @@ global memory access patterns in your kernel are coalesced.
Consider two alternative ways to linearize the global id:
```
-auto rowMajorLinearId = (idx[1] * width) + idx[0]; // row-major
-auto columnMajorLinearId = (idx[0] * height) + idx[1]; // column-major
+auto rowMajorLinearId = sycl::id(globalId[0], globalId[1]);
I don't think this is the original intent, here, since it talks about the linear id presumably the intent is to linearise it manually (to demonstrate the difference in striding). Can we change the other side instead to better match this?
________________________________
In Code_Exercises/Exercise_16_Coalesced_Global_Memory/solution.cpp<#326 (comment)>:
+ auto rowMajorLinearId = sycl::id(globalId[0], globalId[1]);
+ auto columnMajorLinearId = sycl::id(globalId[1], globalId[0]);
Similarly these are 2d ids, not linear ones
—
Reply to this email directly, view it on GitHub<#326 (review)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BE6TABJLX5HC53VYXKKNMSTYRKGQDAVCNFSM6AAAAABBTCWAX2VHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMYTQNJUGU3DMMZUGQ>.
You are receiving this because you authored the thread.Message ID: ***@***.***>
|
Collaborator
|
Update on the status will be appreciated : ) |
EwanC
added a commit
to EwanC/syclacademy
that referenced
this pull request
Jan 16, 2026
This PR is inspired by the Lesson 4 `index.html` changes in KhronosGroup#326 and fixes up the `handler::single_task` invocations in a couple of places to make sure they are correct. * Change `parallel_for` to `single_task` in early example where we haven't defined `bufO` yet to call `bufO.get_range()` on. * Remove `(id<1> i)` parameter for `single_task` lambda invocation, as this shouldn't have any parameters.
EwanC
added a commit
to EwanC/syclacademy
that referenced
this pull request
Jan 16, 2026
This PR tries to resolve the issue highlighted in KhronosGroup#326 where the "Coalesced Global Memory" exercise README was asking a user to manually calculate a linear id that would be used to index memory. However, the exercise code itself uses 2d buffers/accessors which are indexed with a `sycl::id<2>`. So the current exercise request doesn't make sense, or could only be achieved with modifications beyond what is reasonably expected for a beginner. Instead, I've updated the README to state the usage of the 2d buffer/accessors and ask the user to experiment with flipping the dimension indexes, as this is the only difference between this exercise solution, and the solution from the previous lesson ```sh $ diff Code_Exercises/Coalesced_Global_Memory/solution.cpp Code_Exercises/Image_Convolution/reference.cpp 59a60 > 77a79 > globalId = sycl::id{globalId[1], globalId[0]}; ```
EwanC
added a commit
to EwanC/syclacademy
that referenced
this pull request
Jan 16, 2026
This PR tries to resolve the issue highlighted in KhronosGroup#326 where the "Coalesced Global Memory" exercise README was asking a user to manually calculate a linear id that would be used to index memory. However, the exercise code itself uses 2d buffers/accessors which are indexed with a `sycl::id<2>`. So the current exercise request doesn't make sense, or could only be achieved with modifications beyond what is reasonably expected for a beginner. Instead, I've updated the README to state the usage of the 2d buffer/accessors and ask the user to experiment with flipping the dimension indexes, as this is the only difference between this exercise solution, and the solution from the previous lesson ```sh $ diff Code_Exercises/Coalesced_Global_Memory/solution.cpp Code_Exercises/Image_Convolution/reference.cpp 59a60 > 77a79 > globalId = sycl::id{globalId[1], globalId[0]}; ```
This was referenced Jan 16, 2026
Contributor
I've created new PRs to try address the issues highlighted but based on up-to-date tip: |
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
The solution.cpp uses items in the parallel_for but the README.md uses ids. I changed the README.md and added the code to the solution.cpp