[code-reflection] RFR: 1D Matrix Multiplication example for HAT [v2]

[f]

On Mon, 7 Apr 2025 17:51:27 GMT, Gary Frost <gfrost at openjdk.org> wrote:

>> Juan Fumero has updated the pull request with a new target base due to a merge or a rebase. The pull request now contains 11 commits:
>> 
>>  - Merge branch 'code-reflection' into dev/examples
>>  - Minor fix seq-comparison code
>>  - Merge with latest develop
>>  - Merge branch 'code-reflection' into dev/examples
>>  - Merge branch 'code-reflection' into dev/examples
>>  - Merge branch 'code-reflection' into dev/examples
>>  - MatrixMult example moved to matmul directory
>>  - Merge branch 'code-reflection' into dev/examples
>>  - Precision control error down to 1%
>>  - Matrix-Multiplication checks
>>  - ... and 1 more: https://git.openjdk.org/babylon/compare/ee3da036...cd3c7ce9
>
> Hi Sidney
> 
> As Juan mentioned, I don't think we are there yet.  But we have plans.
> 
> We need to add low level primitives to HAT for matrix mul/scans etc to
> allow us to handoff to the vendor backends (GPU drivers) without
> unnecessary copies.
> 
> Gary
> 
> 
> On Mon, Apr 7, 2025 at 5:20 PM Juan Fumero ***@***.***> wrote:
> 
>> Hi @SidneyLann <https://github.com/SidneyLann> , I am not the core
>> maintainer of Babylon. Probably Gary Frost can help you with your
>> questions. From my view, I think you need to access shared memory and some
>> synchronisation primitives to be able to perform reductions. I am not sure
>> if this is implemented in HAT yet.
>>
>> —
>> Reply to this email directly, view it on GitHub
>> <https://github.com/openjdk/babylon/pull/276#issuecomment-2783916596>, or
>> unsubscribe
>> <https://github.com/notifications/unsubscribe-auth/ABBKEN35FIKYUTUIYRDYC5D2YKQWRAVCNFSM6AAAAABSBW7YWWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDOOBTHEYTMNJZGY>
>> .
>> You are receiving this because you are subscribed to this thread.Message
>> ID: ***@***.***>
>> [image: jjfumero]*jjfumero* left a comment (openjdk/babylon#276)
>> <https://github.com/openjdk/babylon/pull/276#issuecomment-2783916596>
>>
>> Hi @SidneyLann <https://github.com/SidneyLann> , I am not the core
>> maintainer of Babylon. Probably Gary Frost can help you with your
>> questions. From my view, I think you need to access shared memory and some
>> synchronisation primitives to be able to perform reductions. I am not sure
>> if this is implemented in HAT yet.
>>
>> —
>> Reply to this email directly, view it on GitHub
>> <https://github.com/openjdk/babylon/pull/276#issuecomment-2783916596>, or
>> unsubscribe
>> <https://github.com/notifications/unsubscribe-auth/ABBKEN35FIKYUTUIYRDYC5D2YKQWRAVCNFSM6AAAAABSBW7YWWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDOOBTHEYTMNJZGY>
>> .
>> You are receiving this because you are subscribed to this thread.Message
>> ID: ***@***.***>
>>

@grfrost
Hi Gray
Are you develope many platforms(ptx,cuda,spirv,hip,etc) simultaneously？How about complete one platform(ie. cuda) first?

-------------

PR Comment: https://git.openjdk.org/babylon/pull/276#issuecomment-2784852485