Skip to content

Conversation

@Critsium-xy
Copy link
Collaborator

Fully make use of the memory of dsp hardware. Can run small cases now but still under testing. DSP memory management is very weird so maybe more time is needed on this.

@Critsium-xy
Copy link
Collaborator Author

Currently this can successfully run on small cases but will cause immediate memory failure on large cases.

@Critsium-xy Critsium-xy marked this pull request as ready for review October 30, 2024 11:47
@mohanchen mohanchen self-requested a review October 31, 2024 04:20
@mohanchen mohanchen added the GPU & DCU & HPC GPU and DCU and HPC related any issues label Oct 31, 2024
@mohanchen mohanchen merged commit ddf990f into deepmodeling:develop Nov 5, 2024
14 checks passed
@Critsium-xy Critsium-xy deleted the mtblas_memory branch November 8, 2024 10:38
Fisherd99 pushed a commit to Fisherd99/abacus-BSE that referenced this pull request Mar 31, 2025
* Initial commit

* Change memory_op construction

* I finally find this

* Fix template bug

* Fix memory header definition

* Optimize memory op usage

* Update diago_subspace

* No change

* Fix MPI Error

* Make the extra memory usage DSP-hardware-specialized. Add some annotations.

* Reorganize dsp codes

* Fix bug 1

* Fix bug 2

* Finish transporting codes

---------

Co-authored-by: Mohan Chen <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

GPU & DCU & HPC GPU and DCU and HPC related any issues

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants