Qinglei Cao

Performance issue on Ookami

12

I tested Blis using dgemm on [Ookami](https://www.stonybrook.edu/ookami/), but the performance for 1 core is less than 50 Gflop/s on matrix with size up to 5000 * 5000. Are there any...

jdf2c issue when dealing with certain case

1

When writing in PTG like this: `READ D super.rank_of(&desc_f_data->super, 0, n); %})` It shows: `#error Expression return desc_f_data->super.rank_of(&desc_f_data->super, 0, n); has not been generated` @therault suggested using a task local...

No data movement to GPU when data from desc with type keyword

7

## Describe the bug In PTG, when data is from memory and a `type` keyword is provided, there will be no data movement to GPU, e.g., `READ A

python site-packages directory is missing in the installed directory

2

## Describe the bug In `parsec.env.sh` in the installed `bin` directory, `site-packages` is provided to PYTHONPATH, but it is missing in the installed `lib/lib64` directory. ### To Reproduce Steps to...

Deadlock on GPU when using NEW and matrix size is large

1

## Describe the bug There is a deadlock when running on GPU. This deadlock happens when using NEW with a large matrix size (memory does not fit into GPU memory)....

## Description When the matrix can not be fitted into GPU memory, performance issues will occur. It's better to find a way to improve the data reuse on GPU, whether...

## Description If the received data type is different from what it needs on GPUs, maybe it's better to do node-level datatype conversion plus caching on devices instead of tile-level.

## Description Right now, the datatype conversion needs to be dealt with explicitly. It would be great to find a way to convert this data type in runtime. The direct...

## Describe the bug This issue is found when running HiCMA on GPU and checking the correctness with `dplasma_potrf`, which is also on GPU, causing an error that is triggered...

Inconsistent function name in RECURSIVE

6

## Describe the bug The first issue: `DPLASMA_WITH_RECURSIVE` and `PARSEC_HAVE_RECURSIVE` are independent; which means `PARSEC_HAVE_RECURSIVE` is always `OFF` by default even if `DPLASMA_WITH_RECURSIVE` is `ON`. The second issue: if `PARSEC_HAVE_RECURSIVE`...