ucx

ucx copied to clipboard

rndv.c:1719 Assertion `sreq->send.rndv.lanes_count > 0' failed

2

### Describe the bug When running osu_bibw or osu_bw Host to Device with 2 processes on the same node, I get the assert: rndv.c:1719 Assertion `sreq->send.rndv.lanes_count > 0' failed The...

amirpavlo

SortIncludes key is duplicated

1

https://github.com/openucx/ucx/blob/ac479601fba6b5d696d95acca204c048b23673af/.clang-format#L121

maxim-banaev

### Describe the bug `ucx_info -d` output for knem shows a bizarrely huge registration time cost not seen in git HEAD version ### Setup and versions - OS version (e.g...

tonycurtis

ucp_mem_map returns success even if NIC registration fails

2

### Describe the bug ucp_mem_map returns success as long as some memory domain succeeds to register the given region. This is misleading for the caller of ucp_mem_map because for instance,...

SeyedMir

I am testing HPCX 2.10 (UCX 1.12, OpenMPI 4.1.2rc4) on a 2-socket EPYC 7742 system using `osu_bibw` benchmark. I test the in-node bandwidth: both ranks are started on the same...

angainor

Extremely low collective performance on UCX IB but not PSM3 when using HDR HCA / FDR switch combinations

5

### Describe the bug During analyses of figuring out low StarCCM application performance, we did some testing with IMB-MPI1 AlltoAll and noticed extremely low performance with UCX communication. Both OpenMPI...

AlexNinaber

### Describe the bug UCP tag match performance of host-to-GPU is 1/5 than that of GPU-to-host, for both intra-node and inter-node. ### Steps to Reproduce - Command line ``` ucx_perftest...

lyu

### Describe the bug `ucx_info -d` reports much slower registration cost for `knem` on a64fx platform after an upgrade from CentOS 8.4 to Rocky 8.4. Not seen with released ucx...

tonycurtis

UCM/CUDA/TEST: Install memory hooks for async Cuda allocations

6

## Why As discussed in #7194 and #7110 , need to add memory hooks support for cuda async allocations. Without this, applications using these allocations may fail to detect Cuda...

yosefe

Failed assertion using active messaging in multi-threaded environment

15

### Describe the bug Using active messaging (`ucp_am_send_nbx`) on the same ~(TCP)~ endpoint from withing two different threads results in a failed assertion, although the library is compiled with multithreading...

krakowski