Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / openucx/ucx issues and pull requests

#9531 - GTest failures when running on the L40G GPU

Issue - State: open - Opened by Alexey-Rivkin 11 months ago - 1 comment
Labels: Bug

#9530 - UCP: Report invalidation error if no rcache

Pull Request - State: open - Opened by ivankochin 11 months ago - 2 comments

#9529 - UCT/CUDA/BASE: Do not use cuda device 0 if no context is active

Pull Request - State: closed - Opened by ivankochin 11 months ago - 3 comments

#9528 - GPU test fails on assertion

Issue - State: open - Opened by iyastreb 11 months ago
Labels: Bug

#9527 - UCP/WIREUP: Define UCT_MD_FLAG_INVALIDATE_RMA description

Pull Request - State: closed - Opened by ivankochin 11 months ago

#9526 - UCX not working with NAT

Issue - State: open - Opened by ziegenbalg 11 months ago - 1 comment
Labels: Bug

#9525 - UCP/AM: Allow AM handlers registration in any order

Pull Request - State: closed - Opened by iyastreb 11 months ago - 1 comment

#9524 - AZP/JUCX: Fix the UCX snapshot pipeline

Pull Request - State: closed - Opened by Alexey-Rivkin 11 months ago - 3 comments

#9523 - jucx 1.15.0 missing arm shared library

Issue - State: closed - Opened by abellina 11 months ago - 4 comments
Labels: Bug

#9522 - could not allocate memory for slow_elems

Issue - State: open - Opened by angainor 11 months ago - 2 comments
Labels: Bug

#9521 - IO-DEMO: fix grep regexp in runner script

Pull Request - State: closed - Opened by evgeny-leksikov 11 months ago - 6 comments

#9520 - multi_rail does not show improvements in bandwidth

Issue - State: open - Opened by rrgargeya 11 months ago - 2 comments

#9519 - BUILD/CONFIG: Introduce Address Sanitizer flag option

Pull Request - State: closed - Opened by ivankochin 11 months ago - 1 comment

#9518 - TEST/UCT/MD: Prevent rkey buffer overflow

Pull Request - State: closed - Opened by ivankochin 11 months ago - 2 comments

#9517 - ucp_rkey_destory encountered a segment fault

Issue - State: closed - Opened by JKLiang9714 11 months ago - 2 comments
Labels: Bug

#9516 - Process hangs when using UCX.

Issue - State: open - Opened by zpcalan 11 months ago - 1 comment
Labels: Bug

#9514 - [1.15.x] segfault in IOV GPU recv

Issue - State: open - Opened by raffenet 11 months ago - 2 comments
Labels: Bug

#9513 - UCS/RCACHE: Synchronize events

Pull Request - State: closed - Opened by Artemy-Mellanox 11 months ago

#9512 - UCP/WIREUP: Consider local distance during slow lanes dropping

Pull Request - State: closed - Opened by ivankochin 11 months ago

#9510 - UCT/IB: fixed reverse_sl feature - removed code-dup and parentheses

Pull Request - State: closed - Opened by roiedanino 11 months ago - 2 comments

#9508 - BINDINGS/GO: Fixed make install.

Pull Request - State: closed - Opened by rakhmets 11 months ago

#9507 - UCP: Fine-grained intra/inter config for rendezvous

Pull Request - State: closed - Opened by iyastreb 11 months ago - 9 comments

#9506 - UCP: Fine-grained intra/inter config for rendezvous

Pull Request - State: closed - Opened by iyastreb 11 months ago

#9505 - BINDINGS/GO/TESTS: Fixed inactive CUDA context failure.

Pull Request - State: closed - Opened by rakhmets 11 months ago

#9504 - UCP: API - adding priority attribute to ucp requests

Pull Request - State: closed - Opened by roiedanino 11 months ago - 5 comments
Labels: Feature, API, WIP-DNM

#9503 - UCT/ROCM: remove rcache_addr_align parameter

Pull Request - State: closed - Opened by edgargabriel 12 months ago

#9502 - UCP: Fix strong fence to always ensure ordering - 1.16.x

Pull Request - State: closed - Opened by brminich 12 months ago

#9502 - UCP: Fix strong fence to always ensure ordering - 1.16.x

Pull Request - State: open - Opened by brminich 12 months ago

#9501 - UCP/PROTO: Unify data type checks

Pull Request - State: closed - Opened by ivankochin 12 months ago - 3 comments

#9501 - UCP/PROTO: Unify data type checks

Pull Request - State: closed - Opened by ivankochin 12 months ago - 3 comments

#9500 - UCT/IB/UD: Completed flush immediately if no AM posted.

Pull Request - State: open - Opened by rakhmets 12 months ago

#9500 - UCT/IB/UD: Completed flush immediately if no AM posted.

Pull Request - State: closed - Opened by rakhmets 12 months ago

#9499 - UCP/PROTO: Do not change multi-fragment perf for 1 fragment range

Pull Request - State: closed - Opened by ivankochin 12 months ago - 3 comments

#9499 - UCP/PROTO: Do not change multi-fragment perf for 1 fragment range

Pull Request - State: open - Opened by ivankochin 12 months ago - 1 comment

#9498 - Invalid Device Context and Seg Fault with UCX+MPI+PyTorch

Issue - State: open - Opened by snarayan21 12 months ago - 5 comments
Labels: Bug

#9498 - Invalid Device Context and Seg Fault with UCX+MPI+PyTorch

Issue - State: open - Opened by snarayan21 12 months ago - 4 comments
Labels: Bug

#9496 - UCP/MM: Fix alignment when rcache disabled

Pull Request - State: closed - Opened by Artemy-Mellanox 12 months ago - 3 comments

#9495 - TEST/RKEY: Check rkey distance with accordance to fp8 precision

Pull Request - State: closed - Opened by ivankochin 12 months ago - 1 comment

#9494 - CONTRIB: Add squash script to use after PR approval

Pull Request - State: closed - Opened by tvegas1 12 months ago - 2 comments

#9491 - AZP: Add L40G test

Pull Request - State: open - Opened by Alexey-Rivkin 12 months ago - 16 comments

#9490 - ucp_config_modify appends variable instead of changing the field

Issue - State: closed - Opened by ramsluk 12 months ago - 2 comments
Labels: Bug

#9489 - UCT/IB/DC: add env for reverse_sl (default: same value of sl)

Pull Request - State: closed - Opened by roiedanino 12 months ago

#9487 - UCT/TCP: Filtered out bridge devices - v1.16.x

Pull Request - State: closed - Opened by rakhmets 12 months ago

#9486 - BINDINGS/JAVA: Set CUDA device before allocating CUDA memory.

Pull Request - State: closed - Opened by rakhmets 12 months ago

#9485 - ucx failed to register memory

Issue - State: open - Opened by skypexu 12 months ago - 3 comments
Labels: Bug

#9484 - UCT/IB: Support select gid by ndev

Pull Request - State: open - Opened by jeynmann 12 months ago - 3 comments

#9483 - UCT/IB/MLX5: using vst4q_u64 vector copy (SIMD) for arm neon

Pull Request - State: closed - Opened by roiedanino 12 months ago - 1 comment

#9482 - Failed to build: "cannot define new methods on non-local type *C.ucp_request_param_t"

Issue - State: open - Opened by banana-bred 12 months ago - 1 comment
Labels: Bug

#9481 - UCT/MM: Add trace for posix transport when shm has no space left

Pull Request - State: closed - Opened by tvegas1 12 months ago

#9480 - UCP/PROTO: Add profiling to memory registration and rendezvous flows

Pull Request - State: closed - Opened by yosefe 12 months ago

#9479 - ARCH/ARM: Add Nvidia vendor and Grace CPU

Pull Request - State: closed - Opened by tvegas1 12 months ago

#9478 - UCS/SYS: Fix loading UCS modules for executable file with dot in name

Pull Request - State: open - Opened by dmitrygx 12 months ago - 2 comments
Labels: Approved pending CLA

#9477 - UCS/TOPO: Report NUMA node distance for sys root only

Pull Request - State: closed - Opened by ivankochin 12 months ago - 2 comments

#9475 - UCT/TCP: Filtered out bridge devices.

Pull Request - State: closed - Opened by rakhmets 12 months ago

#9474 - UCP/WIREUP: Calculate RMA score with regard to local distance BW

Pull Request - State: closed - Opened by ivankochin 12 months ago

#9473 - UCP/EAGER/STREAM: Fix missing proto initializations

Pull Request - State: closed - Opened by yosefe 12 months ago

#9472 - UCT/IB/MLX5: Refactor KSM functions and extend logging

Pull Request - State: closed - Opened by yosefe 12 months ago - 4 comments

#9471 - UCP: Minor code cleanup

Pull Request - State: closed - Opened by yosefe 12 months ago

#9470 - CONTRIB/PR_MERGE: Allow specifying head commit to compare with

Pull Request - State: closed - Opened by yosefe 12 months ago

#9467 - UCP/PROTO: Replace ucp protocol in the middle of send operation (offset > 0)

Pull Request - State: open - Opened by shasson5 12 months ago - 7 comments

#9466 - UCP: Fix lane selection for wireup ack message

Pull Request - State: closed - Opened by brminich 12 months ago - 2 comments

#9465 - UCP/PROTO: Calculate perf for ack messages as parallel stages - v1.16

Pull Request - State: closed - Opened by ivankochin 12 months ago - 1 comment

#9463 - UCP/WIREUP: Make rma score same as rma_bw score - v1.16.x

Pull Request - State: closed - Opened by yosefe 12 months ago

#9462 - Performance issue, NVidia H100

Issue - State: open - Opened by angainor 12 months ago - 1 comment
Labels: Bug

#9461 - Can ucx support to deal with SIGSTOP and SIGCONT？

Issue - State: open - Opened by razor1991 almost 1 year ago - 3 comments

#9460 - AZP: Add wire-compat check for perf caps

Pull Request - State: closed - Opened by brminich almost 1 year ago - 3 comments

#9459 - UCT/IB/MLX5/DV: disable device memory if atomics are not available

Pull Request - State: closed - Opened by roiedanino almost 1 year ago
Labels: Bugfix

#9458 - AZP: Add wire-compat tests with v1.16.x

Pull Request - State: closed - Opened by brminich almost 1 year ago - 2 comments

#9457 - (v1.16.x) CONFIGURE/ROCM: modify HIP_CPPFLAGS for ROCm 6.0

Pull Request - State: closed - Opened by nileshnegi about 1 year ago

#9456 - (v1.15.x) CONFIGURE/ROCM: modify HIP_CPPFLAGS for ROCm 6.0

Pull Request - State: closed - Opened by nileshnegi about 1 year ago

#9455 - CONFIGURE/ROCM: modify HIP_CPPFLAGS for ROCm 6.0

Pull Request - State: closed - Opened by nileshnegi about 1 year ago - 2 comments

#9454 - CONFIG/SPEC: Bump version to 1.17.0

Pull Request - State: closed - Opened by yosefe about 1 year ago
Labels: Documentation

#9453 - UCT/IB: added atomic_mem_types attributes to md_query_v2

Pull Request - State: closed - Opened by roiedanino about 1 year ago

#9451 - AZP: Ensure nv_peer_mem loaded on GPU

Pull Request - State: closed - Opened by Alexey-Rivkin about 1 year ago - 1 comment

#9450 - UCT/TCP: Order network interface listing to produce stable sys_dev number

Pull Request - State: closed - Opened by tvegas1 about 1 year ago - 3 comments

#9449 - UCT/MD/IB: Support MT KSM registration for unaligned buffers

Pull Request - State: open - Opened by ivankochin about 1 year ago

#9448 - GIT/GITIGNORE: Add .noinst directories and test_no_cuda_ctx

Pull Request - State: closed - Opened by ivankochin about 1 year ago

#9447 - UCT/IB: Skip multi-thread memory registration for symmetric key

Pull Request - State: closed - Opened by tvegas1 about 1 year ago - 2 comments

#9446 - UCP/WIREUP: Removed obsolete condition.

Pull Request - State: open - Opened by rakhmets about 1 year ago - 1 comment
Labels: WIP-DNM

#9445 - UCP: Fix RMA lanes selection for exported memh

Pull Request - State: closed - Opened by brminich about 1 year ago - 2 comments

#9444 - GIT/GITIGNORE: Add test_dlopen and test_no_cuda_ctx

Pull Request - State: closed - Opened by ivankochin about 1 year ago - 1 comment

#9443 - UCP/GTEST: Use strong fence with protov2

Pull Request - State: closed - Opened by brminich about 1 year ago - 2 comments

#9442 - cudamalloc fails with "the provided PTX was compiled with an unsupported toolchain"

Issue - State: closed - Opened by angainor about 1 year ago - 6 comments
Labels: Bug

#9441 - UCS/RCACHE: Add support for dynamic region alignment

Pull Request - State: closed - Opened by Artemy-Mellanox about 1 year ago

#9440 - UCP/TAG: Populate tag_info on immediate completion

Pull Request - State: closed - Opened by tvegas1 about 1 year ago - 6 comments

#9439 - UCS/CONFIG: Cleanup the configuration parser after profiler cleanup

Pull Request - State: closed - Opened by yosefe about 1 year ago - 2 comments

#9438 - READTHEDOCS: add recommonmark lib - v1.15.x

Pull Request - State: closed - Opened by Alexey-Rivkin about 1 year ago
Labels: Bugfix

#9437 - UCS/TCP: refactor ucs_netif_get_lowest_device_path

Pull Request - State: closed - Opened by brianplus about 1 year ago - 2 comments

#9436 - AZP: rm Centos8 refs to DockerHub - v1.15.x

Pull Request - State: closed - Opened by Alexey-Rivkin about 1 year ago - 6 comments
Labels: Bugfix