aws/aws-ofi-nccl issues and pull requests

#649 - fix : Fix flexible array member allocation

Pull Request - State: closed - Opened by arunkarthik-akkart about 2 months ago - 1 comment

#648 - .ci/aws: All CI use ami with EFA Installer

Pull Request - State: open - Opened by a-szegel about 2 months ago - 1 comment

#647 - Cherry pick g4dn tests to v1.12.x-aws

Pull Request - State: closed - Opened by AmedeoSapio about 2 months ago

#646 - use C++ for unit tests

Pull Request - State: open - Opened by aws-nslick about 2 months ago

#645 - add `--enable-cpp` build flag

Pull Request - State: open - Opened by aws-nslick about 2 months ago

#644 - feat(ci/gha): enable unit tests for neuron builds

Pull Request - State: open - Opened by aws-nslick about 2 months ago

#643 - Cherry-picks for v1.12.x aws

Pull Request - State: closed - Opened by AmedeoSapio about 2 months ago - 3 comments

#642 - Cherry-picks for v1.11.x aws

Pull Request - State: closed - Opened by AmedeoSapio about 2 months ago - 3 comments

#641 - tuner: add regions for AllGather/ReduceScatter in the one rank per node case

Pull Request - State: closed - Opened by AmedeoSapio about 2 months ago - 1 comment

#640 - fix(rdma): send periodic control messages to sync sender/receiver

Pull Request - State: closed - Opened by rauteric about 2 months ago - 4 comments

#639 - testing

Pull Request - State: closed - Opened by vidsouza about 2 months ago - 2 comments

#638 - Add platform data settings for TRN2N

Pull Request - State: closed - Opened by maxtmann about 2 months ago

#637 - testing

Pull Request - State: closed - Opened by vidsouza about 2 months ago

#636 - vidsouza-p5-ub22-testing

Pull Request - State: closed - Opened by vidsouza about 2 months ago

#635 - only run ub2204 for debugging ssh issues

Pull Request - State: closed - Opened by a-szegel about 2 months ago - 3 comments

#634 - separate out 3rd-party headers

Pull Request - State: open - Opened by aws-nslick about 2 months ago - 1 comment

#633 - enable more warnings

Pull Request - State: open - Opened by aws-nslick about 2 months ago - 1 comment

#632 - feat(build): add -fanalyzer when --enable-werror

Pull Request - State: closed - Opened by aws-nslick about 2 months ago

#631 - Try an ami that passed a few days ago

Pull Request - State: closed - Opened by a-szegel about 2 months ago

#630 - fix: rdma: inverted print statement

Pull Request - State: closed - Opened by aws-nslick about 2 months ago

#629 - fix(init): fix sendrecv fallback logic

Pull Request - State: closed - Opened by aws-nslick about 2 months ago

#628 - fix(ci): prefer ecr to dockerhub

Pull Request - State: closed - Opened by aws-nslick about 2 months ago

#627 - Combined -Wextra -Werror Commits

Pull Request - State: closed - Opened by aws-nslick about 2 months ago - 4 comments

#626 - rdma: Use get_device_from_ep() accessor

Pull Request - State: closed - Opened by bwbarrett about 2 months ago - 4 comments

#625 - aws: Skip the WRITE_IN_ORDER_ALIGNED_128_BYTES check for P5en

Pull Request - State: open - Opened by rajachan about 2 months ago

#624 - feat(build): disable semantic interposition

Pull Request - State: closed - Opened by aws-nslick 2 months ago - 3 comments

#623 - fix(build): ensure -pthread is passed

Pull Request - State: closed - Opened by aws-nslick 2 months ago

#622 - fix(build): add missing AC_PROG_RANLIB

Pull Request - State: closed - Opened by aws-nslick 2 months ago

#621 - fix(rdma): stop setting FI_ORDER_NONE

Pull Request - State: closed - Opened by aws-nslick 2 months ago

#620 - Improve end of process cleanup and reporting

Pull Request - State: closed - Opened by bwbarrett 2 months ago

#619 - .ci/aws: re-Add trainium tests to CI

Pull Request - State: closed - Opened by a-szegel 2 months ago

#618 - feat: add DMA-BUF support

Pull Request - State: closed - Opened by aws-nslick 2 months ago - 6 comments

#617 - Fully destroy endpoints when refcount is 0

Pull Request - State: closed - Opened by bwbarrett 2 months ago
Labels: previously-passed-ci

#616 - fix(m4): set redzone size to 0

Pull Request - State: closed - Opened by rauteric 2 months ago
Labels: previously-passed-ci

#615 - Fix log format string behavior

Pull Request - State: closed - Opened by bwbarrett 2 months ago

#614 - rdma: add separate bounce buffer freelist for data (eager) messages

Pull Request - State: open - Opened by rauteric 2 months ago - 5 comments

#613 - util: Use FI_ENOPROTOOPT to check for a provider's support for option

Pull Request - State: closed - Opened by rajachan 2 months ago

#612 - CI updates

Pull Request - State: closed - Opened by rajachan 2 months ago

#611 - "Request completed with error" log leads to p5e cluster collapse

Issue - State: open - Opened by vmarkovtsev 2 months ago

#610 - Improve protocol selection logic

Pull Request - State: closed - Opened by bwbarrett 2 months ago

#609 - NCCL RDMA expects fi_cq_data_entry, but OPX provider fills CQ with fi_cq_tagged_entry

Issue - State: closed - Opened by lsavers 2 months ago - 2 comments

#608 - feat(ci/github): use docker instead of codebuild

Pull Request - State: closed - Opened by aws-nslick 2 months ago

#607 - fix(valgrind): fix autotools mistake

Pull Request - State: closed - Opened by aws-nslick 2 months ago

#606 - Initialization fails for OPX Libfabric Provider

Issue - State: closed - Opened by lsavers 2 months ago

#605 - fix(tree): import libfabric's container_of macro

Pull Request - State: closed - Opened by aws-nslick 2 months ago

#604 - Add Multiplexed-round-robin scheduler

Pull Request - State: closed - Opened by arunkarthik-akkart 2 months ago - 3 comments
Labels: previously-passed-ci

#603 - platform: trn1 default protocol send receive

Pull Request - State: closed - Opened by hunnorth 2 months ago - 5 comments

#602 - Fix: access domain from ep during mr on device

Pull Request - State: closed - Opened by maxtmann 2 months ago - 1 comment

#601 - feat(build): disable semantic interposition

Pull Request - State: closed - Opened by aws-nslick 2 months ago - 2 comments

#600 - freelist: separate out metadata from user data

Pull Request - State: open - Opened by rauteric 2 months ago - 1 comment
Labels: previously-passed-ci

#599 - Seg Fault during RDMA NCCL Connection with OPX Provider

Issue - State: closed - Opened by lsavers 2 months ago - 4 comments

#598 - fix(sendrecv): fix a memory leak

Pull Request - State: open - Opened by aws-nslick 2 months ago

#597 - No include folder after installation

Issue - State: closed - Opened by YJHMITWEB 2 months ago - 5 comments

#596 - feat(build): better --enable-debug defaults

Pull Request - State: closed - Opened by aws-nslick 2 months ago
Labels: previously-passed-ci

#595 - fix(platform-aws): fill all platform values

Pull Request - State: closed - Opened by aws-nslick 2 months ago

#594 - fix(tree): use empty brace initializers for zero-initialization

Pull Request - State: closed - Opened by aws-nslick 2 months ago - 2 comments

#593 - fix(tracing/nvtx): silence -Wmissing-field-initializer warnings

Pull Request - State: closed - Opened by aws-nslick 2 months ago
Labels: previously-passed-ci

#592 - feat(ci): add package generation

Pull Request - State: closed - Opened by aws-nslick 2 months ago

#591 - feat(rdma): constrain C linkage to init

Pull Request - State: closed - Opened by aws-nslick 2 months ago - 2 comments

#590 - fix(tracing): use header-only nvtx3

Pull Request - State: closed - Opened by aws-nslick 2 months ago
Labels: previously-passed-ci

#589 - fix(build): check features before mangling CFLAGS

Pull Request - State: closed - Opened by aws-nslick 2 months ago - 1 comment
Labels: previously-passed-ci

#588 - feat(build): add -Wextra to "picky" compiler flags

Pull Request - State: closed - Opened by aws-nslick 2 months ago

#587 - fix(test): fix typing issues

Pull Request - State: closed - Opened by aws-nslick 2 months ago
Labels: previously-passed-ci

#586 - fix(rdma): avoid enum/integral comparison

Pull Request - State: closed - Opened by aws-nslick 2 months ago
Labels: previously-passed-ci

#585 - fix(tree): add fallthrough switch markers

Pull Request - State: closed - Opened by aws-nslick 2 months ago - 1 comment
Labels: previously-passed-ci

#584 - register_mr_buffers:544 NCCL WARN NET/OFI Unable to register memory (type = 2) for device 0. RC: -22, Error: Invalid argument

Issue - State: open - Opened by visatish 3 months ago - 9 comments

#583 - fix(tuner): don't choose NVLSTree if nRanks==nNodes

Pull Request - State: closed - Opened by AmedeoSapio 3 months ago - 1 comment

#582 - chore(.github/workflows): constrain push triggers to known branches

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 2 comments
Labels: previously-passed-ci

#581 - fix(cuda): avoid loading stub

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 4 comments
Labels: previously-passed-ci

#580 - .ci/aws: Stop Running ofi nccl functional tests until they are fixed

Pull Request - State: closed - Opened by a-szegel 3 months ago - 1 comment

#578 - chore(build): replace `-Wc++-compat' with `-x c++'

Pull Request - State: closed - Opened by aws-nslick 3 months ago
Labels: previously-passed-ci

#577 - fix(neuron): remove const from ncclNetPlugin_v{4,5} syms

Pull Request - State: closed - Opened by aws-nslick 3 months ago
Labels: previously-passed-ci

#576 - fix(sendrecv): add missing nccl-headers include

Pull Request - State: closed - Opened by aws-nslick 3 months ago
Labels: previously-passed-ci

#575 - fix(tree): avoid sign comparison issues

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 1 comment
Labels: previously-passed-ci

#574 - fix(rdma): use COMM_ID_MASK as invalid id

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 2 comments

#573 - fix(tuner): fix implicit conversions

Pull Request - State: closed - Opened by aws-nslick 3 months ago
Labels: previously-passed-ci

#572 - fix(idpool): avoid sign comparison issues

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 1 comment

#571 - fix(param): move some parameters to unsigned

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 2 comments
Labels: previously-passed-ci

#570 - feat(param): add uint parameter macro

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 1 comment
Labels: previously-passed-ci

#569 - fix(tuner): avoid gotos

Pull Request - State: closed - Opened by aws-nslick 3 months ago
Labels: previously-passed-ci

#568 - feat(test): parse as c++ source

Pull Request - State: closed - Opened by aws-nslick 3 months ago

#567 - chore(build): mpi: set mpicxx, too.

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 2 comments

#566 - chore(build): add AC_PROG_CXX

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 1 comment

#565 - fix(tree): use decltype instead of typeof for cxx

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 1 comment
Labels: previously-passed-ci

#564 - fix(api): avoid mid-function initiializers

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 2 comments

#563 - fix(tree): move declarations to top of function

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 4 comments

#560 - fix(freelist): use uintptr_t for pointer arithmetic

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 1 comment
Labels: previously-passed-ci

#558 - fix(rdma): fi_{send,write}data: do arithmetic on uintptr

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 1 comment

#557 - fix(aws): align declaration and init order

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 1 comment

#556 - feat(tree): add static_assert shim macro

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 4 comments

#555 - fix(tree): add spaces around PRIu64

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 2 comments

#553 - rdma: Eliminate unnecessary ctrl message waits in eager protocol

Pull Request - State: closed - Opened by rauteric 3 months ago - 10 comments

#552 - .ci/aws: Unpin al2 p3dn ami

Pull Request - State: closed - Opened by a-szegel 3 months ago - 3 comments
Labels: previously-passed-ci

#542 - add ci build of rpms

Pull Request - State: closed - Opened by aws-nslick 3 months ago - 6 comments

#541 - Feature/v6 rma ops

Pull Request - State: closed - Opened by maxtmann 3 months ago - 9 comments
Labels: previously-passed-ci

#492 - ci: delete al2

Pull Request - State: closed - Opened by aws-nslick 4 months ago - 2 comments

GitHub / aws/aws-ofi-nccl issues and pull requests