NVIDIA/cub issues and pull requests

#745 - BlockReduce<int, ...> causes "an illegal instruction was encountered"

Issue - State: open - Opened by intractabilis about 1 year ago - 3 comments

#744 - How do I reduce partially filled 2D blocks?

Issue - State: closed - Opened by intractabilis about 1 year ago - 2 comments

#743 - Possible bug in variable naming

Issue - State: closed - Opened by akshit-sharma about 1 year ago - 2 comments

#741 - nvcc fatal : Unsupported gpu architecture 'compute_80'

Issue - State: open - Opened by tyq996 about 1 year ago - 1 comment

#739 - BlockLoad never attempts to vectorize

Issue - State: closed - Opened by iclementine about 1 year ago - 5 comments

#738 - 64-bit indexing for DeviceSegmentedReduce

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

#737 - select_if kernel needs grid boundary or reprogramming tile_idx

Issue - State: closed - Opened by zhaolianshuizls about 1 year ago - 1 comment

#732 - what's the purpose of CUB_SUBSCRIPTION_FACTOR

Issue - State: closed - Opened by zhaolianshuizls about 1 year ago

#731 - Misleading documentation for DeviceSegmentedRadixSort (or I'm using it wrong)

Issue - State: closed - Opened by HapeMask about 1 year ago - 1 comment

#730 - Segmented sorting does not preserve data in-between segments.

Issue - State: closed - Opened by isovic about 1 year ago - 6 comments

#729 - What is the correct compile command in Linux platform to compile a function citing cuh?

Issue - State: closed - Opened by fengwang about 1 year ago - 1 comment

#728 - Fix instances of 'scan' copy-pasted into reduction documentation

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

#727 - Fix multiset erase call

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

#726 - Illegal memory access on trying to use `DeviceReduce::Sum()` to count number of non-zeros

Issue - State: closed - Opened by alexsamardzic over 1 year ago - 2 comments

#725 - Why are inplace overloads for BlockExchange not documented

Issue - State: open - Opened by pauleonix over 1 year ago - 1 comment
Labels: good first issue

#724 - Fix BlockAdjacentDifference documentation

Pull Request - State: closed - Opened by pauleonix over 1 year ago - 3 comments

#723 - Ensure that any CMake re-rooting doesn't break our find_file.

Pull Request - State: closed - Opened by robertmaynard over 1 year ago
Labels: monorepo blocker

#722 - Improve docs for warp-wide primitives

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed, monorepo blocker

#721 - Tune RLE for SM90

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed, monorepo blocker

#720 - Quickselect algorithm to select kth smallest element

Issue - State: open - Opened by mfbalin over 1 year ago - 4 comments

#719 - Can't get correct result when use cub in CUDA12.0

Issue - State: closed - Opened by YuanRisheng over 1 year ago - 24 comments

#718 - Tune select and partition for SM90

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#717 - Compile error: cannot find cuda_bf16.h

Issue - State: open - Opened by peizhang-cn over 1 year ago - 3 comments

#716 - Use `NV_IF_ELSE_TARGET`

Issue - State: open - Opened by senior-zero over 1 year ago

#715 - Fix reduce by key tile state for Pascal

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#714 - Workaround three-way partition compilation issue

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#713 - [EPIC] Design a scheme allowing CUB to process user-defined types of any size

Issue - State: open - Opened by jrhemstad over 1 year ago - 1 comment

#712 - Introduce SM90 tuning policy into scan

Pull Request - State: closed - Opened by senior-zero over 1 year ago - 1 comment
Labels: testing: gpuCI passed

#711 - Tune Decoupled Look-back based Algorithms for H100

Issue - State: closed - Opened by senior-zero over 1 year ago - 1 comment
Labels: cub

#710 - Introduce delay policy to decoupled look-back based algorithms

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#709 - Enable test coverage for int128 in DeviceHistogram::Even

Issue - State: open - Opened by elstehle over 1 year ago
Labels: area: tests

#708 - Fix CDP test wrapper for CTK 11.5

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#706 - Fix radix sort / MSVC 2017

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#705 - Segfault in CachingDeviceAllocator when out of memory

Issue - State: closed - Opened by orjgre over 1 year ago - 4 comments

#704 - Initial port to new monorepo build system.

Pull Request - State: open - Opened by allisonvacanti over 1 year ago - 1 comment

#703 - Use of <array> and <atomic> breaks some CUB algorithms with Jitify

Issue - State: open - Opened by maddyscientist over 1 year ago - 2 comments
Labels: helps: quda, compiler: nvrtc

#702 - Fix dependent template in radix sort

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#701 - Allow analysis script to process multiple dbs

Pull Request - State: closed - Opened by senior-zero over 1 year ago

#700 - Decoupled look-back example

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#699 - Write example for decoupled look-back API

Issue - State: closed - Opened by senior-zero over 1 year ago
Labels: cub

#698 - Implement scripts for detection of performance regressions across CUB versions

Issue - State: open - Opened by senior-zero over 1 year ago
Labels: cub

#697 - Implement VSMem abstraction

Issue - State: open - Opened by senior-zero over 1 year ago

#696 - Implement tuning db merger

Issue - State: closed - Opened by senior-zero over 1 year ago - 3 comments
Labels: cub

#695 - Make decoupled look-back delay part of tuning

Issue - State: closed - Opened by senior-zero over 1 year ago - 2 comments
Labels: cub

#694 - Add policy parameter to allow tuning

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#693 - Static assert arithmetic types in basic radix sort API

Issue - State: open - Opened by senior-zero over 1 year ago
Labels: cub

#692 - Unresolved extern function 'cudaLaunchDevice' error while using NVCC 11.x and cub 2.10 with -G

Issue - State: closed - Opened by lilohuang over 1 year ago - 3 comments

#691 - Provide Run-Length Decode API

Issue - State: open - Opened by senior-zero over 1 year ago

#690 - Add block-wide set operations like merge and intersection

Issue - State: open - Opened by fkallen over 1 year ago

#689 - Add policy parameter to allow tuning

Issue - State: closed - Opened by senior-zero over 1 year ago

#688 - CUB Tuning Infrastructure

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#687 - Fixes narrowing conversion warning in tests

Pull Request - State: closed - Opened by elstehle over 1 year ago
Labels: only: tests, area: tests

#686 - Re-enable extended test coverage for WarpReduce tests

Issue - State: open - Opened by elstehle over 1 year ago
Labels: only: tests, area: tests, cub

#685 - Remove extra fence in radix sort

Issue - State: open - Opened by senior-zero over 1 year ago

#684 - Removes unused variables in DeviceCopy::Batched

Pull Request - State: closed - Opened by elstehle over 1 year ago
Labels: only: tests, area: tests

#683 - Limits tests for generic WarpReduce tests to builtin types

Pull Request - State: closed - Opened by elstehle over 1 year ago
Labels: cub

#682 - Remove add_to_project action as it is no longer needed.

Pull Request - State: closed - Opened by jrhemstad over 1 year ago

#681 - Backport fixes for reordering in CUB member initializer lists into 2.0 branch

Pull Request - State: closed - Opened by ericniebler over 1 year ago
Labels: testing: gpuCI in progress

#678 - Fixes non_void_value_t for ptr-to-void iterator types

Pull Request - State: closed - Opened by elstehle over 1 year ago

#677 - Documentation of warp-wide collectives refers to `syncthreads` instead of `syncwarp`

Issue - State: closed - Opened by fkallen over 1 year ago - 1 comment

#675 - Initial implementation for fancy devicememcpybatch

Pull Request - State: closed - Opened by mfbalin over 1 year ago - 8 comments

#674 - Specialize DeviceMemcpy::Batched to also support iterators

Issue - State: closed - Opened by elstehle over 1 year ago - 1 comment
Labels: cub

#672 - DeviceMemcpy::Batched supports only memory buffers

Issue - State: closed - Opened by mfbalin over 1 year ago - 4 comments

#671 - Support user-defined types in radix sort and introduce initial sphinx docs

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#670 - Add WarpExchangeRegister class

Pull Request - State: closed - Opened by pb-dseifert over 1 year ago - 7 comments

#668 - Backport a few fixes from main

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#667 - Register-only based `WarpExchange`

Issue - State: closed - Opened by pb-dseifert over 1 year ago - 4 comments

#666 - Can cub::DeviceSegmentedReduce::Reduce support self-defined functor for struct variable instead of just integer?

Issue - State: closed - Opened by zlwu92 over 1 year ago - 2 comments

#665 - Fix uninitialized copy in block scan raking

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed

#664 - Optimize merge sort

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed, area: performance

#663 - Performance of small sums could be improved

Issue - State: open - Opened by seberg over 1 year ago - 1 comment

#662 - Apply `ArgMin`/`ArgMax` fix for `infinity` input to `Min`/`Max`

Issue - State: closed - Opened by nolmoonen over 1 year ago - 2 comments

#661 - Reduce RAM usage by radix sort tests

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI in progress, only: tests

#660 - Migrates remaining warp-scope tests to Catch2

Pull Request - State: closed - Opened by elstehle over 1 year ago
Labels: testing: gpuCI passed

#659 - Fix `cub::DeviceSpmv` for empty matrices

Pull Request - State: closed - Opened by senior-zero over 1 year ago - 2 comments
Labels: testing: gpuCI in progress

#658 - Deprecate cub::mutex and fix remaining narrowing conversions in tests

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed, release: breaking change

#657 - Fix conversion issues in block histogram test

Pull Request - State: closed - Opened by senior-zero over 1 year ago
Labels: testing: gpuCI passed, only: tests

#655 - Is there any way that I want to remove duplicates in a array but maintaining the original relative order in array using cuda library? That means not sorting.

Issue - State: closed - Opened by zlwu92 over 1 year ago

GitHub / NVIDIA/cub issues and pull requests