Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Develop stream 2024-10-29 #631

Open
wants to merge 142 commits into
base: develop
Choose a base branch
from
Open

Develop stream 2024-10-29 #631

wants to merge 142 commits into from
+16,012 −2,574

Conversation

Beanavil
Copy link
Contributor

@Beanavil Beanavil commented Oct 29, 2024 •
edited
Loading

  • Remove HIP-CPU support
  • Move ROCPRIM_DETAIL_HIP_SYNC_AND_RETURN_ON_ERROR to a separate header file
  • Add adjacent_find implementation and config tuning
  • Fix data generation for int128 and custom types in tests
  • Add find_end implementation
  • Add search implementation
  • Use std::chrono::steady_clock instead of std::chrono::high_resolution_clock in debug_synchronous calls
  • Add deprecation warnings for C++14 and set c++17 as default
  • Optimize partial_sort making use of radix sort
  • Improve histogram documentation addressing Channels and ActiveChannels restrictions
  • Update batch_memcpy tests to not hardcode the random seed
  • Add new variant of device select, predicated_flag, along with tests, benchmarks and config tuning
  • Add config tuning and dynamic dispatch for device merge
  • Add search_n implementation

@Beanavil Beanavil force-pushed the develop_stream_20241029 branch 2 times, most recently from 149360c to eb5607f Compare October 29, 2024 23:02
@Beanavil Beanavil marked this pull request as ready for review October 30, 2024 04:19
@NB4444 NB4444 mentioned this pull request Oct 31, 2024
Open
@Beanavil Beanavil force-pushed the develop_stream_20241029 branch 7 times, most recently from 929e6e6 to dc63277 Compare November 6, 2024 17:02
Snektron and others added 14 commits November 13, 2024 13:21
@Snektron @Beanavil
remove HIP-CPU support
fdf01cf
@cenxuantian @Beanavil
Resolve: IssueMove ROCPRIM_DETAIL_HIP_SYNC_AND_RETURN_ON_ERROR to sep…
b8037ab 
…erate header file
@cenxuantian @Beanavil
rebase and add RETURN_ON_ERROR to the header
170ff5c
@Beanavil
Added naive implementation for adjacent_find plus tests and benchmarks
d0d46bf
@Beanavil
Improved benchmark by only taking into account relevant processed ele…
feda114 
…ments
@Beanavil
Use a faster reduction operation
cfe1849
@Beanavil
Added block-reduction kernel with early exit
a0d4243
@Beanavil
Improved test with random first pair
32bad70
@Beanavil
Get grid_size for maximum occupancy
b492542
@Beanavil
Improved test coverage
8e4c886
@Beanavil
Implement early exit with sequential blocks execution
f279aee
@Beanavil
Use a dynamic tile_id as in find_first_of for faster stable results
398c92f
@Beanavil
Added documentation for adjacent_find
8691694
@Beanavil
Added tuning for adjacent_difference
111f8dd
cenxuantian and others added 28 commits November 13, 2024 14:16
@cenxuantian @Beanavil
Resolve
9f17434
@cenxuantian @Beanavil
Fix bugs -- several occurrences of consecutive full blocks
7f39798
@cenxuantian @Beanavil
Many modifications, fixed the bugs and edited the tests and benchmarks
874c4e9
@cenxuantian @Beanavil
Optimised the block_search_n_kernel
73bd933
@cenxuantian @Beanavil
2nd version search_n implementation for large input
40ff702
@cenxuantian @Beanavil
Add thread level search_n algorithm
4897960
@cenxuantian @Beanavil
Add optimizations
5518552
@cenxuantian @Beanavil
Edit benchmarks
980054f
@cenxuantian @Beanavil
remove unused variables
11fc510
@cenxuantian @Beanavil
remove unused variables and remove __restrict__
18bcf5e
@cenxuantian @Beanavil
fix the bug on windows
af5f677
@cenxuantian @Beanavil
fix bug and modify benchmakrs and tests
0679cf1
@cenxuantian @Beanavil
fix bugs in benchmarks and search_n_impl
d09f947
@cenxuantian @Beanavil
Oh yes
ac8de49
@cenxuantian @Beanavil
Apply 1 suggestion(s) to 1 file(s)
b236c1e 
Co-authored-by: Beatriz Navidad Vilches <[email protected]>
@cenxuantian @Beanavil
apply some suggestions
c51d895
@cenxuantian @Beanavil
edit doc
e0dfa8c
@cenxuantian @Beanavil
replace search_n_min_kernal by rocprim:reduce
a8d3766
@cenxuantian @Beanavil
fixed some benchmarks bugs
63d315f
@cenxuantian @Beanavil
remove graph support
96ea3c7
@cenxuantian @Beanavil
resolve not compile on win
87f63ff
@cenxuantian @Beanavil
Add graph support and modified the design a little
d403145
@cenxuantian @Beanavil
resolve test fail on windws
84c379b
@cenxuantian @Beanavil
fix gfx960 benchmark dead lock
b8b6f9f
@cenxuantian @Beanavil
Add device_search_n to rocprim.hpp
afad937
@cenxuantian @Beanavil
replace HIP_CHECK by ROCPRIM_RETURN_ON_ERROR
5659468
@Beanavil
fix: fix doxygen error due to __launch_bounds__ macro
64833f4
@Beanavil
Implement 6.3 hotfixes for added/modified tests
b388325
@Beanavil
Copy link
Contributor Author

Rebased onto develop to fix merge conflicts

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@stanleytsang-amd stanleytsang-amd Awaiting requested review from stanleytsang-amd stanleytsang-amd is a code owner

@umfranzw umfranzw Awaiting requested review from umfranzw umfranzw is a code owner

@RobsonRLemos RobsonRLemos Awaiting requested review from RobsonRLemos RobsonRLemos is a code owner

@lawruble13 lawruble13 Awaiting requested review from lawruble13 lawruble13 is a code owner

At least 1 approving review is required to merge this pull request.

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
7 participants
@Beanavil @Snektron @cenxuantian @MiloLurati @NB4444 @parbenc @yssamtu