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

Add GPU sketching #32

Merged

johnlees merged 127 commits into master from gpu_sketch

Aug 24, 2020

Member

johnlees commented Aug 13, 2020

Allows the use of --use-gpu with --sketch (read data only)
128 genomes x 16 CPUs = 55 mins
128 genomes x 16 CPUs + 1 GPU = 4 mins

API breaking, as usual (version release will require corresponding update in PopPUNK)

Closes #25, closes #28

johnlees added 30 commits

August 13, 2020 14:58


          Outline of CUDA sketching (for reads)

6aa7e7c


          Update hash used in countmin

58d35a4


          Add nthash const arrays

359ea3a


          Bump version

1afb96c


          Work on the read device object

23ba16c


          Move reused device memory into objects

7d6b58a


          Connect up with API


          Reorganise pp_sketch cmd line args

3b47810


          Add new .cu files to Makefiles

5aca5c6


          Update supported CUDA compute versions

52a3db8


          Correct cuda.cuh error handling

d5ac365


          Copy nthash tables in constant device memory

5be2fa1


          Use 32 bit int with atomicInc

4b4adc2


          Copy all ntHash tables

4acb590


          Fix pointer type in memcpy

6e2d05c


          Add device seedTab

1bb5959


          Add headers to gpu_api for sketching

7a01f2d


          Add files header

c8a49d3


          gpu_api compiler errors

21d1772


          API headers

428cdc1


          Fix define block variable scope in bindings

c1da674


          More reordering of headers

db8d686


          Correct call of CUDA get_signs

d81a1a0


          Disambiguate progress functions

471bc20


          Correctly add dlto

f6d3edf


          Single definition of initialise device

b67f57e


          Update read_test for GPU

6b5c964


          Add manual GPU option to makefile

c72a8e6


          Add CUDA lib to Makefile

3d21f7a


          Fix max length in SeqBuf

c4c9258

johnlees commented

View reviewed changes

src/gpu/dist.cu Show resolved Hide resolved

johnlees commented

View reviewed changes

src/gpu/gpu_api.cpp Outdated Show resolved Hide resolved

johnlees commented

View reviewed changes

src/gpu/gpu_api.cpp Outdated Show resolved Hide resolved

johnlees commented

View reviewed changes

src/gpu/sketch.cu Show resolved Hide resolved

johnlees commented

View reviewed changes

src/gpu/sketch.cu Outdated Show resolved Hide resolved

johnlees commented

View reviewed changes

src/gpu/sketch.cu Outdated Show resolved Hide resolved

johnlees commented

View reviewed changes

src/gpu/sketch.cu Outdated Show resolved Hide resolved

johnlees commented

View reviewed changes

src/gpu/sketch.cu Outdated Show resolved Hide resolved

johnlees commented

View reviewed changes

src/sketch/seqio.cpp Show resolved Hide resolved

johnlees commented

View reviewed changes

src/sketch/sketch.cpp Outdated Show resolved Hide resolved

johnlees commented

View reviewed changes

src/sketch/sketch.cpp Outdated Show resolved Hide resolved

johnlees added 12 commits

August 24, 2020 11:06


          Add different install options to setup.py

9c7ff96


          Try to pass option to build_ext

54e085f


          Move and parallelise dist host functions

8a1ce8f


          Minor changes and style from self-review

cbfb186


          Use environment variable to change setup type

0780fd8


          Fix some prototypes

88c5aa9


          Patch env in setup.py

ef17e79


          Move warp_size into correct place

e07492a


          Attempt to make GPU sketch async

6a3d0af

Closes #33


          Undo package name in setup.py

ff879a8


          Wrap async args in std::ref

3e41837


          Fix async template type

956b320

const iterator, but iterator is not const

johnlees linked an issue

that may be closed by this pull request

Asynchronous GPU/CPU use in sketching #33

Closed

johnlees added 4 commits

August 24, 2020 15:56


          Fix file_it iteration

1255d3e


          using j outside of its loop

a06b368


          Need to use the next batch size in async call

85bb87e


          i -> idx

edae9c1

johnlees merged commit ef169e1 into master

johnlees deleted the gpu_sketch branch

August 24, 2020 15:51

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

Labels

None yet