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

Experimental AVX512F kernels (protein data only) #1

Open
wants to merge 67 commits into
base: dev
Choose a base branch
from
Open

Experimental AVX512F kernels (protein data only) #1

wants to merge 67 commits into from

Conversation

RCBiczok
Copy link

@RCBiczok RCBiczok commented Feb 6, 2018 •
edited
Loading

Implemented experimental AVX512F kernels for:

  • pmatrix calculation
  • clv calculation
  • likelihood calculation
  • sumtable & derivatives calcualtion

Limitations:

  • No support for nucleotide data, repeats or tip-tip or inner-tip cases
  • Models other then nucleotide data or protein data work, but may don't increase the performance a lot

@RCBiczok
Extend gitignore to ignore build intermediates and generated stuff
b08b768
@RCBiczok
Migrate AVX512F kernels from libpll to libpll2
2db5084
@RCBiczok
Fix missing switch for AVX512 kernel execution
c06f6bb
@RCBiczok
Rename productive tests and move benchmark code to separate folder
43e56a7
@RCBiczok
Add missing stubs for tip-cases / site repeats
b61a46c
@RCBiczok
Ignore eclipse IDE files
f587be2
@RCBiczok
Add AVX512F derivative kernel for handling odd states
8166551
@RCBiczok
Fall back to vanilla AVX512 if tips mode or repeats mode has beedn re…
504fd32 
…quested by the caller
@RCBiczok
Skip tests that are not covered by AVX512F kernels yet
edb7088
@RCBiczok
Fix wrong calculation of rate mask
3aaff41
@RCBiczok
Fix rate_mask calculation
2b7de4c
@RCBiczok
Add instructions for per-site scaling
4c45169
@RCBiczok
Fix wrong per-rate scaling
a8b6423
@RCBiczok
Finish AVX512F kernel imlpementation for general case with per rate
b958938 
scaling
@RCBiczok
Finish per-rate scaling for gereral AVX512F kernel
7cf68bf 
Change-Id: I70b6de28ca931df295b4f40f9cb2f3fe1353ad43
@RCBiczok
Add altered test runner for running tests agains AVX512F kernels
848a59f
@RCBiczok
Execute tests with AVX512F by default (With/without tips / repeats)
237eb2b
@RCBiczok
Vectorize part for proportions of invariant sites properly
bca9b62
@RCBiczok
Remove printing functions
cdb4c13
@RCBiczok
Merge branch 'dev' of [email protected]:xflouris/libpll-2.git into dev
9e7a730
@RCBiczok
Ignore CLion intermediates
4d8f7eb
@RCBiczok
Add "dummy" cmake file (only to make clion work with libpll)
f4d1f5b
@RCBiczok
Add mode for SMID memory layout, and add filter for specific tests only
2fae567
@RCBiczok
Remove redundant macro
a22afb7
@RCBiczok
Add kernels for SMID memory layout
f4f7c33
@RCBiczok
Add benchmark programs and makefile
1343216
@RCBiczok
Remove accidentally added binary files
f096e97
@RCBiczok
Ignore files in obj directories
1f77f9f
@RCBiczok
Reduce number of sites to calculate
f4d947c
@RCBiczok
Fix inefficient clv initialization
3a765e7
@RCBiczok
Add benchmark programs
d995285
@RCBiczok
Fix memory leak
28c06bf
@RCBiczok
Add support for SIMD layout
685f4cd
@RCBiczok
Modify benchmark execution script
4540828
@RCBiczok
Comment out lines of code that causes compiler errors in ICC (only te…
c6674fe 
…mporary)
@RCBiczok
Split benchmark script into a less time consuming one and a full one
d5955ad
@RCBiczok
Add parameter for printing the randomly generated sequences
e1846a9
@RCBiczok
Repeat benchmarks for specific configuration 10 times
2ac0a0b
@RCBiczok
Replace inline function by proper macros
f75404a
@RCBiczok
Add switch for DNA / AA data benchmark
ea67a37
@RCBiczok
Delete old benchmark programs and add micro-benchmark for partials ke…
3588acd 
…rnel. Also add DAZ optimization to partials micro benchmark
@RCBiczok
Add partials kernel for AVX2 sml
10a543b
@RCBiczok
Add error handling for tips setting
ce03564
@RCBiczok
Add test case for partials calculations with DNA data
9eb3300
@RCBiczok
Add SML kernels for partials calculation on DNA data
78807d4
@RCBiczok
Avoid CLV padding for avx2 SML kernels as well
fcc10d4
@RCBiczok
Prefetch clv vectors for one row
9ce778f
@RCBiczok
Fix typo in macro name
e77322d
@RCBiczok
Unroll two innermost loops in DNA kernel
5f7e67e
@RCBiczok
Add kernel variants as comments (TODO only temporary)
bba51d0
@RCBiczok
Add benchmark parameters and add simple benchmark program with an ver…
f57fb11 
…y simple algorithm for AVX2 and AVX512
@RCBiczok
Calcualte full cash footproint
15bfc42
@RCBiczok
Prefetch CLV in protein kernels
20c42d5
@RCBiczok
Add SIMD counting routines / mode
ef8c2ca
@RCBiczok
Update documentation
7fb6488
@RCBiczok
Add parameter for building libpll with SIMD counting
518cbad
@RCBiczok
ADd test program for observing performance degeneration in SIMD kerne…
3013ad7 
…ls with increasing array size
@RCBiczok
Add additional parameters form generating CSV reports
62442d2
@RCBiczok
Add benchmark script for generating reports with varying number of sites
6b6896c
@RCBiczok
Increase number of repeated iterations within one time measurement
e653987
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
1 participant
@RCBiczok