Update mult #66
Merged
Update mult #66
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
added 2 commits
June 25, 2023 12:00
Codecov Report
@@ Coverage Diff @@
## master #66 +/- ##
==========================================
- Coverage 93.97% 93.83% -0.15%
==========================================
Files 8 8
Lines 382 389 +7
==========================================
+ Hits 359 365 +6
- Misses 23 24 +1
... and 2 files with indirect coverage changes 📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
|
Will do, but I am only available after Tuesday next week, sorry! |
|
No rush.
…On Mon, Jun 26, 2023, 10:42 Bernat Font ***@***.***> wrote:
Will do, but I am only after Tuesday next week, sorry!
I will put up a reminder to myself :)
—
Reply to this email directly, view it on GitHub
<#66 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AADSKJ3CDLIFEREVEWWQ4D3XNFDQBANCNFSM6AAAAAAZT2DGUU>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
|
I have remade the benchmarks with Julia 1.9.2, for the serial 7c6b397, master, and this branch (multi-threading and GPU). Interestingly, the old serial results are much slower than the current results, and I have tested both Julia 1.8.5 and 1.9.2 since I am unsure what version I used previously. So now the speedups are not as prominent. I do not know what has changed honestly. Also, running the serial cases with v1.8.5 or v1.9.2 changes the results significantly (1.8.5 is two times faster than 1.9.2 wtf! I may need to redo this test to be 100% sure though). On the other hand, this PR does not affect the speedups w.r.t. Results below contain columns with: TGV
Donut
-- |
|
I have also added a new tag for the serial code before the GPU porting: v0.2.5 |
|
I have found 1.9 to be slower as well, but I hadn't timed it, or tracked it
down to a certain part of the code.
…On Thu, Jul 6, 2023, 10:52 Bernat Font ***@***.***> wrote:
I have remade the benchmarks with Julia 1.9.2, for the serial 7c6b397
<7c6b397>,
master, and this branch (multi-threading and GPU). Interestingly, the old
serial results are much slower than the current results, and I have tested
both Julia 1.8.5 and 1.9.2 since I am unsure what version I used
previously. So now the speedups are not as prominent. I do not know what
has changed honestly. Also, running the serial cases with v1.8.5 or v1.9.2
changes the results significantly (1.8.5 is two times faster than 1.9.2
wtf! I may need to redo this test to be 100% sure though). On the other
hand, this PR does not affect the speedups w.r.t. master so it is okay to
merge.
Results below contain columns with: Speedups: n | routine | CPU | GPU.
*TGV*
update_mult v1.9.2 vs old serial results
n=5 | sim_step! | 003.36 | 004.43
n=6 | sim_step! | 007.32 | 032.41
n=7 | sim_step! | 008.37 | 061.86
n=8 | sim_step! | 009.48 | 077.44
master v1.9.2 vs old serial results
n=5 | sim_step! | 003.27 | 004.53
n=6 | sim_step! | 001.13 | 030.18
n=7 | sim_step! | 008.13 | 063.90
n=8 | sim_step! | 009.39 | 077.55
update_mult v1.9.2 vs serial results v1.9.2
n=5 | sim_step! | 001.11 | 001.46
n=6 | sim_step! | 002.38 | 010.53
n=7 | sim_step! | 002.96 | 021.89
n=8 | sim_step! | 003.40 | 027.78
master v1.9.2 vs serial results v1.9.2
n=5 | sim_step! | 001.08 | 001.49
n=6 | sim_step! | 000.37 | 009.80
n=7 | sim_step! | 002.88 | 022.61
n=8 | sim_step! | 003.37 | 027.81
update_mult v1.9.2 vs serial results v1.8.5
n=5 | sim_step! | 000.63 | 000.83
n=6 | sim_step! | 001.55 | 006.84
n=7 | sim_step! | 001.67 | 012.35
n=8 | sim_step! | 001.90 | 015.49
master v1.9.2 vs serial results v1.8.5
n=5 | sim_step! | 000.61 | 000.85
n=6 | sim_step! | 000.24 | 006.37
n=7 | sim_step! | 001.62 | 012.76
n=8 | sim_step! | 001.88 | 015.52
*Donut*
update_mult v1.9.2 vs old serial results
n=4 | sim_step! | 002.08 | 002.01
n=5 | sim_step! | 015.84 | 025.00
n=6 | sim_step! | 021.77 | 100.17
n=7 | sim_step! | 023.06 | 188.89
master v1.9.2 vs old serial results
n=4 | sim_step! | 001.60 | 001.84
n=5 | sim_step! | 015.53 | 022.36
n=6 | sim_step! | 022.27 | 096.64
n=7 | sim_step! | 023.04 | 191.30
update_mult v1.9.2 vs serial results v1.9.2
n=4 | sim_step! | 000.64 | 000.62
n=5 | sim_step! | 005.14 | 008.11
n=6 | sim_step! | 007.12 | 032.75
n=7 | sim_step! | 007.61 | 062.35
master v1.9.2 vs serial results v1.9.2
n=4 | sim_step! | 000.49 | 000.57
n=5 | sim_step! | 005.04 | 007.25
n=6 | sim_step! | 007.28 | 031.60
n=7 | sim_step! | 007.60 | 063.15
update_mult v1.9.2 vs serial results v1.8.5
n=4 | sim_step! | 000.40 | 000.38
n=5 | sim_step! | 002.86 | 004.51
n=6 | sim_step! | 004.25 | 019.54
n=7 | sim_step! | 004.46 | 036.54
master v1.9.2 vs serial results v1.8.5
n=4 | sim_step! | 000.30 | 000.35
n=5 | sim_step! | 002.80 | 004.03
n=6 | sim_step! | 004.35 | 018.86
n=7 | sim_step! | 004.46 | 037.00
--
PS: Currently, this is all very manual, so we would need to automate it a
bit. Maybe when we start using a GPU in the CI it would be a good time to
do so.
—
Reply to this email directly, view it on GitHub
<#66 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AADSKJ26IUXVI6BEM32TZ5LXO3UMLANCNFSM6AAAAAAZT2DGUU>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Consolidate
multroutine. Less code to maintain (mess-up) and should be a tiny bit faster. Could you do a benchmark before we merge @b-fg?