Perform packing/unpacking of arguments and results in HDL #18
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
martijnbastiaan
force-pushed
the
remove-simulation-hack
branch
from
f51c511 to
f797bbf
Compare
martijnbastiaan
force-pushed
the
remove-simulation-hack
branch
from
f797bbf to
9e480e0
Compare
martijnbastiaan
force-pushed
the
remove-simulation-hack
branch
2 times, most recently
from
4236d7f to
14620c8
Compare
martijnbastiaan
force-pushed
the
use-bbf
branch
2 times, most recently
from
3a7f094 to
6e1f735
Compare
Running three experiments with the old and the new code shows that avoiding packing/unpacking in Haskell yields a ~20-30% performance increase for longer running tests. Table below shows the minimum value for all runs. | Test | main (min) | branch (min) | faster | faster (>1s) | |----------------------------|------------|--------------|---------|--------------| | dyn_dispatch (debug) | 0.34 | 0.31 | 8.82% | | | ebreak_exception (debug) | 1.42 | 1.08 | 23.94% | 23.94% | | fpu_test (debug) | 3.42 | 2.4 | 29.82% | 29.82% | | hello_world (debug) | 0.47 | 0.38 | 19.15% | | | long_string (debug) | 4.97 | 3.37 | 32.19% | 32.19% | | loop_write (debug) | 9.7 | 6.64 | 31.55% | 31.55% | | many_calls (debug) | 0.93 | 0.74 | 20.43% | 20.43% | | print_a (debug) | 0.31 | 0.31 | 0.00% | | | print_b (debug) | 0.39 | 0.32 | 17.95% | | | print_numbers (debug) | 3.8 | 2.71 | 28.68% | 28.68% | | single_write (debug) | 0.82 | 0.6 | 26.83% | | | dyn_dispatch (release) | 0.2 | 0.24 | -20.00% | | | ebreak_exception (release) | 0.27 | 0.2 | 25.93% | | | fpu_test (release) | 0.71 | 0.5 | 29.58% | | | hello_world (release) | 0.2 | 0.19 | 5.00% | | | long_string (release) | 0.35 | 0.25 | 28.57% | | | loop_write (release) | 0.95 | 0.8 | 15.79% | | | many_calls (release) | 0.27 | 0.28 | -3.70% | | | print_a (release) | 0.19 | 0.23 | -21.05% | | | print_b (release) | 0.16 | 0.17 | -6.25% | | | print_numbers (release) | 0.49 | 0.43 | 12.24% | | | single_write (release) | 0.3 | 0.23 | 23.33% | | | JTAG | 33.33 | 25.89 | 22.32% | 22.32% | | JTAG chain | 126.67 | 89.85 | 29.07% | 29.07% | | | | | 15.84% | 27.25% | Figures are taken by observing clash-vexriscv-sim:unittests's times.
martijnbastiaan
force-pushed
the
use-bbf
branch
from
6e1f735 to
582a4a1
Compare
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.
Running three experiments with the old and the new code shows that avoiding packing/unpacking in Haskell yields a ~20-30% performance increase for longer running tests. Table below shows the minimum value for all runs.
Figures are taken by observing
clash-vexriscv-sim:unittests's debug times.I'm overrunning my timebox though, so I'm putting this as "Draft".
TODO
bittide-hardware