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

Cfft #84

Merged
merged 52 commits into from
Oct 12, 2023
Merged

Cfft #84

merged 52 commits into from
Oct 12, 2023

Conversation

mbertuletti
Copy link
Collaborator

@mbertuletti mbertuletti commented Jun 21, 2023 •
edited
Loading

Summarize the changes concisely

Changelog

Added Cooley-Turkey FFT kernel.

Added

Cooley-Turkey radix-4 and radix-2 FFT, parallel and single-core. FFT with a local load-store scheme in memory, to reduce the LSU stalls during computations.

Checklist

  • Automated tests pass
  • Code style guideline is observed

Please check our contributing guidelines before opening a Pull Request.

SamuelRiedel
SamuelRiedel reviewed Oct 11, 2023
View reviewed changes
Copy link
Member

@SamuelRiedel SamuelRiedel left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good to me in general. I have a few questions and comments, and of course, we have to get the CI to pass.

software/apps/Makefile Outdated Show resolved Hide resolved
software/apps/cfft_radix2_q16/main.c Outdated Show resolved Hide resolved
software/apps/cfft_radix4_q16/main.c Outdated Show resolved Hide resolved
software/apps/cfft_radix4_q16/main.c Outdated Show resolved Hide resolved
software/runtime/data/data_cfft_radix4_q16.py Show resolved Hide resolved
@mbertuletti
[apps] Add radix-4 cfft kernel
9fd7baa
@mbertuletti
[apps] Add xpulpimg extensions
fa1905d
@mbertuletti
[apps] Add radix-2 cfft kernel
b6ad9a9
@mbertuletti
[apps] Add log-barriers in code
2b6bcc0
@mbertuletti
[apps] Add partial barriers in code
7b2907d 
[apps] Add partial barrier in code 2
@mbertuletti
[apps] Add cfft with memory accesses to data shaped to MemPool
f7d330b
@mbertuletti
[apps] Add cfft with memory access to twiddles shaped to MemPool
76d97fa
@mbertuletti
[apps] Change folder name to radix-4
95e7591
@mbertuletti
[apps] Correct performance monitoring in parallel cfft
45c3ba6
@mbertuletti
[apps] Correct parallel bit-reversal stage
b4ea04e
@mbertuletti
[apps] Add scheduling of multiple parallel FFTs over different cores
0ce676a
@mbertuletti
[apps] Reduce acc-stalls in Xpulpimg single-core version
33f3a40
@mbertuletti
[software] Add separate header file for butterfly functions
cc9acb1
@mbertuletti
[software] Erase unnecessary test define for scheduled fft
c9ffdff
@mbertuletti
[software] Add unrolling in bit-reversal
50e3a95
@mbertuletti
[software] Hide IPU raw stalls with ASM statements
6766c5f
@mbertuletti
[software] Parallelize initialization functions and simplify main
6ce5bba
@mbertuletti
[software] Add bit-reversal with address calculation
1883412
@mbertuletti
[software] Correct access to columns of scheduled FFTs
030ba88
@mbertuletti
[software] Add function descriptions
89f7a61
@mbertuletti
[software] Correct lint errors in radix-2 FFT
c39864d 
[software] Correct lint errors
@mbertuletti
[software] Rewrite main
fb749d3
@mbertuletti
[software] Fix bitreversal without bitreverse table
cd9c751 
x
@mbertuletti
[software] Add comments, correct lint errors
6191130
@mbertuletti
[software] Substitute pv.pack.h with pv.pack
48c48a0
@mbertuletti
[software] Correct gcc errors
2a5916e
@mbertuletti
[software] Add python golden model
cbdab7d
@mbertuletti
[software] Fix kernels according to golden model
dc167c4
@mbertuletti
[software] Move kernels to runtime/kernels
24002fa
@mbertuletti
[software] Unfold after bitreversal
93fe412
@mbertuletti
[software] Clean up
dfb1471
@mbertuletti
[software] Teporarily disable llvm compilation for xpulpimg in CI
1cd045d
@mbertuletti
[software] Generate data with python scripts
746129b
@mbertuletti
[software] Select correct output
6d6008a
@mbertuletti
[software] Update python generation scripts and add sympy requirement
db12c5a
@mbertuletti
[software] Fix scheduling
995c66e
@mbertuletti
[sotware] Load data unfolded
80f768d
@mbertuletti
[software] Unroll bitreversal in parallel scheduled fft
165aa4b
@mbertuletti
[software] Add macros for twiddles shuffling and storing
4420155
@mbertuletti
[software] Align butterfly loops for folded ffts
1ab79bd
@mbertuletti
[software] Add macros for bitreversal computation
95d073b
@mbertuletti
[software] Fix compilation warning for folded scheduled fft
83d7af6
@mbertuletti
[software] Load data with DMA
9a025e5
@mbertuletti
[software] Disable LLVM compilation
ac904c7
@mbertuletti
Update Changelog
3023765
SamuelRiedel
SamuelRiedel approved these changes Oct 12, 2023
View reviewed changes
Copy link
Member

@SamuelRiedel SamuelRiedel left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM. Thanks.

@mbertuletti mbertuletti merged commit 715fa0b into main Oct 12, 2023
62 checks passed
@mbertuletti mbertuletti deleted the cfft branch October 12, 2023 14:35
nicomar0 added a commit to nicomar0/mempool that referenced this pull request Nov 21, 2023
@nicomar0
Merge pull request pulp-platform#84 from pulp-platform/cfft
74e7ca3 
Add cfft radix-4 and radix-2 kernels
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@SamuelRiedel SamuelRiedel SamuelRiedel approved these changes

@suehtamacv suehtamacv Awaiting requested review from suehtamacv

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

Successfully merging this pull request may close these issues.
2 participants
@mbertuletti @SamuelRiedel