profile
viewpoint
If you are wondering where the data of this site comes from, please visit https://api.github.com/users/jimt/events. GitMemory does not store any data, but only uses NGINX to cache data for a period of time. The idea behind GitMemory is simply to give users a better reading experience.
Jim Tittsler jimt Ito, JAPAN https://WikiEducator.org/User:JimTittsler 7J1AJH / AI8A / ZL2IA @jimt@mastodon.OERu.org

gatsbyjs/gatsby 50241

Build blazing fast, modern apps and websites with React

drowe67/codec2 277

Open source speech codec designed for communications quality speech between 450 and 3200 bit/s. The main application is low bandwidth HF/VHF digital radio.

jimt/docker-mwflow 1

docker Ubuntu image with Mediawiki and Flow extension

DigitalVoiceNZ/xlxd 0

HAM radio multiprotocol dstar reflector server

jimt/atom-vim-mode-plus 0

vim-mode improved

jimt/bacwn 0

clojure + datalog = <3

jimt/barrel 0

Distributed database for the modern world

jimt/bGeigieNanoKit 0

bGeigieNano is a kit version of the bGeigie mobile survey geiger counter designed to fit into a Pelican Micro Case 1010.

jimt/books 0

Repo for collecting all LFE books under http://lfe.io/books/

jimt/Building-Blocks 0

Reusable components for Firefox OS

pull request commentdrowe67/codec2

Optimising cohspk modem

I remember this loop being quite intensive from before. I wouldn't imagine anyone running cohpsk on a small device. Even the SM1000 would probably collapse. But if it can be done, a PI4 might be good enough.

Yes the acquisition system would need re-writing for a very small system like the SM1000. It appears to be running OK (just) on my PI 3B now https://github.com/drowe67/freedv-gui/issues/126

drowe67

comment created time in 2 hours

pull request commentdrowe67/codec2

Optimising cohspk modem

Built on your optimization to compute inner only once.

Thanks @srsampson Did you measure any performance improvement?

I think most of the CPU is being used in this loop, in particular the rate_Fs_rx_processing(). In this modem, we actually do a complete demod at 3 different freq shifts, then do a trial freq estimation on the baseband symbols. So it's pretty CPU heavy, but that's not really a problem for most x86 platforms.

I don't have a PI example ready, but I didn't see any difference on my Ubuntu running the command line. I need to round-up my parts and build a PI for freedv.

I remember this loop being quite intensive from before. I wouldn't imagine anyone running cohpsk on a small device. Even the SM1000 would probably collapse. But if it can be done, a PI4 might be good enough.

drowe67

comment created time in 2 hours

pull request commentdrowe67/codec2

Optimising cohspk modem

100% pass now with the latest commit I pushed. +1

Nice catch :+1: - thanks @tmiw

drowe67

comment created time in 2 hours

pull request commentdrowe67/codec2

Optimising cohspk modem

Built on your optimization to compute inner only once.

Thanks @srsampson :+1: Did you measure any performance improvement?

I think most of the CPU is being used in this loop, in particular the rate_Fs_rx_processing(). In this modem, we actually do a complete demod at 3 different freq shifts, then do a trial freq estimation on the baseband symbols. So it's pretty CPU heavy, but that's not really a problem for most x86 platforms.

drowe67

comment created time in 2 hours

pull request commentdrowe67/codec2

Optimising cohspk modem

100% pass now with the latest commit I pushed. 👍

drowe67

comment created time in 4 hours

push eventdrowe67/codec2

Mooneer Salem

commit sha 30b15db490cfe14eb6bf1fc3083c0d092fc00986

Fix unit test failure on macOS.

view details

push time in 4 hours

pull request commentdrowe67/codec2

Optimising cohspk modem

One test failure on macOS with this branch:

HT-TM05:build_osx mooneer$ ctest -V -R test_OFDM_modem_esno_est_c
UpdateCTestConfiguration  from :/Users/mooneer/devel/freedv-gui/codec2/build_osx/DartConfiguration.tcl
Parse Config file:/Users/mooneer/devel/freedv-gui/codec2/build_osx/DartConfiguration.tcl
UpdateCTestConfiguration  from :/Users/mooneer/devel/freedv-gui/codec2/build_osx/DartConfiguration.tcl
Parse Config file:/Users/mooneer/devel/freedv-gui/codec2/build_osx/DartConfiguration.tcl
Test project /Users/mooneer/devel/freedv-gui/codec2/build_osx
Constructing a list of tests
Done constructing a list of tests
Updating test list for fixtures
Added 0 tests to meet fixture requirements
Checking test dependency graph...
Checking test dependency graph end
test 16
    Start 16: test_OFDM_modem_esno_est_c

16: Test command: /bin/sh "-c" "cd /Users/mooneer/devel/freedv-gui/codec2/octave;
                            echo 'pkg load signal; esno_est; esno_est_tests_c' |
                            PATH_TO_UNITEST=/Users/mooneer/devel/freedv-gui/codec2/build_osx/unittest/ DISPLAY="" octave-cli"
16: Test timeout computed to be: 1500
16: EsNo:  5.00 EsNo_meas:  4.84 EsNo_est:  4.71
16: 
16: Running C version....
16: path_to_unittest = ../build_linux/unittest
16: path_to_unittest_exe = 
16: setting path from env var to ../build_linux/unittest
16: sh: ../build_linux/unittest/tesno_est: No such file or directory
16: error: load: unable to determine file format of 'tesno_est_out.txt'
16: error: called from
16:     esno_est_test_c at line 133 column 5
16:     esno_est_tests_c at line 143 column 5
1/1 Test #16: test_OFDM_modem_esno_est_c .......***Failed  Required regular expression not found. Regex=[AWGN Pass.*MPP Pass
]  0.22 sec

0% tests passed, 1 tests failed out of 1

Total Test time (real) =   0.23 sec

The following tests FAILED:
	 16 - test_OFDM_modem_esno_est_c (Failed)
Errors while running CTest
drowe67

comment created time in 4 hours

pull request commentdrowe67/codec2

Optimising cohspk modem

Saw a couple atan and log that could be optimized.

drowe67

comment created time in 5 hours

push eventdrowe67/codec2

srsampson

commit sha 7eef27030a2bb3a504bf3f20df3e353fa69abe0e

Use atan2f vs atan2

view details

push time in 5 hours

pull request commentdrowe67/codec2

Optimising cohspk modem

Built on your optimization to compute inner only once.

drowe67

comment created time in 5 hours

push eventdrowe67/codec2

srsampson

commit sha cb20faf2555b27c04da3084d9932bd79b75df456

Small loop optimization

view details

push time in 5 hours

PR opened drowe67/codec2

Optimising cohspk modem

Attempting to address https://github.com/drowe67/freedv-gui/issues/126

Test below was initially 9.8s, so 30% faster:

time ./src/freedv_rx 700C ~/freedv-gui/wav/ve9qrp_700d.wav /dev/null
frames decoded: 1404  output speech samples: 898560

real	0m6.810s
user	0m6.810s
sys	0m0.000s
+7 -4

0 comment

1 changed file

pr created time in 11 hours

create barnchdrowe67/codec2

branch : dr-opt-cohpsk

created branch time in 11 hours

push eventphase4ground/phase4ground.github.io

Juno Woods

commit sha fac99895a974ed7221f3954154474f8e59433b9e

A little more clean-up to join page and also added project sponsors to front page.

view details

push time in a day

push eventphase4ground/phase4ground.github.io

Juno Woods

commit sha 0dfd98842ac8be8f6890097d92c10d9b5d43e21c

Perfectionist, wanted to clean things up on the join page

view details

push time in a day

push eventphase4ground/phase4ground.github.io

Juno Woods

commit sha 9cb711ba2a4a978bf2b38be464b50dc76e76220d

Started working on improving website informativeness.

view details

Juno Woods

commit sha ce27cb2de64c092adeeca77fc81d9ad2436c5dab

Merge branch 'master' of github.com:phase4ground/phase4ground.github.io

view details

push time in a day

push eventphase4ground/phase4ground.github.io

Dr. Juno Woods

commit sha 720db0871f1517b9a7b84fad742eb5e3fe9aaa87

Update CNAME

view details

push time in 3 days

push eventphase4ground/phase4ground.github.io

Juno Woods

commit sha 762706e840a0e3fcb52b8388a52a8645f1798e65

Fixed site title and nav problem by giving index a name and not having that name show up in the nav. This may also address some SEO problems.

view details

push time in 3 days

push eventphase4ground/phase4ground.github.io

Juno Woods

commit sha 4879996a33cf791ce5447b31a4dfbd76608e602c

Added Trello board sign-up links

view details

push time in 4 days

pull request commentdrowe67/codec2

Single carrier PSK modem

I just started playing with a symbol tracking filter, but as you say, not really for development.

drowe67

comment created time in 5 days

PR closed drowe67/codec2

Single carrier PSK modem

@srsampson I recall you were interested in single carrier modems (QO-100 is a good use case, or UHF). Turns out the recent data work to the OFDM modem includes the features we need to make single carrier operation work. Only it's not really OFDM any more as there's no other carriers to be orthogonal too!

Uncoded, burst mode Octave demo:

ofdm_tx("test_single.raw","1",1,100,"awgn","bursts",3)
ofdm_rx("test_single.raw","1","packetsperburst",1)
<snip>
BER..: 0.0000 Tbits:   240 Terrs:     0
Packets:   3 Npre: 3 Npost: 0 SNR3k: 16.18

I'm working on another project atm, but if you want to push this fwd I'm happy to assist. Or we can leave it for later :slightly_smiling_face: Obvious addition would be a raised cosine (root Nyquist) filter. We have them already on some other modems (FDMDV & cohpsk). I just took a guess at the waveform design (it's a low symbol rate), feel free to mess around with it.

Screenshot from 2021-04-30 09-39-22 Screenshot from 2021-04-30 09-39-27

+8 -2

5 comments

1 changed file

drowe67

pr closed time in 5 days

pull request commentdrowe67/codec2

Single carrier PSK modem

Will close this for now as it was a prototype and no one has the bandwidth to push it forward right now.

drowe67

comment created time in 5 days

push eventdrowe67/codec2

drowe67

commit sha 600cbfdf28d4cf8650e4a9dabcff3038553d78bb

wip aimed at preventing SNR nans at the freedv-gui level

view details

drowe67

commit sha dceda9e41f679efad496264c70858de4832d9fc6

Merge branch 'master' into dr-ofdm-snr-nan

view details

drowe67

commit sha 2723728ed977ac254508b59e626b276a1779c670

remove warnings about complex doulble to float conversion

view details

Steve Sampson

commit sha 073e799a7d2234cf5e3f5779934c94f312b9c9db

Update ofdm.c Change log10() to log10f()

view details

Steve Sampson

commit sha 009b99267cbadeaaefcb09e30848e0492ac4d6ac

Update est_n0.c Changes to limit to floating point and possibly remove warnings.

view details

drowe67

commit sha f28325bfe449fe1a64429faa1ec1f89e2d6350a9

tweaks to remove some warnings

view details

drowe67

commit sha b73563754d453a65d8877a99bde714f72c13ca5f

tweaks to remove some warnings

view details

drowe67

commit sha 306e424282b0941caf018e408b91e4e3385948d2

Merge pull request #184 from drowe67/dr-ofdm-snr-nan SNR nans from ofdm

view details

push time in 5 days

PR merged drowe67/codec2

SNR nans from ofdm

Bug reported during freedv-gui testing here here of squelch not opening on OFDM modes. Traced to a nan in SNR. Not repeatable on all machines.

+34 -22

6 comments

6 changed files

drowe67

pr closed time in 5 days

push eventdrowe67/codec2

drowe67

commit sha b73563754d453a65d8877a99bde714f72c13ca5f

tweaks to remove some warnings

view details

push time in 5 days

push eventdrowe67/codec2

drowe67

commit sha f28325bfe449fe1a64429faa1ec1f89e2d6350a9

tweaks to remove some warnings

view details

push time in 5 days

pull request commentdrowe67/codec2

SNR nans from ofdm

This looks like an error and should be: k<K

/Users/mooneer/devel/freedv-gui/codec2/unittest/tnewamp1.c:105:18: warning: variables 'm' and 'K' used in loop condition not modified in loop body [-Wfor-loop-analysis]
        for(k=0; m<K; k++)
                 ^ ~
1 warning generated.
/Users/mooneer/devel/freedv-gui/codec2/unittest/tnewamp1.c:105:18: warning: variables 'm' and 'K' used in loop condition not modified in loop body [-Wfor-loop-analysis]
        for(k=0; m<K; k++)
                 ^ ~
drowe67

comment created time in 5 days

pull request commentdrowe67/codec2

SNR nans from ofdm

cexp() is a very slow function. Used the cmplx(float) define and also made changes to reduce warnings.

drowe67

comment created time in 5 days

push eventdrowe67/codec2

Steve Sampson

commit sha 009b99267cbadeaaefcb09e30848e0492ac4d6ac

Update est_n0.c Changes to limit to floating point and possibly remove warnings.

view details

push time in 5 days

pull request commentdrowe67/codec2

SNR nans from ofdm

@drowe67, no more warnings in ofdm.c with the latest changes. 👍 There are warnings for other files (attached) but maybe those can be handled in another PR?

drowe67

comment created time in 5 days