Jarkko Sakkinen
Posts
4802Following
319Followers
489OpenPGP: 3AB05486C7752FE1
Jarkko Sakkinen
repeated
repeated
GCC - GNU Toolchain
gnutools@fosstodon.org
Registration is now open for the 2024 LLVM Developers’ Meeting 16 in Santa Clara, CA.
https://discourse.llvm.org/t/2024-llvm-developers-meeting-registration-workshop-announcement/80643
@llvm
0
1
1
Jarkko Sakkinen
jarkko
Edited 1 year ago
This is Richard Feynman, pioneer of quantum physics and computational models playing bongos :-)
https://www.youtube.com/watch?v=2Ks8gsK22PA&t=31s
https://www.youtube.com/watch?v=2Ks8gsK22PA&t=31s
0
0
0
Jarkko Sakkinen
jarkko
Edited 1 year ago
All my quantum knowledge is based on
1. Physics lab exercises while studying for my MSc (back in the day).
2. Applying numerical methods to cost functions while studying for my MSc.
3. Quantum Computing at https://brilliant.org/
I thought that since I'm ultimately stupid in this topic and want to get the gist of it, Brilliant was exactly in my level :-) It really helped me to get the gist of the topic by combining it to my knowledge of stuff that I learned while studying ages ago at a polytechnic university.
Can highly recommend that course!
And can highly also recommend studying topics that you suck at, not ones that you're good at. It is refreshing!
#brilliant #quantum #computing
1. Physics lab exercises while studying for my MSc (back in the day).
2. Applying numerical methods to cost functions while studying for my MSc.
3. Quantum Computing at https://brilliant.org/
I thought that since I'm ultimately stupid in this topic and want to get the gist of it, Brilliant was exactly in my level :-) It really helped me to get the gist of the topic by combining it to my knowledge of stuff that I learned while studying ages ago at a polytechnic university.
Can highly recommend that course!
And can highly also recommend studying topics that you suck at, not ones that you're good at. It is refreshing!
#brilliant #quantum #computing
0
0
0
Jarkko Sakkinen
jarkko
If you have gone through a basic lab course e.g. in a polytechnic university AND know how to re-formalize an algorithm as a cost function, and apply numerical methods for that, you already have the basic theoretical and practical experience on how to program quantum computers.
That's roughly all there is to it.
#quantum #computing
That's roughly all there is to it.
#quantum #computing
1
0
0
Jarkko Sakkinen
jarkko
Have some catching up to do with #LKML as I was going to job interviews last week and did not have energy to do much else. Also need to some bug fixes to Linus for 5.11.
Anyway I accepted an offer with all the usual suspect hype words except AI: Rust, RISC-V and blockchains ;-) More on that later...
Anyway I accepted an offer with all the usual suspect hype words except AI: Rust, RISC-V and blockchains ;-) More on that later...
0
0
2
Jarkko Sakkinen
jarkko
Edited 1 year ago
The upstream patch that fixes my recent mbsync issues in Fedora is ceb0fa980
[1] https://sourceforge.net/p/isync/patches/19/
[2] https://bugzilla.redhat.com/show_bug.cgi?id=2302132
#fedora #bug #triage
[1] https://sourceforge.net/p/isync/patches/19/
[2] https://bugzilla.redhat.com/show_bug.cgi?id=2302132
#fedora #bug #triage
0
0
2
Jarkko Sakkinen
jarkko
Looking at delays among different CPU and chip companies (not just Intel), maybe there's a CPU bubble coming too in the timeline (long projection)?
E.g. Spectre and its ancestors are so fundamentals flaws in the core design decisions of a CPU that I don't think they will be ever fully fixed in ARM and x86.
E.g. Spectre and its ancestors are so fundamentals flaws in the core design decisions of a CPU that I don't think they will be ever fully fixed in ARM and x86.
1
0
0
Jarkko Sakkinen
jarkko
zkVM is a cool project: https://dev.risczero.com/api/zkvm/
Requires quite selective use tho because it does not scale that well.
#riscv #zkvm
Requires quite selective use tho because it does not scale that well.
#riscv #zkvm
0
0
0
Jarkko Sakkinen
repeated
repeated
Signal
signalapp@mastodon.worldWe're aware of reports that access to Signal has been blocked in some countries. As a reminder, Signal's built-in censorship circumvention feature might be able to help if your connection is affected:
Signal Settings > Privacy > Advanced > Censorship circumvention (on)
1
21
1
Jarkko Sakkinen
jarkko
Act is different same as github-ci-local for Github: https://github.com/nektos/act
0
0
0
Jarkko Sakkinen
jarkko
Edited 1 year ago
Containers cross-talk in standards and even share sub-projects. CI's don't.
I think we as an industry could take the quantum leap and define standard for something like ".runner-ci.yml", and when split to multiple files ".runner-ci.d/*.yml". Then, just define stages and tasks.
That would also fix migration between hosts for Git repositories. Also, in a private company internal CI and external CI can sometimes be different. Only way to address is to support both, which is redundant work.
I think we as an industry could take the quantum leap and define standard for something like ".runner-ci.yml", and when split to multiple files ".runner-ci.d/*.yml". Then, just define stages and tasks.
That would also fix migration between hosts for Git repositories. Also, in a private company internal CI and external CI can sometimes be different. Only way to address is to support both, which is redundant work.
1
1
1
Jarkko Sakkinen
repeated
repeated
DCoder 🇱🇹❤🇺🇦
dcoderlt@ohai.social
Edited 1 year ago
Amazing work, @ryanc !
Edit: Direct link to Ryan’s write-up: https://rya.nc/vpp-hack.html
0
5
1
Jarkko Sakkinen
jarkko
Edited 1 year ago
I've been looking this tool for years and it was there all the time: https://github.com/firecow/gitlab-ci-local
My workflow has been before:
1. Edit .gitlab-ci.yml without knowing what I'm actually doing.
2. git push
3. Hope for the best, fear for the worst.
It has been pretty difficult to learn it when I have to test every edit remotely and once I reach my quota wait for some weeks to continue.
And no, GitLab runner is not a solution:
1. I've never been able to successfully set it up.
2. It does not scale to multiple Gitlab instances easily.
3. Even if two former were somehow addressed I do not want to run random system daemons.
I.e. to summarize: A local Gitlab runner sucks. It sucks unimaginably hard.
#gitlab #ci
My workflow has been before:
1. Edit .gitlab-ci.yml without knowing what I'm actually doing.
2. git push
3. Hope for the best, fear for the worst.
It has been pretty difficult to learn it when I have to test every edit remotely and once I reach my quota wait for some weeks to continue.
And no, GitLab runner is not a solution:
1. I've never been able to successfully set it up.
2. It does not scale to multiple Gitlab instances easily.
3. Even if two former were somehow addressed I do not want to run random system daemons.
I.e. to summarize: A local Gitlab runner sucks. It sucks unimaginably hard.
#gitlab #ci
1
0
2
Jarkko Sakkinen
repeated
repeated
Thomas Sturm
tsturm@famichiki.jpJapan, national morning news show:
An uninterrupted 10 minute segment about a new train running on a line to a popular vacation spot.
A journalist spends most of the segment being absolutely DELIGHTED by every single feature of this tastefully designed train. She sits in every class of car, tries every seat feature, eats a curry in the dining car, enjoys the view out of panorama windows. Arrives rested and gives us a whirlwind tour of town.
This is how you change car culture.
1
8
2
Jarkko Sakkinen
jarkko
Edited 1 year ago
Randomly noticed that one thing I was making noise about while ago is fixed in #Linux 6.10: you can now #BPF when CONFIG_MODULES=n. It was not my main goal to contribute, just make the issue visible, but there was also one of my patches in the final version (nice surprise because it was alien area in kernel for me).
The priority for me was to squeeze down the compilation time for testing/CI while still having up to date tracing capabilities.
The priority for me was to squeeze down the compilation time for testing/CI while still having up to date tracing capabilities.
0
1
1
Jarkko Sakkinen
jarkko
Edited 1 year ago
People say me that #Kirchoff #EQ is great or even better than #FabFilter Pro-Q3 but...
Dynamic EQ is not a great tool for shaping transients as the filters are connected in series. This will result most "bending" transients as EQ points will interact and are interconnected. For shaping, band needs to be split into blocks, and this exactly what a multi-band compressor does.
This reduces the meaningful parameters to exactly one: dynamic range:
1. Define a dynamic range (in dB for an EQ point.
2. Figure out values for the parameters that keep it within that range.
So with the dynamic EQ below this would mean a manual tuning until you find some
values that seem to work, and depending on signal coming this could even mean automating those parameters, depending on how static the signal is. Also any change to the signal coming in would require re-adjusting them.
On the other hand, Pro-Q3 starts from the dynamic range as the input parameter, listens the signal and adjusts compressor parameters dynamically to stay within the range, without user intervention.
Not my cup of tea because the choice is between spending hours on it vs spending less than a minute figuring this all out 🤷
#MusicProduction
Dynamic EQ is not a great tool for shaping transients as the filters are connected in series. This will result most "bending" transients as EQ points will interact and are interconnected. For shaping, band needs to be split into blocks, and this exactly what a multi-band compressor does.
This reduces the meaningful parameters to exactly one: dynamic range:
1. Define a dynamic range (in dB for an EQ point.
2. Figure out values for the parameters that keep it within that range.
So with the dynamic EQ below this would mean a manual tuning until you find some
values that seem to work, and depending on signal coming this could even mean automating those parameters, depending on how static the signal is. Also any change to the signal coming in would require re-adjusting them.
On the other hand, Pro-Q3 starts from the dynamic range as the input parameter, listens the signal and adjusts compressor parameters dynamically to stay within the range, without user intervention.
Not my cup of tea because the choice is between spending hours on it vs spending less than a minute figuring this all out 🤷
#MusicProduction
0
0
0