#26 Futureproofing

Greetings Didier and thank you very much for this :)

I had issues with the 'default' asio out module but none with yours so far, it just works; for a bonus, the sound's improved (albeit ever so slightly, but i have better flow and instrument separation). So again, thank you for your efforts, much appreciated.

Given foobar's "scientific" approach on "how all digits sound the same", i don't see any major updates coming for ASIO from their end; they even warn you not to use it ^^
So your component will probably remain the best solution out there. As such, it would help to futureproof this component a bit?

1) There's currently no support for the 64-bit variant of Foobar2000, we're forced to use the 32-bit version if we want to use ASIO2.
2) Under 'Advanced', the ASIO/input affinity entries currently only accept values up to 255. On an average/budget CPU of today with 16 cores (8+8), you'd need a value of "8000" in order to set affinity on core 15. And that's with 16 cores total. For a 64-core work CPU? A value of, and i quote it accurately, "8000000000000000". So 255's a bit on the short end :)

No clue how much work the above would entail, but since i haven't seen anyone else mention either of them, i thought i should ask ^^
Once again, much obliged for everything so far, truly.

[Didier Galardon]

Hello,

Thank you for you kind message and your wise remarks.
I didn't knew that there was a 64 bit variant of Foobar2000. I will check this.
I could change the affinity range too.
But the problem is when ?
I have made some changes since the last published release but the new version, albeit fitting perfectly my needs, is not yet ready for a bigger audience (e.g, I have added support for native DSD but with some limitations: no volume control, only one DSD format supported. This version is more sensible to the latency tuning too... due to some optimizations)
I don't want to maintain different versions, so I need to find the time (and the motivation) to make the new version more versatile for a larger usage. May be this winter ?

Best regards

Didier

[Anonymous]

I thank you for replying :)

Am aware this is primarily a personal needs thing and more than obliged you take the time to share your work! So time is a non-issue, needless to say.

Appreciate your looking into the 64-bit compatibility.
Regarding your asking "when" (about affinities);
You already know 'why' as it's something you'd introduced yourself; if you're wondering why the need to set affinity wherever we want, err, self-explanatory really.
Some of us may have workloads that are themselves equally sensitive, demanding, or peculiar due to their bad programming. Some of us have chiplet CPUs, or by now Big/Small core architectures (see the new Intel lineups). So now we need -on top of everything else- be aware of latency penalties when switching or splitting a task between chiplet/different cores and balancing all that with our pre-existing needs..
Same thing, same need, it's why you put it in there in the first place.
We just don't have 4 cores nowadays. We additionally no longer have a situation where all cores are "equal".
Or a need to have two PCs, there's that too. For years now, work and 'home'? Same PC. Only now more complicated, more cores, more balls to juggle?

• Sincere apologies if this all sounds a touch pedantic, this is my second language.

My best wishes to you and thanks again for reading all this :)