noteworthy video to see

summary:
1) as example, Reaper is able to fully utilize efficiency cores, Logic and Ableton not
2) M2 Pro is much worse than M1 Max because that's only 6 performance cores vs 8 performance cores

 

I love Ableton, it's my DAW of choice and has been for years but i'll be the first to say its an unoptimized mess on both Windows (STILL no native WASAPI support, still requiring you to use an ASIO wrapper in 2023 for god's sake.) and Mac compared to other DAWs from a performance perspective. lmfao.

 

Also affects Steinberg DAWs and Intel 12th and 13th gen in Windows.

https://helpcenter.steinberg.de/hc/...re-12th-gen-or-newer-hybrid-architecture-CPUs

You'd think that after 30 years of multi-core architectures, programmers would get it right.

 

AMD 7950x & Reaper is a good combo with good single and multicore performance, especially with the recent price drop.

 

Agreed. It's shocking how much more balanced Reason is for cores compared to Bitwig, Live, etc. Have a Xeon 12 core with 64 gig ram and a single instance of Cherry Audio GX-80 can gargle it on the other daws (at any sample block size), but Reason correctly balances the load. Pity it's something that won't be fixed any time soon. Having two kinds of cores just accelerates the issues one can have. I assume they'll have to write preferential core usage and load GUI and other things on the efficiency cores and tell it to not use them otherwise. Headaches all around.

 

Yeah it really comes down to developers cutting corners (Or management denying them the opportunity, a more likely scenario tbh) and not being willing to dig into the code and optimize it for heterogeneous architectures. Ableton is definitely one of the bigger offenders of allowing their software to become more and more inefficient as time goes on because they just want to tack on more features as opposed to optimizing what is already there. I mean, it's been a known issue for years at this point that Ableton 11's plugin scanning is pitifully slow and they, like the WASAPI issue, refuse to do anything about it because they're fat and happy about their market position. No use in optimizing or innovating when you know your position is guaranteed.

 

developers usually focus on creating new versions to CASH IN, they dont care about optimization, their main objective is to sell, there are way better skilled folks out here that dont work for these companies and try to resolve daily compatibility between software and machine, this is the new era of software we are living, a total mess, an experiment, they dont even listen to their customers let alone optimizing stuff ... then I always believed that there is a dark side of hardware that simply block someone's programing ability to reach peak performance but this isnt demonstrated, why I think this? is because otherwise you wouldn't need to buy hardware again and again to stay on top ... processing music should be a less hassle than video or games, and still there seems to be that gaming is way more efficient performing than processing audio ... also do you remember that microsh*it said the 10 will be the last version ? well, how are they gonna cash? spy? ohhh... entering into another dimension of discussion, but still

 

I believe prices drop because 7950X3D and 7900X3D just released?

 

Great points, i have to say tho i think it's more the chicken and egg issue of devs seeing more horsepower to work with leading them to make bigger and more inefficient code because the performance is still there, so you end up with massively wasteful blobs that continuously grow larger with each iteration of chips because as long as the gen-over-gen performance bumps are still present there's no real reason to go back and say: "Hmm, we could make this lighter and faster to run better on this hardware". As hardware gets better, software gets worse. You see the wasted potential all the time in the gaming industry, i mean, shit, look at the nintendo switch. It's running on a 7ish year old smartphone chip the likes of which is completely *destroyed* by any modern smartphone but you still see it running AAA titles at 30fps. They could make things run better if they really wanted to, there's just plainly no financial incentive to do so. lol

Edit: I think the issue with gaming vs audio is the fact that we have dedicated chips to process graphics directly in the system, there's no real internal hardware acceleration for audio workloads outside of those UAD dsp boards and thats unique to that particular ecosystem. It's definitely true that audio workload acceleration is comparatively ignored vs VGA workloads. Tbf tho, 3d Vcache is probably the closest thing to this when you look at the metrics C'T magazine did comparing the 5800x3d vs the M1 Ultra/Max, it seems DAWs (some more than others of course) tend to reallllllly like fast cache as opposed to going out to DRAM.

 

Since Intel introduced those with 12th generation I told myself "this is not good". It also means they're stuck with lithography process and can't squeeze more performance out of their CPUs without using tricks. Multi-threading is also one of their tricks that is known to be not so good for audio due to shared Floating-Point calc-units which are obviously important for DSP processing.

Therefore, if I chose Intel CPU for my next audio computer, I would still buy either 9900K with 8 pure cores and no HT, or go with 10700K. You can find them much cheaper these days and they perform awesome. However, I won't be going with Intel CPUs anymore. AMD CPUs perform better and cost less, use less energy, too.

 

Last edited: Mar 9, 2023

"problem" with 7950X3D is, only half of the cores get 3D vcache, but also that makes them run a bit slower in clocks, so there's controversy again, creating an inconsistency problem (relevant to this topic again), which wasn't present in 5800X3D

 

Could Audiogridder help regarding to this inconsistency issue?

 

It's the 2nd price drop with the new X3D release, and some other components are a little bit cheaper too.

 

Last edited: Mar 10, 2023