The CPU load is completely irrelevant. If you end up using 70% of your CPU but still get more processing done then who cares how much CPU you were using?
Yes, the 7920X is performing better than the 1950x, but it’s also more expensive. When I looked at Newegg the 1950X was $900 and the 7920X $1,100. That’s a 22% difference. At microcenter in the US the 1950x ix $800, which makes the difference 38%.
The performance difference however is much smaller. You linked to the Ryzen 7 review, and not to one including both the 7920x and the 1950x Threadripper. If you look at that comparison instead:
you’ll see that the biggest difference is on the “VI” test, but only at a 128 sample or lower buffer. Rickenbacker said “big mixes”, so if mixing is the workload then the Threadripper offers much better value even at 64 samples: 380 TR vs 410 for the Intel. But, (410-380)/380 = only an 8% improvement for at least 22% more money.
Thanks both of you for your input.
I looked through all of it repeatedly , Wepsta that’s quite a nice build and highly optimized. Wow.
I really am an old AMD fanboy from my first DAW build a 1 ghz Athlon.
The current build , an intel 4790k is my only intel build among many.
I’d like to go back to AMD if I can. I guess it’s the musician in me … thinking with my heart lol.
And yes I am usually mixing , with a few VIs in there but not usually more than a dozen. I can easily freeze them as well.
I tend to do a lot of group processing and use mainly waves stuff , some of their plugins are quite intensive and the 4790 just can’t handle it anymore.
Anyone have direct experience with a Threadripper build?
And can compare it to their old system?
The 1950x is what I would consider.
Are any of you are seeing latency problems with Threadripper and Cubase , which is what I’m wondering about since it’s like 2 Ryzen chips together. There will be latency added.
If not that probably the 7900x.
What buffer sizes do you run when mixing? Threadripper will be competitive with larger buffer sizes (256+) but seems to suffer from real time audio peaks more with smaller buffer sizes (according to benchmarks) I’m like you, I’ve always supported the underdog by building AMD machines and this is my first Intel since the days of overlocking Celeron 300As to a heady 450Mhz (97?)
But you should distinguish more clearly between mixing and running VI.
Rickenbacker wrote "I am usually mixing , with a few VIs in there but not usually more than a dozen. I can easily freeze them as well’ which to me seems to be more about mixing than running a bunch of VIs, and even if he did run a bunch of them he’s open to freezing them. So either a bigger buffer is fine, since it’s mixing, or freezing is an option, and both of those suggestions are only important if he runs out of polyphony. In the case of a 1950x that’s 880 voices at the tested 64 samples buffer, 1680 at 128 samples and 3180 at 256.
If that’s sufficient then what’s the problem?
Waiting for the next version probably means waiting until about August or so before chips are out and reviewed. That’s fine of course, but there’s always a new one around the corner…
Let’s face it, Threadripper CPUs are extremely fast - even if they “only” achieve 85% of a top-end Intel - how many of us actually need that horsepower? You’d be getting a very, very fast CPU for a much reduced system cost compared to a top-end i7 or i9. The only risk IMHO is real time audio spikes caused by immature firmware associated with the various motherboards and interfaces, even then, they’ve been out a while and the issues should be sorted by now, the same could be said of Intel X299.
I refreshed by PC recently and went for X99 on purpose (self-inflicted obsolescence?) I justify my choice (my PC was built with no budget limits but with an eye on the Windows 10 MMCSS thread limit) knowing I’ve gone for rock-solid, reliable, known hardware. Which brings me back to one of my bugbears! It’s about time Steinberg returned to publishing a recommended stable component list for Cubase.
Next to impossible! There’s so much hardware out there it’s almost impossible to predict what will work good or best with Cubase? Configuring a daw for Cubase will probably always remain a painful and time consuming task. I just about concluded the testing of my new Intel I-7900X system and it took me over a month to figure out how to get it optimized! With every new computer build I can’t help to come to the conclusion that the real restriction lies in the audio interface and the quality of it’s drivers. The difference between my (still) current I-4770K and the new I-7900X is not so much the improved latency but rather the ability to be able to handle more instruments with the given latency/buffer size. But then you need to have all the drivers, bios and Windows optimized for Cubase usage. Steinberg will probably never offer you out of the box good, better, best configurations because the variables in components, drivers and firmware/bios versions are just to many to determine that.
The good news is there’s forums like this where thousands of people are struggling with the exact same issue or questions like this and are willing to share information.
It’s not impossible. Pick one single motherboard, CPU and graphics card combo that you know works and publish that (at first) as “Verified For Cubase” (there, invented the term for you) then if motherboard/GPU manufacturers want their boards verified they give Steinberg a few grand and a board/card and the devs build the box and run their standard set of tests. You could even have a latency scale based on peak real time spikes, ie VFC Gold,Silver and Bronze, Bronze being “thanks for your money but performance isn’t really good enough unless for occasional use”
Gold - no real time spikes until CPUs saturated - Pro Users
Silver - some small spikes - Advanced Users
Bronze - real time spikes encountered - occasional users. The manufacturers could opt out of this award being awarded.
It’s still impossible because Steinberg is not affiliated to one brand of motherboard or graphics card and I’m not even beginning to mention bios, firmware and driver versions… And when would Steinberg decide for you to buy this ultimate configuration? It will take time to test and by the time Steinberg releases this, already newer, faster models will see the market.
And furthermore, how will Steinberg know what audio interface you are using and what additional hardware? And what about plugins and additional software you will have installed that could influence Steinbergs recommendation?
So I still say it’s impossible!
Steinberg can go so far as claiming it works on AMD/Intel platforms and common hardware like Radeon/Nvidea and their own line of hardware. From there it’s up to us, with our ‘custom’ configurations, to make it work.
It’s mostly a financial issue I bet. Setup and testing takes time. For us users, or at least most of us, it doesn’t take enough time to be a problem. Most of us aren’t up against the limit of the system of a build people have successfully built so our setup is pretty quick and painless, relatively speaking.
But for Steinberg it’d be really bad to promote certain brands and configurations without saying they’re sure they work, which is completely different level of commitment to testing.
So, possible for sure, but hardly worth it probably…
A lot of third party PS_builders already certify and provide warranties for their models with specific DAW:s
That could be an option and again proves it´s possible.
That said, I doubt Steinberg will do it, but not because it´s impossible.