Anyone want to test different rigs/CPU's? EDITED

[roel]

that seems a correct statement, under the precondition that it uses the maximum offline processing power availeable without errors
but i'm fine with the system like it is. I really don't notice those differences.
i know the market is harsh and they have to resell the product every time over and over again, but this kind of performance is too close to make any sensible difference to put a buy or sell decision on it.

In fact, i have spent a lot of money and time on learning about systems and for me the overall performance is the thing why i did pay more on those, not a benchmarking of one or more of the tasks. You do want to have a system that doesn't give you headaches, red buttons flashing, stuttering, or anything else. So realtime performance is for me the most important thing. What does it do when i play with it.

Just wander the forum and see how many of those high end systems end up with... oooh it outperforms and i payed so much on it.

What you want is not the Ferrari that goes the fastest toward 100 mph but the Mercedes that brings your ideas to the top of the charts without having to deal with technical issues who just hinder your path towards it.



that being said... i find lighting on top or under the keys very important, certainly on a stage

[Stefgro]

Ryzen? Core i9? The UR22 MKII has drop-outs with my Phenom 4 core cpu, 12 GB RAM, SSD, Asus motherboard. And it shouldn't. The thing is to make a driver that acually works, without any noise or drop-outs.

[peakae]

I had the Demo project for C9.5 open and did a few mixdowns @ 44.1kHz
Buffer set @ 256 samples 1m 08sec
Buffer set @ 4096 samples 0m 43sec
Not much difference but still a factor.

[MattiasNYC]

What? That's a huge difference!

It took 37% less time to do the mixdown.

[deeaa]

Yep that'd be very significant in any bigger project, could shave off minutes from the work.

Extremely interesting how badly a dual-xeon 12 core monster fares, which well goes to show how modern instruction sets and such are a much bigger improvement than is apparent in general use. Still would appear that a 8700K *should* be best bang for buck for Cubase use at the moment. Especially overclocked some.

[KHS]

Ryzen? Core i9? The UR22 MKII has drop-outs with my Phenom 4 core cpu, 12 GB RAM, SSD, Asus motherboard. And it shouldn't. The thing is to make a driver that acually works, without any noise or drop-outs.

Phenom 4 isn't exactly a fast CPU.

[MattiasNYC]

The UR22 MKII has drop-outs with my Phenom 4 core cpu, 12 GB RAM, SSD, Asus motherboard. And it shouldn't. The thing is to make a driver that acually works, without any noise or drop-outs.

I don't entirely disagree, but "it depends"... In audio-post I have always used pretty generous buffer settings, 512 samples minimum. Why? Because I didn't need any less. With my Lynx TWO-B I had no issues with drop-outs. But I also adapted my workflow to the capacity of the CPU, which used to be a Phenom Quad Core!

So, I think it's fair to say that a good interface with a good driver should run fine on an older CPU - but - there are still limitations that you can't get around. So without stating what your buffers are and what your project load is like saying that you have drop-outs doesn't really tell us much at all.

[deeaa]

Amen. My friend runs Cubase on an aeons old AMD dual-core Turion laptop which is slow as syrup just running windows. Nevertheless he manages to make 6-8 track mixes using GTR3 for guitar and bass and GrooveAgent for drums.

Of course he always needs to have most of the channels frozen and exports take a while but somehow he still manages. I would never have the patience, but it can be done.

[PeppaPig]

Amen. My friend runs Cubase on an aeons old AMD dual-core Turion laptop which is slow as syrup just running windows. Nevertheless he manages to make 6-8 track mixes using GTR3 for guitar and bass and GrooveAgent for drums.

Of course he always needs to have most of the channels frozen and exports take a while but somehow he still manages. I would never have the patience, but it can be done.

Patience? Have you every tried splicing tapes to make edits or running through mixes with six people crammed round the console trying to "automate" the mix! People have forgotten how good they've got it these days!

[deeaa]

Ahahah, yeah that all hands on console I do remember well and umphteen retries until even the drummer got his faders moved in time and sync

[Stefgro]

The UR22 MKII has drop-outs with my Phenom 4 core cpu, 12 GB RAM, SSD, Asus motherboard. And it shouldn't. The thing is to make a driver that acually works, without any noise or drop-outs.

I don't entirely disagree, but "it depends"... In audio-post I have always used pretty generous buffer settings, 512 samples minimum. Why? Because I didn't need any less. With my Lynx TWO-B I had no issues with drop-outs. But I also adapted my workflow to the capacity of the CPU, which used to be a Phenom Quad Core!

So, I think it's fair to say that a good interface with a good driver should run fine on an older CPU - but - there are still limitations that you can't get around. So without stating what your buffers are and what your project load is like saying that you have drop-outs doesn't really tell us much at all.

I'm getting drop outs with max buffer just from playing music from Spotify, or any other music source.
I also get drop outs when I use my laptops, Asus zenpad UX305F (fanless - not very fast cpu, but should really be fast enough, SSD and 8 GB ram).

I think the driver should be improved.

[deeaa]

Weird dropouts might be related to DPC latency. I mean, I can run standalone Amplitube 4 software on an old Celeron without latency.

Get the DPC latency monitor and find out what is it. For me NVidia display drivers caused terrible dpc issues with several cards. Swapped to Radeon card and zero issues since.

[MattiasNYC]

I'm getting drop outs with max buffer just from playing music from Spotify, or any other music source.
I also get drop outs when I use my laptops, Asus zenpad UX305F (fanless - not very fast cpu, but should really be fast enough, SSD and 8 GB ram).

I think the driver should be improved.

I don't think you're right about that. If I can run Nuendo and produce commercial content on a Phenom 9950 (this was last year, so a 9 year old CPU or whatever it was, Windows 7 x64) then the CPU is of course not an issue. So I agree that it'd be worth looking at the interface at that point. Especially if as you say you have the same experience on other computers with the interface being the common denominator.

However, if the interface and/or the driver was that poor then a lot of people would have a lot more problems. The usage you suggest is very simple. Just playing back Spotify is 'nothing', so it implies that users doing more heavy work should see worse issues. And if they generally did the internet would be filled with criticism of this interface (or the drivers). In fact, I don't even think it would sell at all.

So, I would suggest you look into your setups more in detail and try to troubleshoot it. This isn't the thread for it, so start a new one. Include:

- Used software including version number
- OS version and build
- Computer hardware
- Interface model number
- interface firmware number (if applicable)
- Interface driver version

etc....

[MattiasNYC]

People have forgotten how good they've got it these days!

People can't forget what they've never known.....

[peakae]

True, I don't long for the old days where a project was near finished just a few background vox to sort out, and the tape starts to disintegrate.
I HATE tape. Did sound rather good on some sources.........

[deeaa]

Well, FWIW, I just got my rig updated to an 8700K from 6700K and I now got 8.3 seconds!

[deeaa]

Gotta say, the whole reason I started this thing and testing was that I wanted to know better what kind of overall experience I might get, as based on just CPU tests I would have expected a 30% or so increase which is not enough to warrant 300 bux for update, and by DAW Bench tests I expected more like 60-80% speed increase, but now it's apparent having toyed with the new CPU system that the increase is overall more like 130% for me.

Only changed the mobo (similar as before but z370 board was required so z170 had to go) and the CPU all other parts are the same.

Windows had to be reactivated with digital license bound to my MS Account but all in all I managed to do the swap without any issues with all my current mixes and effects.

Of course before the swap I did de-authorize all kg plugs etc.because they think it's a different machine of course.

But daymn the speed is totally different now. Not only can I use four times smaller buffers and still run projects without dropouts, also it seems like nothing for the CPU to run other software and stuff while mixing. And projects not only export WAY faster they also open about twice as quickly.

I just opened and checked and exported all 15 songs I have in the works and it took me about 15 minutes. Before, I'd have struggled to do a couple of songs in 15 minutes. Amazing increase.

Some of it has to be from overclocking, the mobo has an automatic system to run it at 5.0Ghz which is like a 36% increase over stock settings. And thanks to a HUGE CPU tower cooler with 3 fans it stays at around 70C even when I run stress tests and CPU usage at 100%. The CPU core voltages go rather high to 1.32V with occasional peaks sometimes at 1.4 even but I'm not worried because it stays cool and is solid as a rock. Hell I could open Cubase AND Vegas no problem or heavy delay even while the CPU stress test was going on.

So yeah, going .5Ghz and 2 cores and 2 generations higher really did more than I would have expected. Very happy now. Got 250;- from old parts so was left with just under 350 in costs and it was well worth it.

[-steve-]

Sounds great! Congrats!

[PeppaPig]

Nice one, enjoy your new rig!

[Fly Studio]

Apologies - I haven't read the whole thread, but based on the 1st page posts, here are some things that I've run across while reading about computer setups and DAW usage:
Despite the advances in multi-core usage by new DAW designs, a faster core is better than a lot of cores (not saying a single core is better, but e.g. a Quad Core running at 4.5 GHz is more likely to perform better than an Octo at 3 GHz from the same generation, all other things being equal).

A lot of factors can skew the results of CPU benchmarks vs real world, especially DAW usage. One is the newer instruction sets; another the amount of cache and how it is utilized.

Probably most important for a rig with a lot of cards / components is the number of PCIe lanes, which is both a Motherboard and CPU issue. The new (and very expensive) 7960x has 44 PCIe lanes; I'm pretty sure this translates into more separate lanes for Graphics, add-on cards (UAD-2), Thuderbolt, M.2 storage, more USB 3 & SATA 3 ports, lowering the Interrupts & DPC calls (anyone who has run LatencyMon should know what I am talking about).

Also the way the CPU accesses the RAM is a big factor.

Of course the driver quality will affect the performance of any add-on device, be it USB, FireWire, Thunderbolt, etc. From everything I've read, USB (2 or 3) is generally not the best choice for a converter when one is looking for low recording latency. Possibly due to the way it makes calls to the rest of the system. I still run two RME converters via PCI, as it is very stable and can run at 32 with ease.
Sadly, Firewire has been largely phased out and Thunderbolt is still more the province of Macs.
Many pro audio people think the proliferation of USB converters that can record at 192 kHz is a marketing strategy; USB is cheaper than Thunderbolt, Firewire, Dante, PCI(e), etc.. and do you really need to record at 192 kHz? If the Anti-aliasing (or anti-decimation) filters are good, 96 kHz is more than adequate for capture. But that's for another thread..

[MattiasNYC]

Despite the advances in multi-core usage by new DAW designs, a faster core is better than a lot of cores (not saying a single core is better, but e.g. a Quad Core running at 4.5 GHz is more likely to perform better than an Octo at 3 GHz from the same generation, all other things being equal).

I don't think that's really correct though.

When you add cores you don't get a linear increase in performance. Doubling cores won't double performance, whereas doubling CPU frequency is much closer to doubling performance. However, even if you got only half the increase that Octo-core CPU at 3 GHz would outperform a quad at 4.5 GHz.

A lot of factors can skew the results of CPU benchmarks vs real world, especially DAW usage. One is the newer instruction sets; another the amount of cache and how it is utilized.

But all of the above have to do with the CPU though. So really, looking at the DAWbench results will give a pretty good indication of relative performance of compared CPUs.

(Though incidentally: I actually think the numbers in those charts have changed since I first saw them!!!)

[deeaa]

Goes to show how important clock speed really can be!

[deeaa]

Also, my 6700K @ 4.6 was very much the same, slightly faster but very close. Also can't remember whether it was the first try, could have been the second, if it matters. Nevertheless it appears to give quite consistent results even in different systems.

[MattiasNYC]

Goes to show how important clock speed really can be!

i3 versus i7 though.

[stmad]

Cubase Pro 9.5.10
i5-7600K 4.50 GHz (Overclocked)
16 GB DDR4 3200 MHz

Sata 7200 Rpm
43.9 sec Buffer 96 samples
35.2 sec Buffer 2048 samples

SSD
13.7 sec Buffer 96 samples
8.1 sec Buffer 2048 samples