was thinking about a new built as bigger projects involving some current CPU intense plugins gets the machine down to its knees pretty quick. Anyway, hardware prices suck at the moment, so…
…upgrading the CPU is the other option. Got an i7 7820x running and think about upgrading it to the fastest thing I can find for socket 2066 which seems to be an i9 10980xe.
Anyone here who has done a similar upgrade and wants to share experiences about the increase in processing power that I could expect in reality using Cubase?
I would recommend to stick with current setup till have a budget for new 13th/14th gen i9 13900K/KS or upcoming 14900K. Those CPUs have insane single core performance and superb multicore performance. In combination with some good speed DDR5 it will be perfect machine for music production for many years imo.
For sure paying full price (~ 1200 $/€) isn’t attractive but I’ve seen some used offers around 500 which would maybe extend the life of the machine in a way that makes sense to me. I wouldn’t expect an excessively boosted performance tbh. Can’t go much above 50 % cpu with Cubase right now and a more powerful CPU wouldn’t change that I guess (the serial processing/one core thing) but at least something noticeable just because 18 cores are 10 more than 8
Would really be interested in reports of how much more one could realistically get from an 8 to an 18 core processor while keeping the rest untouched!
For 500 or less usd it could be OK to swap just to that 10980XE. Its not so old (2019) and it could be good improvement in multicore performance. I recommend then check your motherboard compatibility and update to the latest bios which supports this 10980XE.
Got a good deal for € 500 used, fair enough for getting the built through two or three years probably.
As many of us I have tons of software from a multitude of manufacturers and authorization schemes. I‘m pretty confident all iLok auths and UAD will just keep on working. Not so sure about machine based authorization, if replacing the CPU will change my machines‘ ID. Will it at all?
Just reporting back how it went for whomever may come across the same questions:
Swapped CPU, turned on and “yeah, hello, new cpu detected” and that was it. All software authorizations are just there without doing anything. The machine ID hasn’t changed by exchanging just the CPU.
In preparation I had updated my BIOS with the latest version that also explicitely supports the new
CPU (good to know )
Performance boost: it’s definately there!
Working with the system now for three days I can say it’s able to handle buffers of 96 samples in situations before at least 256 samples were needed for cracklefree playback. That’s a good thing while recording.
Real life mix work: the boundaries regarding the length of audio signal chains (needing to be adressed to a single core) aren’t stretched (which was not to expect in the first place). But having more cores to adress the “given length” of a signal chain means I can do what I do more often before I have to switch over to using DOP or RIP to keep a demanding mix project running.
All in all a great update for me while I still wish a sort of processing load distribution to make use of the juice
It’s not satisfying to stare at CPU meters and get a crackly realtime performance while the thing running uses just 11% average
Your question makes me realize that I don’t actually know where I read about it. There are some videos online as well with IT guys explaining what’s happening under the hood. Here’s one: https://www.youtube.com/watch?v=GUsLLEkswzE&t=2s
In practice it means that all plugins and all sorts of realtime processing in a chain - from tracks to groups (maybe several) to master, probably with some sends involved - have to be adressed to a single core of your cpu. So the limit is the capability of one core only. For some tasks a higher core speed may be more useful than plenty of cores.
The way I mix can get complex for an audio engine I guess
Let’s say if I have two kick mics and a trigger, I’ll bus them to a kick bus which will go to a bus together with snare and toms that goes to a drum subgroup that will be bussed with the bass group and from there go to the mains. Several sends used for parallel compression, fx sends, sidechains (like triggering the kick mic gate/expander with the trigger signal). Same procedures continue with all other drum tracks, bass, guitar, synths, vocals and whatever is there.
That’s a bunch of signals forward, backward, sideward with plenty of ‘branches’ coming from and going to almost anywhere.
Can repeat my statement above about performance: it’s not ideal not having a smooth solution for an even load distribution across all available cores. It’s really just a fraction of the whole cpu power that can be used and to my knowledge any DAW basically has this problem, it’s not just Cubase. Whoever solves this may be the one everyone’s looking at how he does it!
The switch from i7-7820s to i9-10980XE is a good upgrade for me. I can reach a little bit further although it’s still easy for me to get to the point where I have to use highest possible buffer (2048 in my case) to finish a busy mix. In theory there’s potential for at least 5x that performance.
I think this type of comparing is too random, not precise and is not revealing real world performance especially to particular hardware setup, bios and OS configuration as well as particular audio software usage scenarios. Those CPUs have different generations, different IPC improvements, different core counts etc… In any mean his new i9 10980XE will be faster in both multicore and single core performance. Knowing that this is a prosumer grade platform it supports quad-channel memory and have a lot more bandwidth then consumer grade dual-channel configs. IMO this CPU upgrade is good step knowing that he swapped only CPU. Congrats on your new CPU btw @marQs
Its just my personal opinion regarding this. I dont personally like this website, for me its too random, incomplete and there is too much questionable/incorrect information or data there. I will trust much more something like Geekbench then kinda this generic cpu comparison websites. And even then for example how good your CPU is performing in some video game or how much points it got in Cinebench doesn mean that your particular setup is good and optimised for real world audio applications. Thats why I dont like those ‘‘cpu comparison’’ websites.
I don’t agree but I am not here to argue and make a hundred posts to prove right or so, I just try to help if possible, so…
I respect that we have different opinions and that’s ok!
Well, I have learned throughout my life to use each tool that gets in my hands the best way according to my needs and always have critical thought.
This video is actually talking specifically about, how the comparisons are affected by the site’s weighting system when you look only at the overall weighted average.
So yeah I agree up to a point with what is being said on video and I advise whoever uses it to dig deep into the results and take everything into account with a grain of salt.
That said, when I am comparing let’s say CPUs, using this site, I will look at the benchmarks in detail so I can make my own conclusions regarding my needs.
I will not skip every detail and just keep the outcome related to “nice to haves” or “single core speed”, etc.
)
