What other factors are you talking about that will determine whether the cpu will work ok with cubase. I want to makesure i have the most suitable cpu, motherboard, and ssd combination.
The motherboard itself, the memory speed and size, the type of USB ports (there have been reports about problematic USB chip sets), the type and size of the SSD and so on.
The CPU is just the processing piece, which is as fast as the surrounding equipment.
Start searching the forum and you find lots of tips and stories from other users and what they have done. Even with a 9th generation CPU it seems to be possible to have a great system
And btw. there is still the workaround to disable the E-cores in case they cause problems or go for AMD CPUs. I still don’t get it why everybody seems to believe only Intel is able to build processors.
I hope you guys dont mind me asking a few general questions as you guys seem to know what your talking about!
- There seems to be quite a few online companies such as scan that build pcs designed for daw use and are using 12 gen intel processors so i wonder if they a disabling the e cores. When you do this are you completely losing these cores, if so seems pointless buying one of these cpus with say 12 cores to then lose half of them!
- Hyper threading-does disabling this always give you better cpu performance with regards to using cubase? Do you sacrifice performance else where such as general pc use and multitasking etc?
- Is it safe to disable this on any machine/cpu?
- My current cpu is a 4 core i7 6700 3.4ghz/4ghz max. So in fact still not to bad cpu with regards to cores and speed. With regards to base and turbo frequency the are some cpus that have quite a low base frequency compared to the turbo and some like mine where there not that different. Are the ones like mine where theres not so much of a difference better as means the cpu works faster in its basic state? Or do they run at the turbo speed as soon as required?
- Is there anything that can be tweaked to force a cpu to run at the turbo frequency permently to increase performance or does it not work like that?
Any help much appreciated.
Thanks Dan
No, why should they do this? There is no reason to do this, unless there is really some obvious performance problem. If you think there is something impacting your work you can do this on your own and re-enable the cores later. It is simply a switch.
No, why? This technology exists for a very long time, I have never disabled the hyper thread cores. Same as above, if you believe it helps, switch it off. It is simply a switch.
It is simply a switch.
This is a Cpu that has 4 hyper thread cores, did you ever think you need to disable it?
Depending on the load they switch to turbo speed, but only if the power scheme in Windows is set to maximum performance.
Sorry, but this is a horrible idea, both from a power consumption perspective as well as in regard to overheating/cooling.
From your questions I’m not sure what you are trying to accomplish. Go to one of these specialized vendors, that build machines for audio work and get a system with one of the latest Cpu (btw. 13k is the current architecture). I don’t know what kind of work you are doing, but I believe these machines will have more power than you need.
Apologies if it sounds like im asking dumb questions. Im just trying to further my knowledge on some of these topics not particularly trying to accomplish anything in particular yet.
So i guess i was asking about 12 gen cpus and newer that use P and E cores due to the fact the OP posted a link of the statement from Steinberg saying that there can be issues and at present steinberg cannot recommend these processors.
Regarding hyperthreading it was just that i saw in filterfreq post he disabled hyper threading so i googled this and it seems to ring true that this can cause problems in some cases and turning if off can help so i was hoping to get other views on this from a daw/cubase users perspective thats all.
As i said in my original post im thinking of upgrading cpu soon which means new motherboard also, so im just gathering information about what will work best and i guess just trying to make sure i am using my existing cpu to full capacity and that there wasnt any tweaks i was missing that could increase its performance as a DAW system. I do already have all of the standard power option tweaks in place.
“1”. You would have to check with each builder if they disable cores by default. My guess is no. There’s plenty of stuff going on that can use the e-cores while the p-cores are doing the heaviest lifting. Also, to date, the only test we have that has been used by DAW builders on multiple CPUs consistently put the newest CPUs at the top of the performance list, despite new architectures. So I think that buying the ‘latest’ is generally going to give you the best performance, even if in some cases you may have to wait a bit before things get sorted out… in some cases.
“4”. As far as I know modern CPUs will adjust clock speeds dynamically depending on the work they have to do. If there’s one “heavy” thread that’s being processed the CPU might pick the best core and boost that as high as possible in order to execute that work. If the workload is more parallel then the CPU will boost a sufficient amount of cores to again do the work. The tradeoff is always heat. If you were to use some monitoring software you can see that the absolute peak speeds can be extremely short for single cores and then the average peak is a bit lower. That’s all fine.
My take on all of this is that you set your budget, you read up on people’s experience with the CPUs you’re considering within your budget, you get the best value, and then you leave it alone unless you need to get more performance out of them. Back in the days when people used to overclock CPUs they were made to run safely within a specific range, and then the users could get more out of them by overclocking. But today when competition is back between AMD and Intel it seems it’s actually the opposite: The CPUs are sold in a lineup where the fastest ones are already “factory overclocked” - hence the high power draws and temperatures we see today. So the tweaks users can entertain today are really the opposite: Limiting power consumption can drastically lower temperatures which reduces the demands on cooling and you’re only losing out on 4-5% performance.
“5”. First of all see the previous answer. Secondly, if you look at the following old graph (which I think is accurate) you can see what the tradeoff is:
I had the Ryzen 1700, shown in grey in the graph. You can see that max boost is 3.8GHz, but only for up to two cores. Back in those days that CPU would automatically drop the boost speed down to 3.1GHz as soon as more than two cores were loaded. What I did on my CPU was overclock it to 3.6GHz, all cores. Some would reach higher all-core OC but it would depend on a) cooling and b) the individual CPU that you bought. So in my case it was a tradeoff since I didn’t want to fiddle any more with the CPU: I got all cores to operate a lot higher, but also lost peak performance for any single thread.
I mix mostly for TV and my projects are mostly parallel in processing, so this was fine. I think for some users it may be that they need a single thread/core or two to boost faster than the rest, which means that locking all cores might actually lower the maximum performance of the system.
If I were you I would just get the best CPU out there and not worry about it, which is what I did with my 5900X I use now. Today the boost logic is much better than it was back then.
Hi Mattius,
Thanks for the in depth response its very helpful.
What is this test you talk of that you say DAW builders use?
I am only a home studio enthusiast so im not running hundreds of audio tracks and very low latency is a priority for me as i like to use vst instruments. Im not particularly having problems with existing setup its just if i do upgrade i want what is going to work best for a daw machine.
What did you mean by locking all cores?
Are you a fan of AMD processors over intel then?
“DAWbench”. It currently has a test for DSP (plugins) and VSTi.
By “locking” I mean that the core always runs at that speed. So instead of the CPU being able to adjust the clocks up and down according to its needs and using the logic that Intel or AMD have programmed you set the frequency to a specific value and it stays there essentially all the time.
I am, but not necessarily because they perform better (which they sometimes do).
Firstly, I don’t like Intel’s business practices over the years. So if I have an alternative all else being equal I’ll take it. And right now the alternative is AMD (I don’t do Apple).
Secondly, AMD typically provides more longevity on their platforms. It’s debatable and individual whether or not that matters, but just as an example the Ryzen 1700 I had before was released in early 2017 and the ‘big’ chipset that went along with that socket (AM4) was called x370. Today, six years later, I could buy the last generation CPU on that platform, the 5xxx series (2022), and it would actually run on that old system. So the AM4 platform has lasted 5 years. If you instead bought for example the Intel 12xxx series then you can use those CPUs as well as today’s 13xxx series CPUs, but then you’re done. So you effectively get two years of CPU architectures on Intel’s platform versus five years for AMD.
Thirdly, I generally like to support the underdog, regardless of point one.
Having said all of that though of course I acknowledge that Intel makes great CPUs and that sometimes they offer better performance, sometimes better performance per dollar, and sometimes it’s the opposite. The good thing these days is that it’s hard to go wrong, especially as an “enthusiast” where demands typically are far lower (from a business perspective).
I like the idea of having the opportunity to upgrade cpu for a longer period of time without having upgrade motherboard as well.
Im not familar with AMD so whats the equivelent to say the 12 gen i5 12600k.
Are all amds unlocked as well for over clocking?
I’m not sure what the equivalent would be. Here are some recent results:
The one above is plugins (DSP Integer)
The one above is the VSTinstrument test.
The 12600 isn’t on there, but I suppose you can maybe figure out roughly what an equivalent of the 12700K would be and take it from there. Prices vary and AMD will have a slightly different version coming out very soon which may affect pricing. It’s always in flux…
Like I said earlier, I don’t think it makes sense to overclock any of these latest generation CPUs. Just leave them at stock behavior and get a good cooler, or lower the maximum power usage and otherwise leave them stock.
But ‘yes’, technically you can. I just think it’s a complete waste of time.
1 Like
I am still running a 2016 Dell XPS with an i7-6700K CPU. This processor is a workhorse, and it will easily handle monster orchestral comps with 100-200 audio/midi tracks. There are plenty of good things out there with the new processors, but it is important to note that digital audio is primarily I/O bound. This means that, although a processor may be faster and might have more parallel CPU cores, it does not necessarily have any faster I/O capability. I/O and interrupt processing remain the bottleneck of desktop computers - and today’s latest CPUs have not really improved much there. A faster CPU may mean the ability to run more VSTs, but only if the systems has the time for them. The amount of time available is determined by your audio buffer size plus all the OS system drivers contending for access. A lot depends on the very particular setup of an individual PC.
I don’t really agree with that.
Memory i/o more than doubled from the 6700K to the 13900K, PCIe is four times faster, which in turn makes it possible to use m.2 NVME drives at massive transfer rates.
Also, while digital audio will be bottlenecked by i/o if all you’re doing is playing back a ton of audio tracks, which even older systems can do just fine, there are a lot of users out there looking for a new system who have maxed out their CPU performance, not i/o.
But a faster CPU makes it possible to do more work with the same buffer. So faster CPU enables smaller buffer enables lower latency… or… faster CPU with same buffer enables more VSTs.
1 Like
Yes, of course you will get more oomph for your VST’s with a later model CPU. No argument there. Go for a new PC. Performance depends on what you are doing - if all you are doing is playback, then yeah, just up let it rip.
It is when you record and playback (with CPU generated effects) in real time that the fun really begins, or not - as this is where system stability issue arise. Although the newer CPU buses have faster throughput, the ability of those CPU’s process that I/O data has not experienced any great jump. It is still a serial queue of events processed by a single, primary core. Sure, there is improvement in the latest CPU’s, they can do more, but it is not a “night and day” difference by any means.
Depends on what you mean by “night and day”.
On the AMD side a new 7xxx series CPU is probably going to have about 70% better performance than my old first-gen 1700. When I replaced that 1700 my system was on the edge, and my new 5900X was a night-and-day difference to me. Whereas advanced metering slowed my system down in the past I have zero such problems now and am using the new one to about 25% capacity generally. At most I’ll hit 40%.
Hi there,
I bought a new config with Core I5-12500 and 16 GO RAM.
If I understand correctly I should not encounter performance issues on Intel® Core™ 12th gen since the Core 15-12500 has no efficient core (only 6 performance cores ?)
Between the last 3-4 years there was mayor Upgrade in single core performance.
I agree, newer generation single core performance is much better than years ago. A 13800k is about double the performance of the 8700k. So if you’re limited by your 8700k (as I am) using all the plugins you need to mix/master and invest in a 13800k, you probably won’t have any performance for a very long time (until you begin changing the plugins for cpu heavier ones ofc).
So to you, should this issue be solved by Windows OS or by DAW developers (Cubase, Nuendo, etc…)?
please give some more info about your projects like quantity of vst instruments, settings?
If the issue is ASIO specific, then the entity that needs to do the work is obvious.
I see lots of DAW users reporting issues on Intel, so it’s not a Cubase-specific issue.
I don’t see as many complaints from people in video or gaming (streamers, etc.).
1 Like