Intel Alder Lake performance

I’m very tempted to upgrade to 12900k but I’m concerned if its novel architecture requires some code tweaks within Cubase.

It’s probably too early to tell but was there any statement by Steinberg in regards to compatibility or any performance issues with it?

3 Likes

I’m not concerned about the new 12th gen CPUs (Alder) causing issues.
Worst case situation: you might need to disable the E-cores for a few months until the software(s) catches up. You already have a bloody fast machine with “only” the P-cores active!

I would advise however to wait at least 6 months (if you can) to let Windows 11 stabilize before jumping in.

And interestingly by that time new AMD CPUs on the current platform will be out which are likely to match or exceed the performance of these Intel CPUs, and not only that but I’d expect at that point to also either have or be close to having AMD’s new platform out.

So in one sense it’s kind of a great time to buy because they’re all great CPUs, yet it’s annoying to wait because of all the upcoming products.

Yeah… there is always a “next-gen” promising more… that’s why I’m still with my 9yr-old PC. :upside_down_face:
Wait till you hear about the Raptor Lake… :wink:

The time to change to a new machine is not when the specs are “great enough”, it’s when you “have to”.

My 2-cent: if your current machine does the job: keep it. If it does not: change it.
(in either case, the decision to change is regardless of what Intel or AMD wants to convince you to buy.)

If you have to change, then take the latest-and-greatest tech, so you won’t have to change
it again for many many years.
Today, that means Alder Lake (i7 or i9) with Thunderbolt 4, NVMe SSDs, and a basic (silent/fanless) GPU.
(TB4 is just gravy for future-proofing, not faster than USB3 anyway)

Recording will be possible with only the E-cores active >> low-power >> fans stopped >> silence >> bliss…

I agree with the last part. Of course if it no longer does the job and you need to do the job then you need something new. I do think though that there’s a case to be made for upgrading when it makes sense financially and technically and professionally, which may mean upgrading before you run into a ‘wall’. If for example you have some headroom left in computation power but you also realize that potential future projects may be too heavy then it might make sense to upgrade “early”. I understand though that that’s ‘debatable’.

Agreed.

The one thing I’d say about this particular point in time is that AMD is currently on PCIe 4.0, latest intel on PCIe 5.0, and the next gen AMD also on 5.0. On top of that the new AMD chips will have more L3 cache which may make a big difference for us in audio. Whether or not that makes a difference is I guess a matter of personal requirements. But point being that now the changes we see are more significant than they were say 2 years ago, technically speaking.

I might disagree here. I think AMD’s Zen 3 Ryzen CPUs offer great performance, and arguably that next release which would be a compatible CPU to upgrade to would see the maximum performance on the generation. Or to put it differently: I think the last CPU AMD will provide for their current socket AM4 will be better than the last CPU Intel will provide for their i7/i9 platform.

Also a quick note that TB3/4 both have twice the bandwidth of “Superspeed” USB 3.2 gen 2x2, 40Gbit/s versus 20.

I’m of the mind that computing of even two generations ago is viable for Cubase right now, I would definitely not jump on any upgrades of working machines until these new CPU designs (performance/efficiency cores, more direct memory access, faster PCIe interfaces, ARM based machines etc) have a chance to stabilize and settle.

In my studio I have ab 8th gen Intel machine, a Ryzen 3 machine and now an M1pro macbook. I have absolutely no desire to touch the AMD machine as it’s working almost too well. I have no plans at all to be using Intel again for a long while, the power drain and cooling requirements of even my existing system make it feel a bit outdated and inefficient, and everything I’ve heard about the new platform is telling me temps and power aren’t going down.

The big thing in all this, though, is that I don’t find myself missing any performance, at least not for how I work. I even “downgraded” my PCs to 64gb of memory from 128gb (to spread DIMMs around to more machines) because with my NVMe setups I found myself being able to use much more aggressive disk streaming with no drawbacks. So if my projects are now using 50-60gb ram instead of 90gb, and ASIO performance never reaches above 75% in even the biggest projects (which for me would be 300-400 track orchestral cues), then the only realm left to consider is power consumption/efficiency/cooling.

I don’t think new Intel CPUs or even new AMD ones are going to be making much impact to audio production work. Video editing/post, and gaming sure, but not really audio.

Right now my little M1 macbook is blowing my mind at the possibilities, where I’m mixing/composing 6-7 hours at a time on battery on a laptop that barely gets warm or makes a noise, and I get lost and think I actually am on a desktop time to time. So if (big if) I ever got back to intel it’s gonna have to be a pretty dramatic shift in their design models to win me over. As much as I prefer using Windows to MacOS I can’t argue with charging this thing once a day and getting a whole day’s work out of it, vs running huge rack machines that act as space heaters (in the case of Intel at least).

1 Like

I have a Ryzen 3950x. Its performance in my projects is unsatisfactory (or Cubase not being optimized).

Seems like my #1 bottleneck is bussing (with heavy processing on them) which to the best of my knowledge is dependent on single core performance. I have quite a few projects where I can’t even play a song without stuttering and having to disable one or two plugins on one of my busses. AL has a 30-50% (don’t remember) faster single core which seems like would solve my problems.

If you’re willing to wait the new 5950x should be out early next year. There should be at least 15-20% uplift from 3950x to 5950x I think, and I wouldn’t be surprised if there was another 10-15% going from 5950x to the one with new/more cache.

Not trying to dissuade you from Intel, they make great products, it’s just that if you’re already on AM4 it would seem like the path of least resistance and maybe the lesser expense would be to just swap the CPU.

Of course - we don’t know what the updated 16-core Ryzen will be like, but it’s a thought worth considering I think.

Like many, you sound more like a shopper for a gaming machine.
I strongly agree with Kotsamanidis that an audio production machine is a complete different world.

The Ryzen is GREAT technology, just not the most power efficient in most workloads.
For recording tasks, the 12th gen i7/i9 will be more quiet/silent (thanks to the E-cores), or even better, the M1 based machines (if you can bear the growing pains).

For shooting armed zombies in 3D, go Ryzen go. For recording it will do a great job too for sure, but just not the best match currently.
But not everyone (admittedly not only gamers) need or value having a silent machine above all else, and that’s fine.

BTW, the 40Gbps mode of TB4 is only available in exotic dual-link scenarios, and valuable only for daisy chaining scenarios or top-tier mass-storage. No single device is or will use it, and no gain for any pro-audio gear in the foreseeable future. The bandwith of USB3.1 Gen2 is already exceeding the needs of the wildest audio dream scenarios (do the maths)… Future-gen audio gear is more in the Dante/TSN/AVB direction. But that’s a whole other debate!!

I believe this feature is coming in Cubase 12…

2 Likes

Sure, the 12th gen Intel will have E-cores which is absolutely great, but the OP didn’t specify recording though. If it was ‘just’ about recording then a far lower tier machine would do just fine. It’s the processing that’s problematic, and the new 12th gen CPUs are really power hungry when under full load. So really the benefit of power efficiency and silence may very well hinge upon using quite a bit less than full capacity - or we’re talking pretty serious cooling solutions. At least as far as I recall from the reviews I saw.

Pretty much a stretch, but depending on where the OP lives it might be worth while seeing if it’s possible to pick up a new Ryzen 5950x and try that to see if it works, and if there is no improvement return it to the store and get store credit to put toward a new Intel system. That way the OP can try out improving performance by spending say 700 (Microcenter) and if that doesn’t work that bigger expenditure is warranted and no money was lost (I’m assuming that a new 12th gen Intel build will be far above $700 taking into account even just the mobo+CPU).

I agree that it’s of little use for audio to have 40Gbs in most cases (maybe raid storage might be different in some situations, though I doubt many encounter that). But either way it’s 40Gbs for both TB3 and TB4 and the doubling dual-link is 80Gbs when trading bandwidth for bi-directionality, if I understand it correctly ( Thunderbolt 4 - Wikipedia).

Like many, you sound more like a shopper for a gaming machine.
I strongly agree with Kotsamanidis that an audio production machine is a complete different world.

I guess for quite a few of us, audio production is not so much about recording (and if recording, then maybe recording multiple vocal takes one after the other) but much more about being able to run many VST instruments and effects.

Exactly the same for me (95% VST+mixing, 5% recording).
But the thing is that a “well-built” machine that runs silent for recording, will be powerful enough for the rest of the job. A gamer’s machine can also do everything well, but will (typically) be too noisy (to mho) for recording if in same room.

Today it’s easy to build a silent machine for recording with plenty of horsepower for the other chores when needed. But to do so, silence must be a priority design goal, not an afterthought.

But I reckon running silent is not a requirement for everyone. If it’s no value to your workflow, then just disregard my post. Some might prefer having a 300W GPU and make it a dual-purpose machine (audio+gaming). That, and everything in between, is perfectly fine/valid!

I don’t game. Not sure where that came from.

3950x to 5950x productive benchmarks provide inconsistent results, some show almost 20% single core increase, some show pretty much nothing. Recent dawbench thread posted on GS suggested there was almost no relative performance increase in DSP between 3950x and 5950x.

3Dcache version of 5950x may or may not benefit audio production. Not sure how much Cubase relies on cache. Could be 0% improvement. There’s a reason it’s called “game cache”.

Just going for 5950x seems like a waste of money. 10-20% may or may not be noticeable. To me personally, 50% is a mental threshold for upgrade where I feel like I’ll be getting some breathing room with DSP (only for mixing).

Also, power consumption is literally one thing I don’t care about at all. I don’t mind paying extra $7/mo in electricity bills as long as I can finish my projects. Saving a few bucks at the expense of not being able to work at all on my project makes zero sense. I’m willing to have a 10,000W PC as long as I can continue working without interruptions.

Well recording is easy. Running heavy VSTi at low latency in a large project is a completely different thing.

I must have missed that. Not that I don’t believe it’s true, it could be, but I don’t recall seeing those two compared in the same test.

I’m in the same boat as you. I think the two things to remember is that higher power draw usually means more heat which usually means more serious cooling which usually means more noise. So that’s a first concern. The second is the long-term impact on the platform. If a new CPU comes out and eats up a ton of power then even if we can live with the noise or pay for better cooling it could spell trouble for that architecture and platform moving forward in terms of building newer faster models. In other words it may be a hint about what we can expect moving forward.

Interestingly AMD’s response to the big-little architectures seemed to be that their fundamental design is so power efficient that their upcoming 5nm cores can be set up in firmware to run super efficiently or draw more power and be super powerful. So a sort of ‘virtual’ big-little design in other words.

Sorry, getting off track now…

The GS Dawbench thread author didn’t compare directly the 3950x to the 5950x. But using a common demoninator in each test (same Intel CPU) someone calculated that both 3950 and 5950x share the same gap in DSP performance to that CPU.

Like I said, power, heat, noise, all can be dealt with very easily. Stuttering, hiccups and inability to finish the project is much worse.

Not so long ago when we had incadescent bulbs, extra Intel power consumption equals to a floor lamp with 3 60W bulbs being on. These days people make it sounds like it equals to an erupting volcano in their room.

I think that is logically the wrong way of thinking about it. The two companies have different lines of CPUs with different types of architectures. Because the test has changed for the latest DAWbench it’s actually not really comparable because the difference in the test can change the performance of the CPU.

To put it differently: It’s possible that both the Intel benefitted from the new test and AMD’s new 5950x is better than the 3950x, but that the former made a bigger difference than the latter. So because the test wasn’t literally the same we simply can’t know.

But anyway, I took the highest performing Intel CPU I found on Scan’s and Vin’s test (“tafkat”), the 10920x, and compared that to the 3950x/5950x.

In the VI portion at a 64 buffer the delta is 13% in favor of the 3950x versus the 10920x. The 5950x at that same buffer size was 25% ahead, which is a 100% improvement for that generation at that buffer size. For the same test but at 128 samples the lead is only 4% for the 3950x, and about 10% for the 5950x. So this time over 100%. At 512 samples it’s 31% vs 7%, which is odd. Almost as if a different bottleneck was reached in the test system.

For the DSP part of the test (% lead 3950x/5950x over 10920x);
64 buffer - 55% / 57%
128 buffer - 54% / 54%
512 buffer - 26% / 56%

Basically there were zero gains on the DSP test at low/medium buffer and 100% at a “high” buffer, and a whopping 100% (roughly) on the VI test at low/medium buffers and a regression at high buffers which was possibly a bottleneck…

So taking those results at face value: for anyone using VIs and needing a very low latency (64 samples) the difference is significant.

But again though, it’s possible that we’re seeing the results of changes in the test and not changes in CPU capability.

I didn’t disagree with that.

Tomorrow I will have 12700. If Cubase does not work well there, then this will be a reason to look in the direction of another DAW)) But I hope this will not happen. The software manufacturer must keep up with the times and technologies!

1 Like