Mesa Boogie 2 CPU Usage

[dmitch]

I've noticed two things about the new Mesa Boogie 2 collection. One, it sounds amazing; this is a really versatile set of amps.

But...all of these new amps are CPU hogs. So much so that even on a reasonably beefy iMac (2019, 3.2 GHz 6-core i7, 32 GB RAM), I actually cannot use them while tracking in Logic Pro. I never head this problem with any other AT5 amps including the original Mesa Boogie collection.

Here are some Performance Meter shots taken with a simple sine wave generator driving one amp. All three configurations use the same speaker, no effects, just a bare amp in mono.

For reference here is a Mesa Boogie Mk 4 from the original collection:


From the new collection, here's a Mk 5:


And a Mk 2C+:


Both of these amps from the new Mesa Boogie collection are using more than 100% of a core. I.e. they need multiple cores just to idle. But in the Logic Pro world - and I suspect in other DAWs too - when multiple tracks are active, each track gets one core. These new Mesa Boogie amps need more than one core and they won't work.

This is with a 256 sample buffer, which should be more than enough. I get CPU overloads all the time when trying to track with these amps, even when I do things like putting the amps on a separate aux channel (rather than the track I'm recording to).

Has anyone else seen CPU overload using these new Mesa Boogie amps?

[emurphy94108]

This is my experience too. None of the other amp models (including the Mesa/Boogie 1 collection) presented any difficulties in terms of CPU usage (I'm using an 8-core iMac Pro with 64 GB RAM). All of the models in the Mesa/Boogie 2 collection will glitch, even when Logic Pro is in pause mode with minimal additional processing (basically EQ, compression, delay, and (non-convolution) reverb).

And on my 2018 MacBook Pro (which admittedly is a dual-core i7, but it's running at 3.5 GHz with 16 GB RAM), none of the models in the M/B 2 collection are remotely usable. Under the same conditions (Logic's transport paused, no other channel strips active), it's mostly glitches.

Too bad, really, because as an owner of a real Mk V, I can say that the difference between the real amp and the model, miked similarly, is basically undetectable. At least there's no glitching on a bounce.

[ibot33]

Same here on a audio optimized Win10 PC. Mesa Boogie 2 is VERY cpu heavy. It's really not usable that way. Even increasing buffer size a lot has not the usual effect, so I think there are some bugs responsible for this behaviour?!

[garfy]

I'm sorry to hear that you are running into some issues.

As this is a user-to-user forum and not direct to IKM support, in case no one else chimes in here with advice or a solution the best thing to do is check the FAQs, via https://www.ikmultimedia.com/contact-su ... ck=QUICKSW

If these do not resolve the issue, there is a link to the Technical Support form at the top of the Search FAQs page, for direct help from Official IK Support. They endeavour to reply within 2 working days, but obviously considering the current situation things may take longer than usual.

And, to help others who run into a similar problem, please let us know what you did to resolve the issues, but without directly quoting any correspondence from official support.

[dmitch]

I would definitely encourage other folks who encounter this problem to submit a support request. I did, and got a prompt reply, asking for more details. This CPU usage anomaly is a known problem, but the folks who are looking at it are looking for more data, which we - the users who see the problem - can give them. The more concrete info the good folks at IK have - details about systems, apps, configuration, etc., the sooner they'll be able to fix it.

[emurphy94108]

Just contacted support, to note that on an iMac Pro with an 8-core Xeon and 64 GB RAM, I get significant glitching, whether the buffer is 32 samples or 1024 samples.

I also noted that on my MacBook Pro with a Core i7 running at 3.5 GHz and 16 GB RAM, the M/B 2 collection is literally unusable. It's nothing but glitches.

[dmitch]

I found something interesting: the problems I reported all were with 96 kHz sampling rate. When I tried 48kHz, the new amps seem to be usable without CPU spikes. Still testing.

[emurphy94108]

Yes, I'd be curious to know. I've been recording at 24/96 for years now without any difficulty. Moreover, the M/B 1 collection, which has been out for a few years now, runs just fine at 96 kHz, as do all my other plugins.

[emurphy94108]

Last night I experimented with recording using the M/B 2 collection models at 48 kHz instead of 96 kHz. While doing so reduced the amount of audio artifacts from the models, it did not eliminate them. Moreover, increasing the size of the sample buffer, making it even larger than it was while recording at 96 kHz, did not improve matters.

[ewangle]

I'm having the same issue and will submit a support request. Any updates to share?

[dmitch]

I'm having the same issue and will submit a support request. Any updates to share?

Alas, no. Last I heard from IK, they were unable to reproduce the problem.

[sylvain.doyon]

Same thing here, strangely it worked well in test mode. I reinstalled everything and I still have the problem. There may have been an update in the meantime though...

[frownin_klown]

so is there anything anyone can do about the glitching with the Mesa boogie 2? really the only one that does it is the "Triple Crown" amp which suck because its a bad ass amp but seem no matter what i try changing buffer or anything it keeps doing the samething

[emurphy94108]

This probably isn't all that helpful to people who aren't willing to spend $4,000 on a new computer, but I've been using the Apple Silicon-native M/B 2 collection on a Mac Studio with M1 Ultra SoC (20 CPU/48 GPU cores), and doing so has completely eliminated audio artifacts, even using multiple instances of M/B 2 models on multiple channel strips and recording at 96 kHz with a 128-sample buffer.

It seems like IK designed these models with the notion that the hardware would eventually catch up with the software.