OK. I have isolated this further. It appears it has something to do with the particular couple of Cubase projects I'm working on. If I start a new project in Cubase...load up the VIP2--assign it a medium workload, whether the GUI is open or not, it makes only a slight difference on the CPU.

So, I will trouble shoot it from here on out as if it's a Cubase project issue and report my findings to Steinberg (and here)...I don't know if it's something to do with some other plug in running in that project that somehow only manifests in VIP2...? Or the VST Expression maps being used in it...? Or just some sort of project corruption....? But, I can duplicate it 100% of the time in those projects, but not in a new one--even with VIP2 under the same or heavier workload. Seriously weird. And, people wonder why I hate computers so much. [:@] It's all I do every day--fix malfunctioning ones. Nothing I hate worse than working on my own when I get home. 

Anyway, it does appear that VIP2 is only one ingredient in the "CPU eating reciepe" here.

Well...new project...same exact midi tracks (exported from the big project) jumps from 5-10% playing back to 25-30% with the GUI open. Not quite the same increase, but more than is acceptable. 

Cutting the display to 16bit color buys a few percentage points, but just a few. 

Im really not feeling good about buying this. If it means I'll need to replace my video card, too...adding up the library extensions, Ssd raid, and vip2, AND a new video card...getting to be quite a pricey upgrade. I do love the VSL chamber/solo sounds, but neither LASS nor the often maligned PLAY version of QLSO give me these system issues. 

I have been working on some GUI optimizations, perhaps you could send a mail to support@vsl.co.at "To Martin", then I can send you this version of VIPro, to see if your situation improves.

Thanks,

Email sent. Thank you. 

 I appreciate Martin's help and "beta build". So much so that I went ahead and bought the app. However, it seems the GUI is still hitting the CPU really hard.

We're tryingto isolate whether it's his i7 just absorbing the calls or his Quadro video chipset takingthe load off the CPU better than my Geforce. Very conflicting reports on the web about how much Win7 really uses the GPU...and how much the drivers for "gaming" video cards do at ALL for 2.5d Win7 acceleration. Lots of benchmarks have intel and nForce integrated laptop graphics besting $700 2 space 3d monsters when it comes to Windows acceleration.

Is this noticable to anyone else?

I just LOVE how CPU light VI/VE has always been (and the underlying engine still is, depite all the cool new DSP it's doing now...I don't break 30% on a full arrangement with whatever audio tracks/processing I'm doing on top of it. But, now, open the VIP2 GUI and it's pushes it up to 50-60% and causes the audio to glitch.

Okay, so VI Pro 2.0 is killing my CPU.

I have not been as good as I should with using EQ and saturation etc. but I started using them when VI Pro 2.0 came out. Therefore, I thought that my use of many plugins was killing my CPU. But after testing everything with and without extra plugins, buses, reverbs, and any combination of things.. I saw no change.

THEN I found the culprit! Just loading multiple VI Pro 2.0 instances instantly causes my CPU usage to jump tremendously. I'm monitoring it in the task manager and the resource monitor in windows. It's bad all around. I wouldn't mind if it was just the task manager, but I can't load NEARLY as many instruments as I did before. I'm currently using about a 65% sized template compared to my old one and I've reached my limits. With VI Pro 1.0 I had many more instances (in some cases double) and I was using less CPU than I am at it's worst now.

Is there any more news than the private testing version listed on here? I wish I had known this was VIP 2.0 earlier to report these results but I've been very busy with personal matters and the holidays.

Thank you,

-Sean

Sorry, I should have clarified. I can still load multiple instances. It's just that I run into performance issues with maybe half of what I used to load. I loaded just the instances with no samples loaded and I still see the same cpu performance. With samples loaded, I just have an unusable glitchy performance the more I load now. I realize VI Pro 2.0 is an upgrade, but I wouldn't imagine I'd have to upgrade my CPU for it also. Not much else is cpu intensive, right?

-Sean

Hi everybody,

Martin has kept on working on the optimization of VI PRO, there will be a new version available as soon as we are through the tests.

Best,

Paul

I just sent him my test results. While still more CPU usesge than any of my other VIs (none of which show significant difference between GUI opened or closed)...it's WAY down from what it was. And in my "real world test" (as opposed the shared test project we've been using) its working fine at 128samples on full arrangements with only occasional minor audio glitches. 

While I CERTAINLY welcome further improvement, I want to thank Martin for working on this so diligently. I did go ahead and buy my permanent vip2 license. Being in IT, I'm glad to help and also understand the development cost calculations that can equal a company having the attitude that it's not worth the development time to make code more efficient. I want to publicly thank VSL for perusing GUI optimization despite it running great on i7/Quadro powered machines. I know I, for one, would rather spend money on more VSL samples for vip2 than on building a new box that wouldn't matter if the GUI were optimized or not. 

Thanks you. I look forward to the official release as well as any further testing I can do for the team at VSL. There are not a lot of software products I honestly LOVE(I am a hardware guy at heart)....these samples played with the vip2 engine...I do. The sound of strings I've always heard in my head are now under my fingers. 

Also note at 256 samples...zippo glitches. Still very playable(live), as the nature of strings and the "no hidden buffer" feeling of the VI engine does not add up to anything to be concerned about. And I'm bugged by latent software instruments. 

Thank you so much Martin. 

Thanks for the kind words. I'm still wondering why your computer is experiencing occasional glitches however. Normally a system should never produce these kind of glitches from any GUI/graphics load, since this is run in lower priority threads. It could be related to your soundcard drivers, or other system components. I doubt it is graphics card related - your card is plenty fast, definitely faster than my laptop chip. Bad soundcards and drivers tend to be a lot more sensitive for this.

Popmann, do you have another computer, another sound card or audio interface that you could try swapping hardware with? I only ask because I completely agree with Martin on the bad sound card/driver point. I've had great hardware where a driver and one bit of software didn't get along. I know you said you're an IT guy and you know your hardware- So you may have already tried switching things out, but sometimes I need hearing the obvious myself as sometimes certain troubleshooting steps seem to slip my mind (sadly more often than not). lol

-Sean

My only other audio interface has no 64bit drivers. Old Maudio 2496 card that had the best GSIF drivers of anything I used back in the day--and I went through a billion. This is the only thing that has made this system even hiccup. And it's just the GUI for it--leave it PLAYING, open any other GUI all day--no glitch at all. 

Technically speaking, I actually don't think my video is as feature rich as the Quadro, which is native DX11 accelerated and has a faster memory bandwidth, and if memory serves, more "cuda core" pipelines. 

My local buddies all have FW, which I guess I could borrow, but does put another layer of driver/controller compatibility into the mix. 

I have the iMac, but them there's nothing similar. I'll see if Echo has updated their drivers since build. Since I've now updated everything ELSE as part of this. 

I have an Nvidia 9800GT which has only 64 Cuda Processors (relatively low) and I'm running fine. Although, I didn't think that VIP or VIP2 used Cuda to its advantage. I'd be interested to know if it did though. If I had to guess, I'd think VSL would have gone the OpenCL route as MIR runs better with ATI.... if I remember reading right? Maybe I'm going insane. lol

I have an Echo pci express card and I think it's phenominal for what it is. It's sitting on a shelf, but I was always very pleased with the echo drivers. That's just my experience anyway and that's not with a FW interface. Just thought I'd mention it.

Either way, swapping one would at least reveal if somehow there are any changes in performance not found through other tests. Although the point about it glitching 'just when the GUI is running' is a bit fishy. Sorry if I'm being redundant. I know you've already troubleshooted a great deal. I just thought I'd add my two bits here.

-Sean

Bad news...have the latest echo drivers. Not been updated since 2/2010. However, it did get me to read the "read me" on the driver set...it says under known issues "if you experiences glitches or distortion while recording, disable unused inputs/outputs in your DAW software". So, I have disabled the ADAT IO, which is only occasionally used. Can't rerun my test though, as it uses the extended libraries, which apparently just ran out. doh! Since my real world projects are not glitching since the update (only the stress test project I made)...I've got no way to see if that helped. 

I also found a setting in Cubase to boost the audio priority. I've set that to boost. 

Also on the graphics card...didn't mention--the drivers for the Quadro cards are meant for more window and openGL acceleration (intended for CAD use) versus the GeForce line. I know the CAD engineers at work require Quadro and FireGL cards even though the "consumer" lines are technically "faster". avid only certifies Quadros for use in HD rigs, too. 

What CPU are you running with the 9800gt? Do you see the CPU jump significantly with the GUI open? I'm running a fanless 9500gt/1gb with a core2quad 2.5ghz.  Martin's testing with a i7/Quadro. I do also believe its using the CPU rather than hardware acceleration. However, the problem gets far WORSE with aero (thus graphic acceleration) disabled. 

Im going to continue to use it with the newly disabled IO. Seems like a weird thing that disabling IO that aren't being used would change performance, but it does say that in the driver read me, so...We'll see...once I actually get the money together to buy the extensions, I can rerun the test project. 

@Another User said:

Seems like a weird thing that disabling IO that aren't being used would change performance

Actually, I don't know a lot about it... but I know that RME has it's actual I/O and virtual channels. The virtual channels are far more than the physical channels. But I know that you can only run so many at different rates. Something like 512 channels at 44.1kHz and 12 channels at 192kHz (or something...?). Anyway, even having channels enabled will effect the ASIO performance of your interface. I'm not sure why. But I know that the reference to the readme saying 'disable unused IO' would likely be accurate. Although, if you don't have this problem with loading other samplers, VST's, etc.. then I don't see that affecting your performance too much with VI Pro. I'm not sure what to tell you unless you have another machine you could use for your setup.

-Sean

Bump to put this original thread right next to two newer, GUI related threads...making it easier to cross reference for common threads.