Simon Phillips Studio Drums and GA Real-Time Peaks

Hello I’m an e-drummer and I’ve just checked out the new SimonPhillips lib. Great sounds! But when I watch the Real-Time Peak meter in CubaseArtist7.5.40 I just cannot believe it: a new project with only one VSTi: GAse4.2.20 with the SimonPhillips library makes my audio machine (a good i5 desktop audio-tweaked PC, RME FF400 with 48 samples ASIO buffer) show red Cubase real time peaks. Most of the time 50% up to 80%, and sometimes 100% with the red peak. Is the meter wrong or is GA extremly bad at ASIO performance??: though it shows red peaks I cannot detect dropouts or stuttering on the i5 machine; but on an older dual-core PC I get massive dropouts where other drum-sampler like EZdrummer2 or Superior Drummer2 even with lots of DSP-Fx stay cool).

Do you (at Steinberg) know drum-tools like SuperiorDrummer2 or BFD2? Compare the Real-Time Peak meter with your GrooveAgent and the SimonPhillips lib. What do you think? Are you for real :question:

Another Simon Phillips Studio Drums question :
I cannot find a variable Hihat articulation for e-drums? Where is it?

“RAM save” seems to be buggy. I could not in every case unload the samples or articulations that I wanted.)

Are there any e-drummers using GA and this new library? What’s your opinion? For me the sounds are outstanding and I like the layering possibilities for main snare and kick and regarding this it is worth the price
BUT the GrooveAgent ASIO performance is just a bad joke! (BTW: I could not really improve anything by altering the performance settings at the GA “options”-page - e.g. “reduce the voices” : nobody wants to have very few voices when e-drumming…)

I ask Steinberg: will this performance issue be improved promptly???

EDIT: I noticed crackling sound now on the i5 machine especially when playing fast rolls :angry: And I’ve tested a very potent audio notebook with a good and newest i7 CPU. Guess what!?: It’s crackling at 64 samples ASIO buffer (RME Babyface)… and whenever another VSTi comes in it becomes completely unstable and crackling.

Hi Craig,

Unfortunately good sound is always expensive on the CPU. There are several reasons why the Simon Phillips drums sound so good and so CPU hungry. The samples are quite long which lead to a great resonance and room sound, but this also means that there are a lot of voices playing at the same time when you are jamming on your DTX. There are also lots of microphone channels including the resonance mic that are great for mixing and let you tweak the sound in lots of different directions, but these also add to the amount of voices. The only thing that you can do is to increase the ASIO buffer or try it on a faster system.

Well, with the experience of audio quality, sample sizes (e.g. libs of 20GB), many many mic and bleeding channels, DSP and therefor the very smart performance of drum libraries in SuperiorDrummer2 it is hard to understand and accept your argument.

I’m sure if the Simon Phillips pack was made for SD2 it would run smart delivering a similar audio quality.

I’m afraid that the e-drum peformance (small ASIO buffer size) of that pack in Groove Agent is quite not competitive because my i5 audio machine (Intel i5 4670K - 4x 3.4 GHz) really is not a slow machine, believe me! :wink: and I know many (drum)sampler VSTi’s and non of them is performing that bad. Sorry. Please get it smarter: you owe it to Simon and his great drum sounds and future development of great acoustic drum packs like that and your competitiveness regarding acoustic drum packs and in the end to e-drummers like me. :wink:

For me as e-drummer GA is a complete joke. Absolutely Plastic sound, very bad response/articulation with e-drums. I can use this sing only with the electronic kits and programmed grooves.
There is absolutely no sense in blaming the “quality” for the high cpu consumption.

For a really good e-drum feeling with great sounds there is no way around SD2 with the Music City SDX and the Vintage EZD


You can’t say this about the Simon Phillips pack. But of what avail is it regarding the GA cpu performance… :frowning:

Most of all bad e-drum features/possibilities: as a matter of fact they forgot a variable hihat articulation (midi-note)…Wow! :angry:

Absolutely agreed: look at the real-time peaks when playing (e-drumming) other market-leading high quality drum sampler VSTi’s!

After having tested it on a potent i7 audio notebook (newest generation i7 6700 , RME Babyface) today and having noticed the red peaks and the crackling I must confirm: This is a complete joke!

Dear Matthias Quellmann and Steinberg, please inform yourself how market-leading drum sampler VSTi’s perform on such potent machines in this day and age. :wink:

Make sure you have the latest version of Groove Agent SE 4 installed.

I just tested this expansion on the full version of Groove Agent 4 and according to Windows, it uses only between 5~10% CPU when played in real time (about 1~5% when playing a fast programmed pattern in Cubase Pro 8.5 with ASIO Guard enabled), about the same as the EZDrummer 2 stock kits. If updating GA SE 4 doesn’t fully fix the issue, try loading the 16-bit versions of the kits and/or increasing the latency of your interface a little tad.

These are my settings:

This is installed on a 1TB 7200 RPM HDD, my CPU is a i5 4690k running at a fixed 3.5 GHz and my interface is 16 years old and has bad drivers.

I have GAse4.2.20 installed. There is nothing about ASIO performance in the ReleaseNotes of 4.2.30

Don’t watch the CPU (your 5-10% says nothing!) but the “Real-Time Peak” meter in Cubase (shorcut: F12). What does it say when you play e-drums at 48 or 64 samples Asio-buffer??? (e.g. play fast (cymbals and hihat) and not only one snare-stroke per few seconds!) I’m talking about ASIO performance and not about the Windows CPU meter and also not about the GA CPU meter. There is a big difference!
What was your ASIO buffer size in samples when you say you played in “real time”?

I play in CubaseArtist7.5. As far as I know ASIO Guard has nothing to do with real time play.

This all helps nothing! Of course I’ve tried everything!

My settings are the same (except “Max Preload” : I have 1200 MB default setting). But there exists no “Multi-Core” setting in GA SE. But your setting is OFF here so I think it will be the same.
…Or not: perhaps the full version of GA performs completley diffrent compared to the GA SE… This would also be a bad joke… :wink:
Then please try it in GA SE and report back if you notice a difference.

I tried Groove Agent SE 4.2.30 (the latest), it seems to perform the same as the full version. My interface’s driver doesn’t report the latency in samples properly, but at my usual settings the realtime peaks (not the average load) while playing as fast and hard as I can are around 20~30%. At the lowest latency possible they never go over 50%.

Note that the VST Performance meter can be a bit misleading. Sometimes it will be constantly almost full but you’re still able to load and use a lot more plugins without issue.

Your interfaces are a fair bit newer and better than mine so I don’t know what could be causing this issue for you. I think these drums were recorded at 44.1kHz, have you tried changing your sample rate to that? Have you tried the Steinberg Power Scheme? What about updating GA SE 4, does it help at all? Keep in mind that updates sometimes have undocumented changes.

So what is your interface and your latency?? Please name it! Almost every ASIO device reports latency in samples or milliseconds. What does Cubase show as input and output latency?? Mine is around 2-3ms input and also 2-3ms output at 48 - 64 samples ASIO buffer. :wink:

I know about a little bit of headroom. But as I experienced at real time ASIO buffer size (48 or 64 samples!) the issue will be at least a little bit of crackling.

Please understand the following: when I’m playing SuperiorDrummer2 (also with many many FX and mic channels) or EZDrummer2 my Realtime Peak meter is around 20-50%.
With the same settings it is around 70-100% (and over) playing GA with the SimonPhillips pack.
I never notice any crackling when playing SD2 or EZD2 - but I do when playing GA SimonPhillips!
I have tried so many different VSTi’s (samplers and synths) and I have never seen such a performance with them.

My project sample rate is 44.1 kHz and of course I set the Steinberg Power Scheme.

I will update to 4.2.30 tomorrow but I’m sure that it won’t help: I’ve found some threads on the internet blaming GA for miserable CPU performance. This issue seems not to be new…

I can hardly believe that you have let’s say 4-10 milliseconds latency and your realtime peak meter never goes over 50% because my i5 CPU is only very little worse than yours (i5 4670K - fixed to 3.4 GHz) and the machine works with RME interfaces (FF400 , Babyface). BTW I also tried the same machine with a Focusrite Saffire Firewire interface: Same thing there!

Can you please try it with Cubase7.5 instead of 8.5? (Seems to be the only difference…)

OK, now I know what causes your 20-50% real-time peak and my 70-100% (and over).

It is Asio Guard!

You cannot play e-drums with Asio Guard enabled because it is of course adding massive latency (minimum 17ms, default 23ms and max 92ms).
How is it possible not to notice this latency (even of course 17ms!) when playing e-drums! :open_mouth: :astonished:

So for real-time play you’ll have to disable Asio Guard in Cubsae 8.5; a latency with Asio Guard enabled is absolutely out of the question for e-drumming!

So disable it and you will experience the “smart” performance of Groove Agent. :laughing:
And remember I get 20-50% real-time peak playing EZDrummer2 or SuperiorDrummer2 or BFD2 or whatever of course without Asio Guard.
That’s simply the difference between the performance of market-leading drum sampler software and Steinberg’s GrooveAgent :wink:

I have tested all this with Cubase8.5.20 and GAse4.2.30 what is the latest update in every case.

I think I will make a youtube video documenting this unbelievable performance issue :exclamation:

EDIT: No, this is wrong! Forget about it. It is not Asio Guard: a different test has proofed it…

I have a Roland SD-90, this is a very old budget MIDI module/interface hybrid which I believe doesn’t even have 100% dedicated drivers. At least in theory this should be much worse than what you have.

According to Cubase, the lowest possible output latency (DAW/VST to monitors/headphones latency) of my interface is 3.175 ms, this should be just 1 or 2 ms higher than your interface set to 48 sample buffer. There’s no way just 1 or 2 ms of extra latency is making such a huge difference in terms performance, specially considering your interfaces are both better designed and are supposed to have great drivers.

Having ASIO Guard on or off makes no difference in terms of performance or latency when playing GA4 in real time, at least in Cubase Pro 8.5 (I tested). ASIO Guard is only enabled on tracks that aren’t Record or Monitor enabled. In other words, if you’re playing a track in real time, Cubase automatically disables ASIO Guard for that track, only tracks playing pre-recorded MIDI or audio are played at a higher latency. I only brought ASIO Guard up because in a project, 95% of the time GA4 will be playing something that you recorded/programmed earlier.

Well that is completely not my experience:
I have GA as a “rack instrument”. Then I rout the output of a MIDI track to GA input. When I set the max Asio Guard latency I can immediately notice the laid back audio (over 90 ms) when hitting a drum pad with of course Monitor on that track enabled.
If I have Asio Guard enabled (even with its minimum latency) it makes a big differnce and I get the result of the real-time peak meter that you report. :wink:

EDIT: this was my experience yesterday on the dual core machine. After a new test something seems to be wrong there…

Now I’ve tested again on the i5 and you are right: Asio Guard does not matter here. So I was wrong in this case…

But I still don’t know why your Real-time Peak meter stays cool. Regarding this I will soon show in a video the difference between EZDrummer2 (stock sounds) and GA (SimonPhillips) running on the i5 machine with same settings in both cases. :wink:

here is the video “GrooveAgent ‘Simon Phillips Studio Drums’ VST RealTimePerformance vs. EzDrummer2”:

Checked out sp drums yesterday and gave it up after two hours. Poor performance within a small “Padshop, Triebwerk” project. Maybe this depends on the quality of samples, maybe on my system, but an extremely difference between sp drums and default acoustic agent sets (ga4 full version) makes it impossible for me to get sp drums. Just uninstalled it :frowning:

“Does anyone know if this works with an electronic drumkit (I have a KAT KT3 that works nicely with EzDrummer) and I’d love to get some Simon sound out of it. I can’t find ANY information about this anywhere, which is weird.”

This is the first comment/question on Youtube (“Introducing Simon Phillips Studio Drums for Groove Agent” - Steinberg Channel)

I tried to answer and guess what? My answer does not appear. And absolutely NO comment by Steinberg regarding the comparison with Toontrack except " Unfortunately good sound is always expensive on the CPU "…

Actually it has to be: Unfortunately bad programmed VSTi’s are always expensive on the CPU :laughing:

And no comment on the eDrum features (like variable Hihat).
Hey Simon, if you read this pleeeeaase comment on the bad/missing eDrum features and the CPU issue.
Can you imagine there are musicians out there who want to play eDrums with your great sound? Don’t you feel sympathy for them?

You can call an idiot :wink: but I’ve opened GA4 with SP-Drums in Studio One 2.6.5 and the performance (same 128/44,1, same computer, same interface, same MIDI events, Plugin as VST3) decreases from 25-30% to 10-12%. Empty song, GA4 only in both situations.

Studio One 2.6.5 :question:
…the performance decreases from 25-30% to 10-12% :question: What does it have to do with the Cubase Real-Time Peak meter??? What performance meter is it? the Studio One performance meter? Windows Task Manager?

same computer :question: - “iMac 27 i7 El Capitan” :question:

same 128 :question: what does it mean??

same MIDI events :question: :question: :question: make a video of your drumming!

I stand corrected after demoing this.
Sound (with the internal groves) is great.

Sadly there is no predefined Map for Roland Drumkits for GA4. That is a shame as its is just too much fiddling for me as enduser.

But the sound is top notch, as is the Nashville one. It gets an expensive month…

At first I was skeptical regarding the quality of the groove agent 4 samples. However after working with it quite a bit on several projects I’m finding that these kits if chosen correctly sit so well in the mix with very little tweaking required. I think Steinberg has gone the extra step to create excellent drum kits ready for production.
Also the variety of kits and options available is impressive.