it’s even smaller: the ASRock Deskmini 310:
Now here’s an interesting discovery:
- I start Cantabile, set audio engine to “Null Audio”
- I load the song “Perftest” - Time load settles at around 32% average on my studio PC
- now I stop the audio engine (on/off switch) - time load (obviously) goes down to zero
- now I turn the audio engine back on - bindings fire again and start playback (since Song->onLoad also fires on turning the audio engine back on)
- now the time load is only at around 14% average and 20% maximum
- same on my old live laptop - average drops from 43% to about 19%
- no such effect on my new live laptop - average stays around 34% after engine cycling
- also no effect from engine cycling on my live cube
@brad: any idea what’s going on here? Maybe you can reproduce this?
Sorry for the late I was “on the road”
My studio PC is "core I7 - 7820X 3.6 GHz, memory access : 3600 Mhz
Time load : 53% - Average 38,6 %
But I wonder if it is not more relevante to observe the CPU load dedicated to Cantabile with the cantabile ‘profiler’
CPU Load (Cantabile) : 1.9 %
What do you think of that ? There is a huge difference between the "time load and the CPU load, what is the more important ?
My gig configuration will be largely under my studio configuration (certainly i5 6th generation) …
Next I will try on my professional laptop (just for the test) it has a Xeon
I have now used this song on four different machines with Tyrell installed in different VST directories: C:\Vstplugins, C:\Program Files\VstPlugins, C:\Program Files\Steinberg\VstPlugins
It loaded and started flawlessly on all four machines - just out of interest: where is the Tyrell dll installed on your machine? And: are you using the 32 bit or 64 bit version of Cantabile? I’m using 64 bit on all machines.
Time load is the more relevant metric for using Cantabile live - it measures how much time it takes Cantabile to fill the audio buffer compared to the time to play the buffer. Once time load crosses above 100% you start getting drop-outs, irrespective of overall CPU load.
I played manually the E-major. Here is the result:
- 43.2% average. 45.3 max (i7-4810MQ@2.8GHz) Null Audio/256
- 42.1% average. 44% max. same processor. Audio Board ASIO/256
I’m unable to reproduce your “19%” conundrum.
I expreminented the same thing…
Stop and reopen and you win some extra %… well I saw that too.
I tried Torsten’s “perftest” song file, with null audio @ 44.1 / 256
Average time load - 68.0%
Maximum time load - 89.4%
I’m using a 10 year old 1st generation i5-750 @ 2.67GHz 4 core desktop.
Ya I know it’s antiquated, but has been reliable. And actually survived (barely) this test.
I did not experience the load % changing after stopping and starting the audio engine like Torsten experienced.
I got better performances in aggressive mode. Not so much as you. About 8%
still better than nothing
I made new test with the same parameters.
Surface pro 2 i5 70% 75%
Surface pro 3 i7 68 % 78%
Surface pro 7 i5 1035G4 48% 60
Dell inspiron i7 8565u 43% 50%
Pc (2014) i7 4770k 42% 45%
This is not directly related to your search for an older system, but I thought I’d mention that early this year I went through an exercise of trying to find a powerful laptop to handle a live VST/etc. laptop. I tried a few “loaded” consumer laptops in sequential order because none of them worked for me and I had to return them as I tried them. Yoga, Thinkpad, Dell. I selected the most current/powerful/fast CPU, 32GB, dual SS drives, etc. The only thing I didn’t spend on was the graphics/monitor. All of the laptops I tried had audio cracks & pops to some unsatisfactory degree. Even if I got a small glitch every 5 minutes, I considered it a failure. I performed all sorts of tuning, cleaning up, … nothing helped. I read, and was told, that the laptop power management (due to having to manage battery power) got in the way of clean-performance, … and that I should try a Clevo chassis (or any other “gaming laptop” type chassis). I ordered one (which did cost 30% more than the consumer laptops). I have not had any glitches at all on my current laptop and am extremely happy with it.
What I run “live” on the laptop:
- 2 instances of standalone NI Kontakt (keyboard VSTs for 2 keyboardists)
- Cantabile, which runs 3 instances of NI Guitar Rig in stereo. This results in 3 vocalists having 6 channels of plugin-effected vocals. Also an occasional Waves Tune Real-time “auto-tune” plugin.
- ShowBuddy Lighting software
*** Both Kontakt and Guitar Rig interface to a QSC TouchMix-30 digital mixer
I’ve had reports of this from some users before but never been able to reproduce it myself. Also never heard of it with the Null Audio driver, but I guess that’s rarely used. Leave it with me I’ll do some tests in the next day or two.
In the meantime it’s probably worth running a CPU reporting tool to see if any of the cores are being throttled or parked before vs after.
I’m on an i9 7700x @ 3.3 GHz (10 cores - HT Off) and hit around 33% not quite the heights one would expect to hit over a lot of the i7s but does show that it’s not necessariy the number of cores that gives you the performance…
Cheers, and thanks to @Torsten as always.
It seems there are very different experiences, and a few processors to evaluate. I’m about to gather our posts to create a small database. I think I’ll use an Excel sheet, in old .xls format (Office 97-2003). So, it will be compatible with G-docs, OpenOffice and more. Here’s a possible screenshot. Any feedback will be appreciated.
In the meantime, I’d like to see more test like the one @Torsten proposed.
More cores can help. The most important parameter is the frequency. The higher GHz the better.That’s why desktop win on laptop.
That is my desktop…
At now, yes
A great way to understand the performance of a processor is a parameter called “thermal design power” (TDP). Hereby the PC’s typical values:
- Ultra portable PC (fanless) : 5W
- Portable PC : 15W
- Gaming PC and Mobile Workstation : 45W
- Desktop PC : 65 to 95W
- Desktop workstation and server : 95 to 250W (and more)
Although the relationship between TDP and performance is not linear, surely the higher TDP the faster PC
Another (@torsten) test. Sorry, the result is really funny, so I posted here.
- 22.1% average. 27.7 max (i7-7700HQ@2.8GHz) Null Audio/256
Well, it’s a MSI gaming laptop, but I think there is something wrong in this result. Time Load is too low. Using my typical test configuration (sampled piano + convolution reverb) I get 10% average. Same test on my older laptop (4th gen. i7) result is about 16% average. Both these PCs are heavily tweaked, I spent days trying turning windows-services off. The wifi is uninstalled, Antimalware is off, same for firewall, indexing, and lots of other services.
@brad, Windows 10 installed in the gaming pc is version 2004. I tried to set the core parking, I set to 0 the right parameter in right keys, but in the power plan i see Min. speed and Max. speed. After the update, core parking is disappeared. Any hint?