Pianoteq 5.2 i386 linux, CPU Overload

You are correct, the cpu usage is higher than expected. We're looking at it.

Not only linux, it also overloads my modest little i-7 laptop when playing the relatively simple embedded midi demo files in windoze 7.

Point is, this isn't a Linux problem.

Most of the time it uses 60 to 80 % on windows, which is WAY more than previous versions used for similar pieces.

I suggest you drop Ubuntu studio in favour of Lubuntu; being lighter you would improve Pianoteq performance; the latest 4.00 kernel improve performance a little bit compared with the distribution release one.

in the meantime that a solution to the high CPU usage issue is released I have switched back to 5.14 release

I had not noticed the regression on my Mac Pro under OSX 10.9 because pianoteq uses very little on this Xeon 3.7 GHz quad-core. Just for comparison, I played through the stock Blues Demo with D4 Daily Practice with version 5.2.0 and then with 5.2.1 after installing. What I got first is 12-14% cpu on the pianoteq GUI with a perf index of 70. On the new version, displayed cpu usage was down to 8-10% and perf was 71.

The displayed cpu usage is less than what I get on the Unix activity monitor for the process, though, about half. I guess this is because the pianoteq displayed value doesn't take into account the kernel and core audio time, and the Unix value is probably for the main thread only (% of one core) not taking into account the total threading for the 4 cores. I guess for less powerful computers, the pianoteq displayed value is conservative, and, depending on operating system and audio interface, some systems may be unusable much below 100% displayed cpu usage.

It is certainly less bad, but still shows over 30% on the D4 daily practice blues demo.
Windows 7, 8 Gig memory Intel I-7 2670QM 2.2 GigHz.

Windows task manager shows it using 4, 5 or 6 % of CPU with a couple of peaks at 9% - I think this is a percentage of the 8 available threads.

It also shows about 145 meg of memory in use, that was a surprise to me - though it is not a problem.

Not working for me...  5.2.1 goes into CPU overload almost immediately.  4.5.5 works fine though.  Someone mentioned rolling back to 5.1.4, do you have a link to download that version for Linux?   I can't find it.  Thanks for your help.

Another option to reduce CPU is to increase latency (done in linux/jack in increasing the buffer size).   Note that CPU usage per se is not the most important issue, DSP load is, and increasing the latency decreases the DSP requirement.
Of course, at some point latency becomes evident in live play so you can only do do much with this.   Also if you are using CPU frequency scaling, make sure to set the CPU to "Performance".

I must say I didn't notice a lot of difference on W7, perhaps a few %, but 5.2.1 is back to how it was before, happy to say. And I'm just running an i5 quadcore here. Playing Chopin's Fantasie Impromptu at 128 samples latency and I get about 10-15% CPU usage (2380P, 3.1 GHz).

BTW, you can always try and reduce the sample rate rather than increase the latency. That would also help reduce the CPU usage.

That said, no matter how much sample libraries get bigger, they can never really replace the playability and response aspect that Pianoteq has. That simply isn't gonna happen, the principle is too different.

aandr, ~30% on your CPU is a good result, I would say, if you didn't adjust CPU-related stuff. Even if it's i7, it's a mobile low-power version. At least make sure that things like core parking and SpeedStep are disabled, so that your CPU is always working at max, providing the best performance it can (2670QM can go up to 3.1 GHz in Turbo mode, you should definitely use that, as per-core speed is actually more important than the actual number of cores, for most of programs). You don't need a hardware upgrade. 30%-ish of CPU usage doesn't mean it's unusable - it's in fact VERY usable, especially if that's the only thing you're running (say, for a live gig). I don't know if you use other stuff along Pianoteq, are you talking about using it live, or using it in a DAW, you didn't give out those details. But it's definitely fairly usable, I would say.

Take your pick:

https://www.pianoteq.com/download?file=...p_v500.exe
https://www.pianoteq.com/download?file=...p_v501.exe
https://www.pianoteq.com/download?file=...p_v510.exe
https://www.pianoteq.com/download?file=...p_v511.exe
https://www.pianoteq.com/download?file=...p_v512.exe
https://www.pianoteq.com/download?file=...p_v513.exe
https://www.pianoteq.com/download?file=...p_v514.exe
https://www.pianoteq.com/download?file=...p_v520.exe
https://www.pianoteq.com/download?file=...p_v521.exe

Hey, another ubuntu studio user! Howdy

Yeah, i haven't been posting a lot on here lately, but i'm still happily twiddling around with pianoteq presets, and slowly but surely becoming a decent piano player.... at least i hope so

For me 5.2.1 is working fine, i3@3.4GHz, 4GB mem, using jack at 44.1KHz with a latency of 11ms (yeah that's good enough for me) .... pianoteq standalone says performance index around 70, and CPU usage around 15%.... never reaches 30% even if i mash all the notes with the sustain pedal all the way down.

I installed V2.1 on the little Eee-PC notebook computer today, in the Windows XP partition.
I'm not going to RAVE about it, but given the crummy little laptop speakers on that notebook, it doesn't sound TOO BAD.
I am NOT trying to get the best possible sound, but for my piano practice I think this might be an adequate music computer.  I will see how much better external speakers and an external D to A module sound.

The demo Chopin piece gets up to about 75% CPU load in the fast parts where the sustain pedal is down, but it doesn't overload or produce pops, drop-outs, etc.
Halving the sample frequency to (I think) 24K reduced that to about 40% without a huge degradation in the sound (OK, still through the same crummy speakers, so maybe it did but I can't hear the difference)

I may try a version 3 and will almost certainly try it in the Linux partition.
BTW, V2.1 was a 25Meg file

When you get that kind of performance out-of-the-box, it's time to stop messing around and leave it at that! Forget about updating kernel if it works!

I experimented with redhat fairly early, it was one of the earliest "distributions", before that we were cobbling everything together ad hoc.
I didn't particularly like it and have managed to forget most of its foibles
Without much to else compare it to redhat "Seemed" OK.

I think dpkg has done a LOT to help stabilize all Debian based distributions. updates/upgrades are well controlled because their dependencies are known.