Hive 1.2 and CPU

[tony10000]

Here on my internet computer which is an old weak i3 running at 3.1 GHz I'm getting 21% average with 3 notes (#36,18, and 60) with Hive "AS Ambient Soundscapes".

512 Samples (12ms) 44.1 Hz FL Studio ASIO. FL is spreading the load across all 4 logical processors quite well.

EDIT: Examining that patch further it's probably towards the more extreme end of CPU demand as it uses 21 total Unison, one WT, almost all the FX and a healthy dose of Release on both Amps (and it sounds awesome).

I am getting a slightly higher average CPU utilization (around 25%) using the FL driver and the same latency setting if just playing a single 3 note chord. However, my test loop is a 3 chord, 4 bar loop with chords held (1 bar each for first and 2nd chords and 2 bars for 3rd chord).

[tony10000]

Using Serum to play the same loop with my original setting using the factory patch "PD Slow Pad [AS]", it is maxing out around 30%. And that patch uses two 7 unison supersaws, a LPF, and two effects (Reverb + EQ).

To be fair, though, I did create a patch with 7 + 14 supersaws in Serum plus sub, a big stack of effects and long envelopes and it topped out at close to 50% CPU.

So, Hive 1.2 is lighter than Serum.

[Urs]

When does the CPU peak happen though?

Even when you set Hive to 3 voices and you play chords, if the next chord overlaps with the old (e.g. long release times), there'll be a brief moment of 6 voices playing. The old voices are being faded out over a few milliseconds to avoid any clicks.

[tony10000]

When does the CPU peak happen though?

Even when you set Hive to 3 voices and you play chords, if the next chord overlaps with the old (e.g. long release times), there'll be a brief moment of 6 voices playing. The old voices are being faded out over a few milliseconds to avoid any clicks.

I just tested it. The peak occurs as the third chord is played, as I would expect. The release time on the amp envelopes is 65 and 67, respectively.

It starts out around 19% when the first chord is played, then goes up to around 28% when the second chord is played, and then up to 33% when the 3rd chord is played.

If I reduce the amp envelope release times to 50, then it starts at 18% maxes out at 28%.

However, I still think Hive would benefit from multi-core switching as the result I observed when using Dune 3 with it on was pretty dramatic.

[EvilDragon]

If Urs says it wouldn't benefit, then it wouldn't. Hive doesn't use the same algorithms as Dune and as such you cannot draw an analogy that if it works on Dune 3 to would also work on Hive as well.

[Teksonik]

Works on DIVA.........

Anyway Hive is one of the those synths that I never even give a second thought about CPU demand. It's always been very efficient here but just like any modern synth it is possible to push it when programming patches at extreme settings (high poly, long releases etc).

I don't think Hive needs Multi-Core support but if other people need it then I sympathize. I'm still in shock that someone is using 44 instances....

[pdxindy]

However, I still think Hive would benefit from multi-core switching as the result I observed when using Dune 3 with it on was pretty dramatic.

The only benefit for multi-core is if the CPU use is high enough that one core is not sufficient for 1 instance. That is not the case in your example. You would likely be better off to leave multi-core off in your Dune example.

[pdxindy]

Works on DIVA.........

It is necessary with Diva because the CPU use is high enough that 1 instance of the synth can max out one core. Since that does not happen with Hive, there would be no use case where multi-core would be a benefit.

[pdxindy]

If Urs says it wouldn't benefit, then it wouldn't. Hive doesn't use the same algorithms as Dune and as such you cannot draw an analogy that if it works on Dune 3 to would also work on Hive as well.

In most cases, it would be better off to not use Multi-core with Dune. However, since you can increase the oversampling, it is possible to use much more CPU in Dune than in Hive. Of course Urs could add the multi-core capability to Hive, there just is not reason to do so.

[Teksonik]

Works on DIVA.........

It is necessary with Diva because the CPU use is high enough that 1 instance of the synth can max out one core. Since that does not happen with Hive, there would be no use case where multi-core would be a benefit.

Did I say it was necessary ? In case you missed it I said.....

"I don't think Hive needs Multi-Core support but if other people need it then I sympathize".

My point was that obviously U-He has Multi-Core code and knows how to use it whether they choose to use it in Hive or not.....

In most cases, it would be better off to not use Multi-core with Dune. However, since you can increase the oversampling, it is possible to use much more CPU in Dune than in Hive.

Just to set the record straight it's Modulation Rate that can be increased not Oversampling. Both have an effect on CPU demand.

I leave Multi-threading on at all times and never give it a second thought even when using multiple instances. No problem with the DAWs I use or my system but I have read of people who use Cubase having issues with MT.

[tony10000]

If Urs says it wouldn't benefit, then it wouldn't. Hive doesn't use the same algorithms as Dune and as such you cannot draw an analogy that if it works on Dune 3 to would also work on Hive as well.

Okay...point taken.

My conclusion is that Hive can no longer be considered a "light on CPU" synth, at least for my purposes. I am sure the additional modulation features of Hive 2.0 will only increase CPU usage. That is fine with me as I have numerous other options and know how to mitigate CPU usage in Hive.

[tony10000]

Works on DIVA.........

Anyway Hive is one of the those synths that I never even give a second thought about CPU demand. It's always been very efficient here but just like any modern synth it is possible to push it when programming patches at extreme settings (high poly, long releases etc).

Ummm...no. I tried the same test with Synthmaster and it barely cracked the low double digits in CPU usage. There are a lot of other modern synths that are very light on CPU. Europa is another recent example.

[Teksonik]

Works on DIVA.........

Anyway Hive is one of the those synths that I never even give a second thought about CPU demand. It's always been very efficient here but just like any modern synth it is possible to push it when programming patches at extreme settings (high poly, long releases etc).

Ummm...no. I tried the same test with Synthmaster and it barely cracked the low double digits in CPU usage. There are a lot of other modern synths that are very light on CPU. Europa is another recent example.

Ummm yea. If you push it Synthmaster can use its fair share of CPU. Try Best quality and 64 note poly then max out the Layers, use a bunch of effects etc....like I said extreme settings.

There are a lot of other modern synths that are very light on CPU

Yes like Hive.....here. Perhaps this CPU Benchmark comparison between our two studio computers will illustrate why.

CPU.png

Anyway I'm sorry you can't use Hive to its fullest but it looks like your request for Multi-Core support will not be granted so.....good luck. I hope you find a solution.

[PieBerger]

Did we jump back in time before multicore processors were cheap and readily available? What year is it, who's the prime minister? Did everyone's accounts get hacked by Lazyer?

[Urs]

Anyway I'm sorry you can't use Hive to its fullest but it looks like your request for Multi-Core support will not be granted so.....good luck. I hope you find a solution.

It's a lot more complicated than that. Until Repro-5, we always had a strict separation between synths that have multi core support and synths which have a mod matrix. The two concepts were mutually exclusive. The reasons are quite obvious once you know what happens.

For Repro-5 we broke that limitation. And if the ModMatrix in Hive was still a simple as the one in Repro, and if the modulation sources were as simple, we could probably enable multicore support within a few minutes. But this isn't the case. The presence of single trigger envelopes and LFOs alone contradicts the concept of multi core support.

I believe that adding multicore support to Hive would open a can of worms. We would have to add a lot of code, like, exceptions and stuff, which most certainly adds more CPU consumption.