u-he

sascha · Post by **sascha** » Wed Jul 24, 2013 2:43 pm

Dip200 wrote:How much CPU power does the plugin need compared to other plugins ? For example Wait Digital VTM, Waves Kramer Tape ?

Apart from the fact that CPU consumption doesn't necessarily say much about sonic quality, public comparison is none of my business.
I've given some measures in past videos during Musikmesse, one of them being around 3.5% per instance on my i7 Win machine (3.6Ghz). My home PC (old Dual-Core 2.4GHz) consumes around 5% per instance, although I haven't checked that again lately.
(These figures apply to 'studio' mode. Delay & Flange need a bit more, but one usually would not use these in a multi-instance / multi-track environment, at least not as many, I suppose)
At 44k1 & 48k, we're constantly running at 8x oversampling (4x at 88k & 96k, 2x at 192k) in the main tape part, and still do half of that whenever nonlinearities are involved in the other sections.

Before anybody asks: no, we don't have an eco mode. Because we're using a HF bias oscillator and have lots of stuff going on beyond normal audio bandwidth.

Dip200 · Post by **Dip200** » Wed Jul 24, 2013 4:01 pm

Thanks for your answer but it doesn't help me. I can't estimate the cpu power consumption (well, asio power consumption is what I am interested in actually) if you don't compare it to other plugins. Does it need more "asio power" than VTM or Kramer tape ?

This question has nothing to do with audio quality by the way. I like to work in real time and I have a very powerful workstation. I'm just wondering if I can squeeze in some instances of it without using the jbridge performance mode.

sascha · Post by **sascha** » Wed Jul 24, 2013 4:07 pm

I've given you all the info I could. If that doesn't help for the moment, I'd suggest you just try out the plugin when it's out and see for yourself.

Agility · Post by **Agility** » Wed Jul 24, 2013 4:18 pm

sascha wrote:I've given you all the info I could. If that doesn't help for the moment, I'd suggest you just try out the plugin when it's out and see for yourself.

......When?

Dunc · Post by **Dunc** » Wed Jul 24, 2013 4:26 pm

I don't think it's down to developers to test out the performance of their "digital offspring" with other developers "offspring".
If that were to happen I could see a lot of nasty flame wars happening because such and such a plugin is supposedly more efficient than the next persons plugin.

Sascha has informed us of the likely CPU useage in a manner that does not compare it to any other plugin of the same type - we get to know roughly how much CPU juice it is likely to need and no other Dev's plug has been harmed to give us that information.

Anyhow, moving on...we are just going to have some more patience and give the u-he guys have some peace to work and get Satin out

sascha · Post by **sascha** » Wed Jul 24, 2013 4:26 pm

Agility wrote: ......When?

We don't know... but we'll try to find out...

kmonkey · Post by **kmonkey** » Wed Jul 24, 2013 4:28 pm

Dip200 wrote:How much CPU power does the plugin need compared to other plugins ? For example Wait Digital VTM, Waves Kramer Tape ?

I laughed and laughed and laughed and i am still laughing...

Wait Digital...dude you made my day...that was so funny..

I understand why you want to call Slate "Wait" but even so it is really funny..

Sorry guys for offtopic post i almost pissed myself..at first i was thinking that it is a typo or that there is some company producing some tape plugin of which i don't know yet then i realized it's Slate digital..hilarious..KVR can be so much fun "sometimes"..

benj3737 · Post by **benj3737** » Thu Jul 25, 2013 1:28 am

Really excited for Satin!
A preset request: I'd love some presets that focus on 'broken' tape machine type sounds. Sending a signal to an old walkman with the batteries about to die can be pretty amazing. Would love to recreate that in the box. I hope such 'ugly' sounds are possible!

marcoskohler · Post by **marcoskohler** » Tue Jul 30, 2013 8:25 pm

bump bump! Almost August! I AM DYING!

Nig · Post by **Nig** » Tue Jul 30, 2013 8:31 pm

sascha wrote:On the release date: we're still aiming for July. I'll have to get our youngest daughter become familiarized with Kindergarten next month, so I can't be all ears to u-he affairs for some unspecified period in August, therefore my wish is have it released before.

So it'll be released tomorrow then?

Urs · Post by **Urs** » Tue Jul 30, 2013 8:35 pm

marcoskohler wrote:bump bump! Almost August! I AM DYING!

We all are... From tomorrow on I'll be hunting remaining bugs (one of which is supposed to be a side effect of fixing TIWLB aka The Infamous White Lines Bug). And I'll also go back to work on AAX. If AAX isn't final, we still want to have a Satin AAX beta from day one.

- Urs

benburling · Post by **benburling** » Tue Jul 30, 2013 11:27 pm

Couldnt help it.

xx

Compyfox · Post by **Compyfox** » Tue Jul 30, 2013 11:40 pm

sascha wrote:We don't know... but we'll try to find out...

Aaaah... Madame Mim, the evil witch from "Sword in the Stone" ("Die Hexe und der Zauberer: Merlin und Mim" in Germany).

Seriously... can hardly wait for SATIN, but knowing that there are several perfectionists behind it, I can wait for a fully fleshed out version.

One question though, Sascha.
You mentioned around 3% per Instance on your i7 with 3,6GHz.

Which i7?
Bloomfield, Gulftown (Hexa-core), Sandy, Ivy?

I recently OC'd my PC to 3,6GHz as well, but that rig is a Bloomfield (920) - so performance might differ greatly. Even at the same CPU speed - and especially in Cubase 7.

sascha · Post by **sascha** » Wed Jul 31, 2013 9:08 am

Compyfox wrote:One question though, Sascha.
You mentioned around 3% per Instance on your i7 with 3,6GHz.

3.5%, currently, in studio mode. Delay & flange use a bit more, but since studio is the typical use case for multi-track, I take it as reference for my benchmarks. Since several years, I benchmark in Samplitude, stacking up 10 instances and using large (WDM) buffer sizes to get away with load spikes and ASIO overhead.

Compyfox wrote:Which i7?
Bloomfield, Gulftown (Hexa-core), Sandy, Ivy?

Frankly, I've never heard of the first two. But I treat these things as volatile and disposable data anyway. I'm over 40, so I seem to begin filtering input data...

My Win7 says I'm using an i7-3820 CPU @3.6GHz. Whatever that 3820 is.
I might still recall various STS-xx mission numbers from my youth, as I swallowed everything on the Space-Shuttle program in the 80s. I still know my first Lego spaceship had LL-928 labeled on it. My C-16 sound chip was called 8360R2 (printed in my memory as I blew up two of these, and they were f*ckin expensive spare parts for a school kid). But computers, these days, I don't care that much, as long as they get the job done. And it doesn't get any better that they put names on the damned chips. And if there's a guy called Rumsfeld or a girl named Mandy doing something inside my computer, I wouldn't want to know...

Compyfox · Post by **Compyfox** » Wed Jul 31, 2013 3:02 pm

sascha wrote:3.5%, currently, in studio mode. Delay & flange use a bit more, but since studio is the typical use case for multi-track, I take it as reference for my benchmarks. Since several years, I benchmark in Samplitude, stacking up 10 instances and using large (WDM) buffer sizes to get away with load spikes and ASIO overhead.

Hm... so I assume the ASIO buffer is around 512 to 1024. Lower buffer sizes, and the CPU might rise up to 5-8% per instance (i.e. 256 samples at ASIO). Interesting bit of information

sascha wrote:Frankly, I've never heard of the first two. But I treat these things as volatile and disposable data anyway. I'm over 40, so I seem to begin filtering input data...

I can relate, but sometimes the details are what interest me.

sascha wrote:My Win7 says I'm using an i7-3820 CPU @3.6GHz. Whatever that 3820 is.

It's a Sandy Bridge E, but not a hexa-core (6-cores) but a quad-core. The smallest of that series. Next would be Ivy Bridge (37xx series) and the new Haswell (4xxx seried if we talk about desktop).

sascha wrote: I might still recall various STS-xx mission numbers from my youth, as I swallowed everything on the Space-Shuttle program in the 80s. I still know my first Lego spaceship had LL-928 labeled on it. My C-16 sound chip was called 8360R2 (printed in my memory as I blew up two of these, and they were f*ckin expensive spare parts for a school kid). But computers, these days, I don't care that much, as long as they get the job done. And it doesn't get any better that they put names on the damned chips. And if there's a guy called Rumsfeld or a girl named Mandy doing something inside my computer, I wouldn't want to know...

I was just curious to somewhat "virtually" compare the CPU load for Satin. And I got to know what I wanted to know. Thanks.

Speaking of number:
C-64 SID Chip (6581, 8580, 6582), several AMIGA types, the Intel chipset "names" (and don't forget AMD), cell phone pins, EC card pins, ... man, this can get really confusing over the years.

u-he

KVR Audio

Satin - Beta