u-he rePro in the works

[EvilDragon]

Yes, it's special. For Urs, at least, since it was his first analog synth.

http://www.kvraudio.com/forum/viewtopic ... 4#p6360824

[fluffy_little_something]

Oh, I see

[fmr]

Take something like the M7 Bricasti for example. It has over 30x (estimated) the processing power of the "state of art i7" and it's all dedicated to one feature: reverb.

30x the processing power of a "state of art i7"? I very doubt that...

[Numanoid]

Kind of an interesting statement

I guess an older pentium processor would have 100x the processing power of i7 if it only was used for notepad wordtyping

http://www.soundonsound.com/sos/feb08/a ... astim7.htm

[Jax Pok]

Take something like the M7 Bricasti for example. It has over 30x (estimated) the processing power of the "state of art i7" and it's all dedicated to one feature: reverb.

30x the processing power of a "state of art i7"? I very doubt that...

Read something about it ages ago it's not more than a single i7 but the way it works for reverb is far more optimised.

[fmr]

Take something like the M7 Bricasti for example. It has over 30x (estimated) the processing power of the "state of art i7" and it's all dedicated to one feature: reverb.

30x the processing power of a "state of art i7"? I very doubt that...

Read something about it ages ago it's not more than a single i7 but the way it works for reverb is far more optimised.

It uses six dual-core Analog Devices DSPs. I had a Scope card that used 15 Analog Devices DSPs (perhaps of an earlier generation). A single i7 was able to do much more than that card. Usually, code for DSP chips is very optimized, but we recently could experience how much more powerful a generic CPU like the i7 is compared to DSPs, by running side by side TDM and native versions of the same plug-ins (and most of them have been ported from TDM to native, so, the code was optimizd on DSP. usually, we could open like 4 yimes the number of instances in native versions.

So, no, I don't believe the processing power of the Bricasti can even beat a single i7, no matter how optimized the code is, let alone 30x. This must be some marketing bullshit.

[pdxindy]

Think about this for a minute. Emulating complex phenomena like analogue and non-linearities is computationally expensive.

If you do circuit modelling, yes, maybe. On the other hand, Diva has seen a few updates, and obviously is not at all as CPU consuming as it was when it came out. Which tells me that there should be always ways to get the same result without frying your CPU. Which also is the whole point of this project as far as i understood it.

Personally, i only can hope that devs will always keep an eye on CPU consumption, and find compromises, because frankly, a plugin i can only use 5 instances until it fries my PC is kind of unusable for me. I don't believe you have to model any single fart coming out of an analog system either to have good sound quality.

I hope that some Devs will push the envelope hard even if it makes our CPU's bleed. No offense, but I don't care that you consider 'only' 5 instances unacceptable. If it gives me results I like, then bouncing is fine with me. Plus 5 instances is an amazing bounty already! I think it is great that different Devs cater to our different interests.

[EvilDragon]

Here's a relevant post for DSP performance vs general purpose CPUs, from Sean (ValhallaDSP):

http://www.kvraudio.com/forum/viewtopic ... 5#p5998245

[fluffy_little_something]

Shouldn't the audio processor do some of the brute number crunching? Like the graphics processor does in the case of modern games, which are also very demanding? Or is that already too far on the output side of things?

[Jax Pok]

Valhalla DSP commented in KVR somewhere about an i7 would not be enough to do what the Bricasti does.

[chk071]

Valhalla DSP commented in KVR somewhere about an i7 would not be enough to do what the Bricasti does.

Hard to compare isn't it? DSP's don't have to run a whole desktop OS, and dozens of processes in the background. I fail to believe that it's more powerful than an i7 also though. Just some business thinking required. Why would they put something which is "30 times as powerful as a i7" in a box worth 3.000 € plus? Would be the bargain of the century, especially as audio hardware is usually pretty expensive. And DSP's are usually not nearly as powerful as what you have in your desktop PC. I laughed hard when i read what hardware is running in a Virus Ti. Comparable with a year 2000 desktop PC.

[Jax Pok]

Shouldn't the audio processor do some of the brute number crunching? Like the graphics processor does in the case of modern games, which are also very demanding? Or is that already too far on the output side of things?

That used to be the case such as the graphics chip in the Nintendo SNES Starfox game cartridge and PC graphics cards 10 years ago. Now though the GPU is faster than the CPU at graphics and at doing the job of the CPU such as rendering CGI and other big jobs.

[nevernamed]

Take something like the M7 Bricasti for example. It has over 30x (estimated) the processing power of the "state of art i7" and it's all dedicated to one feature: reverb.

30x the processing power of a "state of art i7"? I very doubt that...

You can doubt it all you want. It's a fact. FPGA arrays and ASICs blow away anything general purpose. There are some limitations however. It's a specialised computing platform dedicated to only a single purpose: processing reverb. Try cracking DES or WPA on a cluster of intel i7s and see how long that takes compared to something like a FPGA array.

The parallel you might be able to relate to is perhaps the UAD platform (their PCiE cards as well as their interfaces).
The Analog Devices stuff is in the Bricasti and the UAD stuff. Different FPGAs but similar idea. General purpose computing platforms are simply outclassed by specialised, dedicated computing platforms for this kind of stuff.

If you've ever heard the Bricasti reverb in person it might be a little easier to understand this. There is no plugin out there that compares. It isn't to do with software optimisation as much as it's to do with dedicated and specialised hardware. And the FPGAs are modestly clocked. Something like 400MHz.

[Jax Pok]

Valhalla DSP commented in KVR somewhere about an i7 would not be enough to do what the Bricasti does.

Hard to compare isn't it? DSP's don't have to run a whole desktop OS, and dozens of processes in the background. I fail to believe that it's more powerful than an i7 also though. Just some business thinking required. Why would they put something which is "30 times as powerful as a i7" in a box worth 3.000 € plus? Would be the bargain of the century, especially as audio hardware is usually pretty expensive. And DSP's are usually not nearly as powerful as what you have in your desktop PC. I laughed hard when i read what hardware is running in a Virus Ti. Comparable with a year 2000 desktop PC.

i7 is more powerful does more FLOPs but not enough to do an algorithm to match Briscati. Think of it as a super bike verses an old tractor at pulling a giant plough.

[chk071]

But rather due to the reason nevernamed just mentioned in his post. That whole "30 times faster than a i7" is simply marketing bogus.