G5 AMD + UAD-1

[Fayth]

I Did a search on the forum, but I couldn't find if there is a 'fix' for the performance problems with UAD-1 and AMD G5's...

Or is it just the case that AMD and UAD-1 results in a lower performance and there is no solution?

Anybody an idea?

 

[Archmart]

Hey Hey!

There's really no \"fix\" other than turning on AMD mode, which allows you to work with the only real limit being to a large number of instances. A few of the heavy plugs work the same AMD or no. If you need lots of little plugs, you hit a wall.

Check recent WebZines. There were some charts a bit back that detailed the different instances of plugs.

So yeah, it's just a case of lower performance in certain situations... Not that big a deal, though.

Archmart

 

[beal9000]

AFAIK, there's no total fix, but I think that some people have reported improved performance with the AMD controller chips by throwing the UADs into a Magma.

 

[Archmart]

Hey Hey!

From what I can tell, the Magma solution doesn't actually help as it's also plugged into that same AMD-8131 situation. It make the problem a little different, but it doesn't really get any better.

It does, however, enable the use of 4 cards in a G5 with only three open slots.

Archmart

 

[clie]

Hey Hey!

Check recent WebZines. There were some charts a bit back that detailed the different instances of plugs.
Archmart

I'm interested in the FlexiPak but I found the webzine to only discuss the differences in the plugin count on the UAD card itself. UA's website states that the AMD problem also causes your CPU to do more work without the host application's knowledge. So you think your cruising along but your audio starts to crap out.

No one seems to be committed to quantifiing the extra weight on the host cpu. There is some talk of watching out for a "large number of instances" - but that's just talk.

If I'm runnning 12 mono tracks and I want EQ and compression on each - is my host going to stutter with the UAD 1 before it would if I just used my Waves Renaissance channels? Is it equal? At what point do I lose performance - 8 -12 - 16 channels? (I'm on a dual 1.8 using Cubase 3.1)

I'm really interested - I'm close to buying but this question is nagging me (and now I'm nagging you)

Thanks in advance.

Carl

 

[jimmymio]

Good question Carl and i have one more. I've heard that you can make one UAD-1 card work reasonably well on an AMD G5 but additional cards will suffer more. Can anyone confirm this?
Thanks, JP

 

[Holly]

I'm on a PC myself, but in my friend's studio we are working on a project with a G5 and two uads. He uses a lot of cambridge eqs and other little plugins. It isn't possible to get a UAD load of more than 75% without getting the Cubase CPU meter peaking. Disabling one or two plugins gets the Cubase meter back to maybe 40%.

I don't know about any solution, but I think I read somewhere on this forum, the performance should be better when using UAD software 4.0. Can anyone confirm this?
I also don't know how it works with only one card, but this isn't a solution, as we need the extra power for our project with up to 50 tracks.

Holly

 

[Trace]

Performance doesn't decrease with more cards. What i've seen is that my 4 UAD-1's can pretty much max out on DSP with lower instances, but like Archmart said when you try to load a ton of plugs it will effect your host CPU.

here are the results of a test I did a while back on my system:

OSX 10.3.9 - UAD vs 4.0
I set the MIN-GNT PCI Latency to 48
AMD 8131 Comp DMA mode is off
Running 24 Channels in Cubase SX 3.0.1, ASIO Buffer at 512, 44.1k
I've posted the most plug ins I could get to run stable with no Host CPU overload:


24 - Dreamverb - 20% G5 CPU & 95% UAD DSP (24 is max - so that's 100%)

40 - CS-1- 30% G5 CPU & 87% UAD DSP/92% mem (40 is max - so that's 100%)

16 - Plate 140 - 15% G5 CPU & 85% UAD DSP/43% mem (16 is max - so that's 100%)

8 - Nigel - 8% G5 CPU & 70% UAD DSP//48% mem (8 is max - so that's 100%)

28 - 1176LN - 25% G5 CPU and 95% UAD DSP (32 is max - 4 Less than max)

24 - Pultec Pro - 20% G5 CPU & 92% UAD DSP (28 is max - 4 less than max)

40 - Pultec - 28% G5 CPU & 93% UAD DSP (48 is max - 8 less than max)

40 - Fairchild 670 mono - 28% G5 CPU & 92% UAD DSP (48 is max - 8 less than max)

64 - LA2A - 45% G5 CPU & 95% UAD DSP (88 is max - so that's 24 less than max)

72 - 1176SE - 50% G5 CPU & 91% UAD DSP (128 is max - 56 less than max)

80 - Cambridge EQ's w/5 Bands - 58% G5 CPU & 89% UAD DSP (184 is max - 104 less than max)


That is performance I can live with, cuz I tend to use the bigger DSP plugs the most.

TRACE

 

[clie]

here are the results of a test I did a while back on my system:

OSX 10.3.9 - UAD vs 4.0
I set the MIN-GNT PCI Latency to 48
AMD 8131 Comp DMA mode is off
Running 24 Channels in Cubase SX 3.0.1, ASIO Buffer at 512, 44.1k
I've posted the most plug ins I could get to run stable with no Host CPU overload:
TRACE

Thanks TRACE.

Just to verify - you do or do not have a G5 with the AMD 8131 chipset.

Also what is MIN-GNT Latency?

Thanks in advance.

Carl

 

[Trace]

here are the results of a test I did a while back on my system:

OSX 10.3.9 - UAD vs 4.0
I set the MIN-GNT PCI Latency to 48
AMD 8131 Comp DMA mode is off
Running 24 Channels in Cubase SX 3.0.1, ASIO Buffer at 512, 44.1k
I've posted the most plug ins I could get to run stable with no Host CPU overload:
TRACE

Thanks TRACE.

Just to verify - you do or do not have a G5 with the AMD 8131 chipset.

Also what is MIN-GNT Latency?

Thanks in advance.

Carl

Yeah I have a G5 with AMD 8131 chipset. I've been begging UA to make a single Channelstrip plug using all their best analog plugs. This would help to cut down on the number of audio streams across the PCI-X Tunnel.

The MIN-GNT is a PCI Latency setting and it basically adjusts the amount of time that the UAD has a hold on the PCI Bus, at least that's how I understand it. I feel that my system works best when its a bit lower.

TRACE

 

[clie]

Excellent. Thanks very much. I feel better that we're using the same system and host software.

Carl

 

[jimmymio]

here are the results of a test I did a while back on my system:

OSX 10.3.9 - UAD vs 4.0
I set the MIN-GNT PCI Latency to 48
AMD 8131 Comp DMA mode is off
Running 24 Channels in Cubase SX 3.0.1, ASIO Buffer at 512, 44.1k
I've posted the most plug ins I could get to run stable with no Host CPU overload:
TRACE

Thanks TRACE.

Just to verify - you do or do not have a G5 with the AMD 8131 chipset.

Also what is MIN-GNT Latency?

Thanks in advance.

Carl

Yeah I have a G5 with AMD 8131 chipset. I've been begging UA to make a single Channelstrip plug using all their best analog plugs. This would help to cut down on the number of audio streams across the PCI-X Tunnel.

The MIN-GNT is a PCI Latency setting and it basically adjusts the amount of time that the UAD has a hold on the PCI Bus, at least that's how I understand it. I feel that my system works best when its a bit lower.

TRACE

And are you sure the Magma chassis has no effect on performance. i had heard that it did and it is often mentioned as a solution to the AMD/UAD problem.
JP

 

[clie]

Ooops. Good point. Curious myself. To make my decision harder, I downloaded tha Waves SSL plugs and really like them.

Decisions.

C

This is a cool forum BTW.

 

[Trace]

The Magma only makes a difference in that you don't have to use the AMD 8131 DMA Mode, but limitations STILL exist.

Read this Support page for an in depth description of what's going on:

http://uaudio.com/webzine/2005/may/index5.html


TRACE