The Beatsmith said:

so how much difference does the magma chassis make for g5 macs with the AMD chips? can you get 100% out of the UAD's?

The Magma helps but I can tell you that you won't get 100% from every plug. I Just redid some tests with my system:

Dual G5 1.8Ghz PCI-X , 2.5 Gb of Ram

Yamaha O1X

7 Slot 33Mhz Magma Chassis 4 UAD-1's

I set the MIN-GNT PCI Latency to 64

Running 24 Channels in Cubase SX 3, ASIO Buffer at 512, 44.1k

I've posted the most instances of each plug I could get to run stable with no G5 CPU spikes:

Dreamverb - I got 24, on a G4 I should get 24- so that's 100%

CS-1- I got 40, on a G4 I should get 40 - so that's 100%

Plate 140 - I got 16, on a G4 I should get 16 - so that's 100%

Nigel - I got 8, on a G4 I should get 8 - so that's 100%

Pultec Pro - I got 24, on a G4 I should get 28 - so that's 4 less

1176LN - 28 Instances, on a G4 I should get 32 - so that's 4 Less.

Fairchild 670 mono - 40 Instances, on a G4 I should 48 - so that's 8 less

Pultec - I got 40, on a G4 I should get 48 - so that's 8 less

LA2A - I got 56, on a G4 I should get 80 - so that's 24 less

Cambridge EQ w/5 Bands & 2 Filters on - I got 64, on a G4 I should get 112 - so that's 48 less

1176SE - I got 65, on a G4 I should get 128 - so that's 63 less

The max # of instances I was able to get was 74 EX-1's 80 with high host CPU usage.

The good news is that for the plugs that matter most you can get some good mileage out of the G5 PCI-X & Magma combo. I decided to take UA James advice and set the UAD MIN_GNT to 64. I also tried running the system at different buffer sizes and 512 turned out to be where my system worked best. I thought it would be 1024, but I got less host CPU problems using 512.

TRACE