Welcome, Guest. Please login or register.

Author Topic: Zorro DMA vs Accel. DMA: Cyberscsi vs A4091  (Read 4129 times)

Description:

0 Members and 1 Guest are viewing this topic.

Offline Damion

Re: Zorro DMA vs Accel. DMA: Cyberscsi vs A4091
« on: July 19, 2009, 01:14:17 AM »
I suppose it depends how much bandwidth can be utilized before something else chokes. There's barely a 1 FPS difference between my Z2 and Z3 machines in Quake benchmarks, so 20 MB/s for Z3 should be enough. I imagine (non-expert guess here) the PIV or CPU would become a bottleneck before the Z3 bus.
 

Offline Damion

Re: Zorro DMA vs Accel. DMA: Cyberscsi vs A4091
« Reply #1 on: July 19, 2009, 04:20:29 AM »
Quote

@Damion
Sounds interesting (Z2 vs Z3), but you didn't say what video cards / processor cards you have on them...


The A2K has a TekMagic '060/PIV, the A3K also an '060/PIV. I swapped the same CF hard drive between the machines for testing. Regardless of the screen resolution or other settings, Quake was .5 - 1 FPS faster on the Z3 machine. With a 10MHz advantage, the A2K was 1.5 FPS faster. (Didn't overclock the MKII yet.) A1200/AGA was 2 FPS faster than the A2K, but with a 20 MHz advantage (haven't yet tested them at the same clock).

In graphics benchmarks, the A2K/A3K were roughly equal up to 1024x768x16. Only switching to high-res true color modes was the Z3 machine a bit faster, for example, WB backdrops would take a bit longer to draw on the A2K. Writes to the PIV hit the Z2 theoretical maximum (according to Thomas Wenzel's CgxBenchmark), which probably helps.

No difference with RTG demos, either. Running Lightshaft with the -fps tag, they are about equal.