AGP Parhelia 512 Crashing in Photoshop...

[lgeis]

Greetings from Beautiful Boise, Idaho!

I have an AMD 64 3200+ (overclocked to 2.5GHz) running dual Raptors for the OS, 2 GIG of dual channel 2-3-3-6 OCZ Platinum RAM. On this machine I have Photoshop CS within XP Pro SP2. All software, drivers, and so on are the latest (I'm anal).

The Parhelia is an AGP version and drives three 19" LCD monitors in stretch mode.

Occasionally, in Photoshop, when I use a zoom shortcut (Ctl- or Ctl+) the program crashes and asks to send a report on the matter to Microsoft. I've not found any resolution for the matter...even pulling the overclocks off.

Other cards include a TekRAM SATA PCI card, an Adaptec DuoConnect USB/IEEE PCI host, and a SoundBlaster Audigy 2 ZS Platinum with the breakout box. However, I can't imagine why these would contribute to the problem.

I have another computer set up running triple head in independent 1600x1200 with an Matrox PCI P650 driving the third monitor, and it has no problems at all...primary card is an ATI 9800 Pro (overclocked, of course...).

Has anyone else encountered this problem? It only seems to happen when zooming using the shortcuts. My AGP aperture is set to 64MEG and the AGP voltages are stock. The Parhelia has the latest BIOS flash and is running stock frequencies.

Thanks!

[Nowhere]

Parhelia isn't crashing. Photoshop is crashing. Look up for problems with/"inside" Photoshop (reinstall it for a start maybe?)

[lgeis]

...Was the PCI bus...

I had adjusted the PCI latency to accomodate some problems I was having trying to concurrently run a P650 with the Parhelia (PCI supplement to the AGP Parhelia)...wanted big res across three monitors. However, ended up bagging the P650 and running the Parhelia only, so reset a variety of BIOS parameters, and as soon as I returned the PCI latency to normal the problem went away.

Now, don't axe me why the PCI latency would effect the AGP-only setup, but my OS is running on a pair of SATA Raptors connected to a PCI host adapter (my CPU is so overclocked that the HDD's won't run on the MoBo's SATA bus). Perhaps the problem is rooted somewhere in the setup...

[MultimediaMan]

Well, there you have it... you're overclocked.

This is not a Parhelia issue, nor is it a setup issue.

It is a stability issue brought on by running the machine out of it's design specifications.

If you set all of your clock settings to normal, does everything run correctly, even with the PCI latency settings changes?

[lgeis]

Hmmm...I don't know. I'd never be caught running a machine at it's rated frequency or voltage!

Well...actually, that's not true. I'll 'fess up. I have a number of Silicon Graphics Workstations (all but two are "retired," meaning that they are crated up in the garage), and there is no access to voltage or frequency controls on them...in fact, I don't even think anyone has ever tried to work with the PIII SECC2 CPU's used in them. However, I did change out the voltage regulators and installed unsupported CPU's (SGI says max 600MHz, I have installed 1.0GHz CPU's in an SMP arrangement). Perhaps that counts for something! Ha!

And, of course, I used Matrox cards in the SGI's (G450's).

Anyhoo...I don't think the AGP vs. overclocked or not overclocked CPU is an issue...never heard of anyone having such a problem. Fer example, the same Socket 939 MoBo will run a variety of CPUs from 2800+ to 4800+ in spite of their variety of speeds, multipliers, FSB/Hyperthreading busses, factory clock speeds, memory controller versions, and even variance in the number of cores (single vs. dual).

[MultimediaMan]

The AGP and the PCI Bus often times share a clock generator and do not support seperate frequencies.

Variable frequencies are not the issue...the Socket 939/940s have integrated Memory controllers that do not depend on the North or Southbridges. You're probably running an nForce3 motherboard, which have some quirks with their PCI Bus.