Unpowered PCI-E Graphics cards . . . and . .

BonzaiDuck

Lifer
Jun 30, 2004
16,632
2,027
126
I jumped over to this forum to see if anyone would have some insight to a peculiar problem I'm facing.

I'm trying to evaluate and pre-test a P45-chipset motherboard (Gigabyte GA-EP45-UD3R) for use as a file server with an E6600 processor.

Two years ago, I picked up an eVGA (nVidia) 7600 GT PCI-E card for $70 for use in testing. I decided to get some use out of it with this Gigabyte mobo.

I've had terrible trouble, as evidenced on the "Motherboards" forum. The one consistent malfunction of the system would be described this way: Sometimes, the system posts, and appears stable (at default settings). I ran MEMTEST86+ against a 1GB kit of OCZ's through 10 iterations -- no problem -- after going through lockups between the 1st and 3rd iteration after manually setting the mobo to run the processor at the Intel spec. But throughout these troubles, shutting the system down and rebooting shows the proper keyboard initialization signals, floppy-read, etc. But the system doesn't bring the monitor out of sleep state, so I can't really see if the system is posting.

The 7600 GT is an "unpowered" graphics card -- it doesn't require the six-pin power-plug of later, high-end cards.

Is there some discontinuity in compatibility with these newer motherboards and older, unpowered PCI-E graphics cards?



 

chizow

Diamond Member
Jun 26, 2001
9,537
2
0
Are you sure the 7600GT still works? How long has it been since you've verified it does? Do you have a speaker that you can attach to interpret POST beeps? What VGA did you use to test the RAM?

There shouldn't be any problems with a PCIE 1.0 or 1.1 to a PCIE 2.0 board, although there were quite a few initial problems going the other way, with PCIE 2.0 graphics cards on PCIE 1.0 or 1.1 motherboards. Typically a BIOS flash fixed this, although it obviously created problems for people who didn't have a PCIE 1.0 card handy. Personally, if I were experiencing lots of flakiness with so many different areas of a build at stock settings like that, I'd look at the board itself or PSU as the problem.
 

BonzaiDuck

Lifer
Jun 30, 2004
16,632
2,027
126
Originally posted by: chizow
Are you sure the 7600GT still works? How long has it been since you've verified it does? Do you have a speaker that you can attach to interpret POST beeps? What VGA did you use to test the RAM?

There shouldn't be any problems with a PCIE 1.0 or 1.1 to a PCIE 2.0 board, although there were quite a few initial problems going the other way, with PCIE 2.0 graphics cards on PCIE 1.0 or 1.1 motherboards. Typically a BIOS flash fixed this, although it obviously created problems for people who didn't have a PCIE 1.0 card handy. Personally, if I were experiencing lots of flakiness with so many different areas of a build at stock settings like that, I'd look at the board itself or PSU as the problem.

Ah! Chizow! We cross paths again!

It can't be the PSU. I won't go into it in depth, although you can find it at the "Power Supply" forum concerning my brief flirtation with a BFG GS-550. It was quickly replaced with a (new) Seasonic SS-550HT. At least when I was able to boot this Gigabyte motherboard and enter BIOS Setup, the BIOS monitor was reporting the proper voltages, and everything was in order per the PSU.

The graphics card was only used briefly to test a motherboard in 2007, and another motherboard in late 2008. It was handled carefully, and each time after use, I replaced it to its anti-static wrap and stowed it away in a dry place.

I had to come over here to the Video forum to get a take on the PCI-E standard and unpowered graphics cards. The card had worked fine (as I say, with testing) for a 680i board and 780i board, respectively.

I think I'm going to replace the OCZ RAM modules I've tried running with this board and stick in a set of G.SKILL or Corsair modules. If the same behavior persists, I've already sent Gigabyte a message through their web-site. But I understand that their RMA process and promise leaves much to be desired.

EDIT: OF course, there's always the possibility that the processor's "gone south," but I don't see how. I used it about 3 months before replacing it with a Q6600 quad -- storing it also in anti-static wrap -- until I carefully put it in this motherboard socket.

This all seems to be such a P-I-T-A, trying to find out for sure which component is causing trouble with this setup, or whether it's the motherboard.

Getting back to the PSU, the GS-550 likely had a broken solder-joint for its AC ground, showing a site-wiring fault alarm at my battery-backup. How that might damage a motherboard on the DC side of the PSU, I'm not sure. But I think I only switched on the BFG PSU for maybe ten-seconds each of two times before I shut the system down because the UPS alarm was sounding.