Jan. 11th, 2005

plonq: (Me Smash)
It should have been a straight-forward thing. [livejournal.com profile] pierrekrahn wanted to replace his power supply, and update his video card. After some consideration he decided to go with the BFG 6600GT with 128MB onboard. His motherboard was an ASUS A7V8X with an AMD 2400+ onboard. The process should have gone like this.

1) Install new power supply
2) Swap out video cards
3) Reboot and install drivers
4) Enjoy

Everything went fine until step three. For some reason his system would not recognize the new video card. It refused to post, spewing out audio messages about various CPU errors. After extensive frustration he made his first mistake: he brought the whole works over to our place. I tried all of the usual tricks, tweaking the BIOS, removing memory, etc. No love from the video card. It's too bad we didn't have a couple of other computers around that we could have tried the card in.

Oh wait, we did. We're just fucking stupid.

The next step was obvious. We went to the ASUS web site and checked to see if there were any new versions out there for his BIOS. Ah - he was about six versions back, and the latest version specifically made reference to fixing something with the 8X AGP.

1... 2... 3... FLASH!

Replace the card, reboot, no love. Grrr. Wait a minute - it's still showing a revision date of 2002 for the BIOS. Grrr. RELEASE THE HOUNDS! FLASH IT AGAIN!

1... 2... 3... FLAS....ERK! D'oh!

The system froze up solid in mid-flash. Cut to [livejournal.com profile] pierrekrahn and [livejournal.com profile] plonq blinking dumbly at [livejournal.com profile] pierrekrahn's new paperweight.

Fast forward to this evening. He brought everything over again so that we could try a few tricks that we'd found online. No luck. Finally, in desperation, it occurred to us that we should check out the video card in another machine to see if that was the problem all along. We plugged it into my computer and... nothing. It wouldn't post when it powered up. We put my old card back in and everything was fine again (obviously, since I'm writing this). Well what do you know - the card was bad all along!

<sigh>

We bundled up his dead motherboard and video card and made a beeline back to the store where he bought it. They're not close by, and we didn't get there until about 20 minutes before they closed. There was a mob of customers in front of the service desk, and by the time we finally got served, it was about 7 minutes to closing. I related the whole sordid tale to them, and after some quick checking they confirmed that 1) his machine is toast, and 2) the video card they sold him is a dud. Great. We felt dubiously vindicated by that knowledge because it meant that we weren't too stupid to install a working video card. Oh wait, we'd managed to kill his computer in the process, hadn't we? Maybe we WERE too stupid.

After a bit of consulting back and forth, the tech offered him a new ASUS A7V880 at a decent price. That's not a bad little board. It's got dual channel memory (and his would work, since it's matched sticks), SATA RAID, USB 2.0, 8X AGP and Socket A so that he could just move his chip straight across. We decided to follow this path just about the time that the store was closing. Oritinally they were going to have us come back the next day, but when they found out how far away we lived, the head tech said, "This should only take about 10 minutes. We'll stick around."

They unboxed the new MB, swapped out the CPU, swapped out the memory and powered it up. We watched as it went through its usual diagnostics and then blithely announced that it was powered by an AMD 1800+

We protested that we were pretty darned sure that the chip was actually a 2400+, but the tech insisted that "the BIOS doesn't lie." They rebooted a couple more times, and checked the bus speed and multiplier settings, but it all set itself automatically as one would expect for an 1800+ chip. I don't think he entirely believed us about it being a 2400+ chip, but to humour us he cranked up the bus speed to 133 and on reboot it showed as an 1800+ running at 1.9 GHz. Odd. Finally he removed the heatsink again and said, "Hm. I guess it is a 2400+ chip, or that's what it says on the top anyway."

I glanced at the clock. This "10 minute job" has now gone on the better part of 20 minutes, and some of the junior techs were glaring at us for keeping them late. They dropped the chip into a spare Gigabyte board they kept back there for testing components and it promptly reported that it was a 2400+ CPU. The solution was obvious: we needed to reflash the BIOS on the ASUS board. Ten minutes later the chip was back in the ASUS board and the BIOS had been flashed with the latest image. We all watched with bated breath while they powered it down and up again, and then let out our breath when it gleefully reported that we were the proud owners of an 1800+

I think I heard the sound of junior techs banging their heads on various desks.

It was now 30 minutes into our 10 minute install - that is, 30 minutes after these guys should have gone home. By this time [livejournal.com profile] pierrekrahn and I were both applogizing profusely, but they were very good about it. This had become a challenge, and they refused to admit defeat. They fetched a replacement A7V880, swapped out all of the parts and found that it did exactly the same thing. The sad conclusion is that the A7V880 is not compatible with the 2400+ CPU. It only took three technicians and 40 minutes to establish this fact.

They put the CPU into an anti-static case for us, and we are going to return tomorrow to try out a couple of other motherboards. They have an Asrock K7V88 that they think will work. It has similar specs to the ASUS (based on the same VIA chipset), and it's about $20 cheaper.

We plan to get there a bit earlier tomorrow.

April 2024

S M T W T F S
 123456
78910111213
14 151617181920
21222324252627
282930    

Most Popular Tags

Page Summary

Style Credit

Expand Cut Tags

No cut tags
Page generated Jul. 14th, 2025 01:13 pm
Powered by Dreamwidth Studios