I had the same issue with my FX5700. OK without drivers, broken with drivers installed. It was a bad cap for me, so it might worth checking them. If you are lucky, that's it. If not, Radeon 9700s really liked to cook themselves.

Kampfkoloss wrote on 2025-11-06, 12:16:

I've finally found a reference Radeon 9700 for really cheap (25 €) which was working perfectly with and without drivers. Now in […]
Show full quote

I've finally found a reference Radeon 9700 for really cheap (25 €) which was working perfectly with and without drivers. Now in a fresh test install of XP after reboot when drivers are installed output is fullscreen 3 color static garbage. Output at boot time/under DOS is flawless if a little placed to a side.

What I have done so far:
- I've enabled AGP fast writes at driver install this time. So far it was enabled in BIOS but not in driver settings. I know it is a problematic setting, but I've not come across anything else than sporadic instabilities with it so far; screen corruption would be new to me
- I've cleaned the old thermal glue off VPU and cooler - it took half an hour in a zipper bag in the freezer and hours of scraping with WD40 and IPA to get heatsink and chip somewhat clean (stains remain one the chip and scratches on the heatsink -.-') --> the shim does not prevent contact between VPU and heatsink here (I've tested it)
- I've reflashed the card's vbios with a verified (!) Radeon 9700 Infineon 3,3 ns DDR-SGRAM PAL BIOS from techpowerup which was stable for hours on the last test install on the same system (I wanted to get the card to regular 9700 speeds - before they were @ 263/263 like a Dell 9700 TX, but AMD told me it's a HP OEM card???)

The Test-System:
AMD Athlon XP 1700+
Asus A7S333 (SiS 745) set to jumperfree, BIOS at defaults
2x MDT DDR-400 CL2,5
ATI Radeon 9700
PSU Enermax EV365EAX-VE (G) for cromulent 5V-Rail supply
Windows XP SP3 (German version), SiS Chipset Drivers preinstalled by XP, ATI OMEGA Drivers 4.48 for my Radeon (I swear by them; they work flawlessly in my "Main" Athlon XP rig)

I've found some topics regarding screen corruption on 9700 Pros here, but they were a few years old already. That's why I'm posting a new one. If we find a solution other people having similar issues might find answers or hints here in the future. Those working R300-cards have to be saved.

Check out my repair threads:
VGA Repair report collection
Hercules Radeon 9700 Pro repair report

9700s have 4 memory channels.. you might have bad memory and not notice it without drivers. Give R3MEMID a run under DOS and see if it gives you any errors. I've had a card (9800) that looked okay, once the driver was installed, the corruption started with the bootscreen. Safe mode worked fine.

Your PSU looks ample, if the voltages are in spec, it should do just fine.

Hi folks - this is my intro post.

Diagnosis more likely with pictures of the 9700 PCB (both sides) and pictures of the XP artefacts.

My expectation is that some bad upper VRAM is being activated by the drivers but let's work on the assumption that it's not that bad.

The first thing to do is inspect verify the hardware is what it says it is because some old cards are faking it - for example a 9500 with ROM modified to look like a 9700. Sometimes people did that for a free upgrade, but it might backfire you then load drivers tuned for a 9700 retail card. Once the PCB is verified I'd start with Microsoft basic drivers, then try ATI stock drivers, before moving to anything tuned.

Next check you motherboard chipset configuration and drivers because the chipset can be stressed by aggressive Radeon timings causing voltage ripple and signal jitter. Bad caps emerge on motherboards even more frequently than they do on graphics cards, so I'd tone down anything in the main BIOS to reduce voltage stress (minimal AGP bandwidth, minimal shared memory).

And, bad caps are not limited to GPU and motherboard - they also emerge in the PSU. So, I'd reduce the number of expansion cards and HDDs and make sure the 9700s Molex is on a dedicated rail. I'd also check the Molex lines with a multimeter but not convinced that would prove anything as the multimeter isn't going to stress test anything.

Bad caps are the easiest to repair. Bad VRAM is emotionally painful. Bad GPU is rare - bad fan is more likely.

Hey folks, thanks for all of your suggestions! 😁

I've found it to be a purely software related issue -.-' after a reflash with a modified Dell-BIOS and with both MS standard radeon 9700 driver and official Catalyst 7.11 WHQL-drivers with AGP fastwrites disabled corruption has gone away completely (!) I verified it with multiple Aquamark 3 and 3Dmark 2001 and 03@ 1600x1200 6xFSAA fullscreen runs. Temps are warm to the touch, not too hot or anything in any part of the card.

I will however try the ATI ramtester tool, cause I find it pretty nifty 😁

So this was quite the learning experience for me. I hope this will help others having similar issues. It seems not to always be a hardware problem!