What's your motherboard ? As far I can see there is agp and pci-e all together so I presume that agp port is not really agp. It might be AGI or AGX variations . Those ones have compatibility issues.

It's a ASUS P5VDC-MX. Real AGP (8x) and crippled PCIe.
A Ti4200 (AGP 4x) and a Radeon 9600Pro (AGP 8x) do work without issues.
For reference: the X700 does also not post in ASRock P4i65G and ECS K7SEM.

I have tested some of my other Radeon cards, (9550, 9600Pro) and on all of them Vddc is equal Vddq.
Sometimes both rails are even bridged which seems a valid option according to some schematics.

I've also removed the two voltage setting resistors from the X700 and as suspected one is 910 ohms and the other is 1500 ohms, giving 2.12V nominal.
Still can't figure out why in the world on the X700 Vddq is set so low. That's none of any "standard" voltages (meaning any voltage that's generally used for stuff like 1.80/2.50/2.70V) I've seen anywhere and also doesn't conform to the RAMs datasheet.
Anyway, "fixing" this temporarily to ~2.6V doesn't change anything. No signs of life. So I guess another card died from simply switching the machine off.
Damn. Was still thinking about the best way to add additional cooling to the AGP-PCIe bridge before using the card for longer and in a closed case... Case closed.

In regards to the "erroneous" Vddq memory voltage: if this is how the video card was designed / set up originally, then there was probably a reason for it.

Most likely, the card just died... especially if it has one of those Realto AGP-PCIe bridge chips - these are extremely prone to dying, because they run too hot without a heatsink. I know they come without one... but they really DO need one. Whoever the "genius" was at AMD/ATI (or group of such people) that decided the Realto bridge chip does not need a heatsink... that's really what's causing all of these to die quicker than they normally should.

I also can't help but notice the card is a Sapphire. Never liked them in terms of build quality, especially the cooling. They almost always use cheap inadequate coolers... save maybe for the one or two ultra premium top-end models they may have. Unfortunately, the one shown above does not appear to have anything premium, especially in regards to the cooler. According to Techpowerup, the video card is rated for 44 Watts TDP. My guess would be the GPU itself makes up for around 35 Watts of that rating, if not slightly more. In any case, I doubt that cooler can keep the GPU core below 60C or even 70C under full load... which would certainly accelerate the demise of the card. The cooler above should be used on a GPU core with a TDP of 20-25 Watts absolute max.

*Edit*
I see the card two green Teapo capacitors. Probably SC series.
Now, I doubt this would be the problem, as Teapo SC caps almost always bulge when they fail. Still, if you have an ESR meter, perhaps check them anyways. The green Sanyo (WG or WX series??) are probably OK. And the Nichicon HC caps should definitely be fine - at least I've never seen any of these fail before.

My X700 Pro also died unexpectedly: Short somewhere on ATI X700PRO
I bought a different, unfortunately no longer Pro version, just a regular one - I can measure whatever you need, but it's a PCI-E version with a slightly different component layout on the PCB.

Ydee wrote on 2025-02-01, 11:33:

My X700 Pro also died unexpectedly: Short somewhere on ATI X700PRO
I bought a different, unfortunately no longer Pro version, just a regular one - I can measure whatever you need, but it's a PCI-E version with a slightly different component layout on the PCB.

Hmmm.... curious if you still have your shorted x700. It probably has a shorted upper MOSFET, likely either on GPU V_core or RAM V_DDQ rail. If it's the former, typically these old GPUs survive it. If it's the latter, then that's usually a faulty GPU chip that caused this (somewhat common on older ATI cards when they have been overheated or ran too hot all their life.)
I recently got an x800 for super cheap ($1) without a cooler... but all hopes of it working are out the window now, as it has a cracked RAM BGA chip. I'd take a shorted MOSFET card any day over cracked RAM / card that needs RAM replacement.

Unfortunately, the cause of the short circuit could not be found. Capacitors, FETs, coils - everything seemed fine when measured, so the card went on as a source of parts.

momaka wrote on 2025-01-31, 17:39:

In regards to the "erroneous" Vddq memory voltage: if this is how the video card was designed / set up originally, then there was probably a reason for it.

Yeah, but still very questionable to me.

momaka wrote on 2025-01-31, 17:39:

Most likely, the card just died... especially if it has one of those Realto AGP-PCIe bridge chips - these are extremely prone to dying, because they run too hot without a heatsink. I know they come without one... but they really DO need one. Whoever the "genius" was at AMD/ATI (or group of such people) that decided the Realto bridge chip does not need a heatsink... that's really what's causing all of these to die quicker than they normally should.

I also can't help but notice the card is a Sapphire. Never liked them in terms of build quality, especially the cooling. They almost always use cheap inadequate coolers... save maybe for the one or two ultra premium top-end models they may have. Unfortunately, the one shown above does not appear to have anything premium, especially in regards to the cooler. According to Techpowerup, the video card is rated for 44 Watts TDP. My guess would be the GPU itself makes up for around 35 Watts of that rating, if not slightly more. In any case, I doubt that cooler can keep the GPU core below 60C or even 70C under full load... which would certainly accelerate the demise of the card. The cooler above should be used on a GPU core with a TDP of 20-25 Watts absolute max.

In general I don't have a personal acceptance problem with Sapphire products. In fact I've quite a bit of hardware from them and all is working to this day. But that's just my very own experience.
Keep in mind that this isn't a Pro variant but only the bottom of the barrel X700 with low end TSSOP-RAMs which don't even got hand warm.
Wikipedia lists it with 31W which is without the Rialto-chip (i.e. the PCIe variant) I'd guess. So yeah, with Rialto it could go towards 40W or more.
The cooler itself as well as the card as a whole didn't get very warm at all. I really liked it because unlike my 9600pro which always runs the fan full-speed this X700 had a good temperature controlled fan that was really unobtrusive. Even under full load. Hottest part was the Rialto which I was on the way of getting at least some cooling added to. Unfortunately the first try didn't work because the thermal self-adhesive pad didn't hold on the chip and I had no thermal epoxy at hand. And shortly after the card was dead...

momaka wrote on 2025-01-31, 17:39:

*Edit*
I see the card two green Teapo capacitors. Probably SC series.
Now, I doubt this would be the problem, as Teapo SC caps almost always bulge when they fail. Still, if you have an ESR meter, perhaps check them anyways. The green Sanyo (WG or WX series??) are probably OK. And the Nichicon HC caps should definitely be fine - at least I've never seen any of these fail before.

Good eyes. They are SC series. I already put some new Panasonics in parallel but that makes no difference.
The voltages all look ok on a scope. So I doubt the Teapos as well as the other electrolytics are the problem.

I took the time and booted the card one last time with another VGA card in the system to see if there's anything diagnosable.
This is what I got:

Don't know if that tells us anything other than that the card has two video outputs, though.
The first part of the IDs is correct (1002=ATi). But the second part is ... idk.
When it still worked the secord part was 5E4D (didn't make a note about the second video output).
If I had to guess I'd say the Rialto is ok and the X700 chip died. I briefly thought about the BIOS but I think these IDs are silicon hardcoded. If they are wrong the chip itself is gone.

Well, in the end a dead card stays a dead card. Still a bit disappointed because these cards don't show up that often. Exactly when I bought it there was another one on the bay for almost the same money. Decided not to buy it in favor of this one. Maybe it would've lasted a bit longer. Who knows...
And now nothing comparable in sight.

momaka wrote on 2025-02-10, 21:43:

I recently got an x800 for super cheap ($1) without a cooler... but all hopes of it working are out the window now, as it has a cracked RAM BGA chip. I'd take a shorted MOSFET card any day over cracked RAM / card that needs RAM replacement.

I have a dead X800 GTO here. With cooler and eight RAM chips. Dead X800 chip, 🤣.
If you're interested...