X800XT AGP with an R423 core??

[jbodin]

Has anybody else encountered any AGP versions of the X800XT card with the R423 core? I've searched the web and everything I've found indicates that the R423 was the PCI-e core, but I have found some anecdotal evidence that these cores may have been used on AGP cards as well. Also, my X800XT has weird default clocks -- 445 core, 445 memory (890 DDR), but it's detected as an "X800XT" during installation, and GPUz detects it as an "X800XT" (I discovered that it was an R423 core using GPUz, and yes, it does have 16 pipes according to GPUz).

Seems to run well, overclocks to standard XT speeds (500/500), but it doesn't really score any better on benchmarks even when overclocked (3DMark03, 3DMark05 scores remain about the same, might go up slightly when overclocked). Still a fast card even at default clocks, and plays everything well that I throw at it, but it bugs me that it seems like something of an "odd duck."

Any ideas what I might have?

:confused:

[Pinchy]

Might be just a PCIe X800XT with the rialto chip attached.

Thats a punt though, I am not sure at all.

Also, if it runs @ 500/500, just flash it to those speeds :)

[InnocentCriminal]

I'd becareful before flashing y'card. I'd make sure the cooling is adequate and that it can maintain those speeds with stability. If they can, would there be any point in flashing?

I wouldn't be surprised if it is an XT but the core wasn't cutting the mustard for the PCIe counter-parts so it was pushed down to AGP.

[Ripper3]

The small difference in scroes might just mean it is being limited somewhere, with a bottleneck. If it's an R423, which is the PCIe chip, then it might be the Rialto chip that's making a bottleneck. Meanwhile, if it's jsut flashed incorrectly and is showing as an R423, when really it's a native AGP chip, it might have bottlenecks elsewhere, or the card manufacturer probably just decided that the performance increase was minimal when running at the reference XT speeds, and lowered them. Could be a number of reasons. If it works fine, I'd leave it be. If you want it overclocked, like InnocentCriminal said, check the cooling, check the temperatures, before flashing with the higher speeds.

Might also want to take off the cooler, clean up the thermal gloop from stock, and replace it with AS5 or MX-2, etc. While you've got the cooler off, just confirm it's an R423 or not.

[InnocentCriminal]

The small difference in scroes might just mean it is being limited somewhere, with a bottleneck. If it's an R423, which is the PCIe chip, then it might be the Rialto chip that's making a bottleneck. Meanwhile, if it's jsut flashed incorrectly and is showing as an R423, when really it's a native AGP chip, it might have bottlenecks elsewhere, or the card manufacturer probably just decided that the performance increase was minimal when running at the reference XT speeds, and lowered them. Could be a number of reasons. If it works fine, I'd leave it be. If you want it overclocked, like InnocentCriminal said, check the cooling, check the temperatures, before flashing with the higher speeds.

Might also want to take off the cooler, clean up the thermal gloop from stock, and replace it with AS5 or MX-2, etc. While you've got the cooler off, just confirm it's an R423 or not.


Exactly!

[jbodin]

Good call, guys -- there IS a Rialto chip on-board (complete with a heatsink, which is nice -- bridges don't seem to receive much attention from manufacturers in this regard). The card seems stable when overclocked, and the core temps seem to stay low-ish, but I need to check the memory speeds, so I probably will pull off the current cooler to do that. I've got a massive heatpipe cooler lying around that I haven't used -- I'm not interested in massive overclocks, but it would undoubtedly be quieter than the stock fan (kinda loud when it ramps up under temps), and the additional fans would probably do a better job of helping keep the RAM cool as well if I do overclock to "real" XT speeds.

As it is, though, it seems like a good card -- performs slightly better than my X800XL card that I had in the system previously, so I'm not complaining . . . just curious because I seemed to have discovered some type of "mongrel" version of the X800XT card.

;)

[Ripper3]

There's always a few, as mentioned, could have been a chip that didn't quite perform very stable at full speeds, and was cut down, and moved to AGP. I believe the 6600GT was the same, as the AGP version was slower than the PCIe version in clock speeds. Companies do it all the time, but with some additional cooling, more voltage, etc. you can usually get the full performance back.

[InnocentCriminal]

Welcome to the forums by the way.

:rockout:

[erocker]

There are threads, even stickied thread about this same subject. Welcome to the forums and please post in the appropriate subforum. Thanks!:)

[candle_86]

There's always a few, as mentioned, could have been a chip that didn't quite perform very stable at full speeds, and was cut down, and moved to AGP. I believe the 6600GT was the same, as the AGP version was slower than the PCIe version in clock speeds. Companies do it all the time, but with some additional cooling, more voltage, etc. you can usually get the full performance back.

actully that was because the 6600GT AGP at full speed is faster than the PCIe counterpart, i remember reading back in 2004 about AGP having lower latacny compared to AGP, and that AGP cards preform faster than PCIe cards at the same speeds as the AGP counterpart.

[jbodin]

Sorry erocker -- I actually meant to post in the AMD/ATI sub-forum, and I felt pretty dumb when I noticed I had posted in the "main" section. Thanks for moving the message here!