SPCR

Part I looks at contemporary graphics cards on ATI chips

http://www.xbitlabs.com/articles/video/ ... rcons.html

Looks like lots of useful information for enthusiasts. I assume part Part II is to follow.

Wow, this guy is thorough! He even manages to measure AGP power consumption and provides figures for each rail. Thanks for the link, that's an excellent article.

Yes I cant wait to see the followup with results for Nvidia

I'm REALLY impressed with their thoroughness. Take that Tom's hardware!!

I still didn't get how he got to supply the AGP voltages through the shunt, especiall the 3.3V line. Whatever.

Looking at the figures makes me think the 9600XT can also be cooled passively. There's less than 5W of extra heat for the XT, nothing a good heatsink can't take care of

Must say im highly impressed with the article, cant wait for the Nvidia follow up. Its about time definitive mesurements were made.
Ive never used a ATI card(ive only had 1 new card since my GeForce SDR) but the readings on the 9600 are amazing, i knew it was low but not that low, needless to say for silent video cards its where its at.

Bring on round 2

burcakb wrote:I still didn't get how he got to supply the AGP voltages through the shunt, especiall the 3.3V line.

Well, this is what he says:

What to do? I took it easy and just insulated the appropriate pins of the AGP connector on the graphics card with stripes of adhesive tape (pin A1 – 12v; pins B2, B3 – 5v; pins A9, B9, A16, B16, A25, B25, A28, B28 – 3.3v) and applied those voltages directly to the graphics card, through the prepared shunt, taking the voltages from the PSU.

The 3.3V rail is present on the extra mobo connector. He would have needed a separate shunt for the AGP with multiple leads soldered to the graphics card. Pity he didn't provide a picture to prove he really did that.

Inexplicable wrote:Pity he didn't provide a picture to prove he really did that.

True... that sounds like a massive pain in the ass. I wonder though if he just tapped into the motherboard, which would be much easier to perform the tests.

sthayashi wrote:I wonder though if he just tapped into the motherboard, which would be much easier to perform the tests.

You mean tapping the motherboard power connectors? I hope not. That would throw some suspicion over validity of the results in general.

IMHO, the best way to do this would be to prepare a shim card that goes into the AGP slot and provides an AGP connector for the graphics card and all the necessary measurement points. Shouldn't be too hard to do if you have the equipment.

Inexplicable wrote: IMHO, the best way to do this would be to prepare a shim card that goes into the AGP slot and provides an AGP connector for the graphics card and all the necessary measurement points. Shouldn't be too hard to do if you have the equipment.

You mean like an AGP version of a PCI Riser card? im pretty sure you can get right angle AGP adapters out there, I know you can get PCI ones.
If you wanted to be really handy though, i guess if you had an old mobo and an old agp card you were willing to rip apart plus a soldering iron and you were really bored, i guess it could be DIY'ed

Interesting Article...

Uberman1080 wrote:You mean like an AGP version of a PCI Riser card? im pretty sure you can get right angle AGP adapters out there, I know you can get PCI ones.

Sounds like just the thing. There seem to be a bunch of different kinds here:
http://www.mycableshop.com/3rd_Level/AG ... s-Misc.htm

How is the power consumption of the 6800 GT?

Leto wrote:How is the power consumption of the 6800 GT?

I don't have any data to back this up, but my supposition is that they will consume more power than the newest generation of ATI cards.

Even before seeing this article, I knew that the X800 and X800 XT were able to consume less power than their predecessors at idle and no more at load. Quite a feat of engineering.

If somebody has similar information about the 6800 series by Nvidia, please post it at this thread.

X-bit labs have updated their site with the Nvidia benchmarks

http://www.xbitlabs.com/articles/video/ ... power.html

Well, well. Looks like nVidia is not doing quite as badly as people feared.

Inexplicable wrote:Well, well. Looks like nVidia is not doing quite as badly as people feared.

You can't just compare the used wattage. According to the article the problem is that some nVidia cards use a lot of 12 V. So even if a system uses 200 W and have a 350 W PSU, it might not work properly because the PSU can't deliver 12 V enough. That's why a 450-550 W PSU is recommended for these cards.

Mats wrote:You can't just compare the used wattage. According to the article the problem is that some nVidia cards use a lot of 12 V. So even if a system uses 200 W and have a 350 W PSU, it might not work properly because the PSU can't deliver 12 V enough. That's why a 450-550 W PSU is recommended for these cards.

The difference to ATI is hardly that drastic. The 6800U pulls 15W more than the X800XT PE from the 12V rail. While that is not insignificant, a decent 300W PSU can easily supply over 200W at 12V. Those PSU recommendations are just a proactive measure to cut down support requests from people who tend to buy the cheapest crappiest no-name PSU they can find.

The new generation of video cards has shifted power draw away from the 5V and 3.3V lines toward 12V. Total power draw hasn't changed that much under load, but the 12V line is going to be providing most of the power for the GPU and CPU on most modern motherboards. Generic 300W, 400W, etc. PSU power labels didn't mean that much before and they mean even less with newer systems and the latest graphics cards. Read the fine print on the PSUs for the important data.