Heat, power, noise comparison between X1900 & 7900?

[Artic_Kid]

My thinking has turned-around in the last six months. I now see the ATI X1900GT (512MB) as being a better value than either the Nvidia 7900GT (256MB) or 7900GTX (512MB). The ATI has slightly better image quality, better future-proofing (for use on future games that are likely to use more Shader Model 3 and dynamic-branching), better AVIVO (video-in-out) quality, and at a substantially lower price. (Am I correct there?)

However, I still don't have much feeling for the heat, power, noise differences. I understand the Nvidia 7900 has better specs here, but I haven't seen any official comparisons yet. Some people say the ATI noise is barely noticable, except for five seconds when powering-up, and when playing games to the hilt. Other people compare the ATI to a blow dryer (in noise and heat).

Have any labs published comparisons yet? Any thoughts? I'm interested in power requirements (the financial cost of running it), at idle and under full load (I presume that will also tell me the heat output). I'm also interested in the noise (under idle and full load).

[kempez]

Anand have the answer vs: Here

[Smokey21]

I find the whole thing amazing.

The best example is that rubbishrubbishrubbishrubbish D.P at OCUK. He really pisses me off, he's such an Nvidia ****ing Fanboy.

Before the 7900 came out, people were saying how it was gonna be faster than the XTX, with 32 pipes, etc and all that crap. When it was seen it was only on par at best, people starting saying, well at least it runs cooler, and is quiter?

Surely it's for playing games? I don't quite understand the heat part. They still overclock like hell, and the cooler uses an exhaust system, so it keeps the case temp lower, i know it did when i went 7800>X1800XT.

Who cares about how much Power it uses? Nvidiaots clutching at straws me thinks?

[[DW]Cougho]

Its not even worth discussing running costs when comparing graphics cards. I would imagine the difference to be about the same as having a power LED on your computer or not..... aka zero.

Heat and noise, well that may matter. For me the old arctic coolers (for the 9800's etc) is too loud on full but most people say that its silent on full? its all subjective. By the sounds of it both cards would be too noisey for me but thats doesnt bother me as i would likely put a 3rd party cooler on anyway. However having read a bit around the net it seems that the X1900XT's run so hot that practically no 3rd party HSF is capable of dealing with the load (maybe with the exception of the soon to be released zalman thingy), that for me could win the day for a 7900GTX even if it is slightly slower.

So yes heat and noise do matter to some people. To me a silent system is worth alot more than the typical 3-5fps advantage the X1900XT enjoys. Although i must say i have heard neither and i am going off other peoples biased views no doubt.

[kempez]

I will vouch for the quiet of the 7800GTX 512mb/7900GTX cooler. Its very quiet

I will also say thet the back of the PCB get pretty hot on nvidia cards, whereas the back of the PCB does not give off heat on the ATI card. Dunno if that helps at all but there it it.

[vrykyl]

i have the advantage of picking based purely on performance...as i watercool my graphics so noice/heat/power isnt an issue

x1900xtx is the best card out atm imo...and its cheaper than the nvidia too

[Artic_Kid]

Anand have the answer vs: Here

Something about this data is weird. On this webpage, they say the X1900XTx draws at least 146 Watts at idle, and 315 Watts under full load. In other words, a Crossfire gaming rig would use at least 2x315=730 Watts just for the two graphics cards! You'd need a kilowatt powersupply!

Those same folks tested these cards in a Crossfire system, but their powersupply was only 600 Watts. It doesn't add up. What gives?

[Couger]

heat and noise difference to me doesnt bother me the x1900's run quite and cool enough in a proper effiently cooled case

ATI all the way

yeah ill agree they do chew monsterous power,
whist playing FEAR ATI tool displays that my card is using between 20 & 24 amps, thats insane lol

[sawyen]

Something about this data is weird. On this webpage, they say the X1900XTx draws at least 146 Watts at idle, and 315 Watts under full load. In other words, a Crossfire gaming rig would use at least 2x315=730 Watts just for the two graphics cards! You'd need a kilowatt powersupply!

Those same folks tested these cards in a Crossfire system, but their powersupply was only 600 Watts. It doesn't add up. What gives?

2 cards doesnt necessary draw double the power, as both cards arent fully loaded during games.. More often they're still CPU limited.. Unless AMD or Intel comes up with a processor that has 4x the current speed.. maybe.. PERHAPS maybe Crossfire and SLI would be running full load..

[badass]

Something about this data is weird. On this webpage, they say the X1900XTx draws at least 146 Watts at idle, and 315 Watts under full load. In other words, a Crossfire gaming rig would use at least 2x315=730 Watts just for the two graphics cards! You'd need a kilowatt powersupply!

Those same folks tested these cards in a Crossfire system, but their powersupply was only 600 Watts. It doesn't add up. What gives?

Slight flaw there. They are measuring the power draw of the entire system. The GFX card is using a fraction of the total power.

[Artic_Kid]

Slight flaw there. They are measuring the power draw of the entire system. The GFX card is using a fraction of the total power.

Eek! You're right. (I just checked their fine print.) Their entire system (including the X1900XTx, and CPU, mobo, HDDs, and all the rest) uses about 146 Watts at idle, and 315 Watts under full load. That makes much more sense. It suggests that the X1900XTx uses, approximately, 30 Watts at idle, and 140 Watts under full load. ??? (That's a very rough estimate on my part.)

Does anyone have accurate power specs for both the X1900s and the 7900s? I'd like to compare the two. For example, under full load, if the 7900s use, say, 30 Watts less than the X1900s, then that's probably not enough of a difference to justify the substantially higher cost of the 7900s.

[Artic_Kid]

yeah ill agree they do chew monsterous power,
whist playing FEAR ATI tool displays that my card is using between 20 & 24 amps, thats insane lol

That indicates that the ATI X1900XTx is using between 240 and 288 Watts! Yikes! (That's double what I estimated in my previous post.) With power figures like that, it's hard to explain two of these in a Crossfire system using a mere 600 Watt power supply (as the reviewers did here). Their PSU must have been operating on the hairy-edge of shutdown.

[sawyen]

Like I said in my previous post.. its unlikely that both cards are drawing exactly 100% load power...

Crossfire would have distributed alot of the math across both GPUs, therefore instead of 1 XTX running at 100%, you now get 2 cards running 50%.. Given the extra setting u throw at it, probably running 75% of each card with CPU limiting in most situations. If 1 card at 100% load chews 290-300W.. 150% would only be around 450W..

Enuf said..

[Artic_Kid]

... its unlikely that both cards are drawing exactly 100% load power... Crossfire would have distributed alot of the math across both GPUs, therefore instead of 1 XTX running at 100%, you now get 2 cards running 50%.. Given the extra setting u throw at it, probably running 75% of each card with CPU limiting in most situations. If 1 card at 100% load chews 290-300W.. 150% would only be around 450W.

I've been told (on a previous thread here at Hexus) that both of the dual graphics cards perform a fair amount of the same processing in duplicate. Each card uploads 100% of the textures, and each card does 100% of the vertex processing (so I'm told). Moreover, when operating in any non-alternating frame mode, (for example, in split-screen mode, or checkboard mode), there is an overlap of the areas to be computed, so here again both cards do 100% of the processing. In other words, there is some substantial processing-power wasted when doing SLI/Crossfire, and that will show up as extra heat. The change-over to SLI/Crossfire is not 100% efficient.

Also, concerning CPU limiting. People typically do not use SLI/Crossfire to get around CPU limiting. Rather they use it to get better image quality (higher resolution, and better eye-candy effects), while keeping approximately the same frame-rate (and CPU limiting) as before. People crank-up the image quality until their graphics system can barely keep up with the CPU. So I believe it a mistake to assume the two graphics cards are under-utilized (and under-powered).

Given all that, it seems speculative, even doubtful, that the two graphics cards would run at only 75% of their available processing power. Given the power-figures reported so far, it still seems that a 600 Watt power supply would be on the hairy-edge of shutting down.

P.S. We still have no power figures for the 7900XTx. Has anyone posted those?

[Barkotron]

Heat and noise, well that may matter. For me the old arctic coolers (for the 9800's etc) is too loud on full but most people say that its silent on full? its all subjective. By the sounds of it both cards would be too noisey for me but thats doesnt bother me as i would likely put a 3rd party cooler on anyway. However having read a bit around the net it seems that the X1900XT's run so hot that practically no 3rd party HSF is capable of dealing with the load (maybe with the exception of the soon to be released zalman thingy), that for me could win the day for a 7900GTX even if it is slightly slower.

They matter a lot to me as well. I was very reluctant to get the X1900whatever due to the stories about the noise, however in the end I decided to just get one (X1900XT) anyway - much cheaper than the equivalent NV cards, and apparently much better image quality, so a win-win really.

The noise was bad, so I replaced with the Thermalright V1-Ultra, which keeps core temperatures down in the mid-60s instead of the high 80s I was seeing with the stock cooler if it's on full, and mid to high 70s if I turn down the fan (using the Zalman multi-controller thing) to near-inaudible levels. You do have to be happy to take pliers/hacksaws to the RAM heatsinks that ship with the cooler (2 of them need to be cut down to roughly half their height due to the positioning of the heatsink/heatpipe assembly on the front of the card), or you could wait until they release compatible ramsinks, which apparently they are going to do at some point.

The only thing that gets really hot are the power regulators, as they're no longer being cooled by the fan. The Thermalright fan sits on top of the card and sucks air through the heatsink/heatpipe thing, and as the card sits directly underneath my exhaust fan, it gets extracted immediately. For the power regulators I've slapped on some of the Zalman ramsinks which has given me a bit lower temps (high 70s after 2 hours of load instead of mid-90s, (!!!) which it was before) - I need to clear up the airflow a bit from the front fan though, or maybe put a very low-rpm exhaust in underneath the graphics card (am I really ever going to use that PCI-E x1 slot? ) as it looks as if hot air may be pooling under there a bit. Turning up the front fan helps though - I guess plenty of cool air gets added under there, and air can escape via the space on the back panel where the exhaust of the X1900's heatsink used to be.

Now, even if I turn all the fans on full, it's not remotely as noisy as the X1900 stock is when it spins up. Bring on Oblivion!

[GFiSH]

I think I read somewhere that the 7900GTX will use aprrox 120w. Don't remember where I read that, think I googled '7900GTX' and found that a few days before they were released.