Looking at some early info about amds upcoming hd4850 graphics card. Some people have got these cards early and are reporting idle temps at around 68c. I thought these cards were to burn around 10w idle?

The one isn't automatically incompatible with the other.

the cooling solution would have to be almost absurdly bad to hit ~70C with only 10W of heat. also, given they're supposed to pull 100W+ at max load, the card would likely burst into flames if the cooling solution is that bad. seems more likely the temp measurement is just wrong.

All you need to reach these temperatures, is a fan-based cooling solution that has the fan switched off during idle operation.

One possibility is that the people reporting this are using Vista's Aero Glass interface. Since the GUI is rendered by 3D Hardware, the card might not be getting a chance to enter idle mode, and is therefore using much more than 10 watts of power. That's just speculation, of course.

3850's reference single slot cooler had the similar situation, 65'c, at idle. It looks like ATi implement the same temperature-controlling mechanism on 4850 as well. Anyway, I wonder whether 4850 enjoy the same quiet cooler design as 3850 that has been verified by silent authority, SPCR.

Well we have (in this thread at least!) very little information about what the cooling was in the tests reported.

The shots of HD4850 cards I've seen are your typical 1-slot graphics card heatsinks, with a plastic cover and a fan. Quite clearly optimised for high-heat, high-airflow. Turn the fan off, then what? No forced air cooling, no convective cooling, what are you left with? Conductive cooling through a couple of bits of plastic and fibreglass?!

HD3850
HD4850

Spot the differences

(Actually AFAIK the only difference is that there is an opening at the top of the cooler to let hot air out)

I've just realized something...

3870 TDP: 105W
4850 TDP: 110W

Maybe we'll be able to cool the 4850 passively... If that's the case, then I've definitely found my next graphics card.

the 4850 has a larger fan. As to the power consumption...while the 3870 was 105W, Xbit labs tests it as only really consuming ~85W at load. I'm hoping AMD overestimated the 4850 TDP as well.

Or not, if this preview is correct:

http://www.hardware.fr/articles/724-2/p ... -4850.html



But who knows, maybe it was a different system...

That 4850 is quite a blurry card, me thinks
4850 for <100 Watts would be awsome; maybe I could run that together with my (mobile) core duo, and a 2.5" hdd on my 220 Watt brick (more than half the power consumption taken by the card ).
Regarding the test; maybe the 3870 bottlenecks the system such that the cpu can't work at full throttle, possibly causing less power consumption than expected, but not directly related to gfx power consumption

Apparently the fan profile turns on late and turns on slow, so the card will get that hot.

Couple of guys on OCAU have 4850's already, and claim that changing the thermal compound to AS-5 has given a 20deg drop at idle, 10-15 at load for temps.

When it comes down to it it is a pretty basic cooler though, it's not meant to be the high-end even though it performs that way.

Personally i can't wait to see what the 4870 can do, I have a feeling we may see passive versions of both. I figure if my 8800GT can run passive, they can.

The Cat 8.5 drivers (not sure about 8.6) don't have the HD48x0 cards power play activated so the cards voltage and frequency are not throttling down but the fan is.

Good news

I've found out that the mounting holes are the same as on the 38xx cards, so standard coolers will fit.

http://www.pcper.com/article.php?aid=57 ... pert&pid=1

I hope throttling is disabled in AMD Catalyst RV770 Beta used in the PC Perspective preview Matija links to. Otherwise those numbers are less than impressive.

Let's see there's still time for the official launch and new catalysts just came yesterday...

Unfortunately, I think the 110W TDP is correct. The 4850 used more juice than an overclocked 8800... As far as passive cooling goes, I don't think it will be possible.

Another power consumption graph:




Source: Hardware.fr




It doesn't use much more power than a 8800GT... passive cooling is still worth a try IMO.

I linked that one earlier in the thread Notice 352W system load compared to 319W with the HD3870.

But, things just get more and more weird.

http://www.guru3d.com/article/amd-ati-r ... wercolor/7
HD 2900 XT 391 Watt
Radeon HD 3850 259 Watt
Radeon HD 4850 Force3D 269 Watt
Radeon HD 4850 Powercolor 269 Watt

http://www.techpowerup.com/reviews/MSI/HD_4850/21.html




Looks like we're going to have to wait for a while, until Xbitlabs gives us exact power consumption figures.

All things considered AMD seem to have had a bit of a cockup with this one.

Its directly comparable to the 8800GT in the following ways.

Price.
Performance.
Power Consumption.

So why would anyone actually buy one of these over the competition (excluding fanboys and people who prefer red to green).

The 8800GT is an established item in the market, I just dont get what AMD has brought to the table with this item, its not faster as the same price point, and its not cheaper at the same performance.??? Someone please enlighten me.

On that note, the 4870 benchmarks, tests etc etc have not hit the net yet, will that be as intimidating (sarcasm!) to nVidia as the 4850 seems to be.

nVidia are also about to launch (in a months time) a re-hash of the 9800 GTX called the 9800 GTX+, which is really going to make AMD hurt if the 4870 fails to deliver.

http://www.theinquirer.net/gb/inquirer/ ... x9800-tips


Andy

HD4850 comparable to the 8800GT in performance?! What planet are you on? It matches the 9800GTX in performance!





Average performance.

http://www.pcper.com/article.php?aid=57 ... pert&pid=3

Starting on page 3 of the review linked above, there is not a mind blowing performance difference between those 3 cards in most of the tests, and may I add that in some of the tests the 8800GT beats the 9800 GTX.

NOTE:

The 8800GT in question has been factory overclocked to 700MHz (from the standard 600MHz - the RAM has also been given a boost).

The 4850, looks rather good next to the 9800GTX for the money (£90 price difference), but the 9800GTX is a rip off compared to an overclocked 8800GT available for £136 (BFG - 675MHz core).

Either way, it seems that the 4850 (as of right now) look like good value for money for a high-end part so long as it is avaliable at ~ £100 ($200 was the price I saw quoted).

It has my fullest blessing


Andy

the TechPowerUp review shows the 4850 outpacing the other cards in higher resolutions, and in just about every game.

You forgot to mention: In low resolutions. The same thing applies to the new NVidia models. Basically all the introduced cards show significant gains in 1920x1200 and up. For lower res, "old" cards are usually plenty enough. (But only until game designers start cramming more effects in their engines, but that's still ahead of us.)

Jeez....Even the gtx280 burns less power when idle...

tis a bit odd.

what is wrong with the 4850 idle power consumption. I was seriously considering buying one, but after seeing the results I am a bit confused.

More then 40watts idle consumption is outrageous. My current card doesnt even burn that much under load (a seriously overclock 7600gs sonic). 40 more watts under desktop operations is a kill for me considering a whole system power consumption of about 90 watts at idle right now.

At least the 62C idle temp of the card is now explained.

This is from the Tech Report:

http://www.techreport.com/articles.x/14967/10