Dont understand...hd4850 10w idle @ 68c
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
I'm really curious about this card
I'm looking for something to replace my 8800GT, the girlfriend needs a new gfx card (she has a couple months old 8800GT now but it's noisy as hell and impossible to shut it up because of no fan control on this particular card) and I figured I'd put my quiet 8800 in her system and get something new for myself.
The 4850 is already for sale here since friday, but I'd really like to know how silent it is before I start ordering stuff. But if this really fits that is what I'll be using, right now I'm only cooling the CPU with my reserator.
I'm looking for something to replace my 8800GT, the girlfriend needs a new gfx card (she has a couple months old 8800GT now but it's noisy as hell and impossible to shut it up because of no fan control on this particular card) and I figured I'd put my quiet 8800 in her system and get something new for myself.
The 4850 is already for sale here since friday, but I'd really like to know how silent it is before I start ordering stuff. But if this really fits that is what I'll be using, right now I'm only cooling the CPU with my reserator.
Please give me a couple of months notice on this will, because I read the http://www.theinquirer.net/ and those bastards are rarely wrong on dates or figures, or anything else, and I dont work for them.There's almost always a video card refresh around Dec/Jan. I am going to bet that's when the process switch will occur. Either then, or maybe Oct/Nov. Depends on when Nvidia decides to attempt the 280 on 55nm. Maybe Ati would react and do the same, only with 45nm? They'll do it, and they'll do it quick if the yields are better and cost is lower.
Andy
-
- *Lifetime Patron*
- Posts: 1809
- Joined: Sat Apr 24, 2004 1:45 am
- Location: At Home
Interesting, although it shows that the Voltage is the same so I wonder how significant just dropping the clock speeds will be?davemuk wrote:According to this post, the idle freq. should be 160mhz. I'm guessing this would drop the power usuage dramatically.
-
- Friend of SPCR
- Posts: 1439
- Joined: Tue Dec 14, 2004 4:06 pm
- Location: New Hampshire, US
- Contact:
That's exactly what I said over at XS regarding powerplay and the 3870, but members started berating me for not knowing that power draw with the HD3800 series scales more with clockspeed and not with voltage. I asked them to prove it, and they showed me a link to power draw increasing linearly with clockspeed at a static voltage. I asked them why they didn't show multiple voltages, and they said that their results were all they needed for confirmation. Maybe they were on to something though, and that's why the vgpu doesn't drop?smilingcrow wrote:Interesting, although it shows that the Voltage is the same so I wonder how significant just dropping the clock speeds will be?davemuk wrote:According to this post, the idle freq. should be 160mhz. I'm guessing this would drop the power usuage dramatically.
Power consumption of the 4850 and 4870: http://www.expreview.com/topic/2008-06- ... 73_13.html
not bad, but not impressive. Nvidia probably does it better this generation, with higher load power but lower idle power.
not bad, but not impressive. Nvidia probably does it better this generation, with higher load power but lower idle power.
Power consumption in IDLE
Just to redress the power consumption in IDLE for all commenters here who don't want to read the whole thread:
Current drivers are *not yet* supporting the Powerplay power-saving technology in the R48xx cards.
This explains the high idle comsumption in all these "day one" tests.
We will have to wait till ATi bring out the new drivers, then we should see the amazingly low idle consumptions they have promised us. So please people, have a little more patience before you judge.
Current drivers are *not yet* supporting the Powerplay power-saving technology in the R48xx cards.
This explains the high idle comsumption in all these "day one" tests.
We will have to wait till ATi bring out the new drivers, then we should see the amazingly low idle consumptions they have promised us. So please people, have a little more patience before you judge.
Re: Power consumption in IDLE
Eh It seems I got it rightRadiance wrote:Just to redress the power consumption in IDLE for all commenters here who don't want to read the whole thread:
Current drivers are *not yet* supporting the Powerplay power-saving technology in the R48xx cards.
This explains the high idle comsumption in all these "day one" tests.
We will have to wait till ATi bring out the new drivers, then we should see the amazingly low idle consumptions they have promised us. So please people, have a little more patience before you judge.
viewtopic.php?p=419016#419016
-
- *Lifetime Patron*
- Posts: 1809
- Joined: Sat Apr 24, 2004 1:45 am
- Location: At Home
I’ve read speculation that the under-clocking & under-volting feature has been moved from the BIOS to the driver with the HD 4xxx series cards. I hope this is the case as many manufacturers failed to implement it in the BIOS which made choosing a HD 3xxx series card a pain and didn’t reflect well on ATI ultimately. I just hope that a new driver tames the idle power draw of these cards as the 4870 is shockingly high by ATI’s standards.
I have had the Asus EAH4850 for a day now.
The fan spins really slowly on idle.
When I ran 3DMARK06 it still didn't seem to spin faster. At least I didn't spot any difference. This could be a problem.
GPU-Z shows 500MHz for GPU core and 750MHz for memory when on idle.
They sure seem a bit high...
I'm using the HD4800 series hotfix drivers.
Temperatures were too high for my taste (over 70C on idle). I thought the plastic cover would melt.
Accelero S1 lowered the idle temperature to 36C.
My PSU (Silverstone ST30NF) is only 300W (18A for 12V) but it seems to be enough for this card.
The fan spins really slowly on idle.
When I ran 3DMARK06 it still didn't seem to spin faster. At least I didn't spot any difference. This could be a problem.
GPU-Z shows 500MHz for GPU core and 750MHz for memory when on idle.
They sure seem a bit high...
I'm using the HD4800 series hotfix drivers.
Temperatures were too high for my taste (over 70C on idle). I thought the plastic cover would melt.
Accelero S1 lowered the idle temperature to 36C.
My PSU (Silverstone ST30NF) is only 300W (18A for 12V) but it seems to be enough for this card.
-
- *Lifetime Patron*
- Posts: 1809
- Joined: Sat Apr 24, 2004 1:45 am
- Location: At Home
-
- *Lifetime Patron*
- Posts: 1809
- Joined: Sat Apr 24, 2004 1:45 am
- Location: At Home
It doesn’t have to be that way as the nVIDIA GTX 200 series are more powerful and built on the same process size as their previous generation but consume significantly less than them at idle.Modo wrote:Maybe the idle power consumption is a BIOS problem, but the HD 4850 has about twice the processing power of the HD 3870, while being built using the same technology, so I wouldn't count on a large reduction when in use.
ATI were talking about less than 10W at idle so hopefully it is a BIOS or Driver issue. The HD 4870 consumes a lot more at idle than the 4850 which seems to be down to the GDDR5! Not sure how much a BIOS/Driver update will impact on that.
-
- *Lifetime Patron*
- Posts: 1809
- Joined: Sat Apr 24, 2004 1:45 am
- Location: At Home
I’ve noticed that a 6 pin PCIe cable will connect with an 8 pin connector on the card. Provided that the card doesn’t require more than the 6 pin cable can provide is it safe to run that sort of configuration?Scoop wrote:Anyone else find it amusing that you need to buy a big ass PSU for the 9800GTX just for the extra pcie connector when the HD 4850 uses more power and manages with just one.
Sorry if I was not precise enough. I was not referring to idling, but to active use, like 3D gaming. Sadly, that probably won't be helped by new drivers, and it might prevent graphics cards' manufacturers from making parts with factory-designed fanless cooling.smilingcrow wrote:It doesn’t have to be that way as the nVIDIA GTX 200 series are more powerful and built on the same process size as their previous generation but consume significantly less than them at idle.
-
- Friend of SPCR
- Posts: 1439
- Joined: Tue Dec 14, 2004 4:06 pm
- Location: New Hampshire, US
- Contact:
It's not so bad. For a 40+% increase in die size, they only go up 30W at load vs the 3870. Idle power needs to be addressed, as it appears to be some bug. It would be nice to hear an official word from AMD on the matter.Matija wrote:And here we go.
The card does use a whole lot extra power over the 3870
It would be a great choice if its idle consumption was lower... Not so much right now.
I'm disappointed.