EVGA GTX 480 for $250 @ Amazon

[railven]

I just read through the article to understand their methods, and actually the heating/cooling of test cycling I mentioned isn't appropriate here since they're running a constant batch render as opposed to the standard benchmark. However, they do address your question:

http://www.hardwarecanucks.com/foru...eforce-gtx-470-super-overclock-review-16.html

That explaination just baffles me more on why the SOC idles higher than the reference. Everything keeps telling me it is a better built card, with better cooling, and thus should reduce the heat issue that would make the reference card consume more power.

Unless one card is idling at a higher clock or using some crazy fan, I still think it's odd that the Ref idles for 10w less than the SOC.

EDIT; This would have to answer my question:

Another step forward in comparison to the stock heatsink is the Super Overclock’s noise profile. One would expect that the trio of 80mm fans would need to literally scream in order to sufficiently cool the fin array but that doesn’t happen. Due to the surface area of the heatsink, all of the core’s heat is well distributed across a fairly large area which means the fans don’t need to work all that hard. The result is an amazingly quiet experience.

Loaded the review on my phone (man it is a bitch to cut and paste on a smart phone haha) and basically, this thing is running a crazy fan setup haha. Wish the images would load

 

[Grooveriding]

Also worth noting that the claims of a later revisions of GF100 making substantial thermal improvements are... unsubstantiated.

There was a major BIOS flaw with GF100 at release that had a very poor fan curve resulting in horrible temperatures like the 90C+ you saw here at anand and elsewhere. A later BIOS corrected this resulting in better temps and even more noise

 

[railven]

Also worth noting that the claims of a later revisions of GF100 making substantial thermal improvements are... unsubstantiated.

There was a major BIOS flaw with GF100 at release that had a very poor fan curve resulting in horrible temperatures like the 90C+ you saw here at anand and elsewhere. A later BIOS corrected this resulting in better temps and even more noise

Well I use headphones that go to 11 so, I don't mind. Of course, my whole house is electric (damn you hippies!) and of course I sit there and stare at my meter oh and abusively beat the GF if she leaves a light on.

I apologize to my mother who'd yell at me when I left the kitchen light on all night. I'm learning my lesson.

 

[BallaTheFeared]

Also worth noting that the claims of a later revisions of GF100 making substantial thermal improvements are... unsubstantiated.

There was a major BIOS flaw with GF100 at release that had a very poor fan curve resulting in horrible temperatures like the 90C+ you saw here at anand and elsewhere. A later BIOS corrected this resulting in better temps and even more noise

Dunno about that, but any blower fan sounds awful no matter what card it's on.

It looks cool though.

lol @ goes to 11, that phrase will forever be an lulz for me.

 

[railven]

Dunno about that, but any blower fan sounds awful no matter what card it's on.

The blower fan on the HD 7970 is miles better than the blower on the HD 5870. Sure it CAN get louder, just monkeying around with manual options HD 7970@ 20% sounds like HD 5870 @ 50% but as you get higher HD 7970 @ 50% sounds like HD 5870 @ 70-80%, at 100% it sounds like a hair dryer.

Leaving it on auto, my GPU hit about 60-65c so far and never broke 30% fan speed.

I'll take it!

On my HD 5870 I was keeping the fan around 40-50% to keep it under 65c.

 

[Grooveriding]

The blower fan on the HD 7970 is miles better than the blower on the HD 5870. Sure it CAN get louder, just monkeying around with manual options HD 7970@ 20% sounds like HD 5870 @ 50% but as you get higher HD 7970 @ 50% sounds like HD 5870 @ 70-80%, at 100% it sounds like a hair dryer.

Leaving it on auto, my GPU hit about 60-65c so far and never broke 30% fan speed.

I'll take it!

On my HD 5870 I was keeping the fan around 40-50% to keep it under 65c.

I went from 5870 CF to 480 SLI initially and I knew what I was getting into as the reviews had been out for a few months already. It was a night and day difference in terms of noise and heat.

The 5870s were reasonably quiet under load, not inaudible, but nothing outrageous. The 480s were crazy out of hand loud in comparison.

For me performance was always #1 and I'd accept shitty thermals and noise if the cards are faster and better suited to my resolution, but after owning these 480s, it's now a factor for me. I've never had cards so loud or obnoxious. If nvidia releases a 500mm2 die 780 with even worse thermals than a 480 but is wicked fast, I'll buy them, but they're getting water-cooled

 

[BallaTheFeared]

It's all personal preference, personally I couldn't stand my 470's at idle (30%), but as you've read others don't care about it at 80% lol...

I don't use headphones though, or a case, and my comp (hardware anyways) is like 16 inches from my ear so my take on things is a bit skewed already!

I wouldn't do anything but water cool anymore, it's just so nice

 

[Lonbjerg]

Has anyone else been able to back up their findings?

http://benchmarkreviews.com/index.php?option=com_content&task=view&id=607&Itemid=72

Oh, how the mighty have fallen...
...to more reasonable power consumption levels. When Benchmark Reviews first tested the NVIDIA GeForce GTX 480 engineering sample, the idle power draw was a thirsty 52 watts of electricity, and perhaps among the highest idle power draw we've measured for any single-GPU DX11-generation graphics cards. This level of consumption is slightly higher than the 48W we measured for the dual-GPU AMD Radeon HD 5970, and more than twice the demand of AMD's Radeon HD5870 and HD5850.

Similar to the dramatic decrease in temperature due to mature firmware, the Fermi GPU cores have been tamed to use less idle power on the ASUS GeForce GTX 480 than before... down to 39 watts at idle. The GF100 GPU surely has a big power appetite, but it's nice to see ASUS has trained it to snack on only a few watts.

Once 3D-applications begin to demand power from the GPU, electrical power consumption really begins to climb. Measured at full 3D load, the NVIDIA GeForce GTX 480 engineering sample set a new maximum power record and consumed 370 watts. With a more refined BIOS controlling the ENGTX480 unit, the ASUS GeForce GTX 480 slimmed down to 315W... a decrease of 55W.

For comparison, here are the GPU voltages the original engineering sample and the ASUS ENGTX480:

NVIDIA GeForce GTX 480 Engineering Sample
MSI Afterburner reported 0.962V GPU at idle, and 1.025V under load.
GPU-Z reported 0.953V 15.0A 14.3W idle VDDC, and 0.980V 75.0A 74.0W at load.
ASUS ENGTX480
MSI Afterburner reported 0.962V GPU at idle, and 1.075V under load.
GPU-Z reported 0.955V 11.0A 10.5W idle VDDC, and 1.033V 70.0A 72.3W at load.

Hence why sites that only do reviews at launch are useless.

 

[hans030390]

http://benchmarkreviews.com/index.php?option=com_content&task=view&id=607&Itemid=72

Hence why sites that only do reviews at launch are useless.

That was the site originally posted, no? I was wondering if anyone else found the same thing they did.

Either way, I already ordered a card. I'll have a flop of a rig! Got my FX6100 highly overclocked (draws so much power) AND a GTX 480! Haha.

 

[BallaTheFeared]

lol I hope you have a 1.21 gigawatt psu!

But on a serious note, report back here with your thoughts on it after you get it.

 

[MrK6]

http://benchmarkreviews.com/index.php?option=com_content&task=view&id=607&Itemid=72
Hence why sites that only do reviews at launch are useless.

Sites that do shoddy reviews and leave out important information are also useless. First, they use a Kill-a-Watt to test for power consumption, which isn't the most accurate tool. Secondly, they use Furmark to test the load power of the card, a program which NVIDIA purposefully limits in their drivers. One driver release could have taken down the power consumption, not any engineering change. Finally, they never even bothered recording fan speed or dB levels - the new cards could simply have the fan run faster to keep the card cooler and take a few more watts off.

If this was a bigger issue, AT or another real review site would have discovered it.

 

[BallaTheFeared]

http://benchmarkreviews.com/index.php?option=com_content&task=view&id=607&Itemid=72



Hence why sites that only do reviews at launch are useless.

Yeah and despite what some uninformed people might have you believe just about everyone uses a kill-a-watt type unit and 4 series never even had built in driver throttling for furmark. 5 series did, however 4 did not.

 

[hans030390]

lol I hope you have a 1.21 gigawatt psu!

But on a serious note, report back here with your thoughts on it after you get it.

Haha. Well, I'm currently running my FX6100 at 4.3GHz with 1.45V. I haven't tried taking it further or seeing if I can run a lower voltage at that speed. Temps are fine, thanks to my Antec Kuhler 620 in push/pull configuration.

With that and my 6870, my Rosewill 530W PSU (continuous, and I believe 80 Plus certified) pulls 553W from the wall at full load. If I were being generous and said my PSU was 85% efficient at best, that means my components use about 470W.

My full load test consists of running OCCT (CPU, all cores, small data set) and Furmark (1080p, 2x AA, xtreme burn-in). This combo draws more power than any other stability test I tried, and as such is in NO way indicative of a normal power draw at load. For example, playing Skyrim or Battlefield 3 at highest settings for a while only brings wall power consumption to 365W (about 310W at 85% PSU efficiency).

I just purchased a new Antec EarthWatts 650W PSU. It has 2 12V rails rated at 38A each, and the box says the 12V combined output is 650W (54.2A). My Rosewill 530W PSU, on the other hand, only has 41A on the 12V rail (492W).

I'm hoping I will be able to drop the GTX 480 in with my new PSU and not run into any issue. I think I should be fine, though I might be pushing it if I were to do the same power consumption test mentioned above.

I do not plan on overclocking the GTX 480. I actually plan on undervolting it at stock clocks to help with power consumption and heat. If I still have headroom on my PSU, I plan on overclocking my FX6100 more. If not, I'll either leave it the same or try to find a nice overclock at a lower voltage.

 

[BallaTheFeared]

Overclock it!

Also from another thread blackend linked a consumption chart which showed another Gigabyte SOC card, this time a 480, using less power than their initial review card while being overclocked from 700MHz to 820MHz.

It's a pity we can't get an absolute answer on this and more cards weren't reviewed after the fact

 

[wbynum]

Here's my wattage numbers at idle on my new EVGA 480 SLI setup. Clean install on Win 7 x64 with latest Nvidia drivers (296.10). Wattage measured with a kill-a-watt at the wall.

Before:

i7 system oc'ed to 3.8 ghz with cheap AMD HD4550 card: ~155 watts

After:

Same i7 system with 2xSLI 480 in 2d down clocked state: ~210 watts

So take away ~25 watts for the AMD HD4550 card and that leaves 210 - 130 = 80 watts that the two 480's are drawing in total at idle. Around 40 watts per card. Pretty spot on with the 39 watts at idle reported by the "6 months into launch" article by Benchmark Reviews.

 

[ArchAngel777]

Well, so far so good. One thing I will say: Don't leave the fan profile to 'auto'... I almost jumped out of my char when I let the fan do it's own thing... It would randomly shoot up to 100% to bring temps in line, then would repeat after another 5 minutes. Rather than gradually ramp up, it would do a knee jerk reaction.

The fan is very loud at 70% at 80% is like a vaccum cleaner in the next room and at 100% it is like putting your ear right next to the dust buster. No joke!

Anyway, 70% of tolerable when gaming and perhaps even 80% if you turn up the volume but no way in freaking God's great earth should anyone find 100% even for 1 second acceptable. This card is major FAIL in regards to fan noise. No doubt.

But... For $210, I am satisfied. I'd have felt ripped off if I purchased this thing at anything greater than $250.

 

[BallaTheFeared]

Haha Arch, what kind of temps are you getting at 70% fan?

 

[ArchAngel777]

Haha Arch, what kind of temps are you getting at 70% fan?

70% keeps the temps right around 80C - But, I also run V-Sync, which lowers GPU loads during times where I exceed my refresh rate. So my GPU usage tends to average around 85%...

 

[BallaTheFeared]

Dang that's a hot one, have you considered removing the HSF and reapplying thermal past?

 

[Rvenger]

Dang that's a hot one, have you considered removing the HSF and reapplying thermal past?

It will knock it down about 3c at the most unfortunately. These cards are just naturally frying pans.

 

[hans030390]

Overclock it!

Also from another thread blackend linked a consumption chart which showed another Gigabyte SOC card, this time a 480, using less power than their initial review card while being overclocked from 700MHz to 820MHz.

It's a pity we can't get an absolute answer on this and more cards weren't reviewed after the fact

Oh wow, that is promising! I might overclock it if I find I have room on the PSU. My CPU has first priority, as it's the weakest link in my rig. The GTX 480 at stock offers plenty of performance for me.

 

[Mezzanine]

An aftermarket cooler does wonders for the 480, mine is fitted with a VF3000F and only gets up to around 67c in BF3 with the fans running at minimum (and silent) speed. My card has a low default voltage of .988v but I run it even lower at .963v so that probably helps a lot.

It's a different story when overclocked to 850-900mhz though, the card throws out some serious heat and it makes my case creak as it heats up in there.

 

[hans030390]

Unfortunately, I don't think I have room for an aftermarket cooler. My sound card is below my GPU. I might be able to swap them, as they're both using a standard PCI-E slot. Just not sure if the bottom one runs at a slower speed or not...

 

[hans030390]

Just got mine in today! Did some quick power consumption measurements with my Kill-a-watt. Here's my system info:

CPU: AMD FX 6100 @ 4.3GHz (1.36V Idle/1.43V Load CPU, 1.3V CPU/NB, 1.2V NB)
RAM: DDR3 1600 (1.65V)
GPU: EVGA GTX 480 (stock)
Mobo: Asus M5A97
PSU: Antec EarthWatts 650W (Green, newest model)

Power consumption measurements were taken from the wall/Kill-a-watt at idle, full GPU load with Furmark (Burn-in and Xtreme burn-in, Dynamic Background, Post FX all enabled, 1080p, 8x MSAA), and while playing Skyrim at high settings.

With HD6870:

Idle - 141W (Did not test to see if long idle would make a difference)
GPU Load - 332W
Skyrim - 320W

With GTX 480

Idle - 160W (long idle brought this up to 172W after running some applications)
GPU Load - 458W
Skyrim - Averaged around 385W, peaked at 405W a few times (very brief)

Note: I installed a new 120mm fan in the side of my case AFTER running these tests and BEFORE installing the GTX 480. I know fans draw comparatively little power, but I figured it would be good to make that known.

I have not tried doing a power consumption test at 100% load for both CPU and GPU, but my calculations/estimations place that type of power consumption around 705-720W from the wall. Taking the PSU's efficiency into account, this should still be perfectly safe and within the 650W rating for the 12V rails alone. That, and no real world situation will ever draw nearly as much power as a combo CPU + GPU stress test.

With my rig, the GTX 480 draws 126W more than the HD6870 with the strongest possible load you can place on it. If you compare the GTX 480 and the HD6870 on the Anandtech GPU11 benchmark comparison page, the difference is 173W. Quite a difference. Of course, I'm not 100% sure they use the exact same rig for these benchmarks. Still, that might be further proof that later GTX 480s have better power consumption than the early review models.

Realistically, power consumption won't be an issue with the GTX 480. The most stressful thing I do with my computer on a regular basis is gaming, and we're looking at an "average" difference of 65W between the HD6870 and GTX 480. Big deal! Oh, and for reference, Battlefield 3 drew less power than Skyrim from my system with my HD6870. This was tested long before I had my GTX 480.

I'm well aware that you can get the same performance for much less power consumption, but the additional electrical costs I'll experience from my rig aren't an issue. I got the FX 6100 for $100, the M5A97 for $80, and the GTX 480 for $210. I think it would be difficult to beat that performance for the price, and power consumption, while not great relative to other parts, is fine for me.

Next up, I'm going to start undervolting my GTX 480!

 

[n0x1ous]

Just got mine in today! Did some quick power consumption measurements with my Kill-a-watt. Here's my system info:

CPU: AMD FX 6100 @ 4.3GHz (1.36V Idle/1.43V Load CPU, 1.3V CPU/NB, 1.2V NB)
RAM: DDR3 1600 (1.65V)
GPU: EVGA GTX 480 (stock)
Mobo: Asus M5A97
PSU: Antec EarthWatts 650W (Green, newest model)

Power consumption measurements were taken from the wall/Kill-a-watt at idle (long idle is specified), full GPU load with Furmark (Burn-in and Xtreme burn-in, Dynamic Background, Post FX all enabled, 1080p, 8x MSAA), and while playing Skyrim at high settings.

With HD6870:

Idle - 141W
GPU Load - 332W
Skyrim - 320W

With GTX 480

Idle - 160W (long idle brought this up to 172W after running some applications)
GPU Load - 458W
Skyrim - Averaged around 385W, peaked at 405W a few times (very brief)

I have not tried doing a power consumption test at 100% load for both CPU and GPU, but my calculations/estimations place that type of power consumption around 705-717W from the wall. Taking the PSU's efficiency into account, this should still be safe for my PSU. That, and no real world situation will ever draw that much power.

With my rig, the GTX 480 draws 126W more than the HD6870 with the strongest possible load you can place on it. If you compare the GTX 480 and the HD6870 on the Anandtech GPU11 benchmark comparison page, the difference is 173W. Quite a difference. Of course, I'm not 100% sure they use the exact same rig for these benchmarks. Still, that might be further proof that later GTX 480s have better power consumption than the early review models.

thanks for the update and power consumption test! :thumbsup: