[Jawed]

Trying to keep up with HD5770.

[CRoland]

Originally Posted by Jawed

Why is it showing D3D10.1?

Maybe because only Vista SP1 is installed?

[mczak]

Originally Posted by Jawed

Trying to keep up with HD5770.

Apparently. And so far it looks quite promising. Ok it required high clocks (though considering what GF104 could easily do not extraordinary high, as a guess I'd suspect it will have only very slightly higher voltage), and even with that the vantage GPU score can't quite keep up with HD5770, but the same is true there for GTX460 vs HD5830... I still got some feeling though GTS450 will have trouble keeping up with HD5770 in games with that configuration but I might be wrong. Memory clock is quite modest too.

[Alexko]

Originally Posted by mczak

Apparently. And so far it looks quite promising. Ok it required high clocks (though considering what GF104 could easily do not extraordinary high, as a guess I'd suspect it will have only very slightly higher voltage), and even with that the vantage GPU score can't quite keep up with HD5770, but the same is true there for GTX460 vs HD5830... I still got some feeling though GTS450 will have trouble keeping up with HD5770 in games with that configuration but I might be wrong. Memory clock is quite modest too.

Yeah, the part that really worries me is the memory bandwidth. Only 128-bit GDDR5 compared to 448-bit GDDR3 for the GTX 260, with the same number of shaders…

[trinibwoy]

Originally Posted by Jawed

So, only 128-bit and only 4 SMs?

Maybe it's so huge because there are two GPCs?

If the full chip is 192 shaders have we ruled out the possibility of 6 SMs, 32 shaders / 8 texture units each? I don't know how much Nvidia is playing around with the unit counts but it's a possibility and could explain the chunky die size.

[Jawed]

Originally Posted by trinibwoy

If the full chip is 192 shaders have we ruled out the possibility of 6 SMs, 32 shaders / 8 texture units each?

I was going to mention this kind of possibility, e.g. with a single despatch unit per SM that issues two instructions: 1 to MAD + 1 to either load/store or SF.

Then there's the matter of whether a GPC can support more than 4 SMs.

I don't know how much Nvidia is playing around with the unit counts but it's a possibility and could explain the chunky die size.

ATI has lower ALU:TEX for the smallest GPU, perhaps NVidia's doing the same. It seems like an amazingly heavy price to pay.

But it's worth remembering this was probably meant to be a 32nm chip. That would have made it about the same size as Juniper, I suppose. Too small for 256-bit physical interface, without doubling.

[Jawed]

Originally Posted by Alexko

Yeah, the part that really worries me is the memory bandwidth. Only 128-bit GDDR5 compared to 448-bit GDDR3 for the GTX 260, with the same number of shaders…

It's competing with Juniper, which has the same bus width.

[Alexko]

Originally Posted by Jawed

It's competing with Juniper, which has the same bus width.

Yes, but it doesn't really seem to bother it, from what I've seen it's not any more bandwidth limited than Cypress.

On the other hand, I do remember a test of the GTX 285 which showed that when you overclock either the core, the shaders or the memory by 10%, you get roughly the same speedup, meaning the card was fairly well balanced. The GTX 260 had pretty much the same flops/bandwidth ratio (I think) so this drastic reduction in bandwidth might have a significant impact.

The card shown in the screenshot only has 60.16GB/s of bandwidth, compared to 76.8GB/s for the HD 5770.

Then again it has more or less the same flops/bandwidth ratio as the GTX 460 768MB…

[ECH]

Alleged GTX 480-512 mini review. I looked at the power consumption and wondered if it's fact or fiction.

[PatrickL]

Maybe time for a GF100 POST-release thread ?

[psolord]

Originally Posted by ECH

Alleged GTX 480-512 mini review. I looked at the power consumption and wondered if it's fact or fiction.

It is a FACT!

[TKK]

Originally Posted by ECH

Alleged GTX 480-512 mini review. I looked at the power consumption and wondered if it's fact or fiction.

strange, only the first page opens for me. Have the other pages been taken down?
How much power does it consume?

[Alexko]

Originally Posted by TKK

strange, only the first page opens for me. Have the other pages been taken down?
How much power does it consume?

The pages opens fine for me: http://en.expreview.com/2010/08/09/w...80/9070.html/6

Anyway, the system based on the 512-SP GTX 480 draws 204W over what the system based on the standard GTX 480 draws.

So the card itself probably draws close to ~ 300 + 200 = 500W, which is so huge that I wonder whether it's even possible.

[BRiT]

Their text on that page is below.

As we had expected, the stand-by power consumption of 512SP GTX 480 was 17W higher.

Under full load the GPU voltage of reference GTX 480 was 1.0V, while 512SP edition was 1.056V. Surprisingly, the full spec’ed GTX 480 sucked 644W power, which was 204W higher than 480SP GTX 480!

[Chalnoth]

Originally Posted by BRiT

Their text on that page is below.

As we had expected, the stand-by power consumption of 512SP GTX 480 was 17W higher.

Under full load the GPU voltage of reference GTX 480 was 1.0V, while 512SP edition was 1.056V. Surprisingly, the full spec’ed GTX 480 sucked 644W power, which was 204W higher than 480SP GTX 480!

That's quite a lot. But it seems a bit fishy to me, almost as if they'd just taken a 480SP part and found a way to unlock the remaining SP's. Because otherwise the additional SP's, which make up only a 6.7% increase in SP's, and thus less than a 6.7% increase in die area used, should not lead to more than a 7% increase in total power (at least, not without a corresponding significant increase in clock speed). Unless the additional SP's had major process problems that dramatically increased their power draw (and thus should have been disabled in the first place).

Although bear in mind that a couple of things could exacerbate this. First, if their load situation made use of a GPU-limited benchmark, higher performance on the GPU side would mean that the rest of the system would have to work harder, and thus may lead to higher power draw from other components. Second, this is total system power draw, which is increased further by the inefficiencies of the power supply. I don't think these things can fully explain the large discrepancy, though.

[chavvdarrr]

I doubt the cooler is able to dissipate 500W.

[mczak]

Originally Posted by Chalnoth

That's quite a lot. But it seems a bit fishy to me, almost as if they'd just taken a 480SP part and found a way to unlock the remaining SP's. Because otherwise the additional SP's, which make up only a 6.7% increase in SP's, and thus less than a 6.7% increase in die area used, should not lead to more than a 7% increase in total power (at least, not without a corresponding significant increase in clock speed).

Don't forget the higher voltage - now their conclusion doesn't make much sense, since almost certainly the higher voltage was needed for the higher clocks the card had (before it was downclocked for the review), not for the additional SM. That could easily cause another 10-20% increase in power draw. Still, all together (voltage, additional SM, PSU efficiency) that could only explain about half the difference of 200W. And that the cpu has to work harder should only make a tiny difference.

[Chalnoth]

Originally Posted by chavvdarrr

I doubt the cooler is able to dissipate 500W.

Power supply to the card of that magnitude would also be problematic. But it wouldn't be 500W, perhaps 400W-450W. Still, you're right. Way too high for operation.

[BRiT]

It is total system power draw they measured. Despite their system being quite beefy, I can't imagine their load-test scenario putting much strain on the CPU portion at all since it's Furmark. Though despite all the issues etc, a delta 204 watts is insane. The power difference was only 5.6% higher (1.056 vs 1.00 volt). I figure there's definite leakage and power issues in the core. It's definitely not a 'fixed' product.

The test bed includes Intel Core i5-750 (oc’ed to 4.0GHz), 4GB DDR3 dual-channel memory (oc’ed to 1600MHz), Windows 7 Ultimate 64-bit OS, and ForceWare 258.96 WHQL drivers.

Record the maximum value as stand-by power consumption 5 minutes after the PC entered Windows 7 system interface; Load the GPU with FurMark V1.8.2 Multiple-GPU mode, and then record the maximum value as loaded power consumption.

[Psycho]

Originally Posted by chavvdarrr

I doubt the cooler is able to dissipate 500W.

It is. Almost the same cooler is pretty good doing it's job on the 5970 Toxic, even overclocked: http://www.overclockersclub.com/revi..._hd5970/18.htm
78 degrees, when the card is sucking 230W more than a regular 5970. Or here, where the most powerhungry card is also the coldest: http://www.bit-tech.net/hardware/gra.../11/sapphire/8
That cooler deserves it's name

[ECH]

Originally Posted by mczak

Don't forget the higher voltage - now their conclusion doesn't make much sense, since almost certainly the higher voltage was needed for the higher clocks the card had (before it was downclocked for the review), not for the additional SM. That could easily cause another 10-20% increase in power draw. Still, all together (voltage, additional SM, PSU efficiency) that could only explain about half the difference of 200W. And that the cpu has to work harder should only make a tiny difference.

I was wondering if the 512 version was intended for a much lower clock rate.

[Tridam]

Originally Posted by Chalnoth

That's quite a lot. But it seems a bit fishy to me, almost as if they'd just taken a 480SP part and found a way to unlock the remaining SP's. Because otherwise the additional SP's, which make up only a 6.7% increase in SP's, and thus less than a 6.7% increase in die area used, should not lead to more than a 7% increase in total power (at least, not without a corresponding significant increase in clock speed). Unless the additional SP's had major process problems that dramatically increased their power draw (and thus should have been disabled in the first place).

Although bear in mind that a couple of things could exacerbate this. First, if their load situation made use of a GPU-limited benchmark, higher performance on the GPU side would mean that the rest of the system would have to work harder, and thus may lead to higher power draw from other components. Second, this is total system power draw, which is increased further by the inefficiencies of the power supply. I don't think these things can fully explain the large discrepancy, though.

There could be many reasons but most of the time the most obvious one (the results are not correct) is the right one. Obviously the reviewer wasn't puzzled enough by these strange results to dig a bit deeper there.

With one of my test systems, at some point while wondering why some results were strange, I noticed that the power supply I used (high-end CoolerMaster) was basically jumping from 135 to 155 watts (total system power draw).

[Chalnoth]

Originally Posted by Tridam

There could be many reasons but most of the time the most obvious one (the results are not correct) is the right one. Obviously the reviewer wasn't puzzled enough by these strange results to dig a bit deeper there.

True, that's a good point. It would be fantastically difficult for a video card to actually draw that much power, and especially to do it without failing almost immediately. More likely it's something wrong with the test.

[Moloch]

Common sense says they dun goof'd

[DavidGraham]

http://overclockzone.com/tor_za/year...top/index5.htm

the card is not competitive with HD5770 performance wise , that job is for GTS 455 they say .