NVIDIA GF100 & Friends speculation

But you shouldn't ignore power consumption and realated headroom in possible voltage adjustment. GTX460 at 700MHz consumes as much power as HD5870 at 850MHz. I think GF104 would hit the 225W limit sooner than performance level of mildly overclocked HD5870.

Will nVidia go beyond 225W with single-chip GF104? They had to be quite desperate to do that...
That's not what the results look like. Even OC GF104 stay well below the power draw of HD5870 (for instance here: http://ht4u.net/reviews/2010/zotac_geforce_gtx_460_amp/index12.php). By my guesstimate a 8 TPC GF104 running at 800/1600/1000Mhz would have quite a similar power draw to HD5870, while being slightly slower.
 
http://www.guru3d.com/article/msi-geforce-gtx-460-hawk-review/

Nice card!

But it can truly draw the juice when the voltage and clocks get bumped.

http://www.guru3d.com/article/msi-geforce-gtx-460-hawk-review/20

[FONT=verdana,geneva]With the overclock our temperature now rises to roughly 77 degrees C under load. DBa levels rose to 44 DBa under full stress and power consumption went up from 342 Watt towards 435 Watt. When we add full maximum GPU/AUX/MEM voltage we noticed the power draw rising to 503 Watt (for the entire system, crazy really) But let's re-check some numbers shall we?[/FONT]
That 342 Watt number refers to peak, out of the box, consumption. Unless I'm reading the article incorrectly.

http://www.guru3d.com/article/msi-geforce-gtx-460-hawk-review/8

Not many will try to overclock that high, let alone succeed, but it's nice to see such a robust card.
 
That's not what the results look like. Even OC GF104 stay well below the power draw of HD5870 (for instance here: http://ht4u.net/reviews/2010/zotac_geforce_gtx_460_amp/index12.php). By my guesstimate a 8 TPC GF104 running at 800/1600/1000Mhz would have quite a similar power draw to HD5870, while being slightly slower.
Be careful in such comparisons. Processes change and some of the differences can just be due to the point in time that they were profiled.
 
Hmmm, and they claim it draws only 50w, 8w less than the GT220. Will be interesting to see how it fares against that card as it has similiar specs. It's not a full GF108 though. According to Nvidia's product images the 96 shader mobile parts and the 48 shader GT415m share the same chip.
 
TDPs :

400M-9.jpg


http://www.hardwarecanucks.com/forum...ntroduced.html

something isn't right in that table , wasn't GTX 480M a 100w part ?
 
...

On the Asian market where warranty is worth how much exactly?

I have no idea :) But PcLab.pl suggested that that card will be initially available on Asian market but later will be available in Europe as well. I don't know if thats true but i have seen many Colorful cards on Polish market so this is not Asian exclusive brand.
Anyway i was just trying to show to Arun that higher clocked models are/will be available.
This is really strange because it looks like that this particular Colorful board will be at least as fast as GTX 470 if not faster. I am wondering how much will it cost :)
 
something isn't right in that table , wasn't GTX 480M a 100w part ?

nVidia marketing and nv-loving sites like hardwarecanucks falling for it. Sure, they didn't lie here, but they sure as heck didn't tell the whole truth.

the GPU itself might have a TDP of 70W, but the whole MXM module+GPU has a TDP of over 100W (see what they did there? "w/o MXM module")

http://www.notebookcheck.net/Review-NVIDIA-GeForce-GTX-480M-Fermi-in-the-Clevo-D901F.32251.0.html

it's like putting notebook weights on your website without the battery.

5870M has a 55W TDP total (GPU+MXM.)
 
the GPU itself might have a TDP of 70W, but the whole MXM module+GPU has a TDP of over 100W (see what they did there? "w/o MXM module")

5870M has a 55W TDP total (GPU+MXM.)
oh , I see thanks for the clarification ..

On a lighter note , the site you mentioned states this :

Beware, the TDP values of ATI cards usually does not include the whole MXM board, but the chip itself only. Therefore, the power consumption is not easily compareable.
So that means ATi does the same too ?
 
oh , I see thanks for the clarification ..

On a lighter note , the site you mentioned states this :

So that means ATi does the same too ?

Yes, I mentioned it there as well. the 5870M has a TDP of 50W according to NV's slide, but as far as I'm aware (and reviews show) the package Wattage of the 5870M is 55W. i.e. on a Crossfire 5870M setup the total TDP is rated at 110W.
 
So I guess Nvidia mobile chips are 30%~40% more power hungry than ATi chips .

that is bad for OEMs , I don't think Optimus would be enough to promote them .
 
So I guess Nvidia mobile chips are 30%~40% more power hungry than ATi chips .

that is bad for OEMs , I don't think Optimus would be enough to promote them .

IF they already have a 150W power brick, who cares? It's just about that the performance can be tapped when called upon. NV is only feeling the backlash on it's own campaign of focusing on better better performance per Watt/mm2 in this community.

Optimus is doing an excellent job and if your products are power hungry, the efficiency just gets better (85% power savings!)
 
There aren't any additional SMs for redundancy in GF100, so I doubt there are on GF104 and especially GF106.

Not additional SM's. The practice of disabling SM's in general is a form of redundancy. Each GTX 480 and 460 had a "redundant" SM. But yeah, I don't think GF106 has 5 SM's either. Sure is big though.

Personally I'd expect 384SP/825MHz/1GB 4.5GHz GDDR5 with the same 215W TDP as the GTX 470, although maybe that's too optimistic overall.

Yeah, that's a tad optimistic. My vote is for 384SP/775Mhz/1GB 4Ghz GDDR5. That's still good for a 25-30% perf increase.

the GPU itself might have a TDP of 70W, but the whole MXM module+GPU has a TDP of over 100W (see what they did there? "w/o MXM module")

Haven't kept up with mobile stuff. Why is the "+MXM" delta for Nvidia's chips so much higher?
 
Haven't kept up with mobile stuff. Why is the "+MXM" delta for Nvidia's chips so much higher?

If it's about the whole setup including coolers, the GTX480M requires at least two fans, that include some extra power usage right there.
 
But his point is irrelevant; many cards from both vendors with TDPs well below 225W have two power connectors. Case in point: the HD5850.
>225W products require 6-pin + 8-pin (e.g. GTX480). Sub-225W boards are fine with two 6-pin power plugs. Hypotetical GF104-based board going over 225W would require 6-pin + 8-pin like Fermi. 2 6-pins are common in mainstream, but 6+8-pin aren't...
That's not what the results look like. Even OC GF104 stay well below the power draw of HD5870 (for instance here: http://ht4u.net/reviews/2010/zotac_geforce_gtx_460_amp/index12.php). By my guesstimate a 8 TPC GF104 running at 800/1600/1000Mhz would have quite a similar power draw to HD5870, while being slightly slower.
The review at guru3d shows, that factory OCed 1GB models of GTX460 are already ower 200W...
 
Nice, they snuck GF108 in as well: 48 shaders, 2GB RAM, 900Mhz
http://www.nvidia.com/object/product-geforce-gt-420-oem-us.html
I very much doubt they created 3 SKUs with 96 SPs, all of them cut-down from a 192 SP or even 240 SP chip. It's much more likely GF108 is a 96 SP part which is cut-down to 48 SP in GT415M. The texture numbers look... wrong though.

no-X: Argh, yes. That's an embarassing brainfart, thanks; but anyway if GTX 470 with a 215W TDP and arguably even higher peak power doesn't need the 8-pin, then I'm not sure we should be overly worried about a potential GTX 475.
 
Back
Top