NVIDIA GF100 & Friends speculation

[neliz]

FWIW, this article here http://ht4u.net/reviews/2010/nvidia_geforce_gtx_470_480_cebit/index5.php says nvidia is NOT planning on a low-end fermi derivative, because it would be pointless (tell that to AMD...). That doesn't exactly inspire confidence on perf/mm^2 for lower end parts...
Though this really only affects the bottom of the barrel, even gt220 isn't low end according to that anymore. But it should mean we won't see a fermi part with 32SP (since gt220 was already 48SP), if they only scale SMs across GPCs and GPCs it could mean the lowest end fermi part planned is something along 1 GPC / 2 SMs (with 32 "cores" each), which would smoke Cedar but look awful compared to Redwood, so maybe it will be something completely different after all...
BTW I'm looking forward to see clocks for GTX480 as they were supposed to be leaked today (or rather, nvidia is supposed to give this information to partners today ).

NV's Fermi presentation from last year specifically talks about top-to-bottom Fermi and I was under the assumption that there were 5 members of the Fermi family, excluding the Fermi2 part.

 

[rpg.314]

So , I guess the comparison between version 1.0 and 1.1 is completely invalid , right ?

Only if you keep the gpu the same, then no. Otherwise with different gpu's, it makes sense to use the same version.

 

[fellix]

GTX470 stock cooling

Direct contact heat-pipe design, 12V/1.8A fan... the thing is going to be really hot!

 

[mczak]

NV's Fermi presentation from last year specifically talks about top-to-bottom Fermi and I was under the assumption that there were 5 members of the Fermi family, excluding the Fermi2 part.

Interesting. Well apparently the no-low-end part came from a nvidia representative, though of course he could have been wrong too (he also said they'll do a dual-gpu card, and I'm not convinced there neither). Or maybe they have indeed dropped a part (with 5 members you mean chips, right, not cards?), certainly without bottom-of-the-barrel part I can only really see 4 chips max. Also, I find the "very soon" part for those follow-up chips interesting, though it remains to be seen what that really means, without any leaks whatsoever until now...

 

[trinibwoy]

GTX470 stock cooling

Direct contact heat-pipe design, 12V/1.8A fan... the thing is going to be really hot!

Actually, doesn't that mean it's going to be cool?

Besides, the 2GB 5870 has a similiar direct contact heatpipe design and is a 8+6 pin. Does that mean it's going to be OMG HOT! as well?

http://www.xtremesystems.org/forums/showpost.php?p=4276602&postcount=2200

Hopefully beefier cooling translates into less noise. The cooler itself has no direct bearing on power consumption - the card is limited to 225w after all.

 

[nAo]

Fermi being supposedly faster at rendering highly tessellated geometry could also be a byproduct of higher efficiency at rasterization and shading of very small primitives.

EDIT: With help from the 'usual suspects': higher setup rate, more tessellation engines, etc..

 

[jimbo75]

HD5770 is considerably more than half as fast as HD5870 on the benchmark.

I'm not sure what performance is like with HD5770 on the worst-case close-up of the dragon though.

Jawed

I ran the benchmark a few times with my 5770 at 1680x1050, same settings otherwise.

With tesselation on and no AA, it scored 25fps while with tesselation off and 8xAA it scored 23.6fps, so there is more of a hit with 8xAA than with tesselation on the 5770. Thats the overall benchmark fps but I can check the dragon later at any settings you need to know, except I'm stuck with either 1680x1050 or 5040x1050.

 

[GZ007]

Actually, doesn't that mean it's going to be cool?

Besides, the 2GB 5870 has a similiar direct contact heatpipe design and is a 8+6 pin. Does that mean it's going to be OMG HOT! as well?

http://www.xtremesystems.org/forums/showpost.php?p=4276602&postcount=2200

Hopefully beefier cooling translates into less noise. The cooler itself has no direct bearing on power consumption - the card is limited to 225w after all.

That 12V/1.8A fan will eat 20+W alone. Thats a very powerfull fan for 70mm.

The 92mm diameter 5870 blower is rated at 12V/0.8A at full power, and it resembless a vacuum cleaner in that stage.

 

[neliz]

Actually, doesn't that mean it's going to be cool?

Besides, the 2GB 5870 has a similiar direct contact heatpipe design and is a 8+6 pin. Does that mean it's going to be OMG HOT! as well?

fan on the Cypress family is 0.8A/12v. you do the TDP calculation. Why did they opt for a near 1Amp cooler on the 2900XT and not something like 0.5? Does nV care if the card stays within operating limits or bragging rights because it only becomes 70 degrees under load?

 

[Jawed]

I ran the benchmark a few times with my 5770 at 1680x1050, same settings otherwise.

With tesselation on and no AA, it scored 25fps while with tesselation off and 8xAA it scored 23.6fps, so there is more of a hit with 8xAA than with tesselation on the 5770. Thats the overall benchmark fps but I can check the dragon later at any settings you need to know, except I'm stuck with either 1680x1050 or 5040x1050.

Interesting results. It'll be interesting to see the results for dragon close-up with/without tessellation. Have to admit I'm surprised by the 8xMSAA performance - with so few triangles in the benchmark, overall and assuming the engine sensibly only uses MSAA where/when it's needed.

It's a shame the B3D article doesn't show HD5770 results as well. Oh well.

Jawed

 

[trinibwoy]

That 12V/1.8A fan will eat 20+W alone. Thats a very powerfull fan for 70mm.

The 92mm diameter 5870 blower is rated at 12V/0.8A at full power, and it resembless a vacuum cleaner in that stage.

fan on the Cypress family is 0.8A/12v. you do the TDP calculation. Why did they opt for a near 1Amp cooler on the 2900XT and not something like 0.5? Does nV care if the card stays within operating limits or bragging rights because it only becomes 70 degrees under load?

This is the 470 cooler we're talking about right, with rumoured maximum board power of 225w - lower than the GTX 280? I guess I'm missing why a powerful fan and more elaborate heatsink than what we saw on the GTX 280 is cause for alarm. Or are you guys trying to say that a 225w 470 board will dump more heat than the 236w TDP 280? If I'm not mistaken that sort of defies the laws of physics.

 

[AlexV]

It's a shame the B3D article doesn't show HD5770 results as well. Oh well.

Jawed

Welp, there's no 5770 on the premises. I'll eventually snag one (hopefully sooner rather than later), and by the time that happens we might even have a better framework to do this sort of investigation in place (Fermi will be out too by then, one would assume).

 

[Squilliam]

That 12V/1.8A fan will eat 20+W alone. Thats a very powerfull fan for 70mm.

The 92mm diameter 5870 blower is rated at 12V/0.8A at full power, and it resembless a vacuum cleaner in that stage.

It'll only eat 20+W if its spinning at 100%. Since its a very powerful fan obviously its enough to both keep the chip cool and lift the skirt of any lovely tech show lady who happens to walk by. Surely they would never let it spin at greater than 40% power?

 

[Dave Baumann]

Or are you guys trying to say that a 225w 470 board will dump more heat than the 236w TDP 280? If I'm not mistaken that sort of defies the laws of physics.

Well, the hotter the chip the more power it will draw. Keeping it cool will reduce power (at our operating temps we saw about a correllation of 1W to 1 degree; this may be much more with Fermi), but the cooler it is the more air it may need to move.

 

[trinibwoy]

Well, the hotter the chip the more power it will draw. Keeping it cool will reduce power (at our operating temps we saw about a correllation of 1W to 1 degree; this may be much more with Fermi), but the cooler it is the more air it may need to move.

Perhaps, but that doesn't answer the question if why it should be harder to keep it cool in the first place (which is what I believe neliz and GZ007 were getting at). Or in other words given a particular level of power consumption how do you define an inherently "hotter" chip?

 

[neliz]

Perhaps, but that doesn't answer the question if why it should be harder to keep it cool in the first place (which is what I believe neliz and GZ007 were getting at). Or in other words given a particular level of power consumption how do you define an inherently "hotter" chip?

It could be (hint) by the higher voltage required to run it in the first place? If you are missing your targeted low voltage point, you'll end up with a chip requiring a higher voltage to run and thus generate more heat.

 

[trinibwoy]

Power is converted to heat neliz, not voltage. Physics 101. No matter what voltage targets they missed the result has to come in under their total power budget.

 

[itsmydamnation]

yes because volts X amps like doesn't equal watts or anything..........

i see you did a ninja edit as i was typing

who knows what there total pwoer budget was in the first place?

 

[mczak]

Well, the TDP might be at a low temperature. For the sake of the argument let's say the specified TDP (which also includes power of other chips, hence we don't know if the chip power is actually lower than GTX280, but at least it should be similar) is at 50 degrees celsius (but was 95 degrees for GTX280), and your power increases by 2W for each degree. So at 90 degrees you'd already have 80W more TDP hence you'd need a ridiculous cooler to cool it down to 50 degrees keep it within TDP (even though that ridiculous fan will of course also increase TDP).
Now I'm not saying this is the case (would be disaster really), just from the theoretical side it would be possible.
Might be more likely though they just got a better deal on that fan .

 

[trinibwoy]

yes because volts X amps like doesn't equal watts or anything..........

i see you did a ninja edit as i was typing

who knows what there total pwoer budget was in the first place?

What ninja edit? I simply added the obvious fact that no matter how you do your fancy multiplication it can't exceed the board's input power. Or was that not obvious?

Well, the TDP might be at a low temperature.....

Ah, this is probably what Dave was getting at. Yes, that would be a tragedy of comedic proportions.