NVIDIA GF100 & Friends speculation

[rjc]

From the australian atompic pc site's unboxing above is an image of the memory.

(was going to hot link it but decided that was bad manners)

Anyway from the photo the memory looks to be:
Samsung K4G1035FF HC04
The last 'F' and importantly the last '4' are very hard to be sure of. Decoding:
K - Samsung
4 - DRAM
G - GDDR5
10 - 1G
32 - x32
5 - Banks 16 maybe???
FF - 7th Gen
H - GDDR 170FBGA
C - Commercial
04 - 5Ghz

As you can see from the above last digit is very important, it is definitely not a 5 though indicating 4GHz memory.

Interesting that they appear to be using higher bins than they have to. Maybe this memory has some other attribute that they need other than speed.

 

[LordEC911]

As you can see from the above last digit is very important, it is definitely not a 5 though indicating 4GHz memory.

Interesting that they appear to be using higher bins than they have to. Maybe this memory has some other attribute that they need other than speed.

Looks like a 4 to me in the next pic, picture #19.

I see K4G10325FE HC04. Only difference is the E which is 6th Gen and it is 8banks.

Edit- Maybe they purchased the high binned GDDR5 before they knew they messed up the MC(Charlie) or that they couldn't use the full clocks due to a power/TDP wall and decide to leave it on there to allow users/AIBs to overclock them.

Samsung GDDR5 product sheet

 

[aaronspink]

The BFB0712HF is a custom fan as only the BFB0712HD is listed on their website. It draws 0.3A and is rated at 4000rpm and 42dBa. And the fan used by GTX480 draws 1.8A current

Going by other delta blowers in the BFB series, the fan should be in the 6-7k rpm range and the 60-65 db range.

 

[trinibwoy]

I hope they provide some free cheese with all that whine.

 

[CRoland]

I don't think their hand was forced, I think it was in the plan all along to do a shrink and X2. They did it with G71, G92 and GT200 so why not Fermi too?

What do you mean by shrink? Process shrink to 28nm? That would probably mean out early next year. However, at that time they would need to do more than just shrink to remain competitive.

 

[trinibwoy]

What do you mean by shrink? Process shrink to 28nm? That would probably mean out early next year. However, at that time they would need to do more than just shrink to remain competitive.

Yeah first batch of 28nm. G92 ended up with 40% higher clocks than G80 (though not 40% higher performance due to lower bandwidth). Of course, that comment was made before the latest round of evidence suggesting Fermi is going nowhere fast....

 

[seahawk]

Let's wait for the official reviews, shall we?

Sure. But I can say, it won´t be an eye opener either. Little faster, some more energy consumption, although idle load could be fixed. The cooling solution is decent. Not a Vapour X but also not a hair dryer.

 

[GZ007]

Going by other delta blowers in the BFB series, the fan should be in the 6-7k rpm range and the 60-65 db range.

The bigger problem could be that the fan is rated at 0.3A and 4000 rpm. Thats only 16% fan power.
We will need to wait for the reviews to know more.

 

[aaronspink]

The bigger problem could be that the fan is rated at 0.3A and 4000 rpm. Thats only 16% fan power.
We will need to wait for the reviews to know more.

no the fan is 1.8A. The marking on the fan don't lie.

 

[GZ007]

no the fan is 1.8A. The marking on the fan don't lie.

Yep, with my english it sounded strange. I justs want to say that even at something like 16% it will have 4000 rpm.

 

[stevem]

The BFB0712HF is a custom fan as only the BFB0712HD is listed on their website. It draws 0.3A and is rated at 4000rpm and 42dBa. And the fan used by GTX480 draws 1.8A current

Same 70mm fan as the GTX470, then. It looked a size bigger to me but smaller than the 92mm on the 5870.

no the fan is 1.8A. The marking on the fan don't lie.

It's going to be higher than 6-7k rpm at max current draw...

 

[Silus]

nVidia is struggling to beat ATI with 50% more of everything so why would a change of strategy help?

They just cant compete with ATI in efficiency so they have to go big every time. nVidia is losing due to being fundametally inferior and a change of strategy isn't going to change that.

LOL "fundamentally inferior" is very funny. You need to read less Charlie articles

Also, you are seriously suggesting that by re-thinking their strategy of a big chip for the high-end, nothing will change ?
As I stated before, if that same strategy still works for them, they will keep it. If not, they will change it. You really need to think outside of your red box and go back to another time where ATI was up and NVIDIA was down.

 

[GZ007]

Same 70mm fan as the GTX470, then. It looked a size bigger to me but smaller than the 92mm on the 5870.

It's going to be higher than 6-7k rpm at max current draw...

If u compare the size of the bearing and motor case on gtx480 and the pdf drawing than u see they are diferent in size.(also the rotor blades are much shorter on the gtx4xx fan)
It seems they did go with a stronger blower. U can increase also the torque not just rpm to get more air and pressure.
So those numbers in the pdf doesnt say to much.

 

[AlexV]

It is fake , look at the catalyst(or catalytic ) driver version .. here is an old legitimate one ..:]

I'm looking, what should i be seeing? The version is fine...

 

[DavidGraham]

I'm looking, what should i be seeing? The version is fine...

PcGamesHardware would never benchmark with an old driver , they will use 10.3 at least !

 

[Silus]

This looks like one of Charlie's GF100 has no hardware for tessellator bits. If true, the hilarity on Nvidia's part ..

So, even after reading about the architecture specs, you are still somehow supporting "Charlie's" FUD about that ?!...

 

[Ailuros]

LOL "fundamentally inferior" is very funny. You need to read less Charlie articles

Also, you are seriously suggesting that by re-thinking their strategy of a big chip for the high-end, nothing will change ?
As I stated before, if that same strategy still works for them, they will keep it. If not, they will change it. You really need to think outside of your red box and go back to another time where ATI was up and NVIDIA was down.

For the time being NVIDIA is anythng but "up" but that's besides the point.

How about not changing their strategy entirely but merely modify it? Something like keeping the single high end chip strategy, but develop a performance/mainstream chip in parallel with the high end chip?

 

[psolord]

PcGamesHardware would never benchmark with an old driver , they will use 10.3 at least !

Driver packaging version 8.712.3 belongs to Catalyst 10.3X mate.

 

[Silus]

For the time being NVIDIA is anythng but "up" but that's besides the point.

My point was "look back at a similar time when NVIDIA was down and ATI was up...and what happened next".

The silly argument that "they're down now and won't be able to come back" is truly...silly.

How about not changing their strategy entirely but merely modify it? Something like keeping the single high end chip strategy, but develop a performance/mainstream chip in parallel with the high end chip?

I never suggested an entirely different strategy. I did say though that if the same strategy that they've been using this far, isn't working for them anymore, they will change it, is it the way you suggested; simply going the route ATI did with the high-end represented by two chips on a PCB or simply completely different.

 

[AlexV]

PcGamesHardware would never benchmark with an old driver , they will use 10.3 at least !

It's good then that that driver is actually the latest one provided by ATI, the famous 10.3a!