The official GT300/Fermi Thread

Printable View

Show 100 post(s) from this thread on one page

10-09-2009, 04:39 AM
onethreehill

Has this been posted?
Nvidia GT300 whitepaper disclosed
http://en.inpai.com.cn/doc/enshowcont.asp?id=7137
10-09-2009, 05:17 AM
Farinorco

Quote:

Originally Posted by Florinmocanu

the layering usually is
Application
Direct
Drivers
video card

If you do tesselation through cuda, you have

Application
Directx
Drivers
Cuda
video card

With hardware tesselation, the hardware already knows how to do the job, you don't need another software (in this case Cuda) to tell the shader how to do it, the shader just receives the data and does the math.

I thought they would simply decompose the DirectX tesselation calls into CUDA processors (formerly shader processors) instructions instead of some specific tesselator instructions in the drivers, since I've always supposed that the instructions set of each architecture wouldn't directly fit into the external interface that the drivers expose, but I'm not surprised if I'm wrong, since I'm more in the software side of things than in the hardware side :D

Quote:

Originally Posted by ajaidev

@Farinorco implementation of physx over cuda is not as efficient and effective as the real thing "PPU". Nvidia gpu loses more than a few shade units when it simulates the PPU

Yeah, that part I knew about and I have already mentioned it. What I don't know is how many CUDA processors they are going to have to use to do the tesselation, and therefore how much of an impact it will have on performance when used in real world apps, that's why I said that it can be seen when we have some real world data.

Of course it might turn into a huge performance impact :yepp:. Then... ouch!

Quote:

Originally Posted by onethreehill

Has this been posted?
Nvidia GT300 whitepaper disclosed
http://en.inpai.com.cn/doc/enshowcont.asp?id=7137

Yep, the whitepaper is accesible through the Fermi's website along some analysts white papers that NVIDIA has payed to write about Fermi, and I think it was in this forum where I got the link to the white paper.

Here is the link to the Fermi's site, and here the white paper itself...
10-09-2009, 05:58 AM
Florinmocanu

Quote:

Originally Posted by Farinorco

I thought they would simply decompose the DirectX tesselation calls into CUDA processors (formerly shader processors) instructions instead of some specific tesselator instructions in the drivers, since I've always supposed that the instructions set of each architecture wouldn't directly fit into the external interface that the drivers expose, but I'm not surprised if I'm wrong, since I'm more in the software side of things than in the hardware side :D

Well yeah, but how do they decompose DirectX tesselation calls ? through Cuda, the aditional software layer i was talking..

While if you have hardware tesselation, you just send those calls directly to the tesselators, no need for intermediaries.

At least that's how i assume it will be done, we will wait to see the reviews in dec/ian for more info on this issue.
10-09-2009, 06:14 AM
BenchZowner

Quote:

Originally Posted by Florinmocanu

assume

KEYWORD

p.s. I really enjoy reading this topic.
Easily puts the latest cinematic drama & trash-level comedies to sleep :D
10-09-2009, 06:50 AM
Florinmocanu

well, everyone is "assuming". We do not know any facts, so we can just "assume". No harm in doing that.

But if you have some "hard facts" you want to contribute to clear some matters, please do, we would like that.
10-09-2009, 06:58 AM
BenchZowner

Seems like assuming the worst is easier than good nowadays.

How naive could somebody be to think that nVIDIA or any other manufacturer would decide upon something without weighting the effect of their decision with several performance tests.

Anyways... I stopped giving a f*, why would I want to share any information now ?
No matter what people could show or not, it would start a flame war and a conspiracy theory as usual.

No way of stopping this negative assumptionism, and TBH I wouldn't like to see this stop, it's quite entertaining :D
10-09-2009, 07:04 AM
Florinmocanu

Well, the fx series did not properly support the DX9 standards, and it was shown later that it did poorly in DX9 games, so i think nvidia already showed it can make such a "bad" decision, overlooking the full implementation of a Directx API.
10-09-2009, 09:29 AM
JohnZS

Quote:

Originally Posted by Florinmocanu

Well, the fx series did not properly support the DX9 standards, and it was shown later that it did poorly in DX9 games, so i think nvidia already showed it can make such a "bad" decision, overlooking the full implementation of a Directx API.

Very true, I belive that was because nVidia put all their eggs into the OpenGL and Cg basket where the FX series did do quite well. I recall that Tomb Raider Angel of Darkness (DirectX9 game) brought the FX cards to their knees and even ran choppy in some areas on the Radeon 9700, however using the Cg shader path it ran quite well on the FX (almost as good as the 9700).

I might be wrong though...

As for the tesselation ATi used to do hardware tesselation on the Radeon 8500 (Truform I belive it was once called), this was really good in Return to Castle Wolfenstein and Serious Sam, however the Radeon 9700 series cards dropped the hardware tesselation and opted for software Truform II, this killed performance... it would be interesting to see how the GT300 copes with tesselation as IMHO I can see nVidia having the edge with DriectX11 compute, but ATi having the edge with the tesselation features.

John
10-12-2009, 06:12 PM
Chumbucket843

fermi to be in petascale supercomputer.

http://www.brightsideofnews.com/news...rcomputer.aspx
10-18-2009, 02:46 AM
mindfury

http://www.evga.com/forums/tm.asp?m=...ey=&#100973314

Quote:

I was forced to use usb monitor as GPUs haven't any video output (this engineering samples of Fermi are Tesla like, but they have 1.5GB of memory each like GT300 will).
Because of the new MIMD architecture (they have 32 clusters of 16 shaders) i was not able to load them at 100% in any other way but to launch 1 F@H client per cluster and per card. Every client is GPU3 core Beta (Open MM library). I supose it is much more efficient then previous GPU2. In addition they need very little memory to run. Having 16GB of DDR3 and using Windows 7 Enterprise I've managed to run 200 instances of F@H GPU and 4 CPU (i7 processor HT off). The 7th card is not fully loaded. This could also be an issue with EVGA X58 mobo.
I use together two Silverstone Strider PSU's 1500W each that is probably too much but
now I experiment with overclocking (cards are factory unlocked). Max power consumption
I've noticed was 2400W.
The whole system is cooled by my own construction of liquid CO2 which is heavy an inconvenient and I have to supply a new cylinder every 5 days.

Quote:

It's probably actually closer to:
(30) 275's + CPU now a push = (7) Fermi

(30) 275's / (7) Fermi = 4.285 X as fast...

Power considerations:
He has (2) 1500W PSU's, using 2400 watts.

So lets say his i7 CPU uses 150 watts...

2400 watts - 150 = 2,250 GPU watts total.

2,250 / 7 Fermi = 321 watts per GPU...
10-18-2009, 03:06 AM
Helloworld_98

Quote:

Originally Posted by mindfury

http://www.evga.com/forums/tm.asp?m=...ey=&#100973314

then add in energy used by the motherboard, memory, hard drive, and it comes to about 313w per card.

those fermi's use an awful lot of energy, we're definitely not seeing a dual GPU card fermi until a die shrink, even then energy usage might be a bit high.
10-18-2009, 04:10 AM
trinibwoy

Heh, so I guess you guys figure the PSU is 100% efficient too :) In any case there's no reason to believe him until he provides some kinda proof.

Quote:

A little bird told me 4 SMP, 31 GPU, and the rest are CPU clients (at least in the last 7 days).

http://foldingforum.org/viewtopic.ph...=11717#p114890
10-18-2009, 08:45 PM
saaya

Quote:

Originally Posted by mindfury

http://www.evga.com/forums/tm.asp?m=...1&key=�

Quote:

The whole system is cooled by my own construction of liquid CO2 which is heavy an inconvenient and I have to supply a new cylinder every 5 days.

:rofl:

you might not get the hint, but hes just bsing to make fun of people who desperately wait for gt300 :D
10-18-2009, 09:47 PM
gOJDO

Quote:

Originally Posted by saaya

:rofl:

you might not get the hint, but hes just bsing to make fun of people who desperately wait for gt300 :D

Today it is hard to be stupid! The competitiion is very strong...:yepp:
10-19-2009, 03:17 AM
zalbard

Quote:

Originally Posted by saaya

:rofl:

you might not get the hint, but hes just bsing to make fun of people who desperately wait for gt300 :D

LOL, and I actually believed it... Was fishy, though, he mentioned USB monitor also, very weird.
Funny btw, quite a few Nvidia fanboys stopped posting almost completely on these forums. Perhaps they're just banned from the news section, though...
10-19-2009, 03:33 AM
VoodooProphetII

Or perhaps they just decided not to take part in the discussion which would be the better choice. I mean we have som many "know it all" people on this forum.
10-19-2009, 03:38 AM
gOJDO

Quote:

Originally Posted by VoodooProphetII

Or perhaps they just decided not to take part in the discussion

Fans may stay out of the discussion...
But fanboys...no chance! They will enter every discussion regarding their beloved brand(not prouct) or its competition and will blindly deffend/attack the beloved brand/cometition in every opportunity they will have, without facts, arguments and logical explanations.
10-19-2009, 04:49 AM
VoodooProphetII

I don't know. I just see a fair amount of assumptions from both sides of the fence.

But as BenchZowner mentioned in an earlier post, it is easier to assume the worst.
10-19-2009, 05:12 AM
XSAlliN

Nvidia's CEO, gets informed about ATI's Evergreen series :shrug:
10-19-2009, 05:19 AM
WeeMaan

Quote:

Originally Posted by XSAlliN

Nvidia's CEO, gets informed about ATI's Evergreen series :shrug:

old...
10-19-2009, 05:24 AM
N19h7m4r3

Lord some this guy seems like the definition of fanboy...

Quote:

Originally Posted by Reason4termination

So evergreen is the fastest single chip right now. Big deal.

While ATi is continously resort to smaller fab (which has a limit) and faster mems, nVidia is actually mitigating the limit with newer architecture and efficient processing. Once their next gen grabbed the foothold on the game development scene, ATi will come in following the trend of the architecture (much like they did with G80 and its unified shader). Monkey see monkey do with the architecture, and fail at it on 2 generations.

GDDR4, GDDR5, 55nm. etc.. They resort to those brute force to have higher clocks so that they can counter nvidia's architecture. nVidia can integrate GDDR4, GDDR5, 55nm fab etc if they wanted to. But they do not because that will leave the competetion to the dust. DX10.1 is a joke. and we have yet to see DX11 titles proliferating. Because whatever DX9 consoles release, that is the limit to what the PC games graphics' have.

Gotta love youtube comments.
10-19-2009, 05:27 AM
Jamesrt2004

Quote:

Originally Posted by N19h7m4r3

Lord some this guy seems like the definition of fanboy...

Gotta love youtube comments.

seen that it made my day lol :ROTF:
10-19-2009, 06:13 AM
OutSider

http://img520.imageshack.us/img520/1365/54632720.png describes this guy
10-19-2009, 06:42 AM
Vit^pr0n

Quote:

Originally Posted by OutSider

http://img520.imageshack.us/img520/1365/54632720.png describes this guy

Now that is hilarious.

I'll be saving that for future purposes :ROTF::clap:
10-19-2009, 07:56 AM
trinibwoy

Heh, looks like a lot of guys are Nvidia fanboys and don't know it yet.

Show 100 post(s) from this thread on one page