The GT300/Fermi Thread

[Chumbucket843]

wow this whole time i thought it was saturday. though you guys were crazy.

i think the best counter for ATi is not more performance but lower price point. just like HD4000. its quite clear that is the ultimate goal of small die strategy. nvidia puts absolute performance first as it is good for their image. nv needs to get their power consumption higher to compete with ATi's dual gpu strategy. this makes a 250 watt single gpu comparable to a 300 watt dual gpu. its like nehalem EX of gpu's.

[kpablo]

wow this whole time i thought it was saturday. though you guys were crazy.

i think the best counter for ATi is not more performance but lower price point. just like HD4000. its quite clear that is the ultimate goal of small die strategy. nvidia puts absolute performance first as it is good for their image. nv needs to get their power consumption higher to compete with ATi's dual gpu strategy. this makes a 250 watt single gpu comparable to a 300 watt dual gpu. its like nehalem EX of gpu's.

That is not going to happen... Nvidia's chip are bigger (has more transistors) so is more expensive to make them and produce less chips per waffle then RV870.
Unless they want to lose money, lower final price then ATi cards is not going to happen.
And also, i am pretty sure, the day GF100 hot the street, ATI will lower their current prices.

[BababooeyHTJ]

no one is gonna be awake at 9pm pacific time, so lame

30 minutes to 9 PST

I love the will it play crysis and 000 tags.

[TurboDiv]

no one is gonna be awake at 9pm pacific time, so lame

I will be since I am in the PST time zone. Hehe, you go ahead and have a good night.

[randomizer]

no one is gonna be awake at 9pm pacific time, so lame

I will be I don't expect much info to actually show up though. I don't even expect real benchmarks.

[570091D]

no one is gonna be awake at 9pm pacific time, so lame

i will be, here at work pressing F5 to pass the time...

[sbattosai]

Charlie's own interpretation of the infamous NDA:
http://www.semiaccurate.com/2010/01/...anufacturable/

enjoy the read!

[jazkat]

Charlie's own interpretation of the infamous NDA:
http://www.semiaccurate.com/2010/01/...anufacturable/

enjoy the read!

you gotta enjoy charlies articles just for the laughs LOL

i find him abit crazy...i hope hes not right about fermi tho

[Redsand426]

you gotta enjoy charlies articles just for the laughs LOL

i find him abit crazy...i hope hes not right about fermi tho

I hope he is. 60% increase over the HD5870 is great, and he says that like it's a bad thing

[Boissez]

I hope he is. 60% increase over the HD5870 is great, and he says that like it's a bad thing

Yeah because it's 60% in cherry-picked benchmark scenarios. Real life performance should be significantly lower.

Anyway, as much as I despise his writing style he does raise some very relevant points. And he has been rather reliable as of late...

[Burninator]

lol the tags on this thread are hilarious

[SKYMTL]

Yeah because it's 60% in cherry-picked benchmark scenarios. Real life performance should be significantly lower.

Yeah, because everything is "cherry picked" when the outcome doesn't give you the results you were hoping for. /sarcasm.

[tool_462]

Yeah, because everything is "cherry picked" when the outcome doesn't give you the results you were hoping for. /sarcasm.

Beat me to it

Calling "cherry picked" on benches that aren't even out yet?

Feels like AMDZone in here sometimes.

[annihilat0r]

this is from a nvidia fanboy at DH, you gotta admit it's funny:

FERMI


ATI 5000

[GoldenTiger]

Articles are about to come up... http://www.hardforum.com/showthread.php?t=1487318 HardOCP shows that their article will be up at midnight EST (19 minutes from now).

[Boissez]

Yeah, because everything is "cherry picked" when the outcome doesn't give you the results you were hoping for. /sarcasm.

No I said cherry picked because of what was actually written in Charlies article:

When showing off the tessellation capabilities last week, Nvidia was very careful to show the Unigine/Heaven benchmark with the dragon statue, claiming it beat the ATI 5870 by 60%. While technically true, this number is purposely misleading on two accounts. First is that this is a best case for showing off tessellation and and only tessellation. If they were to show off any of the other Heaven tests, the margins would narrow significantly.

Now - i'm no position to assess whether the lack of hardware tesselation is going to have any performance impact - but *usually* emulation takes more processing cycles than hardcoded funcions.

[TurboDiv]

15 more minutes !!! Im tired of speculations. Nvidia better be ready or I'm jumping ship...

[SKYMTL]

No I said cherry picked because of what was actually written in Charlies article:


Now - i'm no position to assess whether the lack of hardware tesselation is going to have any performance impact - but *usually* emulation takes more processing cycles than hardcoded funcions.

His statement is semi-accurate which is what I was referring to in my statement. I'll be able to talk more in 15 minutes.

[jfromeo]

It's 5:45am in Spain and I have to go to work in 2 hours :P

[annihilat0r]

looks like the slides did their job, ppl who have them talk about how great Fermi is

[Periander6]

Charlie's own interpretation of the infamous NDA:
http://www.semiaccurate.com/2010/01/...anufacturable/

enjoy the read!

His usual spin seems strained this time. I'll bet Fermi is going to end up delivering after all.

[randomizer]

5 minutes before we find out what we probably already know

[GoldenTiger]

Article up! http://www.pcper.com/article.php?aid=858

http://www.pcper.com/images/reviews/858/wp-aaperf.jpg

[jfromeo]

Thanks Golden

[randomizer]

http://www.tomshardware.com/reviews/...x-11,2536.html