The GT300/Fermi Thread

[jaredpace]

It's a fairly accurate guess in that shop. I'd put it at about 85-90% range.

[saaya]

perhaps thinking too far ahead is why the arch performed poorly.

perhaps? of course it is!

The reason I said R600 was a pretty scalable design is based on how it's progressed

hmmm well i so far saw it as ati doing a bad job at creating r600 and doing a great job at fixing r600

Wow, a devastating article on Fudzilla about Fermi! I thought I would never see the day... (he was really insisting on a Dec 2009 paper launch for a while)

well he insisted on the nov launch, then he insisted on a december launch... i guess he got tired of people mocking him and is now spanking nvidia for playing games with him

this was such a typical nvidia pr move, and this is so exactly why no press in the world likes and trusts nvidia heh... i wonder if they will ever change

the Q3 part (aka the Daddy), is going to be competitive with the HD5970.

well it better be considering it comes out almost a year after 5970 :P

nVidia have hit the proverbial nail on the head with this one, think of it as more of a G80 than a FX.

huh? G80 stomped ati into the ground and offered a huge performance boost over the previous generation... fermi is already late, will come out even later, will not stomp ati, and will not offer a huge performance boost... i dont see any corelations :P

Whether or not the card is a success, you have to hand it to nVIDIA.
This card has caused the most drama in the history of computing, i'm also very impressed with the secrecy surrounding this graphics card.
This really is some grade A marketing.

rofl, what? are you serious? bsing the investors, end users, press, holding up DIY prototypes claiming they are the real deal, and not releasing any information whatsoever is good marketing?

Personally, i have to buy it as my 280 GTX is blowing up and i refuse to go ATI as i had a X800XT P.E on pre-order for 8 months, then being told it's discontinued..

what does the 800xt pe have to do with this? it was a slightly overclocked version of a card that was available everywhere... that was some terrible attempt at marketing as well, although im not even sure that was intentional... and i dont get why so many people complained about the availability of the 800xtpe and the 850xtpe... i wanted an 850xtpe and bought not only one but two... all while several websites claimed it wasnt available anywhere and impossible to get

Oh, it's not just that single excuse, i was a very loyal/happy ATI customer since radeon 7000.
But after a POS 9800, two overheating 9800XT's, my non-existant X800XT P.E and changing the control panel to that God awful catalyst control center.
I lost the faith and love.
Don't get me wrong, i have a hd4670 in a HTPC and its brilliant at doing that, but it would take a lot for me to consider them for performance related needs again.

If this GF100 is the equivalent to two 280 GTX's in SLI i'll be more than happy with that.

yes... i HATE CCC... god i hate it so much...
but just when i was about to give nvidia another try, they switched to a browser style driver interface...
sigh... i dont see why its so hard to NOT SCREW UP A WELL WORKING DRIVER INTERFACE...
i can see why many companies struggle to create one, but having one and screwing it up... amazing work there nvidia and ati

What's up with the weird contradictions, one source says Fermi is going to be superfast, and the next says it's only like 20% faster then the 5870.

I wish we could be done with it already, it's getting on my nerves.

i think they all have the same infos, its just that some consider faster than 5870 SUPERFAST and others dont

Amazing gaming performance? Even a GTX 260 has pretty much amazing gaming performance except Crysis. You can play pretty much any game at any resolution you want. Except Crysis. BTW; when discussing game performances the two words "Except Crysis" are used so much that I propose one of the following against time wasting:

1. Use a short hand notation. E.C. should do fine.
2. Don't count Crysis as a game.

totally agree!
i got 2 260s in sli and i gotta say...i keep wondering why i have a second 260 in the rig
sure, it helps to keep the fps steady above 60 in far cry2 in 1920x1080 with max eye candy... but i dont feel a difference... the only difference i feel are the weird sli bugs and stuttering and flickering every now and then... which i didnt have with two G92s in sli btw!

so yeah, the real question right now is... do we really need a faster gpu right now? if nvidia or ati would ask me what i want, id say:
get me G92/rv770 performance at below 20W, or G200/rv770 performance at 99$

[~CS~]

so yeah, the real question right now is... do we really need a faster gpu right now? if nvidia or ati would ask me what i want, id say:
get me G92/rv770 performance at below 20W, or G200/rv770 performance at 99$

4-6 month = 5770 100$

[hurleybird]

yes... i HATE CCC... god i hate it so much...
but just when i was about to give nvidia another try, they switched to a browser style driver interface...
sigh... i dont see why its so hard to NOT SCREW UP A WELL WORKING DRIVER INTERFACE...
i can see why many companies struggle to create one, but having one and screwing it up... amazing work there nvidia and ati

QFT. It baffles the mind.

[saaya]

4-6 month = 5770 100$

without any competition from nvidia? i dont know... :/

QFT. It baffles the mind.

especially since they could have supported both, the old and the new, but after some time they removed and blocked the old interface...
just dont support it, and if something is broken, then its broken... but how can you castrate your product and remove a feature so many customers use?

[Nedjo]

Wow, a devastating article on Fudzilla about Fermi! I thought I would never see the day... (he was really insisting on a Dec 2009 paper launch for a while)

If the article has any truth in it, then NV30 might be repeating itself with Fermi. I hope that Fud gets it wrong this time as well!

well it's obvious now that NV has fed Fudzilla with launch FUD as a base for their articles. They end up ridiculous (as I've warned them) and naturally they can't feel comfortable now... sobering is never easy and w/o consequences

[Nedjo]

I hope the writer realizes they have to ship the cards/parts to board partners quite some weeks before the launch date.

or... you can cut number of partners to feed remaining?!

Fudzilla is writing this about Nvidia? I did not see that coming

it was a matter of time! How would you feel if NV was telling you that GT300 will be out in Nov/Dec, and then turned up being spit-wall protecting the real source of FUD - NV!

Performance is likely to be ~20-25% faster than HD 5870... but the Q3 part (aka the Daddy), is going to be competitive with the HD5970.
John

What's "Daddy"?

Thats not quite what I'm seeing yet. Given it's rumored heat and power and die size, being only 20% after being 6 months late is a bit dissappointing

that's not how Hohn and Fiskov are seeing it

Whether or not the card is a success, you have to hand it to nVIDIA.

hand them what? the price for the most postponed HW after R600, LRB, and Boeing Dreamliner

being loyal to a manufacturer is retarded..

those are harsh words! that kind of loyalty is actually quite common... it's almost natural for the people!

When I dig dipper in to my choice of loyalty I can come out with bunch of reasons, but key ones are the fact that hardware that I'm loyal to I've cached out from my own money in the years of grates crises, and that HW served me without glitch for many years... when t I've gathered more money I've again went for AMD... and when I've got in the line of business that allowed "hardware promiscuity" I've switched sides from time to time, but eventually always returning to the roots

If 20% faster means like 10fps more, (30/36fps for instance) I would call that a serious failure.

I would call that pure and simple math

[InCredible]

those are harsh words! that kind of loyalty is actually quite common... it's almost natural for the people!

When I dig dipper in to my choice of loyalty I can come out with bunch of reasons, but key ones are the fact that hardware that I'm loyal to I've cached out from my own money in the years of grates crises, and that HW served me without glitch for many years... when t I've gathered more money I've again went for AMD... and when I've got in the line of business that allowed "hardware promiscuity" I've switched sides from time to time, but eventually always returning to the roots

didn't mean to sound harsh but its true..and its not the fact that you switch from time to time..i spend money on whatever is worth it at the time..usually the best of either side..but when some people are so loyal as to throw their hard earned money on an inferior product just because of the badge it wears, that i think is retarded...both manufacturers have their ups and downs.. and when you jump on a new tech/arch theres going to be issues..even as great as the 8800gtx was it still had issues on launch..same goes for the 4870x2(ive owned both around launch time and had issues with both) however both pulled through after some driver updates and performed great...however i just killed my 4870x2 with a water accident so now my next card will be whichever seems best for the money..i might possibly wait til fermi release since its not TOO far away plus i dont have 400-500 to drop on a card right now..but ill consider both to see which will get my hard earned money...

[trinibwoy]

Depends on how you define "inferior product". If it's just based on benchmark graphs, then yeah that's easy. But in reality people place value on experience, familiarity and peace of mind. That's what brand loyalty gets you.

[Pillo-kun]

as i still own an 8800gtx i find myself in more and more trouble than when 8800gtx was new. I actually feel that nvidia only concentrates on fixing the newer cards and don't care about the older ones. Right now i cant play Mw2 when my tv set is connected cus i get blue screen.. I dont trust either of companies cus they only wants to make money...

[Aerwidh]

it was a matter of time! How would you feel if NV was telling you that GT300 will be out in Nov/Dec, and then turned up being spit-wall protecting the real source of FUD - NV!

Point taken, guess even FUD has to face reality at some point

Depends on how you define "inferior product". If it's just based on benchmark graphs, then yeah that's easy. But in reality people place value on experience, familiarity and peace of mind. That's what brand loyalty gets you.

Yup, agree completely

[Tim]

well it's obvious now that NV has fed Fudzilla with launch FUD as a base for their articles. They end up ridiculous (as I've warned them) and naturally they can't feel comfortable now... sobering is never easy and w/o consequences

Why would it be nV's fault? Lets assume for a second that they would have said to Fudo : we might launch it there and then, then you can be pretty sure Fudo would fantasize on that and make a big story about it how it will definitely come out at said time. I don't necessarily trust both, and I think they can both be full of it, I just do not think it's solely nVidia's doing.

[Wesker]

as i still own an 8800gtx i find myself in more and more trouble than when 8800gtx was new. I actually feel that nvidia only concentrates on fixing the newer cards and don't care about the older ones. Right now i cant play Mw2 when my tv set is connected cus i get blue screen.. I dont trust either of companies cus they only wants to make money...

News flash: EVERY company wants to make money.

I'm expecting big things from Fermi, but the more I wait, the more I'm reminded of the hype surrounding R600.

[saaya]

Why would it be nV's fault? Lets assume for a second that they would have said to Fudo : we might launch it there and then, then you can be pretty sure Fudo would fantasize on that and make a big story about it how it will definitely come out at said time. I don't necessarily trust both, and I think they can both be full of it, I just do not think it's solely nVidia's doing.

fudo doesnt make up stuff... ever... he misunderstands things, he exaggerates, but most of the time he gives back pretty exactly what somebody has told him... and its no secret that nvidia pr told EVERYBODY that they will have a hard launch in november, and then said they will have a hard launch in december... i think you can even find those old posts on nvidias twitter account, lol...

its not like fudo made up the november or december fermi launch lol...

[SKYMTL]

... and its no secret that nvidia pr told EVERYBODY that they will have a hard launch in november, and then said they will have a hard launch in december... i think you can even find those old posts on nvidias twitter account, lol...

Say what? Link to NVIDIA's Twitter post or it didn't happen.

[ulticool]

vaporware

[LedHed]

this thread contains no new information from page 30 to 35...

[shoopdawoopa]

this thread contains no new information from page 30 to 35...

This post helps with the stated problem.
O wait so does mine!

[Solus Corvus]

This screen i found on TPU forum, represent GF100 power truly REAL!

http://www.czechgamer.com/pics/clank...6-19_fermi.jpg

Between 20-23 percent above HD5870 and only little under HD 5970

Either you don't have inside information as claimed, or performance is worse then I'd hoped (>=30% faster). As to which is true, I'll wait and see.

As for that graph, it's horrible. Why would you compare all resolutions on high end cards when they are going to be CPU bound at anything but the top resolutions? Midrange cards cover lower resolutions just fine. These cards are for cranking up the res/AA. 20-23% faster then 5870 isn't really "a little under 5970", more like half way between the two:

[Manicdan]

Either you don't have inside information as claimed, or performance is worse then I'd hoped (>=30% faster). As to which is true, I'll wait and see.

As for that graph, it's horrible. Why would you compare all resolutions on high end cards when they are going to be CPU bound at anything but the top resolutions? Midrange cards cover lower resolutions just fine. These cards are for cranking up the res/AA. 20-23% faster then 5870 isn't really "a little under 5970", more like half way between the two:

there is definilty some truth that a person who spends 600$ on video cards, should be running them at 1920x1080 minimum

[zerazax]

From what it looks like, he's just taking the PR slides and transposing the information to the chart...

[Zed_X]

everywhere in Europe cost HD5970 USD 800! and more!

http://www.overclockers.co.uk/produc...=56&subid=1577

and in USA - 600-700! Where is USD 500?

http://www.newegg.com/Product/Produc...on%20HD%205970

[Jamesrt2004]

everywhere in Europe cost HD5970 USD 800! and more!

http://www.overclockers.co.uk/produc...=56&subid=1577

and in USA - 600-700! Where is USD 500?

http://www.newegg.com/Product/Produc...on%20HD%205970

England isn't really a good basis for Europe... it's overpriced on EVERYTHING..


they will drop to 500 i expect the moment gf100 is coming (maybe a few days before) to make price/perf dominate.....

i dont think ati where really expecting nvidia to take this long... therefore no need to drop from launch prices as people still buying them..

[LordEC911]

everywhere in Europe cost HD5970 USD 800! and more!

http://www.overclockers.co.uk/produc...=56&subid=1577

and in USA - 600-700! Where is USD 500?

http://www.newegg.com/Product/Produc...on%20HD%205970

Why are you looking for it at $500?
That isn't MSRP...

[annihilat0r]

HD5970 is around $700 here in Turkey.

A few points about GTX 380. No doubt it'll be compared to 5970 like hell, so pre-emptive strike:

The Facts
1. It's gonna draw about 50W lower than 5970 in load, which is good.
2. It's a single GPU without all the microstuttering / scaling problems of 5970, which is good.
3. It has massive GPGPU performance and PhysX, which is good.

Now, there are hundreds of rumors but the "25% faster than HD5870" looks consistent. It would make GTX 380 about 15% slower than HD5970. In this thread, Zed_X mentioned earlier that it performs great on Unigine benchmark so maybe its tessellation performance is better than Cypress / Hemlock too. Don't think we're gonna see hundreds of tessellated titles in the next years, but still it's an advantage.

If everything I wrote above is correct, I'd say that 380 would be a good product if it was priced about HD5970 pricepoint. But if Nvidia does that and AMD lowers 5800/5900 price points, Fermi fill indeed be a (huge) flop.