GF100 (Fermi) previews and discussion

[verndewd]

they just got running samples to flaunt as of this month writing a driver for a core that massive and testing it out on everything its used for takes time; when it can change your cars oil it will be released and then you can wire it up to make your bed and toast your bread.

 

[JAYDEEJOHN]

If it could only.... nahhhhhhhhhhhhh

 

[randomizer]

i stumbled on this on toms couple of day ago and they quickly removed it in a matter of minutes.. i prey that this is true and it goes with mine asumptions witch have always bean +,- 90% corect. it would bee nice that this is true and i bet it would.


[url]http://www.freeimagehosting.net/uploads/th.3f7801e497.png[/url]

[url]http://www.freeimagehosting.net/uploads/th.609089c0b1.png[/url]

If those were real, don't you think the staff at THG would know something about how Fermi performs already? Don't you also think that giving a card the name "GTX380" when no naming scheme exists is also a bit fishy? If they were real, I'd know about it 😉

 

[IzzyCraft]

If it could only.... nahhhhhhhhhhhhh

shoot lazer light show yeah that'd be great!

imo about the heat well then who cares then you can use your computer as a space heater i know i used to.

and to mr_moon along with everyone else we all know that fake esp with the poorly done memory sizes.

 

[EXT64]

I need a space heater here (I knew I would regret this 5850 sooner or later :lol: ) as my stupid heat pump is no good below 20F or so.

 

[mr_moon]

Well, the AA is supposedly 2.33 times, and who knows what that means, as G200 did poorly at 26x16 res with 8xAA, and didnt have the DX10.1 1 less pass, which makes it 10-20% faster already.
In other words, add 20% to Fermis, plus 60% perf, plus a better high res solution for 8xAA, and you can see how theyd get 2.33 times better.
As for the DP compute, thats only on Tesla, as theyve removed the DP ability for the gfx desktop cards, and isnt needed for games

True,but what about physics compute,cuda and direct compute as the requirment for DX11,c++ etc. i think that the Fermi's compute is not entirely shut of and the remark of tom's that this is more cpu than gpu i thought that tesla only had EPP on memory enabled and double the memory buffer than fermi? everything else what you said is somekind of logical the way you put it if that 2,33x faster is like you already said.all of this just makes me more impatient to see what is this Fermi all about either way i cant wait the release of it and to compare ati and nvidia and than to wait little for price stabilization and finaly get one of their's and first of mine DX11 card no mater of the "colors" just to performe as i wish and to be more afordable but this is gona be realy hard cause i think that this time we'r gona have competition in the real meaning of that word atleast i hope so in sake of astro prices,well i'm gona see if i was corect or not with not buying until now and this long long waiting are you still waiting to see conclusion or have you already decided few month's ago?

 

[mr_moon]

If those were real, don't you think the staff at THG would know something about how Fermi performs already? Don't you also think that giving a card the name "GTX380" when no naming scheme exists is also a bit fishy? If they were real, I'd know about it 😉

quote- IzzyCraft-and to mr_moon along with everyone else we all know that fake esp with the poorly done memory sizes.

i alredy said that i dont know if it's realy truth or photoshop i just found that few days ago and i posted it in the Fermi tread as one more info false or not just one curiosity nothing more with no bad intention and if Fermi end up like that it will and only than justify the delaying and i truly hope that we'r gona have tough competition from which we all are going to benefit in regard of astro prices these days.

I'm alowed to hope for christ sake guys and i cant wait anymore to see which card is better and buy my first DX11 gpu.

 

[randomizer]

Well, keep waiting, because Fermi is not "just around the corner." It's around a few blocks :lol:

 

[IzzyCraft]

Well, keep waiting, because Fermi is not "just around the corner." It's around a few blocks :lol:

but if we keep taking the right wont we come back to where we were

 

[AMW1011]

The price for a IN STOCK 5970 = 800......But you all knew that.
You use this line, go away alot, I'm not your mother asking you to clean your room. I'm showing you proof that ATI will gouge you just as much as Nividia might for a video card.
800 dollars when they told reviewers it would be 600. Thats worse than jacking the price up on the 5850 to 320 from 269.00. But ATI is for the people, yeah right.

You don't get it, so some carrier kept a few 5970s by charging a lot more and making more by being one of the few retailers to have one in stock. How the hell does ATI's pricing scheme have anything to do with retailer inflation? I'll remind you that ATI stops having a say in prices after they release a card with an MSRP, manufacturer's suggested retail price, and the MSRP of the 590 is $599. Anything over that is purely retailer inflation.

But you knew that, at least I hope so because if you didn't then you have no idea how markets work...

 

[randomizer]

but if we keep taking the right wont we come back to where we were

Isn't that how the waiting game always goes?

 

[RealityRush]

Isn't that how the waiting game always goes?

Nah, sometimes you take a wrong left turn and end up in Ampsterdam drunk off your arse, then wake up the next day with a unionized hot blonde escort telling you to pay up or face legal action. Those situations can be awkward sometimes... not that I'd know or anything, I would never do such a horrible thing. I clearly have too much character to let alcohol get the better of me.

 

[JAYDEEJOHN]

but if we keep taking the right wont we come back to where we were

Everybody waiting on Fermi needs to wait just a lil more
http://www.fudzilla.com/content/view/17387/1/

 

[RealityRush]

Everybody waiting on Fermi needs to wait just a lil more
http://www.fudzilla.com/content/view/17387/1/

... that's 2nd gen, first gen is still slated for march.....

 

[JAYDEEJOHN]

If it doesnt come out in ngood numbers in March, itll be April May before people can get them, and by Oct? therell ne new ones? Plus the ATI refresh/new arch.
That leaves 5 months, or less time than nVidia has had everyone waiting, to use the card before its old, not even counting a ATI refresh

 

[RealityRush]

If it doesnt come out in ngood numbers in March, itll be April May before people can get them, and by Oct? therell ne new ones? Plus the ATI refresh/new arch.
That leaves 5 months, or less time than nVidia has had everyone waiting, to use the card before its old, not even counting a ATI refresh

Which is why at this point I'm 90% sure I'll be buying the cheapest 58x card in a month or two that I can and then wait for the 2nd gen Fermi's/ATI refresh so that when more games come out that use DX11 that actually require some hardcore graphics hardware, I can get a new ATI card or SLI a couple 2nd gen Fermi's or something.

I'll be really pissed though if when Fermi comes out the 5800 cards aren't cheaper than when they first came out... 'cause I was kinda counting on that to save me money.... if that doesn't happen, somebody is going to get hurt real bad.

 

[mindless728]

@RealityRush, hope for the best, expect the worst

an ati 58xx price drop is pretty much contingent with fermi being better performance and near the same price

 

[L1qu1d]

What interests me is this:

From Guru3D:
So to the left the GeForce GTX 280 with 16xCSAA and to the right the new GF100 (Fermi) with 32xCSAA. Now what I need to tell you here is that the above example comes from an NVIDIA PDF presentation. We can not verify this whatsoever until we have the final product in our hands. 32xCSAA should be possible with a minimum performance hit, in fact it should be as fast as say the GT200 at 16xCSAA. Actually, if we may believe NVIDIA, 32xCSAA takes roughly the same hit as normal 8xAA, which admittedly would be really good.

If true, it seems to be a fair good trade off.

 

[mindless728]

What interests me is this:

http://www.guru3d.com/imageview.php?image=21886

From Guru3D:
So to the left the GeForce GTX 280 with 16xCSAA and to the right the new GF100 (Fermi) with 32xCSAA. Now what I need to tell you here is that the above example comes from an NVIDIA PDF presentation. We can not verify this whatsoever until we have the final product in our hands. 32xCSAA should be possible with a minimum performance hit, in fact it should be as fast as say the GT200 at 16xCSAA. Actually, if we may believe NVIDIA, 32xCSAA takes roughly the same hit as normal 8xAA, which admittedly would be really good.

If true, it seems to be a fair good trade off.

good if true, though that is blown up, the question is if it is really noticeable on say a 21" 1080P monitor while playing, i have noticed that the only time i notice those jaggies is when i am not moving, as soon as there is motion it is much harder to see them

 

[notty22]

... that's 2nd gen, first gen is still slated for march.....

He doesn't like to get the story straight. This does not fit in his endless Nvidia fails scenarios earlier in the page.

 

[AMW1011]

What interests me is this:

http://www.guru3d.com/imageview.php?image=21886

From Guru3D:
So to the left the GeForce GTX 280 with 16xCSAA and to the right the new GF100 (Fermi) with 32xCSAA. Now what I need to tell you here is that the above example comes from an NVIDIA PDF presentation. We can not verify this whatsoever until we have the final product in our hands. 32xCSAA should be possible with a minimum performance hit, in fact it should be as fast as say the GT200 at 16xCSAA. Actually, if we may believe NVIDIA, 32xCSAA takes roughly the same hit as normal 8xAA, which admittedly would be really good.

If true, it seems to be a fair good trade off.

Great 32x CSAA, just what we need. Because anything over 4x AA is SO noticeable! Sorry L1qu1d I think you need to take a step back first. Second, there is no way in hell Fermi will be able to do that without it being a slide-show, we just don't have the tech. right now. 32x CSAA, if it is ever relevant, is at least 2 years off. Also, I can't stand AA over 8x because it makes things blurry, I'm not so sure the "improvements" in those blown up slides are actually improvements.

 

[AMW1011]

He doesn't like to get the story straight. This does not fit in his endless Nvidia fails scenarios earlier in the page.

Its funny seeing you point out someones mistake when you are mistaken most of the time. :lol:

 

[daedalus685]

He doesn't like to get the story straight. This does not fit in his endless Nvidia fails scenarios earlier in the page.

Oh come now... He was just making a joke about how we have had to hear about "wait for (random thing), it comes out only a few months later." for so long now, it is logical the smae folks should now expect us to wait for rev 2.

Not every comment is doom and gloom, BS.. despite how many seem to take them as such.

 

[AMW1011]

Hi, it is not 32xAA, it is CSAA.

Although I will admit, depending on what is being displayed, a bit of grittiness it okay. Some things look too smooth, like rocks and whatnot.

Your right it was a mistype.

 

[RealityRush]

Oh come now... He was just making a joke about how we have had to hear about "wait for (random thing), it comes out only a few months later." for so long now, it is logical the smae folks should now expect us to wait for rev 2.

Not every comment is doom and gloom, BS.. despite how many seem to take them as such.

I'll prob be buying an ATI card soon so don't shi t on my soul for saying this....

It might be worth it waiting 'til rev 2 to actually buy a significantly powerful card.

If the tom's hardware requirements for AvP3 are true, beyond a 5770 wont seem to matter very much. And I doubt there'll be anything much more hardware intensive 'til BF3 or some such. Than again, there's very few DX11 games coming out I seem to care about so that could just be me spewing BS, but it seems likely.

So waiting 'til rev 2 Fermi and ATI's HD6000s seems like a worthwhile wait, by that time lots of DX11 games should be in the pipeline and you can actually invest in higher powered hardware without getting screwed by water blocks, etc. being exclusively for rev 2 cards.

But I'm focusing solely on gaming... I guess if you just wana have the best benchs all the time, then it doesn't really matter....