Geforce 300 series discussion

I'm pretty sure this has been covered before but the 6gb version will most likely be for workstations and not meant for general gaming use.
 
On paper it seems epic. But with TSMC's incompetence with 40nm chips and the sheer size of the GT300 die it almost seems like an impossibility seeing it ever come to market.

However if/when it ever does I'm in for whatever the GTX260 or 275 equivalent is.
 
thenewrick said:
http://en.wikipedia.org/wiki/GeForce_300_Series

Looks like it might blow away the 5800 and 5900 series from ATI. 6gb of memory seems like a ton. Think any of the 300s that come out will be under 400 bucks?
Click to expand...

I wonder about that, if anything I am wondering if its going to keep up. Its going to be a MONSTER of a card but the white papers make it sound like " it will do all this, and it can do games" Maybe I am see this wrong but its coming across to me like they have made the most powerful GPGPU ever and not worried so much about what use to be its core market.

hopefully I am wrong and it can do both and beat the 5870 (price wars for the win) but I don't see it being under 400, there just isn't any way.
 
thenewrick said:
http://en.wikipedia.org/wiki/GeForce_300_Series

Looks like it might blow away the 5800 and 5900 series from ATI. 6gb of memory seems like a ton. Think any of the 300s that come out will be under 400 bucks?
Click to expand...

Competition is good for everyone and I'm hoping Nvidia hits this outta the park despite having a 5870 myself.

But here's what I don't get. AMD is receiving the lion's share of publicity with its entire 5 series lineup. Why hasn't Nvidia or anyone even remotely connected to the Fermi released even a tiny bit of gaming benchmarks to counter AMD's commanding lead? We constantly hear about the Fermi's memory bandwidth, the architecture itself, the number of shaders, ECC capability, yadda yadda. How about a few Crysis benchmarks? How about some Vantage numbers?

NVidia keeps saying this beast is due for release before year's end. Someone surely has to leak some numbers by now, if only to stem AMD's tide, huh?
 
Valset said:
I wonder about that, if anything I am wondering if its going to keep up. Its going to be a MONSTER of a card but the white papers make it sound like " it will do all this, and it can do games" Maybe I am see this wrong but its coming across to me like they have made the most powerful GPGPU ever and not worried so much about what use to be its core market.

hopefully I am wrong and it can do both and beat the 5870 (price wars for the win) but I don't see it being under 400, there just isn't any way.
Click to expand...

Consider "Fermi" to be 2 x 285 GPU's..and added GPGPU capabilities.

If people think "fermi" (with 2 times the ASIC for gaming than a GT200b) is not going to perform well in games and only be good good a GPGPU and that NVIDIA is leaving the gaming market...well then these people should shut down their PC's, disconnect the internet and go outside...because IT is cleary not your strong side....
 
Atech said:
Consider "Fermi" to be 2 x 285 GPU's..and added GPGPU capabilities.

If people think "fermi" (with 2 times the ASIC for gaming than a GT200b) is not going to perform well in games and only be good good a GPGPU and that NVIDIA is leaving the gaming market...well then these people should shut down their PC's, disconnect the internet and go outside...because IT is cleary not your strong side....
Click to expand...

Have any proof that the woodscrewed Video card will perform the way you say it will?

Evidence man evidence!
 
Atech said:
Consider "Fermi" to be 2 x 285 GPU's..and added GPGPU capabilities.

If people think "fermi" (with 2 times the ASIC for gaming than a GT200b) is not going to perform well in games and only be good good a GPGPU and that NVIDIA is leaving the gaming market...well then these people should shut down their PC's, disconnect the internet and go outside...because IT is cleary not your strong side....
Click to expand...

Sentences and blind faith > proof

/thread
 
Atech said:
Consider "Fermi" to be 2 x 285 GPU's..and added GPGPU capabilities.

If people think "fermi" (with 2 times the ASIC for gaming than a GT200b) is not going to perform well in games and only be good good a GPGPU and that NVIDIA is leaving the gaming market...well then these people should shut down their PC's, disconnect the internet and go outside...because IT is cleary not your strong side....
Click to expand...

um, given how much the changed every thing your statement of it being 2 X GT200 is not even in the ball park. we really don't know yet. This is hardly the next simple step but a real change.

Maybe with your vast IT skills you over looked some of ATI high paper spec cards. HD 2900 XT might ring a bell. on paper it killed the competition. too bad it didn't work out like that. save the condescending attitude for when the card comes out and exceeds all expectations. in the time being I am really interested in just what this can do all round. but for gaming performance I still wonder
 
The other point about "IT Skills" is that IT skills and Enthusiast computing don't go hand in hand. Here at my company (its a fortune 500 enterprise) I would say 99.9% of the workstations have integrated graphics. Of the 200 or so that have an external grapics solution 100% are ether Quadro or FireGL based workstation cards. Reality is that you are not going to hear an end user complain "ZOMG I'm only getting 35fps in Excel!!". Even if I ever did hear that I really doubt I'd ever get a PR for new hardware past a director much less a controler with that type of request.
 
6GB would do very little for games. You have to remember with regards to memory for games you either have enough for the data they want or you don't. Once you have enough, more doesn't do anything. Now yes, game devs could simply make larger and larger textures but that isn't really likely. In part we are moving more to procedurally generated content, but more to the point you aren't going to go nuts for a single card. If one card has 6GB and all the rest have 1GB, well games will be designed for 1GB.
 
Wish they bring it out sooner than next year. So ATI could drop some price on their card(good for us).
 
dorky82 said:
Wish they bring it out sooner than next year. So ATI could drop some price on their card(good for us).
Click to expand...

There was a "slip" from TSMC that Nvidia would be launching a 40nm card in december. Combine that with the roadmaps would say "wheee GF100/GTX 300" but who really knows.

I will say this card is a huge change architectually. And it's really hard to say what effects that will have.

I could build a dream list:
1) "down-spikes" go away
2) "free AA" like we get "free AF" now.
3) Magically perfect SLI scaling
4) Much improved Physx preformance, i.e. less taxing on fps.

But there is no basis to assume any of those things. Like many things, this is a wait and see. On paper it looks to be ~2x a 285. While powerful, won't be a game changer compared to the 5870. Except in folding of course, where it should experiance a SERIOUS advantage over the 5870. The real problem with saying "on paper" though is it assumes clock speeds. If Nvidia can send out this part with at 800mhz and price it on top of the 5870 they will take all the non-fanboy sales. But that is still speculation. I might as well be saying well if Nvidia sells a ferrari at the price of a civic it's going to outsell the civic. Well no shit sherlock!
 
Atech said:
Consider "Fermi" to be 2 x 285 GPU's..and added GPGPU capabilities.

If people think "fermi" (with 2 times the ASIC for gaming than a GT200b) is not going to perform well in games and only be good good a GPGPU and that NVIDIA is leaving the gaming market...well then these people should shut down their PC's, disconnect the internet and go outside...because IT is cleary not your strong side....
Click to expand...

I'm beginning to think you work for nVidia or something along those lines with the comments you've been posting around the forum. Seems they always end with a rude comment too. :mad:

maybe YOU should go outside for a breath of fresh air and relax some eh?
 
Until TSMC get their act together I'm not getting my hopes up for a release of cards we can actually get our hands on, which is what we need at the end of the day, with ATIs dual GPU solution just around the corner and the lack of 40nm chips.

Still it sounds like a very powerful card, 3Gb variant would be very nice, I guess I have a few more months to save for a new graphics card :)
 
MoparMadness said:
I'm beginning to think you work for nVidia or something along those lines with the comments you've been posting around the forum. Seems they always end with a rude comment too. :mad:

maybe YOU should go outside for a breath of fresh air and relax some eh?
Click to expand...

240 streaming cores vs 512 "CUDA" cores...that is more double the ASIC right there.

And no, I don't work for or with NVIDIA...but do I know to see a GPGPU conference for what it is...and especially to see what it isn't.
 
Atech said:
240 streaming cores vs 512 "CUDA" cores...that is more double the ASIC right there.

And no, I don't work for or with NVIDIA...but do I know to see a GPGPU conference for what it is...and especially to see what it isn't.
Click to expand...

Come on, two get 2x the preformance they would have to clock those cores to an insane 540mhz. They'll never get that kind of crazy clocks with a die shrink involved. Stop drinking the koolaid.(Yes, this is sarcasim)
 
alxnet2003 said:
But here's what I don't get. AMD is receiving the lion's share of publicity with its entire 5 series lineup. Why hasn't Nvidia or anyone even remotely connected to the Fermi released even a tiny bit of gaming benchmarks to counter AMD's commanding lead?
Click to expand...

I sometimes feel that they already know it will be so stupidly powerful that they are letting AMD make asses of themselves.

Other-times I feel like they are just sitting out this round and dedicating their efforts into future releases.
 
AMD isn't making as ass of themselves until we can buy something from nvidia. All the numbers in the world won't matter
 
a techinicalish user wil speculate how this is proof nvidia is going to release 'another G80' :rolleyes:

kudos to those who know what I mean.
 
ivandurago said:
I sometimes feel that they already know it will be so stupidly powerful that they are letting AMD make asses of themselves.
Click to expand...

Doesn't make sense from a business point of view. Sure you never want to play your cards too early, but NVidia's been speaking for a while about a November/December release (still do).

This close to release you'd think they might want to divulge benchmark numbers far superior to AMD's current offerings if only to squash the immense position ATI has at this point - that is, if they really have numbers far superior to anything ATI has on the market.

All this talk about how many cores the beast has, about its GPGPU prowess, massive quantity of memory, bandwidth, etc, and not one benchmark, not one inkling of how it trashes ATI in games.

I want numbers dangit!
 
jeremyshaw said:
a techinicalish user wil speculate how this is proof nvidia is going to release 'another G80' :rolleyes:

kudos to those who know what I mean.
Click to expand...

NV30 repeat *cough

half a billion transistors added for Cuda perhaps and some awesome scientific calculating and simulating power most likely. In JHH and nvidia fan's wet dreams there is an abundant amount of die space to accommodate all added stuff for Cuda GPGPU chips and some left over for gaming. The limitations of the route they took with die size will bite them here.

Of course all speculation, and if it kicks 5870 ass by 30% I'll pony up $600 + it will probably cost. :p
 
There's not much to discuss about the Geforce 300 series till we see some benchmarks. Otherwise, were just sitting here doing the equivilent of trying to spot shapes in the clouds. Some of us will see a powerful dragons and others a sheep. Really, without benchmarks nobody knows how this card will perform in games.

I'll admit I'm getting more skeptical with each day that passes though. No benchmarks, a week later after competition's release, still no benchmarks. A month after the 58xx release, still no benchmarks? A month and a week or so now, still no benchmarks?

I'm wondering whether the 5870x2 benchmarks/full release occurs before we see any benchmarks of this Fermi card nvidia likes to rumor...i mean...speak of? I mean, I think the 5870x2 is suppose to release around the 2nd or 3rd week of November? November has started as the countdown towards the 5870x2's launch begins. 14-21 days ish. Personally, I'll jump on a 5870x2 if nVidia can't show me benchmarks by then.

If we don't see at least leeked benchmarks, ATI will have the performance crown, will it not? I think a single 5870x2 should beat a GTX295 since crossfire 5870x2s beat a GTX295, right?
 
kadozer said:
Of course all speculation, and if it kicks 5870 ass by 30% I'll pony up $600 + it will probably cost. :p
Click to expand...

If it kicks the 5870 by only 30%, would you prefer a $600 Fermi or a $499.99 5870x2? Since I guess the 5870x2 should beat a 5870 by at least 30% even on a bad day(not crossfire optimized game)?
 
vengence said:
Come on, two get 2x the preformance they would have to clock those cores to an insane 540mhz. They'll never get that kind of crazy clocks with a die shrink involved. Stop drinking the koolaid.(Yes, this is sarcasim)
Click to expand...

You mean like R700 55nm (850Mhz) to R800 40nm (850Mhz) :p ;)?
 
Atech said:
You mean like R700 55nm (850Mhz) to R800 40nm (850Mhz) :p ;)?
Click to expand...

or g92 across 3 generations (on 8 different cards based on that now:p;):eek:)

9600gso (96sp version)
8800gt
9800gt
8800gts (512 version)
9800gtx (and the '+' version, too)
gts250
9800gx2
8800gs

More direct, eh? went from 65nm - 55nm (called the g92b, but still the same), and is currently looking for a 40nm refresh... in the gts240!
 
jeremyshaw said:
or g92 across 3 generations (on 8 different cards based on that now:p;):eek:)

9600gso (96sp version)
8800gt
9800gt
8800gts (512 version)
9800gtx (and the '+' version, too)
gts250
9800gx2
8800gs

More direct, eh? went from 65nm - 55nm (called the g92b, but still the same), and is currently looking for a 40nm refresh... in the gts240!
Click to expand...

So a R800 is a R700 refresh? ;)
 
WBurchnall said:
If it kicks the 5870 by only 30%, would you prefer a $600 Fermi or a $499.99 5870x2? Since I guess the 5870x2 should beat a 5870 by at least 30% even on a bad day(not crossfire optimized game)?
Click to expand...

Jury is still out on what msrp + Resellers will price 5870x2. I gotta see what waterblocks vendors come out for either one too.
 
jeremyshaw said:
nope.

However, a g92b is just a g92 refresh ;)
Click to expand...

So a 65nm G92b @ 600/1500/1800 Mhz is the same as a 55 nm G92b @ 675/1620/2200?

You could go all then way and then say a 65nm G92 600/1500/1800 (8800GT) is the same as a 55 nm G92b @ 675/1620/2200(GT240)...I have no problem with that.

Gamers won't miss out any gaming features from a 8800GT to a 240GT
But the consumer has a lot easier task at hand:

First generation
8600GT
8800GT
8800GTX

First refresh
9600 GT
8800 GT
9800 GTX

Second refresh
8800 GT
GTX 260
GTX 280

Third refresh
8800 GT
GTX 260 (216)
GTX 285

Unlike this:
First generation
8600GT
8800GT
8800GTX

First refresh
9600 GT
9800 GT
9800 GTX

Second refresh
9800 GT
GTX 260
GTX 280

Third refresh
GT 240
GTX 260 (216)
GTX 285

It makes sense, not to much interruption of cycles performance relative numbering...and AMD has done it too:
Radeon 8500 LE -> Radeon 9100
Radeon X300 -> Radeon X550
Radeon X1300 Pro -> Radeon X1550
 
You must log in or register to reply here.
Back
Top