8800GT Spec is out! 65NM CONFIRMED !!

[quadnad]

now all we need is some info on those g100's or whatever their calling them,

yup....the high end is still the only thing I'm interested in. I wish someone would leak some info

[BenchZowner]

Thanks Benchzoner- do you know what cards will be PCI-E 2.0 and/or when they'll be out?

On another note...still no-one agrees on the specs of this card. There isnt even agreement on what card we're talking about.

Controversy in details and stuff can be funny and at the same time irritating

Making all the new cards PCI Express 2.0 would be a good idea, and that's what it looks like it will be from now and on.
With the PCI Express 2.0 being fully compatible with PCI Express 1.0 you can't go wrong here

I'm currently having a chat with a VGA manufacturers' personnel, and looking for new info on cards to be released, and specs.
I've put my best bait on my data-fish-hooks and looking for a big fish to have a go at 'em

The 8800GT looks like a very good deal for the price asked ( if the given MSRP is true ).

Oh wait! I felt some vibrations from my data fishing rod hold on

[GoldenTiger]

Controversy in details and stuff can be funny and at the same time irritating



Oh wait! I felt some vibrations from my data fishing rod hold on

[BenchZowner]

Don't remember/know what's been posted here and the rest 40 threads about the same thing ( not funny ), here's some things that I've heard from a contact in a VGA manufacturer:

-nVIDIA GeForce 8800GT: Release Date, 29 Oct 2007
512MB, 256BIT, 600MHz Core Clock
-ATi HD2950Pro & HD2950XT will be ready & launched in mid November

That's the "best" of what this contact could give me at the moment...gotta check another contact later on though for some real juice.

[NH|Delph1]

Don't remember/know what's been posted here and the rest 40 threads about the same thing ( not funny ), here's some things that I've heard from a contact in a VGA manufacturer:

-nVIDIA GeForce 8800GT: Release Date, 29 Oct 2007
512MB, 256BIT, 600MHz Core Clock
-ATi HD2950Pro & HD2950XT will be ready & launched in mid November

That's the "best" of what this contact could give me at the moment...gotta check another contact later on though for some real juice.

I've heard the same

Also heard 96 SP again from another source.

//Andreas

[turtle]

OCW has been saying 112, and a 96 part might be in the wings depending on sales/Rv670.

I am in the faction whom (now) believes 96 was the original target for this part, but rv670 changed the game, and the 640MB GTS. Extremely possible this is a G80 shrink with a 256-bit bus like RV670, and we'll see a fully decked-out or GX2 version come Q108.

[NH|Delph1]

OCW has been saying 112, and a 96 part might be in the wings depending on sales/Rv670.

I am in the faction whom (now) believes 96 was the original target for this part, but rv670 changed the game, and the 640MB GTS. Extremely possible this is a G80 shrink with a 256-bit bus like RV670, and we'll see a fully decked-out or GX2 version come Q108.

The last I heard from them they said 96 too.

//Andreas

[Start]

Just going soley based on the Nvidia marketing scheme, theoretically shouldn't the GT be slower than the GTS. Bumping the GT to 112 would put it faster than the GTS (assuming the GTS doesn't get any bump).

[cookerjc]

http://www.xtremesystems.org/forums/...d.php?t=161754

here is somethin' about 8800GT(and GTS)

[perkam]

Title edited for this:

The one thing that didn't change on G92 is the process node. NVIDIA's foundry partner, TSMC, forwarded a second memo to DailyTech confirming G92 is in mass production at the company's Fab 12 with samples available now on 65nm process node. NVIDIA's GeForce 8500 and GeForce 8600 are manufactured on TSMC's 80nm node; GeForce 8800 GT will be the company's first 65nm graphics processor.

Blazingo

Perkam

[RPGWiZaRD]

Ahh, I feel relieved altho I always felt it just has to be 65nm but how dissapointed it would have been if not. lol 2½ weeks remaining eh, feels like 2 months at this rate with all 2950 and 8800GT talk. I wonder if 2950 Pro and 8800GT doesn't take the top position when it comes to anticipated midrange cards?

Now the remaining thing is the amount SPs.

[DavidP]

Nothing is confirmed until there is a press release from Nvidia or a partner. Until then I say BS.

[[XC] gomeler]

Well if it's 65nm maybe it'll clock like a beast. Loving these new midrange cards, this is what we should have had 8 months ago.

[NH|Delph1]

Nothing is confirmed until there is a press release from Nvidia or a partner. Until then I say BS.

Doesn't that kill the fun of trying to figure it out?
I mean it's like a huge puzzle where you always miss at least a few pieces of the entire picture, but now and again you get a new piece and with the help of that one you get a better understanding for the whole picture, but there is still no way to be entirely sure
And sometimes they decides to draw a new picture without finishing the first one and you have to start all over again. The puzzling and the guessing is what makes it all fun.

//Andreas

[WeaponX]

Speculation is not fun, puzzles however are. I'd rather have the facts whenever things are annouced, that way I wouldn't have the worrying feeling or any doubt on my mind, I would know exactly what I'm getting and exactly when I'm getting it and I should expect no less. 100% Facts also weed out all the crazy fanboys and people with completely false information, I don't know about you, but I can live happily without any of that garbage.

[NH|Delph1]

Speculation is not fun, puzzles however are. I'd rather have the facts whenever things are annouced, that way I wouldn't have the worrying feeling or any doubt on my mind, I would know exactly what I'm getting and exactly when I'm getting it and I should expect no less. 100% Facts also weed out all the crazy fanboys and people with completely false information, I don't know about you, but I can live happily without any of that garbage.

But the facts are revealed when things are announced
Everything up until then you either ignore or try to make the most of, you choose to read or not to read the rumors stories.
That's what so great about the freedom of mind, you can actually ignore stories like this if you want and never have to get worried or excited about things before they are either enforced or demolished

You do have a great point on the fanboy issue however.

//Andreas

[K404]

With the PCI Express 2.0 being fully compatible with PCI Express 1.0 you can't go wrong here

Is it absolutely 100% working both ways? If a PCI-E 2.0 GPU is plugged into a PCI-E 1.0 board, isnt there theoretically 1/3 of the possible power missing?

Also- on a PCI-E 2.0 mobo, which I think allows 150W to be taken through the PCI-E slot, wont that be a challenge for a lot of PSUs, especially once we start overclocking? The 12V rails often aren't really tuned for that distribution are they...especially for a new SLI board? Will the reliability depend strongly on whats powered by each PSU rail? e.g... if 24-pin ATX and molex are on the same rail...uh-oh?

getting specific- if I plug a 200W stock PCI-E 2.0 GPU that has 1 additional 6-pin power plug into my PCI-E 1 mobo, wont it work in LP mode, if it works at all?

Sorry for asking the same thing lots of different ways, but I want it to be crystal clear for everyone...and me

[KTE]

Why do the GPUs need more power potential at lower process node with lower voltages (i.e. PCIe 2.0)? It should be exactly the reverse, so all mainstream and performance cards should work fine with basic PCIe 1.0 power available, and PCIe 2.0 power potential should be what is kept in-safe for higher power/rocket priced enthusiast cards, as well as those bearing the dual-GPU frameworks. That's what it was originally implemented and appealed for.

[Shintai]

Is it absolutely 100% working both ways? If a PCI-E 2.0 GPU is plugged into a PCI-E 1.0 board, isnt there theoretically 1/3 of the possible power missing?

Also- on a PCI-E 2.0 mobo, which I think allows 150W to be taken through the PCI-E slot, wont that be a challenge for a lot of PSUs, especially once we start overclocking? The 12V rails often aren't really tuned for that distribution are they...especially for a new SLI board? Will the reliability depend strongly on whats powered by each PSU rail? e.g... if 24-pin ATX and molex are on the same rail...uh-oh?

getting specific- if I plug a 200W stock PCI-E 2.0 GPU that has 1 additional 6-pin power plug into my PCI-E 1 mobo, wont it work in LP mode, if it works at all?

Sorry for asking the same thing lots of different ways, but I want it to be crystal clear for everyone...and me

The PSU is not really a problem.

And for the PCIe 2.0 vs 1.0. It will be quite some years before there is 2.0 only GPUs. I cant even imagine it within the first 3 years from now. Until then they will have all the 6/8pin conenctors that they will need for working in 1.0 aswell.

Also atleast nVidia prefers to use non slot power. And ATI also did quite abit in the past.

[SnipeR²]

Good question there dude. was wondering along those same lines....

[aiya]

http://www.theinquirer.net/gb/inquir...ittle-g80-65nm

Inq says that G92 originally has 128SPs but will have 16 disabled for the 8800GT

yeah, its Inq so its just more wood for the fire of rumors

a part of me is saying G92 is just like G71 was to G70, die shrink and replaces current G80 cards. logically thinking it makes more sense to since G92 is cheaper to make.

a G92 based GTS is just like 7800GT(20Pixel Shaders) to 7900GT(24Pixel Shaders + die shrink).

just my thoughts.

[XS Janus]

PCIE 2.0 card will draw all it's needed power through PCIE slot when pluged into PCIE 2.0 capable board.

When pluged in PCIE 1.0 board the card will operate just as it does today, using external power connectors to the card itself.

Its logical! Just think!
There will be virtually NO 2.0 capable boards sold by the time these cards appear.
Did you really think they would be so stupid and screw up their sales with the lack of user base using the new 2.0 standard - NOT POSSIBLE

[Bobsama]

It seems to me that this is how it'll all be distributed...

8800GT - G92 with 64 SP, 256-bit, PCI-E 2, and $220
8800GTS 320 - G92 with 96 SP, 320-bit, PCI-E 2, and $280
8800GTS 640 - G92 with 112 SP, 320-bit, PCI-E 2, and $370

I'd really like to hear more about G98, however. If it's the 8600 Ultra, it'll be a dream finally come true.

I wonder about SLI performance, however--just how well will these newer cards perform?

[KTE]

AFAIK, the 8800GT MSRP was $250 not $220.

[LordEC911]

And Nvidia is releasing a new 8800GTS with a 1200mhz shader domain and 112SPs to replace the current 640mb GTS.