Furmi (GT300)

[3DGuy]

Originally Posted by Skorp

Well, the site claims that the GTX 260, 275, and 285 were killed off months ago. They don't seem killed off to me.

Production was, there are still units in teh warehouses, and stocks though to filter through.

Originally Posted by Zefram

I think the issues is with the memory bandwidth. 4870x2 have 2X256-bit memory buses.....

ATi engineer think about yields, they think about less complex GPU, thus the decision to go with 256-bit memory buses and using GDDR5 to cover back the bandwidth.

If nVidia stays true to their 448-bit (or more) memory bandwidth, and by pairing those with GDDR5, the memory bandwidth would be huge. It would ate ATi for breakfast

The point is. That more than 256bit isnt needed. The gpu's can only use so much bandwidth that they can handle. If you stuck a 448 on a 4870X2 it wouldnt boost teh performance that much to use it all as it is. So by using a bigger memory bus it just adds extra cost which is why Nvidias pricing always has to be higher... They use more Complex PCB's Hell with 256 bus i bet AMD/ATI could get away with a 4-6 layer PCB... The 2900 on 512 bit needed 12 layer PCB for the hundreds of memory tracers = Higher cost...

Hell. The current G200 series i bet doesnt use all the memory bandwidth that is there and thats on 384... 448 bit + GDDR5 is more than teh new Nvidia chip will need. = Waste of resources if its not used. sure marketing bull$************************ sounds good but your paying for stuff that wont be used and it will just jack the price up more..... Tards. lol

[Zefram]

Originally Posted by 3DGuy

The point is. That more than 256bit isnt needed. The gpu's can only use so much bandwidth that they can handle. If you stuck a 448 on a 4870X2 it wouldnt boost teh performance that much to use it all as it is. So by using a bigger memory bus it just adds extra cost which is why Nvidias pricing always has to be higher... They use more Complex PCB's Hell with 256 bus i bet AMD/ATI could get away with a 4-6 layer PCB... The 2900 on 512 bit needed 12 layer PCB for the hundreds of memory tracers = Higher cost...

Hell. The current G200 series i bet doesnt use all the memory bandwidth that is there and thats on 384... 448 bit + GDDR5 is more than teh new Nvidia chip will need. = Waste of resources if its not used. sure marketing bull$************************ sounds good but your paying for stuff that wont be used and it will just jack the price up more..... Tards. lol

That's all about GPU architecture efficiency in using those memory bandwidth.
If Fermi was to be anything like the rumor suggest, yes, huge memory bandwidth wont hurt

[Comixion]

Nvidia is really pissing me off just about, right now.

[Nova]

Originally Posted by Comixion

Nvidia is really pissing me off just about, right now.

What's hilarious is how huge this thread is over a non existing and probably won't exist for another 4-6 months card.

[Skorp]

Originally Posted by 3DGuy

The point is. That more than 256bit isnt needed. The gpu's can only use so much bandwidth that they can handle. If you stuck a 448 on a 4870X2 it wouldnt boost teh performance that much to use it all as it is. So by using a bigger memory bus it just adds extra cost which is why Nvidias pricing always has to be higher... They use more Complex PCB's Hell with 256 bus i bet AMD/ATI could get away with a 4-6 layer PCB... The 2900 on 512 bit needed 12 layer PCB for the hundreds of memory tracers = Higher cost...

Hell. The current G200 series i bet doesnt use all the memory bandwidth that is there and thats on 384... 448 bit + GDDR5 is more than teh new Nvidia chip will need. = Waste of resources if its not used. sure marketing bull$************************ sounds good but your paying for stuff that wont be used and it will just jack the price up more..... Tards. lol

Actually, I have to disagree. In my opinion, performance on the Radeon 5870 is held back by the memory bandwidth; it has the same setup as the 4870, and you remember how much of a performance gap there was between the 4870 and 4850, thanks to the GDDR5 on the 4870 doubling the memory bandwidth. The 5870 is exactly double the 4870, but the bandwidth hasn't met the increase in power.

Fermi is much more than double a GTX 280. Shrinking the bus from 512-bit to 384-bit or 448-bit, then adding on GDDR5, doesn't quite double the bandwidth there, but it's an excellent trade-off. The 512-bit bus on the 280 was one of the reasons it was so huge.

[Avid6eek]

It's official....nVidia has confirmed the delays in their conference call with financial analysts:

http://xbitlabs.com/news/video/displ...y_in_2010.html

Expect card production to ramp up in nVidia's first financial quarter (Jan 26th - April 26th). Considering that they couldn't even say early Q1 has me a bit worried.

[TechZeal]

Originally Posted by Avid6eek

It's official....nVidia has confirmed the delays in their conference call with financial analysts:

http://xbitlabs.com/news/video/displ...y_in_2010.html

Expect card production to ramp up in nVidia's first financial quarter (Jan 26th - April 26th). Considering that they couldn't even say early Q1 has me a bit worried.

This sucks... I would like to sell my 4870X2 and move on but want to see what NVidia brings to the table before I make a purchase. Maybe I will sell it off and buy a cheaper ATI now so I have a backup video card if needed later. hmmm.

[TechZeal]

For decent prices, there will be a delay for ATI also... Do to shortages, the 5800 series will be low supplies and will have price increases. Predicted changes aren't due until the beginning of next year. Check out AnandTech here.

For Comixion, this is some thoughts you may be interested in to help with you choices of video cards...

We can’t say we’re happy with any of this, but we can’t accuse AMD and their vendors of acting irrationally here. It’s a lousy situation for consumers, but that’s a shortage for you. When has there ever been a good shortage?

Finally, with these price hikes, our product recommendations are changing some. The 5870 is still the card to get if money is no object, but the 5850 is far more situational since it’s no longer the great bargain it once was. We can get 1GB 4890s for $170 right now, which have become downright cheap compared to our projected $300 for a 5850. Certainly the 5850 whips the 4890 by upwards of 40%, not to mention DX11 and Eyefinity, but at that level it’s commanding a 75% price premium. It’s a $300 card and performs accordingly, but don’t break the bank in order to get a 5850 at these prices.

If you want a cheap 5800 series card, then it looks like you’re out of luck until 2010.

[Comixion]

$300 or so for a 5850? The wise choice both ways would be if you have the extra dough, get the 5870, if not...then the 4890. That's not a good "split-the-difference" monetarily wise on that 5850.
I can get a 4890 locally at a retail store for $175..XFX

[Avid6eek]

The Inquirer's guess on Fermi's arrival is next May....I think that may be a bit late, but April would be a good guess:

http://www.theinquirer.net/inquirer/...-fermi-delayed

That makes it official....assuming a sngle 5970 can perform as well as my current setup, I'm ready for a new set of ATI GPU's.

[3DGuy]

Originally Posted by Avid6eek

The Inquirer's guess on Fermi's arrival is next May....I think that may be a bit late, but April would be a good guess:

http://www.theinquirer.net/inquirer/...-fermi-delayed

That makes it official....assuming a sngle 5970 can perform as well as my current setup, I'm ready for a new set of ATI GPU's.

Yea i saw that also. To be honest i can see it being May... really due to delays in 40nm TSMC process and AMD using it and Nvidia, also the A3 silicon delay, not to mention they need fermi for 3 high end product lines and the Quadro and Tesla lines create more revenue using the same silicon.

Add all these factors together and also tehy want to have enough chips made for a blanket launch of mass avaliability it's going to take some time.... A3 Silicon Start of Jan, Competing for dies's to be produced in mass numbers March, Building the boards and shipping out to partners AIB's etc End April/Start of May

Also a 5970 would be very nice. Surly that reduce power overall, heat and such. Have you checked your case size? Rumour has it that card is 14.5 inches long and a guy struggled to fit that in a large tower case already....

Any news on 5970 Release? i think they wanted end of november time? start of Dec?

[Avid6eek]

I'll have no problems fitting any size graphics card in my system....it's on a benching platform and not in a case.

I'd like some info on ATI's next generation card and when we could expect that. I hope they do a tech demo as soon as Fermi launches just as nVidia did to them. TSMC should have their 40nm process fixed in the near future. It was partially ATI's fault for not ordering enough wafers...they were expecting Fermi competition.