[pocketmoon_]

'GeForce FX 5800 and 5800 Ultra are in production'

http://news.com.com/2100-1040-966103.html?tag=fd_top

[Evildeus]

Originally Posted by pocketmoon_

'GeForce FX 5800 and 5800 Ultra are in production'

http://news.com.com/2100-1040-966103.html?tag=fd_top

Just before me

Plus that on NV30 and NV31/34 production

http://www.digitimes.com/NewsShow/Ar...ages=04&seq=28

[maskrider]

At US$499, it may have 256MB on board.

[RussSchultz]

The fact that it requires flip chip packaging may suggest that it has a 256 pin external memory bus.

[Grall]

Russ,

There are other reasons you may want to use FC packaging instead than just higher pin count. For example, thermal dissipation is better for FC than the older style used previously. You're also no longer restricted to only having solder pads at the edges of the chip like with older technology, you can have them wherever they may be needed. Clock signal, power and ground distribution is probably rather critical in a 500MHz (unconfirmed?), 125 million transistor chip, so that would probably be a good reason in itself to go to a FC package. You can add clock, power and ground wherever it is needed, plus you don't need to route I/O all the way out to the edge to the chip either. Probably makes things easier when designing it.

*G*

[Sabastian]

Originally Posted by RussSchultz

The fact that it requires flip chip packaging may suggest that it has a 256 pin external memory bus.

That is a good point... but I would suggest that on the same hand it will limit, as it does in the Radeon 9700s case, how low the price of the card will drop as a result of a more expensive PCB. I have been wondering about the design of the Radeon 9700 PCB it seems that it is well designed. Consider the size of the card (relatively small compared to even the Geforce 4 family.) itself and you can conclude that the PCB design was intended to be made as cheaply/efficiently as possible and still retain the needed 256 bus. If you look at the samples provided by nvidia in their video you can see how the NV30 PCB looks considerably more complex.. this could only result in a more expensive PCB? Correct me if I am wrong here.(This of course assumes that the cards we witnessed in the video were actually NV30 reference cards.)

Indeed if the NV30 does use a 256bit bus with a core running @ 500mhz on the .13micron process and utilizing DDRII Nvidia will likely hold the performance crown with this monster for some time..... but what of ATIs potential response to this product? I doubt very much that ATi would manage an over clock of the R300 core to anywhere near 500mhz so it may be that there is no response to the NV30 using 256bit bus at all..

[SlmDnk]

http://money.cnn.com/2002/11/15/tech...idia/index.htm

Story + two tech demo pictures, I think...

[jb]

Originally Posted by Sabastian

Indeed if the NV30 does use a 256bit bus with a core running @ 500mhz on the .13micron process and utilizing DDRII Nvidia will likely hold the performance crown with this monster for some time..... but what of ATIs potential response to this product? I doubt very much that ATi would manage an over clock of the R300 core to anywhere near 500mhz so it may be that there is no response to the NV30 using 256bit bus at all..

Cost. Just like they did when the GF4 came out with their 8500 style cards. They will drop the price. Currently I dont believe the volumes are there yet to mass produce the NV30 on .13u products to lower the cost of the core. Sure someday (maybe soon) they will. But today I doubt it. They probably both then will have the same complexity of PCB (10 layer) since they are both use 256 in this setup. Thus the NV30 will have higher cost factors since they will be using DDR2 which is a bit more costly than DDR.....

[pocketmoon_]

Originally Posted by SlmDnk

http://money.cnn.com/2002/11/15/technology/comdex_nvidia/index.htm

Story + two tech demo pictures, I think...

" It's also the first chip to combine the efforts of nVidia and former employees of 3dfx"

and a picture of the fairy from the videos

[Laa-Yosh]

NV30 has a 128 bit memory bus.

[pocketmoon_]

Originally Posted by Laa-Yosh

NV30 has a 128 bit memory bus.

Do we get a link ?

[Nappe1]

I doubt that NV30 would have 256Bit bus. They have so many times stated that it's not for them yet.

IMO, with DDR-II running on 500MHz (DDR1000) with brand new HSR tech and good Memory Controller is enough for 500MHz core. I also expect to see 8x1 configuration, because it seems to be most efficient for memory bandwidth available. (and besides, after ATI introduced chips with only one "TMU", the pixel fill rate has been became more important "hype" factor than texel fillrate.)

[Joe DeFuria]

500 Mhz is certainly very impressive if nVidia can pull it off! Though not quite as impressive if it means a $500 price tag.

500 Mhz core + 500 MHz, 128 bit DDR Ram vs. 325 Mhz Core and 310 MHz 256 bit ram...should be an interesting exercise in finding out relative bottlenecks.

Though the (currently still a rumor, IMO) naming convention is strange: 5800 and 5800 Ultra?

nVidia has never used the "ultra" moniker since the GeForce2 architecture...before they started using numerics to indicate relative performance (like Ti 200/500, 4200/4400, etc.)

One would think that if there's a difference in clock speed between these two high-end nVidia chips, nVidia would just stick with numerics: possibly 5600 and 5800? Is the 5800 Ultra a 256 MB version?

Definitely some odd and even conflicting rumors flying around even at this late stage. Trying hard to not look into things for a few more hours (I assume) when the NDA is lifted and everything is official...

[Randell]

Originally Posted by Joe DeFuria

Though the (currently still a rumor, IMO) naming convention is strange: 5800 and 5800 Ultra?

nVidia has never used the "ultra" moniker since the GeForce2 architecture...before they started using numerics to indicate relative performance (like Ti 200/500, 4200/4400, etc.)

One would think that if there's a difference in clock speed between these two high-end nVidia chips, nVidia would just stick with numerics: possibly 5600 and 5800? Is the 5800 Ultra a 256 MB version?

I dont know Joe, it look like a cleaner naming convention than the 3 MX and 3 Ti names with the Gf4 launch.

One main product, with where 5800 > 4600 and an ultra version makes more sense to me and should be clearer with the public following the Gf4 naming criticisms.

Does this hint though that ATI will hold the DX9 mainstream market though for a while with the 9500Pro?

[Laa-Yosh]

I'm sure that sooner or later you will... but right now, you'll have to trust me on that. It's 128 bit and not 256 ;)

[tieros]

Originally Posted by SlmDnk

http://money.cnn.com/2002/11/15/technology/comdex_nvidia/index.htm

Story + two tech demo pictures, I think...

New nVidia chip to bring cinema-quality graphics to Sims, EverQuest programs.

ROFL I laughed so hard I think I hurt myself ! I guess the "real" name of the card is going to be "SimsFX" ?

Never let financial guys review a tech product

[Joe DeFuria]

I dont know Joe, it look like a cleaner naming convention than the 3 MX and 3 Ti names with the Gf4 launch.

Well, I don't know if it's "cleaner" or not....because we don't know what the differences are between the 5800 and 5800 "ultra". I was just noting that this would be a change in naming convention for nVidia, and trying to figure out what that might mean in terms of product line up.

[malcolm]

Originally Posted by Laa-Yosh

I'm sure that sooner or later you will... but right now, you'll have to trust me on that. It's 128 bit and not 256

After reading it is only 25-50percent faster than the radeon 9700 im glad its a 128bit bus...

[DadUM]

Originally Posted by Evildeus

Plus that on NV30 and NV31/34 production

http://www.digitimes.com/NewsShow/Ar...ages=04&seq=28

Let's see, with the 100 day tap-out to production level nVidia has done in the past, that would put NV31/NV34 for April release. The next few months should be very interesting for graphics cards.

[malcolm]

Originally Posted by Joe DeFuria

Well, I don't know if it's "cleaner" or not....because we don't know what the differences are between the 5800 and 5800 "ultra". I was just noting that this would be a change in naming convention for nVidia, and trying to figure out what that might mean in terms of product line up.

The difference will probably be faster clock speed
I agree its much better naming this way

[RussSchultz]

Originally Posted by malcolm

After reading it is only 25-50percent faster than the radeon 9700 im glad its a 128bit bus...

Why are you glad?

[Joe DeFuria]

I would be glad too...

in the sense that if it is in fact 20-30% faster (across the board) than the Radeon 9700, and it's doing that with significantly less raw bandwdith....think of what that means for the NV3x architecture on a 256 bit bus.

[Sabastian]

Originally Posted by RussSchultz

Why are you glad?

I was asking the same question. But "only" 25-50 % faster is a bit of a funny thing to say as well. I can only presume that he infers that the NV30 would be considerably faster if it had a 256 bit bus. Of course assuming that the NV30 was designed with the 128bit bus in mind there should be no 256bit bus NV3X anything ever.

[malcolm]

Originally Posted by Sabastian

I was asking the same question. But "only" 25-50 % faster is a bit of a funny thing to say as well. I can only presume that he infers that the NV30 would be considerably faster if it had a 256 bit bus. Of course assuming that the NV30 was designed with the 128bit bus in mind there should be no 256bit bus NV3X anything ever.

I meant the 2x speed increase of a 256bit bus is still to come in the future...
Maybe nv35 or maybe nv40, doesnt mather, but it hasnt been used yet
Anyway it would be a disapointment if it would be only 25-50percent faster with almost twice the bandwidth as the radeon 9700 if it would have a 256bit bus.

[Psikotiko]

http://www.theregister.co.uk/content/3/28153.html

"First up, a faster processor: 500MHz v. 325MHz in used in the TI4600. Next, there’s more memory bandwidth 1GHz DDRII DRAM (600MHz DDR1); new anti-aliasing technology, dubbed Intellisample, designed to double the efficiency of memory reads and writes; and AGP 8X bandwidth (double that of AGP 4X). Built using 0.13micron production technology, the GeForce FX also handles eight pixels per clock cycle, against 4 for its immediate predecessor, by definition doubling the theoretical fill rate."