• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Early NV50 Specs!

Status
Not open for further replies.
H

HeavenX

Limp Gawd
Joined
Jul 25, 2004
Messages
204
350 Million Transistors on 90-nm process
800MHz Core
16MB Embedded DRAM (134M trans.)
1.6GHz 256-512MB GDDR3 Memory
32 Pixel Pipelines
16 Vertex Pipelines
204.8 GB/sec Bandwidth (eDRAM)
51.2 GB/sec Bandwidth (GDDR3)
12.8 Billion Pixels per Second
1.6 Billion Vertices per Second
Shader Model 3.0 (maybe Shader Model 4.0)
Graphics Multi-Threading
Virtual Memory Management
PCI Express 16x
Release Q2/Q3 2005
 
Fuck....FUCK....that thing might just be powerful enough to take 2 6800Us in SLI. So where did you get this info from anyways?
 
HeavenX said:
800MHz Core
1.6GHz 256-512MB GDDR3 Memory
32 Pixel Pipelines
16 Vertex Pipelines
Click to expand...
thats all i see when i read your post
:p

god i hope thats true.
but umm...
batty said:
So where did you get this info from anyways?
Click to expand...
is what i wanna know also ;)
 
Heh, just imagine these babies in SLI. Elder Scrolls 4 Oblivian will get obliviated. :D
 
doh-nut said:
ill take 2 in SLI please, for 64 pipelines. thx
Click to expand...

2? Shit...

With specs like that, and NV40s not depreceating due to SLI, that beast would probably cost like 1k American per card!!!! I guess a guy can dream though
 
It will not cost 1k. It will be around $600-$700 dollar range. It will not be over $700 dollars thats for sure.
 
HeavenX said:
It will not cost 1k. It will be around $600-$700 dollar range. It will not be over $700 dollars thats for sure.
Click to expand...

How do you know this? Where's your source.
 
HeavenX said:
350 Million Transistors on 90-nm process
800MHz Core
16MB Embedded DRAM (134M trans.)
1.6GHz 256-512MB GDDR3 Memory
32 Pixel Pipelines
16 Vertex Pipelines
204.8 GB/sec Bandwidth (eDRAM)
51.2 GB/sec Bandwidth (GDDR3)
12.8 Billion Pixels per Second
1.6 Billion Vertices per Second
Shader Model 3.0 (maybe Shader Model 4.0)
Graphics Multi-Threading
Virtual Memory Management
PCI Express 16x
Release Q2/Q3 2005
Click to expand...


GDDR3??? WHY GDDR3 if tis is THAT advanced...why not ddr4? since thats what ati will most likely be using...

PCI-e 16x - DUH...

SM4? We dont even use SM3 yet...but still...thats would be cool...

204.8 GB/sec BW??? wow...that will be affected by some other system bottlenecks...but then again...that is a THEORETICAL limit..

Graphics Multi-Threading? ooohh....HyperThreading on a graphics chip...oooh...id rather have dual processors on the graphics chip...but still would be bottlenecked somewhere else in the system...

32 PX pipelines.... is that 32 HW lines or 32 implied? (implied = 16 going twice as fast)
 
zapian_ said:
cpu and system memory bottleneck? anyone?
Click to expand...


no...at these speeds...i dont even think a 3gbps sata2 would help releive any bottlenecks...or if you use DDR2 667MHz ram...even at dual channel...

Bottleneck: Priceless
 
i dont belive these specs for a moment. 1.6 Billion Vertices per Second ... er.. yea right.
 
Those numbers don't make sense..........

350 Million Transistors on 90-nm process
16MB Embedded DRAM (134M trans.)
Click to expand...
So, take away the EDRAM and it's a 216M transistor part - less than NV40 AND with a 32x pixel pipeline architecture and a bunch of new stuff??

800MHz Core
32 Pixel Pipelines
12.8 Billion Pixels per Second
Click to expand...
Huh? 800MHz x 32 pipes (assuming 32 x 1) = 25.6GPixel/s

Looks like BS.
 
Hmmm, these look a LOT like the same rumours that were doing the rounds for NV40........

http://www.vr-zone.com/Home/news181/news181.htm#2922

- 0.09 Micron Process
- 300-350 Million Transistors
- 750-800 MHz Core
- 16MB Embedded DRAM (134M trans.)
- 1.4 GHz 256-512MB DDR-II Memory
- 16 Vertex Shader Engines
- 204.8 GB/sec Bandwidth (eDRAM)
- 44.8 GB/sec Bandwidth (DDR-II)
- 25.6 GigaTexels per Second

Conclusion? Total BS.
 
bran(d)muffin said:
GDDR3??? WHY GDDR3 if tis is THAT advanced...why not ddr4? since thats what ati will most likely be using...
Click to expand...
Uh.. last I heard ATi was using GDDR3 memory on the R5xx but that it's being made compatible for GDDR4 for when GDDR4 starts shipping in sufficient quanities. I would suspect that NV is doing the same thing. :rolleyes:
 
coz said:
Those numbers don't make sense..........

So, take away the EDRAM and it's a 216M transistor part - less than NV40 AND with a 32x pixel pipeline architecture and a bunch of new stuff??

Huh? 800MHz x 32 pipes (assuming 32 x 1) = 25.6GPixel/s

Looks like BS.
Click to expand...

exactly what i thought...
 
maybe some !!!!!! just got jealous that there's r5xx news before nVidia's nv5x?.......

I don't think a Pentium M @ 3ghz would bottleneck something like this(if it were real):D:D
 
Released Q2/Q3? that means you won't be able to buy one until 2006, the way production has been.
 
tbh i dont think nvidia even needs to make an "nv50" new architecture to compete with r520. if they just jack up the pipes of the nv40 it would do the job.
 
SocketA said:
Released Q2/Q3? that means you won't be able to buy one until 2006, the way production has been.
Click to expand...
You could say the same thing for R520 due for launch in H1 '05. R420 was 'launched' in Q2 '04 and how easy is it to buy them now? You may also be waiting until the very end of 2005 and early 2006 to buy an R520 card. I hope not but I'm just saying it's possible

Besides, these 'specs' are so obviously fraudulent that speculating on them is a total waste of time.
 
tranCendenZ said:
tbh i dont think nvidia even needs to make an "nv50" new architecture to compete with r520. if they just jack up the pipes of the nv40 it would do the job.
Click to expand...
There is a rumour going around that NV are going to make some changes to the pixel pipes on NV48. We may see an extra TMU per pipeline for a 16x2 arrangement? Personally I think it'll stay at 16x1 and just the clocks will be increased though.
 
You really cant believe anything HeavenX says. Hes proven this time and time again.
 
lol if anyone takes these specs as gospel you might be disappointed. The numbers just don't add up. The specs are self-contradictory.
 
Maybe 18-22 pixel pipes, but 32? Whatever. This is BS out the butt... You wouldnt even come close to needing 32 until something like Unreal Engine 3.0...
 
remember the unreal 3 promo's at E3? it ran on a 6800 ultra, which was prolly modded, but it still ran on it. i dont htink it would need 32 lol
 
Well, my point was saying, maybe you'd see UE 3 come CLOSE to using all the pipes at 1600x1200 with max AA and AF. Bottom line: The gaming world isnt ready for and doesnt need 32 pixel pipelines. Therefore, Nvidia will probably not be putting that many in the next card.
 
yeah 800 mhz is rediculous, that would require constant LN2 supply to operate at those speeds all of those specs are wrong, i dont think you would need 16 vertex pipes either lol
 
Status
Not open for further replies.
Back
Top