NVIDIA GeForce 8800 GT: The Only Card That Matters
by Derek Wilson on October 29, 2007 9:00 AM EST- Posted in
- GPUs
G92: Funky Naming for a G80 Derivative
If we expect the G9x to represent a new architecture supporting the GeForce 9 series, we would be wrong. In spite of the fact that part of the reason we were given for NVIDIA's move away from NVxx code naming was to bring code name and product name closer to parity (G7x is GeForce 7, G8x is GeForce 8), it seems NVIDIA has broken this rule rather early on. Code names are automatically generated, but how we only ended up with three different G8x parts before we hit G9x is certainly a mystery. One that NVIDIA didn't feel like enlightening us on, as it no doubt has to do with unannounced products.
While not a new architecture, the GPU behind the 8800 GT has certainly been massaged quite a bit from the G80. The G92 is fabbed on a 65nm process, and even though it has fewer SPs, less texturing power, and not as many ROPs as the G80, it's made up of more transistors (754M vs. 681M). This is partly due to the fact that G92 integrates the updated video processing engine (VP2), and the display engine that previously resided off chip. Now, all the display logic including TMDS hardware is integrated onto the GPU itself.
In addition to the new features, there have been some enhancements to the architecture that likely added a few million transistors here and there as well. While we were unable to get any really good details, we were told that lossless compression ratios were increased in order to enable better performance at higher resolutions over the lower bandwidth memory bus attached to the G92 on 8800 GT. We also know that the proportion of texture address units to texture filtering units has increased to a 1:1 ratio (similar to the 8600 GTS, but in a context where we can actually expect decent performance). This should also improve memory bandwidth usage and texturing power in general.
Because NVIDIA was touting the addition of hardware double precision IEEE 754 floating point on their workstation hardware coming sometime before the end of the year, we suspected that G92 might include this functionality. It seems, however, that the hardware behind that advancement has been pushed back for some reason. G92 does not support hardware double precision floating point. This is only really useful for workstation and GPU computing applications at the moment, but because NVIDIA design one GPU for both consumer and workstation applications, it will be interesting to see if they do anything at all with double precision on the desktop.
With every generation, we can expect buffers and on chip memory to be tweaked based on experience with the previous iteration of the hardware. This could also have resulted in additional transistors. But regardless of the reason, this GPU packs quite a number of features into a very small area. The integration of these features into one ASIC is possible economically because of the 65nm process: even though there are more transistors, the physical die takes up much less space than the G80.
If we expect the G9x to represent a new architecture supporting the GeForce 9 series, we would be wrong. In spite of the fact that part of the reason we were given for NVIDIA's move away from NVxx code naming was to bring code name and product name closer to parity (G7x is GeForce 7, G8x is GeForce 8), it seems NVIDIA has broken this rule rather early on. Code names are automatically generated, but how we only ended up with three different G8x parts before we hit G9x is certainly a mystery. One that NVIDIA didn't feel like enlightening us on, as it no doubt has to do with unannounced products.
While not a new architecture, the GPU behind the 8800 GT has certainly been massaged quite a bit from the G80. The G92 is fabbed on a 65nm process, and even though it has fewer SPs, less texturing power, and not as many ROPs as the G80, it's made up of more transistors (754M vs. 681M). This is partly due to the fact that G92 integrates the updated video processing engine (VP2), and the display engine that previously resided off chip. Now, all the display logic including TMDS hardware is integrated onto the GPU itself.
In addition to the new features, there have been some enhancements to the architecture that likely added a few million transistors here and there as well. While we were unable to get any really good details, we were told that lossless compression ratios were increased in order to enable better performance at higher resolutions over the lower bandwidth memory bus attached to the G92 on 8800 GT. We also know that the proportion of texture address units to texture filtering units has increased to a 1:1 ratio (similar to the 8600 GTS, but in a context where we can actually expect decent performance). This should also improve memory bandwidth usage and texturing power in general.
Because NVIDIA was touting the addition of hardware double precision IEEE 754 floating point on their workstation hardware coming sometime before the end of the year, we suspected that G92 might include this functionality. It seems, however, that the hardware behind that advancement has been pushed back for some reason. G92 does not support hardware double precision floating point. This is only really useful for workstation and GPU computing applications at the moment, but because NVIDIA design one GPU for both consumer and workstation applications, it will be interesting to see if they do anything at all with double precision on the desktop.
With every generation, we can expect buffers and on chip memory to be tweaked based on experience with the previous iteration of the hardware. This could also have resulted in additional transistors. But regardless of the reason, this GPU packs quite a number of features into a very small area. The integration of these features into one ASIC is possible economically because of the 65nm process: even though there are more transistors, the physical die takes up much less space than the G80.
Facebook
Twitter
RSS
90 Comments
View All Comments
DukeN - Monday, October 29, 2007 - link
This is unreal price to performance - knock on wood; play oblivion at 1920X1200 on a $250 GPU.Could we have a benchmark based on the Crysis demo please, how one or two cards would do?
Also, the power page pics do not show up for some reason (may be the firewall cached it incorrectly here at work).
Thank you.
Xtasy26 - Monday, October 29, 2007 - link
Hey Guys,If you want to see Crysis benchmarks, check out this link:
http://www.theinquirer.net/gb/inquirer/news/2007/1...">http://www.theinquirer.net/gb/inquirer/.../2007/10...
The benches are:
1280 x 1024 : ~ 37 f.p.s.
1680 x 1050 : 25 f.p.s.
1920 x 1080 : ~ 21 f.p.s.
This is on a test bed:
Intel Core 2 Extreme QX6800 @2.93 GHz
Asetek VapoChill Micro cooler
EVGA 680i motherboard
2GB Corsair Dominator PC2-9136C5D
Nvidia GeForce 8800GT 512MB/Zotac 8800GTX AMP!/XFX 8800Ultra/ATI Radeon HD2900XT
250GB Seagate Barracuda 7200.10 16MB cache
Sony BWU-100A Blu-ray burner
Hiper 880W Type-R Power Supply
Toshiba's external HD-DVD box (Xbox 360 HD-DVD drive)
Dell 2407WFP-HC
Logitech G15 Keyboard, MX-518 rat
Xtasy26 - Monday, October 29, 2007 - link
This game seems real demanding. If it is getting 37 f.p.s. at 1280 x 1024, imagine what the frame rate will be with 4X FSAA enabled combined with 8X Anistrophic Filtering. I think I will wait till Nvidia releases there 9800/9600 GT/GTS and combine that with Intel's 45nm Penryn CPU. I want to play this beautiful game in all it's glory!:)Spuke - Monday, October 29, 2007 - link
Impressive!!!! I read the article but I saw no mention of a release date. When's this thing available?Spuke - Monday, October 29, 2007 - link
Ummm.....When can I BUY it? That's what I mean.EODetroit - Monday, October 29, 2007 - link
Now.http://www.newegg.com/Product/ProductList.aspx?Sub...">http://www.newegg.com/Product/ProductLi...18+10696...
poohbear - Wednesday, October 31, 2007 - link
when do u guys think its gonna be $250? cheapest i see is $270, but i understand when its first released the prices are jacked up a bit.EateryOfPiza - Monday, October 29, 2007 - link
I second the request for Crysis benchmarks, that is the game that taxes everything at the moment.DerekWilson - Monday, October 29, 2007 - link
we actually tested crysis ...but there were issues ... not with the game, we just shot ourselves in the foot on this one and weren't able to do as much as we wanted. We had to retest a bunch of stuff, and we didn't get to crysis.
yyrkoon - Monday, October 29, 2007 - link
Yes, I am glad instead of purchasing a video card, I instead changed motherboard/CPU for Intel vs AMD. I still like my AM2 Opteron system a lot, but performance numbers, and the effortless 1Ghz OC on the ABIT IP35-E/(at $90usd !) was just too much to overlook.I can definitely understand your 'praise' as it were when nVidia is now lowering their prices, but this is where these prices should have always been. nVidia, and ATI/AMD have been ripping us, the consumer off for the last 1.5 years or so, so you will excuse me if I do not show too much enthusiasm when they finally lower their prices to where they should be. I do not consider this to be much different than the memory industry over charging, and the consumer getting the shaft(as per your article).
I am happy though . . .