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
Nick388 - Wednesday, September 24, 2008 - link
Never thought the 8800 GT was this good! I found some really low prices at this website (http://www.videocardemporium.com)">http://www.videocardemporium.com), but I'm still unsure if I want to try to get two of these in SLI or go for the 9600 GT. Or maybe just slurge for the GTX 280 >.<sickish - Monday, February 25, 2008 - link
Just ordered mine Friday, should be here Wednesday.I was lucky, ordered it from Newegg at $239.99 and it seems totally worth it after reading these comments.
Oh and I checked again today and it seems Newegg is all sold out =]]
mgambrell - Monday, November 12, 2007 - link
geforce 8800 gt: the only card that crashes my PC in nv4_disp over and over and over.no matter which motherboard it is in or which OS is installed or which other expansion cards are installed
no matter whose 8800 I use.
nvidia you are now on my shitlist.
gleblanc - Friday, November 9, 2007 - link
It'd be awfully nice to have the axes of the graphs labeled. For the first set, I can guess that they are screen resolution on the horizontal axis, and frames per second on the vertical, but I could be wrong, since there's no labels.I also couldn't follow the page on comparing the 8800 GT to the 8800 GTX. Your conclusion seems to state that, "the 8800 GT doesn't touch the GTX." However, I can't come up with that conclusion from the graphs. They look roughly comparable in most of the tests that you've shown, with only a slight advantage to the GTX at very high resolutions.
On the "Out with the old..." page, there is a typo in the second paragraph. In the last sentence, "fasted" should be "fastest".
True Strike - Thursday, November 8, 2007 - link
From what I have read here and else where this seems to be THE card to get. Before I make a purchase though, I would very much like to see more data comparing the cards offered by the different card manufacturers.Is Anandtech going to be doing an 8800 GT roundup any time soon? Do I have to beg?
Manch - Monday, November 5, 2007 - link
How much of an improvement if you SLI'd a 8800GT with a GTX? I know mix n matching is not optimal but the price difference makes me wonder. Would it fall between two GT's and two GTX's? I dont have any experience with SLI. I've avoided it because it's never been a decent upgrade path.hydrasworld - Friday, November 2, 2007 - link
Ordered mine from Scan.co.uk - Gainward Bliss - £170 GBPO.M.G - it really is amazing.
I bought 2x 7800GTX's (just before the 7900GTX's came out) at £660 for the pair and this card just blows them away.
On my 24" dell monitor at 1920x1200 with 4xAA on an opteron 175 @ stock 2.2ghz i get average 125fps in Team Fortress 2. Everything else I've tried has been very smooth - world in conflict, bf2142. A *very* noticable performance increase at a billiant price!
If you're considering an upgrade buy one of these NOW, play todays games at awesome speeds. then get a nice new intel 45nm quadcore + x38 + ddr3 in january when the products are released and prices will be lower.. Then if needed sell the gfx card and buy whatever nvidia are offering in january - if you really need too (which i doubt you will).
Highly Recommended.
rap - Friday, November 2, 2007 - link
Info on the RV670:
http://www.penstarsys.com/editor/so3d/q4_2007/so3d...">http://www.penstarsys.com/editor/so3d/q4_2007/so3d...
ksherman - Wednesday, October 31, 2007 - link
A great review, and a great part! Finally a video card that excites me. Too bad I ditched the desktop in favor of a laptop.Screammit - Wednesday, October 31, 2007 - link
Most online retailers have pulled these items off their websites entirely, as I'm sure these cards have been picked up ravenously by gamers wanting The Holy Grail of video cards, as it seems this is.My question is, "Why so cheap, and why now?" for injection into the market, NVIDIA could have raised the price at least $50 (which most retail shops have already done to capitalize on its popularity) and still have a product that sells like crazy. This makes me wonder what is next, and if a better product is in the works that makes them want to get rid of this inventory as quickly as possible before the next big thing comes out. It may be (and yes, I'm reaching) that this card is on the low side in NVIDIA's new product line, and they can clear inventory at a price premium now as opposed to when the full line is released. They have little reason to throw out their best until AMD has shown their hand, and are playing the same game that Intel is, with their 3.0ghz processor that can easily be clocked higher.
With this in mind, I plan on holding on to my money for now, partially because I can't even find one in stock yet, and partially because having this card at this price point doesn't seem to make much sense unless a full line refresh is coming, and this card is the weakest link, which is an incredible thing to think about, considering how good this card appears to be