Fall '06 NVIDIA GPU Refresh - Part I: GeForce 7900 GSby Derek Wilson on September 6, 2006 9:00 AM EST
- Posted in
It's been a while since we've seen any truly new graphics architectures from ATI or NVIDIA. G7x and R5xx GPUs have had quite a long lifespan, dating back to almost 12 months ago for ATI and 15 months for NVIDIA. When you consider that the G7x architecture is really just a refinement of NV4x (similar to how R4xx was a refinement of R3xx), we could even say that NVIDIA's current architecture is almost 2 1/2 years old! That's practically an eternity in the world of GPUs, where for a while we were seeing new architectures every 18 months or less. Things seem to have settled down a bit in terms of architecture progression, but the war for the fastest GPU(s) certainly hasn't tapered off.
During the past two years, we have seen the introduction of SLI and CrossFire as ways to eke more performance out of the current architectures. NVIDIA has lately taken that a step further with quad SLI configurations, or cards like the 7950 GX2 which feature "SLI on a single card". If we were to conservatively estimate the number of variants of the different graphics cards that have been launched in the past two years, it would certainly be over 40. This proliferation of GPU models has been done in an attempt to target all of the various price points, from the $50 budget cards all the way up to the $600 super cards. Microsoft seems intent to get in on the action as well, as Windows Vista will launch with nine different variants at last count.
We bring up Windows Vista for more reasons than one. With Windows Vista will come the launch of DirectX 10, bringing with it an expanded feature set and improved performance. In order to fully support DirectX 10, however, NVIDIA and ATI need to do a decent amount of work on new architectures. This work takes time, although with the delays of Windows Vista perhaps that isn't a bad thing. We expect to see new graphics architectures from both ATI and NVIDIA either late this year or early in 2007, just in time for Windows Vista to start shipping. In the meantime, the battle for consumer dollars wages on.
In order to keep filling in the gaps, improving value, and competing, NVIDIA is following ATI (who followed NVIDIA...) in releasing another run of rebadged graphics solutions. The two new solutions are the 7900 GS launching today, with the 7950 GT scheduled to launch next week. NVIDIA is targeting the $200 and $300 price points (respectively) with these new offerings in two different ways. For the 7900 GS, NVIDIA is disabling vertex/pixel pipelines, while the 7950 GT will be taking advantage of the high overclockability of the G71 chips. Let's take a closer look at what will be offered with the new GPUs.
What's New Today?
Two new cards are making their debut today, but only one will be available as of this publication. The card available today is the GeForce 7900 GS, for which we will also be able to talk about performance. We will have to wait at least a week for the second part of this two part series to talk about the performance of the other new part: the GeForce 7950 GT. The GeForce 7950 GT will be available on September 14th.
Don't be confused by the 7950 moniker: the 7950 GT is a single GPU solution that is slower than a single 7900 GTX. NVIDIA intends for the GX2 postfix to indicate a dual GPU solution, and those who don't do their research might think a higher marketing number means higher performance. GeForce 7950 GT is a horrible name for this card no matter how we slice it. The plan is for this new card to compete with the X1900 XT (512MB) and highly overclocked versions of the 7900 GT, at a price point above $300 USD.
For today's tests, we will have to make do with only the GeForce 7900 GS. This card will come in between $200 and $250 and compete directly with the X1900 GT. Until now, as we saw in our earlier midrange roundup, the X1900 GT had no direct competitor from NVIDIA at its current price point -- at least if you're talking about MSRPs and prices without mail-in rebates. At present, the X1900 GT is selling for $225-$245, with street prices of the 7900 GT ranging from $262-$312 before rebates, depending on clock speed and features. With rebates, 7900 GT cards can be had for as little as $236. That puts the 7900 GS in an interesting position, as it will need to compete with not only the ATI offerings, but it will also have to go against some of NVIDIA's own products. The 7900 GS will certainly have to fall closer to the $200 price point rather than the $250 price point if it is to be successful.
Both companies have some other products falling close to the $200 price point. ATI offers the X1800 GTO, and you can sometimes find X1800 XL and XT cards carrying similar prices. Availability of some of these parts is sketchy at best, so they are not the primary focus of our current GPU articles. NVIDIA also offers 7600 GT cards ranging in price from $167 ($132 with MIR) to $219. Depending on the performance the 7900 GS offers, we may see 7600 GT prices drop in the coming weeks.
This is obviously a crowded market, and NVIDIA is absolutely hoping that the 7900 GS (and factory overclocked variants) will be able to reclaim the performance and value lead at this very important price point. It is somewhat interesting (humorous?) to note that NVIDIA's literature compares the 7900 GS to the X1800 GTO, which obviously isn't the right comparison to make. Of course we will be running our own performance tests, and in our attempt to discover whether or not NVIDIA has succeeded in recapturing the lead of the mainstream $200 segment, we have an XFX 7900 GS that we will run through our benchmark suite.
Post Your CommentPlease log in or sign up to comment.
View All Comments
munky - Wednesday, September 6, 2006 - link
FEAR is a DX9 game, not OpenGL...
DerekWilson - Wednesday, September 6, 2006 - linkI'm looking into this at the moment but having trouble finding documentation on it.
I suppose, as I was recently testing quad sli and saw huge performance increases, I assumed the game must be using the 4 frame afr mode only possible in opengl (dx is limited to rendering 3 frames ahead). I'll keep looking for confirmation on this ...
MemberSince97 - Wednesday, September 6, 2006 - linkJupiter EX is a DX9 rendering engine...
DerekWilson - Wednesday, September 6, 2006 - linkcorrected, thanks ... now I have to figure out why FEAR likes quad sli so much ...
MemberSince97 - Wednesday, September 6, 2006 - linkNice writeup DW, I really like the mouseover performance % graphs...
PrinceGaz - Thursday, September 7, 2006 - linkSo do I, but there is one error
That should be 14% and 25% advantages
The 7900GS has 20 PS while the 7900GT has 24 PS. That makes the 7900GS 20% slower than the 7900GT, but it makes the 7900GT 25% faster than the 7900GS. It's important to remember which one you're comparing it against when quoting percentages.
Hopefully the percentage performance difference in the graph itself was calculated correctly, or at least consistently.
PrinceGaz - Thursday, September 7, 2006 - linkOoops sorry, please ignore my post. For some reason I thought for a moment the 7900GS had 16 PS and the 7900GT had 20 PS (despite writing the correct values in my comment). The article is correct, I was just getting confused.
PS. an edit function would be nice.
Frackal - Wednesday, September 6, 2006 - linkThere is no way an X1900xt gets 75fps at 1600x1200 4xAA, at that same resolution and AA setting I get well over 120-130fps average with an X1900xtx. Most sites show it hitting at least 100+
DerekWilson - Wednesday, September 6, 2006 - linkif you use the built in demo features to run a timedemo with dice's own calculations you will get a very wrong (skewed upward) number. Dice themselves say that results over 100 fps aren't reliable.
the problem is that they benchmark the load screen, and generally one card or the other will get better load screen performance -- for instance, the x1900 gt may get 300+fps while the 7900 gt may only get 200fps. (I just picked those numbers, but framerates for the load screen are well over 100 fps in most cases and drastically different between manufacturers).
not only does no one care about this difference on a load screen, but it significantly interferes with benchmark numbers.
the timedemo feature can be used to output a file with frametimes and instantaneous frames per second. we have a script that opens this file, removes the frame data for the load screen, and calculates a more accurate framerate average using only frame data for scenes rendered during the benchmark run.
this will decrease over all scores.
we also benchmark in operation clean sweep which has a lot of fog and water. we use a benchmark with lots of smoke and explosions and we test for some ammount of time in or near most vehicles.
splines - Wednesday, September 6, 2006 - linkOwnage approved.