Sager 9860 benchmarks ATI vs NVIDIA

It will be interesting to see what the Nvidia GDDR3 scores will be like. But those will have to wait until February. *sigh*

I'm (pleasantly) surprised by the OpenGL (Doom3) results.

I'm returning my current 9860 and have a new one on order. I was a little worried about the GL performance. I'm more optimistic now and have to wonder how the 6800 with DDR3 will perform?

Great job Adam, thanks for the info!

When Adam said the X800 has 350MHz memory clock, is that before or after applying the 2x DDR factor? I ask because I think the 600MHz memory clock of the GeForce 6800 Go is already multiplied by the x2 factor, right?

As a Linux user, I think there's more than just benchmarks at stake. One company support Linux the other only provides lip service (bad driver architecture, etc.). There's still very good reasons to use the Nvidia chip. Once the DDR3 version is released, perhaps even Windows users might like it.

Regardless, every time, EVERY time, I go back to ATI I get burned... I'm not going to do it again.

The drivers used with the 6800Go make a huge difference to 3DMark '05 scores, I achieved over 3900 overclocked using v71.20 and other users have achieved more than 4100 stably. Basically Adam's review shows what a number of us suspected - an OC'd 6800Go is at least as fast (in some cases faster) than the stock x800 mobility. Add to this the appearance that the x800 mobility doesn't overclock nearly as well giving only marginally better scores over the OC'd 6800Go and it's questionable whether it's worth choosing it over a 6800Go (considering that you give up Shaders 3.0 support, Linux support and some 3D modelling/animation performance).

Adam's benchmarks show that the x800 mobility overclocked is only 9% faster on average than the current 6800Go overclocked. To put this in perspective - running the latest games at max res. and max settings, FPS on the 6800Go can drop down to 20-30fps at some points in a game, with an average being somewhere in the range of 40-60fps (depending on the game), if you add 9% to a base score of 30fps you get 32.7fps, add 9% to a score of 60fps and it's still only 65.4fps, not that much of a difference when you think about it. Admittedly there are other factors like optimised code for certain games such as Half Life 2 which still favors ATI over NVidia, or Doom 3 which favors NVidia over ATI.

Definitely throws some light on the subject for those considering an upgrade, it may make more sense now to wait for the 6800Go revision 2 cards with DDR3 and higher clock rates - these now look like trumping even the x800 mobility!

Comparing an overclocked 6800Go to a stock X800? Linux support is a valid argument, but seriously.........

Adam also made a typo on his comparison, which was the memory speeds. The X800 had GDDR3 running stock @ 350 Mhz, which is a 700Mhz effective memory speed. The 6800Go has DDR with a 600Mhz effective, or 300Mhz DDR base setting. Overclocked, the X800 had an effective memory of 818Mhz (409Mhz DDR), with the 6800Go an effective memory of 680Mhz (340Mhz DDR).

When the refreshed 6800Go comes out with GDDR3 instead of DDR(2?), then maybe the 6800Go will be better for gaming, but until then, the X800 just wipes the floor with it.

BTW, you should use the game FPS when making your argument, not 3DMark05. '05 is not very indicative of game performance.

This isn't an entirely unreasonable thing to do considering that the MR X800 overclocked is only 9% faster on average (based on Adam's benchmarks - which include game benchmarks) than the stock MR X800 (compared with a 22% on average and over 30% at best overclock with the 6800Go).

That's simply NOT what Adam's benchmarks (which include game benchmarks) indicate - a 9% lead on average (13% at best) when comparing the OC'd X800 to the OC'd 6800Go is not what I would call "wiping the floor" with the 6800Go.

This really is only partially true, 3D Mark '05 is probably the best synthetic benchmark of latest release and upcoming game performance, granted there are other factors including optimized code which can give one chipset an advantage over another. Besides which, I wasn't using 3DMark '05 to make an argument about 6800Go vs. MR X800 performance - I was just correcting the false assumption that there is such a large discrepancy between the two cards in this particular benchmark. The 6800Go scores significantly better than the graph indicates when coupled with the latest drivers (this incidentally applies to other benchmarks too - but I wasn't being pedantic at the time).

Arent there another set of drivers for the ATI card aswell. I thought I had read that Omega drivers significantly improved other ATI cards performance. Is it not safe to assume there will be some of these drivers released for the x800. So really the 71.20 comment is apples to oranges. I personally would like to see a benchmark where both cards are clocked at the same speeds, high and low. I would also like to see a benchmark with both maxed out and one where the play Everquest 2 on Extreme Quality.

The one thing I really wanted to see in the comparison was the power draw, although, I do not know how one would test it in a notebook. Heat is always something everyone here is concerned about, & everything the low-k chip w/ GDDR3 indicates it runs cooler.

Adam, could you post the stock & overclocked core temps for both as well?

Andre, 9% to 12% is large margin in GPU's. The desktop 6800GT is within 10% of the 6800Ultra. ATi released a refresh of the X850XT PE, which was only within 5% or less of the X800XT PE. When it comes to notebook gaming, that same percentage equals FPS that you will actually see in game.

When it comes to drivers, who thought the X800 would be within 0.2% of average FPS in Doom3 compared to a 6800? ATi really has been working hard on those OpenGL drivers, & if parity doesn't good enough, then I don't know what would be. The DX9 drivers have been superior to nVidia for years, now the OpenGL is nothing to sneeze at also.

As for the Linux support, no one can argue with that. But for serious gamers, not workstation/programmers, it's a non-issue anyway.

That is partly true, there probably will be/are Omega drivers that will work with the MR X800 and these probably would improve it's performance, however the performance increase of using these sorts of drivers is not usually in the order of 19% such as the performance gain from using the later NVidia drivers - this indicates to me a flaw in the earlier drivers which Adam uses for the review.

I agree this would be interesting but really what matters is how fast they CAN go when running at optimal performance - this is what Adam's review gives us an indication of, which as I've already stated points to the MR X800 being only 9% faster OC'd than the 6800Go OC'd (we can only presume the MR X800 Overclock Adam uses is comparatively close to the top end of it's performance band as the 6800Go's OC is to it's).

Adam's review does pit the two cards overclocked, the 6800Go is OC'd near it's limit and we can only presume that the OC Adam chose for the X800 was comparatively near it's limit. He's given us a fairly broad spectrum of results to consider across both current game benchmarks and synthetic ones, EQ2 would make a good benchmark as it's highly demanding on the GPU (some might argue - not very well coded), a Half-Life 2 comparison would also be interesting as it tends to favour ATI over NVidia.

Actually 9-12% is not a large margin; the performance gain of the 6800Go over the MR9700 was up to 135% - that's a large margin.

I doubt that most gamers are going to see the difference between 30fps and 32.7fps in demanding scenes or between 60fps and 65.4fps in less demanding scenes - that a 9% performance gain achieves.

The comparison is always between cards of the same generation, andre, not 2 completely different generations. The 6800Go is 550% better than the MR9000, too, but it isn't meant to compete with the MR9000, or the MR9700, either. That's what the 6600Go is for.

Serious gamers don't use Windows.

To each his/her own I suppose.

I've been gaming on Linux for years.

Having an Ati-favored game isn't a good comparison when an nVidia-favored game is (Doom3)?

Andre, you have your 6800Go & you like it. Why does it matter to you what the X800 does? If Adam's benches aren't changing your mind, what else would you need?

I think nVidia did a great job with their 6800Go.
Any ideas or rumors when the the 6800 with DDR3 memory is supposed to be released ?

EDIT:
Sorry, just saw the post from Adam (or was it Luke?) about that.

LOL. "Serious gamers don't use _________".

How's the Linux Doom3 going for you? According to the latest comparison, windows is ~25% faster than linux:

Anandtech on Linux gaming compared to Windows (Oct 2004)

Some people, including myself, don't like Doom3 as a game, but it sure is a game engine that is being used as a template for the future, if there was any.

I'm sure there are some Linux games that run on par with windows, but it just isn't as much a priority for the programmers as Windows XP is. But hey, to each their own I suppose.