NVIDIA GeForce 8600M GT Mobile GPU Review

[Dashken]

The NVIDIA GeForce 8600M GT is the mobile version of the NVIDIA GeForce 8600 GT. However, instead of being based on the G84 GPU used by the GeForce 8600 GT, it is based on the smaller G86 GPU used by the GeForce 8500/8400 graphics cards. The G86 GPU only packs 210 million transistors, which 79 million or 27% less than the G84 GPU. The result is a smaller and cooler mobile GPU.

Even though the GeForce 8600M GT is based on a different GPU, it is similar to the GeForce 8600 GT desktop GPU in specifications. In fact, the only difference is the GeForce 8600M GT runs at 475 MHz while the GeForce 8600 GT runs at 540 MHz. So, is the GeForce 8600M GT almost as fast as its desktop counterpart? Read on and find out.

Here's a quote from the review :-

"Like its desktop counterpart, the NVIDIA GeForce 8600M GT looks better on paper than in real life. On paper, it is roughly between the desktop GeForce 8600 GT and the GeForce 7600 GT in performance. That makes it a rather powerful mobile GPU, especially compared to integrated graphics solutions like the Intel Graphics Media Accelerator X3100."

Link : NVIDIA GeForce 8600M GT Mobile GPU Review

[doppler_pulse]

Granted while the mobile 8600 GT will definitely be slower, the test rigs were not comparable in their basic configurations. The desktop used a quad core @ 2.66 ghz and 1066 fsb while the mobile cpu was a dual core limited @ 2.00 ghz and an 800 fsb. Furthermore, the memory clocks, hard drive spindle speeds, and hard drive densities would all take their toll on gaming performance.

It would have been best had the reviewer at least reduced the clock speed of the qx6700 if not swapped in an e4400 with similar specs to the t7300. Although many games are not multithreaded, the reviewer's selection included titles such as supreme commander and bioshock which are known to be able to utilize them.

Thus, while I agree the 8600M GT is slower, I do not believe that it is slower by that great a margin.

[Chai]

Originally Posted by doppler_pulse

Granted while the mobile 8600 GT will definitely be slower, the test rigs were not comparable in their basic configurations. The desktop used a quad core @ 2.66 ghz and 1066 fsb while the mobile cpu was a dual core limited @ 2.00 ghz and an 800 fsb. Furthermore, the memory clocks, hard drive spindle speeds, and hard drive densities would all take their toll on gaming performance.

It would have been best had the reviewer at least reduced the clock speed of the qx6700 if not swapped in an e4400 with similar specs to the t7300. Although many games are not multithreaded, the reviewer's selection included titles such as supreme commander and bioshock which are known to be able to utilize them.

Thus, while I agree the 8600M GT is slower, I do not believe that it is slower by that great a margin.

I would agree with you with some of the lower res results, but at high res, they are most likely fill-rate limited, so none of those should make such a big difference anyway.

[odin243]

C'mon guys. I had to register just to comment on this.

I seriously have to inquire whether the author and/or editor of this so-called "article" knows anything about notebook technology. Anything at all? I didn't think so.

Starting at the beginning: The supposed test laptop, the Dell 1501, does not feature an 8600M-GT, and never did, and never will. I can only assume that that is a typo, and that what was actually used was the Dell Inspiron 1520. If that's the case, then it should be noted that it uses a budget version of the 8600M-GT with 400MHz DDR2 VRAM, rendering your listed specifications invalid. It is documented that the DDR2 8600M-GT runs on average 30% slower than the 700MHz GDDR3 8600M-GT, whose specifications are listed in the review. The budget version is extremely bandwith starved, which was not helped by the fact that the notebook tested had only 2GB RAM running on Vista, where it is again documented that at least 3GB is needed for the hardware Turbocache to really come into full effect.

I won't even talk about the processor difference, that's just disgusting. Any professional reviewer would be ashamed of that kind of unaccounted for differential.

Next off, the 8700M-GT, 8600M-GT, and 8600M-GS are all based on the G84M chip. You can claim that that chip is more similar to the G86 than the G84 if you like, but the fact is it's still called the G84M. The G86M would be the chip which the 8400M-GT, 8400M-GS, and 8400M-G are based off of.

Also, what the hell is the "GeForce 7600M GT" supposed to be? No nVidia card ever existed with that name. You may be referring to the Geforce Go 7600, or possibly the Geforce Go 7600GT, both of which are different cards.

Also, about it's DX10 support: It's far from useless, as long as you don't run games on maxed or high settings. The 8600M-GT is a mid-range mobile card, meant to play the newest games at WXGA or WXGA+ with medium settings, which it does admirably.

Your subjective conclusions are absolutely worthless considering the card wasn't tested in it's intended environment (the settings on most of the test games were too high to provide any kind of meaningful review of what kind of playability consumers can expect), not to mention the fact that you didn't even manage to get it's paper specifications right, so you make it seem like it's horribly underperforming, which is not at all the case.

Sigh. I can't do this anymore right now, but if any readers would actually like some info on mobile graphic cards, please head over to NotebookReview.com and either look me or one of the other knowledgeable members up.

-Thomas

[Adrian Wong]

Originally Posted by doppler_pulse

Granted while the mobile 8600 GT will definitely be slower, the test rigs were not comparable in their basic configurations. The desktop used a quad core @ 2.66 ghz and 1066 fsb while the mobile cpu was a dual core limited @ 2.00 ghz and an 800 fsb. Furthermore, the memory clocks, hard drive spindle speeds, and hard drive densities would all take their toll on gaming performance.

It would have been best had the reviewer at least reduced the clock speed of the qx6700 if not swapped in an e4400 with similar specs to the t7300. Although many games are not multithreaded, the reviewer's selection included titles such as supreme commander and bioshock which are known to be able to utilize them.

Thus, while I agree the 8600M GT is slower, I do not believe that it is slower by that great a margin.

Hi doppler_pulse,

I agree with your points. However, it is not easy to review a mobile GPU because it's almost impossible to compare them on the same testbed. This is why this time, I decided to compare the GeForce 8600M GT against desktop graphics cards.

Originally, the plan was to test using a E6600 dual-core processor, but that blew away when the processor died in the line of duty. So, that only left us the quad-core QX6700. Granted, it is NOT a fair comparison. We never said it would be. In fact, it would NEVER be a fair comparison UNLESS we can swap the MXM modules in and out on the same notebook.

With that said, we intentionally ran the games at the highest graphics settings possible to stress the GPUs instead of the CPU. Hence, at the higher resolutions, the CPU is much less of a factor.

In fact, there is little sign of CPU limiting the frame rate of the GeForce 8600M GT. If you check the resolution comparisons of all four games, you do not see a plateau in its frame rates as the resolution changes.

If you check the GeForce 8600M GT vs. GeForce 8600 GT/7600 GT comparisons, you will see that the GeForce 8600M GT's curves are slightly flatter than the desktop cards. That's the extent of the CPU's effect on the frame rates.

[Adrian Wong]

Originally Posted by odin243

Starting at the beginning: The supposed test laptop, the Dell 1501, does not feature an 8600M-GT, and never did, and never will. I can only assume that that is a typo, and that what was actually used was the Dell Inspiron 1520.

You are right. It was a typo. It is an Inspiron 1520.

Originally Posted by odin243

If that's the case, then it should be noted that it uses a budget version of the 8600M-GT with 400MHz DDR2 VRAM, rendering your listed specifications invalid. It is documented that the DDR2 8600M-GT runs on average 30% slower than the 700MHz GDDR3 8600M-GT, whose specifications are listed in the review.

Actually, it is NOT a budget GeForce 8600M GT with 400MHz DDR2 VRAM. It is the same GeForce 8600M GT but with 256MB of 400MHz DDR2 SDRAM. I agree that we listed the wrong clocks and that will be corrected immediately.

Originally Posted by odin243

I won't even talk about the processor difference, that's just disgusting. Any professional reviewer would be ashamed of that kind of unaccounted for differential.

I would disagree with this. If you bother to check the curves, you will see only a slight flattening. If the CPU is such a big effect, we would have seen a much more significant flattening of the frame rate curves.

As I posted in my earlier reply, it's not easy reviewing a mobile GPU. It's not reviewing an ENTIRE notebook where you can get away with a simple PCMark score. The most ideal solution would be to compare all GPUs on the same notebook by swapping the MXM module. But that's just not possible for us, so this is the best solution we can come up with.

This is also why we chose to go with real world game benchmarks, instead of synthetic benchmarks like PCMark. These benchmarks show the actual performance of the GPU, no matter whether the comparison is good or not. When all is said and done, the GeForce 8600M GT will still produce the same frame rates no matter whether we are comparing it against a desktop graphics card... or another mobile GPU.

Originally Posted by odin243

Next off, the 8700M-GT, 8600M-GT, and 8600M-GS are all based on the G84M chip. You can claim that that chip is more similar to the G86 than the G84 if you like, but the fact is it's still called the G84M. The G86M would be the chip which the 8400M-GT, 8400M-GS, and 8400M-G are based off of.

Hmm.. We made a mistake here, I think. Looks like we will have to change both the review and our Mobile GPU Comparison Guide. Thanks for the heads-up.

Originally Posted by odin243

Also, what the hell is the "GeForce 7600M GT" supposed to be? No nVidia card ever existed with that name. You may be referring to the Geforce Go 7600, or possibly the Geforce Go 7600GT, both of which are different cards.

Again, a typo. It's the NVIDIA GeForce Go 7600 GT.

Originally Posted by odin243

Also, about it's DX10 support: It's far from useless, as long as you don't run games on maxed or high settings. The 8600M-GT is a mid-range mobile card, meant to play the newest games at WXGA or WXGA+ with medium settings, which it does admirably.

This is where I would have to disagree with you. I believe games should be played at the best image quality possible. Gaming, for me at least, isn't about getting the highest frame rate possible or the highest PCMark/3DMark score. It's about enjoying the game. This is precisely why we test at maximum image quality. Of course, doing so also stresses the GPU, preventing the CPU from becoming a bottleneck.

Yes, you can play DirectX 10 games at lower resolutions (as I mentioned in the review) or with a lower rendering quality but it is certainly NOT the way the game is meant to be played. With DirectX 10 offering no real advantage in current games, I would say the huge drop in frame rate is an unacceptable trade-off for the "ability" to run in DirectX 10. If you can run the game in DirectX 9 with better image quality and a much higher frame rate, why bother with DirectX 10? This is precisely why I say the DirectX 10 support in the GeForce 8600M GT is useless.

Originally Posted by odin243

Your subjective conclusions are absolutely worthless considering the card wasn't tested in it's intended environment (the settings on most of the test games were too high to provide any kind of meaningful review of what kind of playability consumers can expect), not to mention the fact that you didn't even manage to get it's paper specifications right, so you make it seem like it's horribly underperforming, which is not at all the case.

I would disagree with that. It reflects precisely the frame rates you can expect if you wish to run the games at full image quality, if you wish to play the game the way it was designed to be played. In fact, that's what DirectX 10 is all about. If it's not about delivering better rendering quality and effects, why bother?

I agree though that I got the specs wrong. I did note that the GeForce 8600M GT in this notebook was running at a lower clock speed, but I totally forgot about it because unlike a simple quick run of PCMark or 3DMark, I had to test 10 different graphics cards in addition to this mobile GPU. By the time I got to this review, it slipped my mind.

Thanks for the reminder though. All mistakes will be corrected, as always.

[Dashken]

The NVIDIA GeForce 8600M GT is the mobile version of the NVIDIA GeForce 8600 GT. However, instead of being based on the G84 GPU used by the GeForce 8600 GT, it is based on the smaller G86 GPU used by the GeForce 8500/8400 graphics cards. The G86 GPU only packs 210 million transistors, which 79 million or 27% less than the G84 GPU. The result is a smaller and cooler mobile GPU.

Even though the GeForce 8600M GT is based on a different GPU, it is similar to the GeForce 8600 GT desktop GPU in specifications. In fact, the only difference is the GeForce 8600M GT runs at 475 MHz while the GeForce 8600 GT runs at 540 MHz. So, is the GeForce 8600M GT almost as fast as its desktop counterpart? Read on and find out.

Here's a quote from the review :-

"Like its desktop counterpart, the NVIDIA GeForce 8600M GT looks better on paper than in real life. On paper, it is roughly between the desktop GeForce 8600 GT and the GeForce 7600 GT in performance. That makes it a rather powerful mobile GPU, especially compared to integrated graphics solutions like the Intel Graphics Media Accelerator X3100."

Link : NVIDIA GeForce 8600M GT Mobile GPU Review Rev. 2.0