If I didn't know better, it might appear that we haven't had anything better to do than test NVIDIA's new G84/G86 graphics boards in the past week. In reality we are still as swamped as ever but I felt it was important to get as much information out about this product launch quickly before the relevancy of it fades. So, we are finishing up our 8600 GTS and 8600 GT product reviews with a set of cards from XFX.
I covered the technology and detailed specifications of the GeForce 8600 GTS and 8600 GT graphics cards in our previous reviews, so I won't bother repeating it here. But I would highly recommend you read them over first (BFG 8600 GTS OC, EVGA 8600 GT) so you can get a good idea of the GPUs we are testing today, their market segments and competition from ATI and previous NVIDIA generations.
The XFX 8600 GTS XXX
Despite what you might be hoping, the "XXX" model of the XFX card is nothing more than their highest overclocked model; sorry to disappoint!
XFX GeForce 8600 GTS XXX
The XFX 8600 GTS XXX uses the same cooler as the reference 8600 GTS cards but as the name indicates, overclocks the frequencies well over the default speeds setup by NVIDIA:
730 MHz core clock versus 675 MHz reference speed
2.26 GHz memory clock versus 2.0 GHz reference speed
These are healthy overclocks and will result in performance past the stock cards on the market in nearly all cases.
XFX GeForce 8600 GTS XXX - PCIe power connector
Just as with the other 8600 GTS cards, the XFX XXX model requires a single 6-pin PCIe power connector
XFX GeForce 8600 GTS XXX - SLI connection
The SLI connection lives on, but only a single one, for connecting a second card for improved gaming performance. We'll cover SLI performance with the 8600 GTS cards in our benchmarks so be sure to catch that!
XFX GeForce 8600 GTS XXX - dual dual-link DVI and HDTV output
The GeForce 8600 GTS cards all feature two dual-link DVI connections for support for the ultra-high-res monitors like the new 30" ones that run at 2560x1600. The TV output supports HDTV via an included dongle.
The extras included with the XFX cards (both of them in this case) include a single VGA-to-DVI adaptor, a power adaptor, S-video cable and instructions. The XFX 8600 cards will also include a free copy of the orignal Ghost Recon: Advanced Warfighter -- not the newest title out there but free is always good. Notice though that XFX did NOT include an HDTV dongle with these cards; quite confusing indeed.
Interestingly, the XFX 8600 GT XXX card uses a different heatsink cooler than the other 8600 GT cards we have seen -- it's a bit smaller but didn't seem to have any problems cooling the GPU efficiently enough to run at its overclocked speed.
XFX GeForce 8600 GT XXX
And again, the clock speeds on XFX's GeForce 8600 GT XXX card are pushed beyond the NVIDIA settings:
620 MHz core clock versus 540 MHz reference clock
1.60 MHz memory clock versus 1.18 GHz reference clock
XFX GeForce 8600 GT XXX - no PCIe power connector
The GeForce 8600 GT cards do NOT require an external power connection and as such are friendlier for HTPCs and SFFs and users with small power supplies.
XFX GeForce 8600 GT XXX - SLI connection
SLI technology support is of course still here -- and we test it out in the coming benchmark pages.
XFX GeForce 8600 GT XXX - dual dual-link DVI and HDTV output
Even the 8600 GT cards have two dual-link DVI connections and an HDTV output.
8600 GTS GPU
8600 GT GPU - same markings
On these two cards from XFX, both of the GPUs are actually labeled "G84" indicating that the G86 8600 GT is actually the same chip with the clocks turned down, not a separate spin at all.
Testing Methodology
Graphics card testing has become the most hotly debated issue in the hardware enthusiast community recently. Because of that, testing graphics cards has become a much more complicated process than it once was. Before you might have been able to rely on the output of a few synthetic, automatic benchmarks to make your video card purchase, that is just no longer the case. Video cards now cost up to $500 and we want to make sure that we are giving the reader as much information as we can to aid you in your purchasing decision. We know we can't run every game or find every bug and error, but we try to do what we can to aid you, our reader, and the community as a whole.
With that in mind, all the benchmarks that you will see in this review are from games that we bought off the shelves just like you. Of these games, there are two different styles of benchmarks that need to be described.
The first is the "timedemo-style" of benchmark. Many of you may be familiar with this style from games like Quake III; a "demo" is recorded in the game and a set number of frames are saved in a file for playback. When playing back the demo, the game engine then renders the frames as quickly as possible, which is why you will often see the "timedemo-style" of benchmarks playing back the game much more quickly than you would ever play the game. In our benchmarks, the FarCry tests were done in this matter: we recorded four custom demos and then played them back on each card at each different resolution and quality setting. Why does this matter? Because in these tests where timedemos are used, the line graphs that show the frame rate at each second, each card may not end at the same time precisely because one card is able to play it back faster than the other -- less time passes and thus the FRAPs application gets slightly fewer frame rates to plot. However, the peaks and valleys and overall performance of each card is still maintained and we can make a judged comparison of the frame rates and performance.
The second type of benchmark you'll see in this article are manual run throughs of a portion of a game. This is where we sit at the game with a mouse in one hand, a keyboard under the other, and play the game to get a benchmark score. This benchmark method makes the graphs and data easy to read, but adds another level of difficulty to the reviewer -- making the manual run throughs repeatable and accurate. I think we've accomplished this by choosing a section of each game that provides us with a clear cut path. We take three readings of each card and setting, average the scores, and present those to you. While this means the benchmarks are not exact to the most minute detail, they are damn close and practicing with this method for many days has made it clear to me that while this method is time consuming, it is definitely a viable option for games without timedemo support.
The second graph is a bar graph that tells you the average framerate, the maximum framerate, and the minimum framerate. The minimum and average are important numbers here as we want the minimum to be high enough to not affect our gaming experience. While it will be the decision of each individual gamer what is the lowest they will allow, comparing the Min FPS to the line graph and seeing how often this minimum occurs, should give you a good idea of what your gaming experience will be like with this game, and that video card on that resolution.
Our tests are completely based around the second type of benchmark method mentioned above -- the manual run through.
System Setup and Benchmarks
A big area of contention for me with this review was my choice to use Vista 64-bit instead of Windows XP. While indeed most gamers are probably still on Windows XP, with the DX10 games right around the corner, the move to Vista is going to be picking up for the enthusiast crowd. As such, I thought it was only responsible for me to start testing on Vista full time with gaming so that my results with it are more reliable and stable; in addition, since we have been spending so much time listening to both graphics companies tout their dominance in Vista, we really should be putting it to the test on a regular basis.
In addition to my move to Vista 64-bit for our testing, you'll also notice a DRASTIC change in the game titles that are used in benchmarking. Only a couple of titles (Prey and FEAR) remain from the previous set and we have thrown SIX new games to reflect the gaming of our users. I have included FEAR, Prey, HL2 Episode 1, Supreme Commander, Company of Heroes, Oblivion, Battlefield 2142 and Rainbow Six Vegas. This of course added a TON of time to our testing cycle so a lot more work had to go into getting the test setups ready for the new onslaught of performance metrics. I think it turned out very well though and pushes the hardware and drivers to areas that other reviews might not investigate.
The XFX cards are here on their own today; if you want to see how the 8600 GTS and 8600 GT compare to cards from ATI's lineup check out our initial reviews of them. For this review I decided to include some SLI testing using the XFX cards and a card from a different vendor: BFG for the GTS and EVGA for the GT. I didn't run into any compatibility problems using different vendor boards -- so all went well!
You'll notice that I took both FEAR and HL2: Episode 1 out of the testing suite for this article -- with the big hitching problems we are seeing in the current driver release, benchmarking them is basically pointless. Once the new driver that NVIDIA says is only days away shows up, we'll put them back in the set of games tested. Sorry!
Battlefield 2142 (DirectX 9)
Battlefield 2142 is the sequel to the extremely popular Battlefield 2. It uses some much more detailed shaders yet still provides the same wide-open areas for fighting that the BF series is known for.
SLI performance on Battlefield 2142 is somewhat limited it would seem: at 1280x960 we see 28% and 15% scaling on the 8600 GTS and 8600 GT cards and at 1600x1200 we saw 33% and 25% scaling respectively. That means that by adding the second card to the system, we are only seeing 15-33% better performance over a single GPU working. Obviously, that performance gain doesn't really justify the additional cost of a second card. Maybe some other titles will improve the outlook.
Supreme Commander (DirectX 9)
This real-time strategy game boasts an incredibly beautiful graphics engine that really stress our GPUs here as well as being one of the first games to be really multi-threaded as well.
Supreme Commander is a very CPU-heavy game and as such performance between the GTS and GT cards from XFX are very close. The SLI performance numbers are even more disappointing than the last set. There was no gain at 1280x960 and a NEGATIVE gain (losing performance) at 1600x1200. My guess is that the SLI processing overhead is diving into the CPU cycles the game needs and thus performance drops slightly.
Company of Heroes (DirectX 9)
Company of Heroes is another very compute-intense RTS that also has some unique features in it for fans of the genre including a fantastic graphics engine and a real-time physics engine built by Havok.
Company of Heroes performance is very refreshing with the XFX GeForce 8600 GTS and GT cards and the SLI performance is also improved. At the lower 1280x960 resolution, we see a 40% and 39% gain on the GTS and GT respectively and at 1600x1200 we see 27% and 28% gains. Those numbers aren't exceptionally high, and we are used to seeing numbers over 50% in our Windows XP testing of SLI, but these gains are at least something.
Rainbow Six Vegas (DirectX 9)
The Rainbow Six series has always been a high quality set of games that center on action-based tactical assualt. The new Vegas iteration takes a bit more of the action approach similar to what we have seen in Ghost Recon: Advanced Warfighter. This game is the first to use the new Unreal 3 engine and you can tell! The graphics are amazing and really stress our systems.
Rainbow Six is definitely a GPU-intensive game and the new 8600 series of cards does very well compared to the competition in this price range. SLI performance is somewhat limited now though, at 18% and 15% at 1280x960 on the 8600 GTS and 8600 GT respectively and at a slightly more reasonable 27% and 21% across the two test platforms.
Prey (OpenGL)
Prey is the latest title to be released using the Doom 3 engine, and since we have learned how to tweak the engine a bit more since then, it makes sense to include this major title in our testing suite.
Moving onto Prey we see some impressive numbers from these mainstream cards being able to run at 1600x1200 resolution with 4xAA and 8xAF enabled without too much of a problem. Adding in a second card improves there performance significantly here as well. At the 1280x1024 resolution test we saw a 32% and 45% increase in performance and was probably limited by the 60 FPS limit the Doom 3 engine has during real gameplay. At 1600x1200 we see 55% gains from both the 8600 GTS and 8600 GT cards with NVIDIA's SLI technology; not bad!
Oblivion (DirectX 9)
Probably one of the most demanding and graphically customizable games you will ever see, Oblivion can stress any system that exists for gaming today. With its wide open areas and high quality textures the graphics system for Oblivion produces spectacular results.
Now we're talking! These SLI performance numbers are more what we were looking for from the 8600 GTS and GT cards and reaches up to the levels we have seen in the top Windows XP numbers previously. At 1280x960 the 8600 GTS is able to get 84% scaling in performance while the 8600 GT sees a 75% gain in average frame rate. Not only that, but the minimum frame rates improved just as dramatically. At 1600x1200 those numbers go down slightly to 57% and 43%, but are still very good.
The new Futuremark 3DMark06 benchmark was released just a few months ago and has already become a popular test for GPU benchmarks and CPUs as well. You can see my full overview of the new benchmark, what it has to offer over the 2005 version and some other initial scores from my previous article.
3DMark06
The performance numbers from XFX's 8600 cards are very good here and the SLI numbers improve in the areas we would expect and since these are synthetic numbers, the performance scaling is pretty high. While the 3DMark06 score only goes about 30% on each SLI configuration the fill rates and pixel shader tests show 80-95% increases in theoretical GPU performance. If only we could have seen those performance values translated into real world gaming results.
Graphics are quickly becoming the most power hungry component in your system today and that can be somewhat of a cause for alarm as we begin to receive 1200+ watt power supplies in our offices! How do the new 8600 GTS and GT stand up against each other?
None of these configurations really uses a LOT of power compared to some other test beds we have seen and the 8600 GTS in SLI mode still only uses 288 watts of power at full tilt.
Conclusions
The XFX GeForce 8600 GTS XXX and 8600 GT XXX cards are fantastic additions to NVIDIA's mainstream graphics card line up.
Performance
The performance on the 8600 GTS XXX and 8600 GT XXX cards was pretty good in comparison to the other 8600 cards we have seen and tested thus far. The clock speeds on the XFX models were actually the highest we have seen to date and as such, it makes sense that these XXX models (hehe) bring in the fastest performance in their respective families.
Separating the two though, the 8600 GT model is still my favorite pick of the two as it has a better price and is able to outperform the similarly priced competition from ATI by a larger margin. The 8600 GTS is still a great card, but as we saw in our initial review of it, the ATI X1950 Pro card could consistently run games faster at the same or lower price point.
SLI Performance
Testing dual graphics cards in SLI configurations with the 8600-series of cards on Windows Vista 64-bit turned out to be a bit of a letdown. There were really only a couple of games that showed dramatic improvement: Prey and Oblivion. The other four titles didn't show enough of a performance gain to warrant the purchase of a second 8600 GTS or GT card; in fact even on games like Prey and Oblivion you'd be hard pressed to convince me that $450 for a pair of 8600 GTS cards is a better option than a single 8800 GTS 640MB card that you might be able to find for about $400.
Vista is still young, and 64-bit is still maturing as well, so we'll give NVIDIA a push for now, and say we'll check back in another driver release or so to see if the performance gains that we once expected during our Windows XP gaming days can find their way into our Vista-based machines.
Features and XFX Warranty
XFX has been around for some time now as an NVIDIA board partner and has moved around a bit on its warranty policies. Most recently, XFX has updated its lifetime warranty to "alleviate the concerns that enthusiasts may have" over overclocking their graphics cards. Basically, XFX is stating that the warranty protection on the graphics cards will still be honored as long as there is no physical damage to the card and no original components are missing. This is important for their double lifetime warranty that lets users sell their older cards to other enthusiasts and still maintain the full lifetime warranty.
Here's how the warranty is stated on their website:
Heres the deal: We will repair and service your 6, 7, and 8 Series Graphics Card for as long as you live. Even for those of you who know how to push our cards to the limits, if anything goes wrong, well service it free of charge. All you have to do is register the card with us online. If you ever decide to sell or give the card away to someone, well still honor the protection plan for the second owner as well. They just need to register the card with us. You get the protection AND the added value of being able to pass along a full warranty.
That's a pretty good deal. XFX does not have a 24/7 support line but does offer 24/7 email support and some forums as well. Warranties and support are very important for some people and if you are one of them, be sure to do you research on the various companies before spending your hard earned cash.
Final Thoughts
XFX has come up with a pair of excellent graphics cards for the 8600-series and their "XXX" series. The XFX GeForce 8600 GTS XXX offers the highest clock and memory frequencies we have seen here at PC Perspective but the XFX GeForce 8600 GT XXX stole the show with high clock speeds, lower prices and superior performance compared to other 8600 cards.
XFX GeForce 8600 GTS and GT Review - SLI Tested
![](http://ak.buy.com/buy_assets/v5/img/clear.gif"){kind=spacer}
