by Alexey Stepin , Yaroslav Lyssenko
03/18/2008 | 02:53 PM
The modern GPUs with unified architecture are known to have a far higher computing capacity than the previous generations of GPUs, but not all games can put their resources to full use. A good example is our recent test of the Nvidia GeForce 9600 GT: in many games its 64 shader processors proved to be enough to deliver performance similar to the Nvidia GeForce 8800 GT 512MB. It means the Nvidia GeForce 8800 GT 512MB can be as fast as the GeForce 8800 GTS 512MB if it is overclocked to the same GPU and memory frequencies.
We have tested a number of pre-overclocked models of GeForce 8800 GT 512MB and found them to be equal or even superior to the GeForce 8800 GTX. And now we’ve got the opportunity to see how the GeForce 8800 GTS 512MB and the GeForce 8800 GT 512MB differ in performance if their clock rates are identical. For this test we’ve taken a curious sample of 8800 GT 512MB manufactured by Elitegroup Computer Systems, the well-known mainboard maker.
The increased frequencies are not the only difference of the ECS N8800GT-512MX DT from Nvidia’s reference card. A special feature of this product is its ability to work silently thanks to the Arctic Cooling Accelero S1 system that already showed its worth on the ATI Radeon X1950 Pro. If our supposition is true and the retail price of the ECS card is lower than that of the GeForce 8800 GTS 512MB, it can be viewed as an excellent choice for a user who is interested in modern games and cares about the noise produced by his gaming platform.
Following Gigabyte, PowerColor and Albatron, ECS uses an upright box for its product. It is not large:
The box is painted black and silver, its front side bearing a picture of a dragon-rider. This fantasy style is quite a hackneyed theme among hardware makers. You can see all sorts of dragons, monsters, knights and elves so often on the boxes with graphics cards and mainboards that your eyes can’t really be surprised and attracted to them.
You can learn the basic information about the product, except for the GPU and memory frequencies, from the front side of the box. Besides the graphics card proper, it contains the following accessories:
The accessories aren’t rich or numerous, yet the ECS N8800GT-512MX DT is quite a unique thing in its own right for the user to need additional stimulation, like free games for example, to make the purchase. There is everything you need to use the card in your system – that’s enough for a majority of gamers. Some won’t even use the additional fans – we’ll tell you about their purpose below. Note that the fans can be connected to the mainboard’s 3-pin header or to a 4-pin Molex connector thanks to the included adapter.
Overall, there’s nothing extraordinary about the packaging and accessories of the ECS N8800GT-512MX DT. The box looks nice, although doesn’t stand out among other manufacturers’ products, and the accessories are just sufficient. The only thing we don’t like is the lack of a software multimedia player for reproducing Blu-ray, DVD and HD DVD formats. If you are going to use this noiseless card in your HTPC, you’ll have to purchase the software directly from the developer, CyberLink or InterVideo.
The ECS N8800GT-512MX DT uses a unique PCB. It differs from Nvidia’s reference design as well as from the PCB developed by Gainward:
The differences aren’t significant, however. Except for the PCB color, they concern the layout and components of the power circuit. This circuit is supposed to have three phases, but only two phases are installed, just like on the reference GeForce 8800 GT, although the graphics core of the ECS card is overclocked by the manufacturer. Like in Gainward’s design, the power circuit is governed by a Richtek RT8802A controller. An RT9259 chip is responsible for the memory chips. The card has a standard 6-pin PCI Express 1.0 connector for additional power supply.
The GPU is revision A2. Its configuration is standard for Nvidia GeForce 8800 GT: 112 shader processors, 28 (56) TMUs, and 16 raster operators. The frequencies are those of the Nvidia GeForce 8800 GTS 512MB: 650MHz for the main domain and 1620MHz for the shader domain. So, we can expect the N8800GT-512MX DT to deliver similar performance in games that don’t use up all the computing resources of modern GPUs. The GPU has a metallic frame to protect the core. It is a called-for safety measure considering the large cooling system the card is equipped with.
The memory chips carry heatsinks. When we removed them, we saw standard Qimonda HYB18H512321BF-10 chips of GDDR3 memory with a rated frequency of 1000 (2000) MHz. The card clocks the memory at a frequency of 950 (1900) MHz, which is 50 (100) MHz above the memory frequency of the reference GeForce 8800 GT 512MB. It is lower than the frequency of the GeForce 8800 GTS 512MB, yet provides a nice 3.2GB/s addition to the memory bandwidth.
The card is equipped with two dual-link DVI-I ports, a 7-pin mini-DIN port, and a standard MIO connector. Besides it, there is a 2-pin connector near the MIO port – it is connected to the sound card to enable the audio-over-HDMI feature.
The cooling system installed on the N8800GT-512MX DT should be familiar to our readers as it is nothing else by the Accelero S2 cooler we saw installed on the PowerColor X1950 Pro SCS3.
The second revision of the cooler is only different from the first one with the improved fastening mechanism that enhances the cooler’s compatibility with modern graphics cards. Otherwise, it is the same simple design consisting of a copper base with four heat pipes carrying 32 aluminum plates that make up a heatsink with a dissipation area of over 2500 sq. centimeters. The heatsink measures 140x215mm and protrudes far beyond the PCB’s dimensions. As a result, the N8800GT-512MX DT may not be installed in many compact system cases.
The cooler’s base is fastened to the PCB with four screws. Although it weighs a mere 360 grams with the Turbo Module, you still should be careful while handling the card. High-quality thermal grease (Arctic Cooling MX-2) is used as the thermal interface between the GPU die and the cooler’s sole. The memory chips and the load-bearing elements of the power circuit are cooled with individual aluminum heatsinks.
The Accelero S1 heatsink easily copes with a thermal load of 60-65W, but the GeForce 8800 GT 512MB generates more heat even when working at the reference GPU and memory frequencies. The N8800GT-512MX DT produces even more heat due to the increased clock rates, so the developer included the so-called Turbo Module into the box. It consists of two caseless fans with 80mm impellers.
The fans are installed above the cooler’s heatsink and hitched to the ends of the ribs with plastic locks. They have a common power cable that ends in a standard 3-pin connector for plugging into the mainboard but you may find the cable too short to reach to it. In this case you can use the included adapter to connect the Turbo Module to an ordinary Molex. The fans run on slide bearings at a speed of 1500rpm and should be quite noiseless. Their performance is as high as 45CFM, however. The Turbo Module should indeed improve the performance of the Accelero S1/S2 greatly.
The Turbo Module has one drawback. Although the height of the fans is only 15 millimeters, it is enough to block one more slot. The fully assembled cooler thus occupies three slots, which may be a problem for some users especially those who have microATX system cases.
The ECS N8800GT-512MX DT is expected to be used together the Turbo Module by default, so we decided to check out Arctic Cooling’s claim about the complete noiselessness of these fans. We measured the level of noise produced by the card with a digital sound-level meter Velleman DVM1326 using A-curve weighing. The level of ambient noise in our lab was 36dBA and the level of noise at a distance of 1 meter from the working testbed with a passively cooled graphics card inside was 43dBA. Here are the results:
The Turbo Module proved to be very quiet indeed. Even with the fans installed the N8800GT-512MX DT is nearly silent and surpasses such cards as GeForce 8800 GTX and GeForce 7900 GTX whose coolers are considered among the quietest. Alas, the tradeoff is compactness. With the Turbo Module attached, the revision 2 Accelero S1 blocks two neighboring expansion slots instead of one as most dual-slot coolers do. As we noted already, the N8800GT-512MX DT won’t be a good choice for compact system cases and microATX mainboards.
The Accelero S2 with the Turbo Module is very effective, though. According to the latest version of RivaTuner, the CPU temperature was 44.5°C in Idle mode and 52-69°C under load. This is much better than the performance of the reference coolers of GeForce 8800 GT 512MB: over 90°C of the first version and 85-88°C of the second, improved, version. And the reference card had lower frequencies than the N8800GT-512MX DT! So, the ECS card won’t overheat if you install the Turbo Module on it. Moreover, you can do without that module and save one expansion slot if your system case is ventilated well: the large dissipation area of the Accelero S1 provides some ground for such experiments, especially if you’ve got a system fan on the side panel. Keep an eye on the GPU temperature, though, if you dare to experiment this way.
Our attempt to overclock the card was not much of a success. We increased the core frequency from the default 650/1620MHz to 700/1744MHz while the memory refused to be overclocked at all, the highest stable memory frequency being 970 (1940) MHz. We won’t benchmark the card in the overclocked mode due to the small frequency growth – its performance is going to be limited by the memory frequency anyway. Of course, you can be more successful, especially if you use extreme overclocking methods.
As opposed to early samples of GeForce 8800 GT 512MB, the N8800GT-512MX DT is free from compatibility problems. It started up on every PCI Express 1.0a mainboard we tried it with.
To test the performance of ECS N8800GT-512MX DT in games we assembled the following standard test platform:
According to our testing methodology, the drivers were set up to provide the highest possible quality of texture filtering and to minimize the effect of software optimizations used by default by both: AMD/ATI and Nvidia. Also, to ensure maximum image quality, we enabled transparent texture filtering - Adaptive Anti-Aliasing/Multi-sampling for ATI Catalyst and Antialiasing – Transparency: Multisampling for Nvidia ForceWare. As a result, our ATI and Nvidia driver settings looked as follows:
For our tests we used the following games and benchmarks:
First-Person 3D Shooters
Third-Person 3D Shooters
We selected the highest possible level of detail in each game using standard tools provided by the game itself from the gaming menu. The games configuration files weren’t modified in any way. The only exception was Enemy Territory: Quake Wars game where we disabled the built-in fps rate limitation locked at 30fps.
Games supporting DirectX 10 were tested in this particular mode. With a few exceptions, the tests were performed in the following most widely spread resolutions: 1280x1024/960, 1600x1200 and 1920x1200. If the game didn’t support 16:10 display format, we set the last resolution to 1920x1440. We used “eye candy” mode everywhere, where it was possible without disabling the HDR/Shader Model 3.0/Shader Model 4.0. Namely, we ran the tests with enabled anisotropic filtering 16x as well as MSAA 4x antialiasing. We enabled them from the game’s menu. If this was not possible, we forced them using the appropriate driver settings of ATI Catalyst and Nvidia ForceWare drivers
Performance was measured with the games’ own tools and the original demos were recorded if possible. Otherwise, the performance was measured manually with Fraps utility version 2.9.1. We measured not only the average speed, but also the minimum speed of the cards where possible.
We have also included the results for the following graphics accelerators:
This game doesn’t support display resolutions of 16:10 format, so we use a resolution of 1920x1440 pixels (4:3 format) instead of 1920x1200 for it.
Working almost at the same frequencies as the reference GeForce 8800 GTS, the ECS card delivers the same performance in Battlefield 2142, which is not a heavy application. Both cards can be used to play at resolutions up to 1920x1200 and are likely to provide a high speed even at 2560x1600 notwithstanding the enabled FSAA.
BioShock doesn’t support FSAA when running in Windows Vista’s DirectX 10 environment. That’s why we benchmarked the cards without FSAA.
The game engine creates advanced special effects, so there is a gap between the GeForce 8800 GTS 512MB and the ECS N8800GT-512MX DT. It is no bigger than 6% at high resolutions, though. At 1280x1024 there is no difference between the two cards at all. The cooler of the ECS card can work in passive mode, but the reference cooler of the GeForce 8800 GTS 512MB is almost silent, too. So, the retail price is the decisive factor when it comes to choosing between these two graphics cards.
The Nvidia GeForce 8800 GTS 512MB is somewhat faster than the ECS N8800GT-512MX DT, yet this has no practical value since the average frame rate is about 20-22fps. Neither card can provide a comfortable speed if you select the highest graphics quality settings and enable 4x FSAA.
The Nvidia GeForce 8800 GTS 512MB ensures a bigger reserve of speed at resolutions above 1280x1024 than the ECS card, but the difference is rather small. Theoretically, both cards can be used to play at 1920x1200, yet the most comfortable mode is 1280x1024.
The game being too hard at its Very High level of detail, we benchmarked the cards without FSAA to get a more playable speed.
With the release of Crysis Crytek supported its reputation of a developer of tomorrow-oriented games it had earned with Far Cry. Crysis cannot be played at the highest level of detail on today’s single graphics cards even if you give up FSAA.
The frame rate is fixed at 30fps in this game as this is the rate at which the physical model is being updated at the server. Thus, this 30fps speed is the required minimum for playing the game.
The ECS card is a mere 5% slower than the GeForce 8800 GTS 512MB. Both cards deliver high performance in this game, allowing you to play it at 1920x1200 with 4x FSAA.
It’s the same in Half-Life 2: Episode Two. The ECS N8800GT-512MX DT and the GeForce 8800 GTS 512MB are identical in terms of performance, and your shopping choice must be determined by their pricing. There is no sense in paying extra for the additional 16 shader processors and 4 TMUs of the latter card if ECS’ product is cheaper.
The game doesn’t support FSAA when you enable the dynamic lighting model, but loses much of its visual appeal with the static model. This is the reason why we benchmarked the cards in S.T.A.L.K.E.R. using anisotropic filtering only.
The ECS N8800GT-512MX DT is again almost as fast as the Nvidia GeForce 8800 GTS 512MB, but its minimum speed is below comfortable level at 1920x1200. So, you should prefer the GeForce 8800 GTS 512MB if you’ve got a 23” monitor.
Forcing FSAA from the graphics card’s driver doesn’t produce any effect as yet. That’s why the game is tested with anisotropic filtering only.
The difference is negligible again, 4-5% in favor of the GeForce 8800 GTS 512MB. Well, every graphics card in this test session has a superb speed at any resolution in this test.
The ECS card is only 7% behind the GeForce 8800 GTS 512MB at 1280x1024 and less than 1% behind it at the other resolutions. But like in Crysis and Call of Juarez DX10 Enhancement Pack, this has no practical value since the speed is too low.
Like in many other games, the ECS N8800GT-512MX DT is competing successfully with the Nvidia GeForce 8800 GTS 512MB. The GPU and memory frequencies being the same, the difference in the amount of functional subunits, shader and texture processors, matters but seldom to the gamer. In this case, both cards can be used to play at 1920x1200. Well, the same is true for the reference GeForce 8800 GT 512MB as well.
We’ve got a different picture here. The increased frequencies don’t help the ECS card at 1600x1200 and its minimum speed bottom out below comfortable level. The GeForce 8800 GTS 512MB is also close to the dangerous level, though. On the other hand, it is 17% ahead of the ECS N8800GT-512MX DT at 1280x1024, giving the gamer more confidence for the most complex scenes.
The current version of the game doesn’t support FSAA, so we performed the test with anisotropic filtering only.
There is less than 10% of difference between the Nvidia GeForce 8800 GTS 512MB and the ECS N8800GT-512MX DT in any mode. Both cards deliver comfortable performance at resolutions up to 1920x1200 inclusive.
The game loses much of its visual appeal without HDR. Although some gamers argue that point, we think TES IV looks best with enabled FP HDR and test it in this mode.
The difference between the cards is smaller in TES IV and the ECS N8800GT-512MX DT can be preferable if it costs lower than the GeForce 8800 GTS 512MB.
The new add-on to Company of Heroes is tested in DirectX 10 mode only since it provides the highest quality of the visuals.
Although this test is run in DirectX 10 mode, the additional shader processors of the GeForce 8800 GTS 512MB do not make it any better even than the reference GeForce 8800 GT 512MB, let alone the pre-overclocked card from ECS. The low minimum of speed still means the lack of comfort for the gamer.
The game having a frame rate limiter, you should consider the minimum speed of the cards in the first place.
Every graphics card in this review belongs to a high enough class and easily hits the frame rate limit. They all have almost the same minimum of speed – the difference of 1fps cannot be spotted with a naked eye.
The ECS card is exactly as fast as the Nvidia GeForce 8800 GTS 512MB excepting the slightly lower minimum of speed at 1280x1024. The performance is low, however, even at that resolution for comfortable play.
3DMark05 obviously cannot be a truthful measure of performance of modern graphics cards. This benchmark runs at 1024x768 without FSAA and anisotropic filtering, so the result can be limited by the CPU performance. Moreover, it doesn’t use SM3.0, let alone SM4.0.
The individual tests are closer to real-life applications. We can’t see any difference between the Nvidia GeForce 8800 GTS 512MB and the ECS N8800GT-512MX DT: these tests cannot use up all of a modern GPU’s computing resources while the memory bandwidth is nearly identical for both cards.
3DMark06 produces a surprising result: the GeForce 8800 GTS 512MB scores only 39 points more than the ECS card, which is within the measurement error range considering that the scores are over 11,000 points.
There is a certain difference in the groups of tests, though. It is somewhat odd to see the ECS N8800GT-512MX DT win the SM2.0 tests because it has fewer TMUs than the GeForce 8800 GTS 512MB. The latter’s advantage in the SM3.0/HDR tests provokes no questions since it has 16 shader processors more than the ECS card.
In the individual tests the ECS is always a little behind the Nvidia GeForce 8800 GTS 512MB. This may be due to the use of FSAA and the slightly lower memory frequency of the ECS card.
We are really pleased with the version of GeForce 8800 GT 512MB from Elitegroup Computer Systems we have tested today, largely due to its superb cooler Arctic Cooling Accelero S1 (also see our article called Roundup: AeroCool Double Power, Thermaltake DuOrb, Zalman VF1000 LED and a Few Other Good VGA Coolers). Thanks to the Turbo Module, which consists of two additional fans, the cooler coped even with a seriously overclocked GeForce 8800 GT 512MB.
We guess the ECS N8800GT-512MX DT can also be used in passive-cooling mode in a well-ventilated system case like the Chieftec LCX with system fans on the side panel. You can remove the Turbo Module then – the single drawback of the card is that it occupies as many as three slots with its cooling system fully assembled.
As for the performance, the test results suggest that there is no significant difference between the GeForce 8800 GT 512MB and GeForce 8800 GTS 512MB in most games if their GPU and memory frequencies are identical. The latter card provides a higher minimum speed at high resolutions in such games as Call of Duty 4, S.T.A.L.K.E.R. and Hellgate: London. Otherwise, the cards are identical for the gaming fans. Of course, you should prefer the GeForce 8800 GTS 512 if they cost the same money. But if the ECS card is cheaper, it may be a better buy, especially as it comes with a very quiet and highly efficient cooler.