by Alexey Stepin
05/13/2005 | 09:23 AM
Graphics cards from ASUSTeK Computer stand out on the market due to their original exterior, highest quality of manufacture and support of numerous exclusive technologies. The company is trying to bring something innovative into each new product, and regularly succeeds.
Just recall the ASUS 9800 XT that was more like a piece of pure designing art rather than a utilitarian computer device (for details see our article called ASUS RADEON 9800 XT and Leadtek WinFast A380 TDH Ultra: The Battle for AquaMark3), or the ASUS V9999 Gamer Edition with the controllable cooling system and the performance head above its GeForce 6800 prototype (for details see our article called ASUS V9999 Gamer Edition Graphics Card Review).
Even low-end solutions acquire more appeal if implemented by ASUS (for details see our article called ASUS Extreme AX600 XT/HTVD Graphics Card Review). Of course, the company is also offering mainstream products: the Extreme N6600 GT graphics cards we used to check NVIDIA’s SLI technology with made a very good showing (for details read our article called NVIDIA Multi-GPU SLI Technology: New Approach to Old Ideas). Although they were based on the reference design, ASUS equipped them with copper coolers and also added memory heatsinks that many other GeForce 6600 GT do not have. Coupled with the deep blue color of the PCB, this resulted in a truly beautiful device.
Our today’s review is about the ASUS N6600 GT graphics card, too. But this time we’re going to talk about its AGP version – so welcome the ASUS N6600 GT/TD!
Like nearly all products from ASUS to have ever entered our test labs, the N6600 GT/TD came to us in its full retail rig. This graphics card is manufactured in two versions: ordinary and Top Limited Edition, the latter having higher clock rates. The design of the box copies what we saw with the ASUS V9999 Gamer Edition, but the box itself is smaller.
We opened it up to find the following:
We had got the ordinary version of the card, while the Top Limited version is also accompanied with a copy of the game XPand Rally. The accessories are not actually lavish. For example, if you happen to own a TV-set with an S-Video input, you’ll have to buy an appropriate adapter separately. There is also no Molex power splitter among the accessories which is usually enclosed with graphics cards that demand additional power. Well, it is possible that we dealt with a presale version of the product and the off-the-shelf version will be quipped with both an S-Video-to-RCA adapter and a power splitter. This presale supposition is also indirectly confirmed by the lack of cardboard compartments and of the traditional orange box with CDs that you usually see in the packages of ASUS products.
The software CD contains a full version of the user manual as well as ASUS’ drivers for GameFace Live technology. The point of this technology is in using web-cams and microphones when you are playing a game, so you can communicate with other players easily. More expensive graphics cards from ASUS come with a special web-camera with an inbuilt microphone, but the owners of mainstream solutions have to buy these devices separately to enjoy GameFace Live. This is hardly a drawback, though: the camera from ASUS can’t boast a high image quality and the inbuilt microphone isn’t sensitive enough, in our opinion.
The ASUS N6600 GT/TD for the AGP platform looks as cute as the PCI Express version: the deep blue of the PCB goes well with the golden color of the cooler.
The cooling system is an exact copy of the one we described in our NVIDIA SLI review (for details see our article called NVIDIA Multi-GPU SLI Technology: New Approach to Old Ideas). Like on the Extreme N6600 GT, the fan is not highlighted, although it has transparent blades and looks as if having some highlighting. The cooler is mounted at an angle of 45 degrees, like on many other GeForce 6600 GT AGP cards. It is fastened with two spring-loaded clips which are not as tight as those on the Leadtek card, and the fastening is rather slack as a result. To avoid damaging the open die of the GPU we recommend you to be very careful during the installation of this graphics card into the system case. There is no protective frame on the heatsink or on the GPU case, so you must not take the card by the cooling system’s heatsink lest you chip the GPU die, killing the device altogether.
The design of the N6600 GT/TD makes the cooler partially cover the memory chips, so the latter have no real cooling at all. The HSI bridge is, on the contrary, cooled with a separate copper heatsink. Its efficiency is questionable, though. Unlike the one on the Leadtek WinFast A6600 GT TDH, this heatsink is not blown at by the main cooler and becomes very hot at work. This defect is common among many GeForce 6600 GT cards equipped with coolers of that type.
This being not the Top Limited version of the product, the GPU is clocked at 500MHz and the memory at 450 (900) MHz. The accelerated version of the N6600 GT/TD works at 520/550 (1100) MHz and is equipped with 1.6ns memory. The memory chips on our sample were slower, having 2.0 nanoseconds access time. The manufacturer of the chips is Samsung, the company that makes memory for almost all present-day graphics cards.
On the whole, the design of the ASUS N6600 GT/TD fully repeats the reference design with one DVI-I and one D-Sub connector, with such minor deviations as the copper cooler and a different color of the PCB.
The acoustic characteristics of the ASUS N6600 GT/TD are acceptable, yet its cooler is louder than the one installed on the Leadtek WinFast A6600 GT/TDH. Anyway, the noise remained always in a comfortable range and didn’t practically stand out against the noise from the power supply and the Thermaltake Venus 12 cooler.
We got as high as 580/530 (1060) MHz during our overclocking experiments. The GPU frequency gain is fairly big, while the memory frequency gain is expectably small due to the use of 2.0ns chips on an intricately wired PCB.
The quality of the 2D image outputted by the card was crystal-sharp in all resolutions up to our monitor’s maximum of 1800x1400@75Hz. Since the overwhelming majority of graphics cards today deliver an impeccable 2D quality we will touch upon this subject in our upcoming reviews only if there are really some problems with a graphics card’s 2D output.
We performed our tests on a testbed configured like follows:
Software:
We compared the ASUS N6600 GT/TD with the following graphics cards:
Following our standard procedure, we enabled the ForceWare optimizations, except the Anisotropic mip filter optimization. The Image Settings slider was set to “Quality”. In ATI’s Catalyst driver we set the Catalyst A.I. option to “Standard” and Mipmap Detail Level to “Quality”. The VSync option was disabled for both ATI Catalyst and NVIDIA ForceWare.
If the game allowed turning full-screen anti-aliasing and anisotropic filtering on, we used this option. Otherwise we forced the necessary mode from the driver. We didn’t edit any configuration files. We selected the highest graphics quality settings in each game, the same settings for cards from ATI and NVIDIA. The following games and applications were used:
First Person 3D Shooters:
Third Person 3D Shooters:
Simulators:
Strategies:
Semi-Synthetic Benchmarks:
Synthetic Benchmarks:
The ASUS N6600 GT/TD is on the same level with the RADEON X800 PRO because Doom III prefers NVIDIA’s GeForce6 GPUs due to some of their specific functions. Besides, this is an OpenGL game and NVIDIA’s driver for this API has always been the best.
In the “eye candy” mode, i.e. with enabled full-screen anti-aliasing and anisotropic filtering, the RADEON X800 PRO outperforms the ASUS card in 1024x768. Then the gap gets smaller in higher resolutions, despite the 128-bit memory bus of the N6600 GT/TD.
It’s generally all the same on the d3dm4 map as on Hellhole. The N6600 GT/TD rivals the RADEON X800 PRO and leaves the RADEON 9800 XT far behind.
Here, the ASUS card cannot keep up the pace of the RADEON X800 PRO, but it anyway outperforms the RADEON 9800 XT easily, even though providing a comfortable frame rate in 1024x768 resolution only.
It’s difficult to talk about one graphics card being faster than another in the “pure speed” mode since all the participating devices are almost equally fast. Anyway, it’s clear the N6600 GT/TD is ahead of the RADEON 9800 XT in 1600x1200.
The “eye candy” mode suits better for our benchmarking needs. You can see the ASUS card successfully crunching through the low resolutions, but then slowing down almost to the level of the RADEON 9800 XT in 1600x1200 as the load on the memory subsystem becomes unbearable. On the other hand, the slightly overclocked N6600 GT wins this round, too.
Without full-screen anti-aliasing and anisotropic filtering, the ASUS N6600 GT/TD is always ahead of the RADEON 9800 XT on the Metallurgy map. Overclocking helps to make the gap bigger but the ASUS card still cannot overtake the RADEON X800 PRO.
FSAA and anisotropic filtering activated, the ASUS is still faster than the RADEON 9800 XT but loses its ground suddenly in 1600x1200 due to the narrow 128-bit memory bus.
This shooter runs generally faster on NVIDIA’s GeForce 6 cards, and the ASUS N6600 GT/TD does outperform the RADEON 9800 XT across all the resolutions, including 1600x1200. The advantage of the ASUS card amounts up to 35%. Unfortunately, the GeForce 6600 GT cannot provide a playable frame rate in resolutions above 1024x768 at the maximum graphics quality settings. The RADEON 9800 XT is not capable of even that, though.
None of the graphics cards with eight pixel pipelines is capable of running this game at an acceptable speed with enabled anisotropic filtering and anti-aliasing. The absolute speeds of the ASUS N6600 GT/TD and the RADEON 9800 XT are similar.
The Pier map depends on the performance of the whole system, but it anyway puts the graphics cards with a high pixel shader performance on top. Particularly, the RADEON X800 PRO is faster than the GeForce 6800 GT in low resolutions. As for the ASUS N6600 GT/TD, it works like the RADEON 9800 XT at the default frequencies. When overclocked, it almost equals the GeForce 6800.
The ASUS N6600 GT/TD is slightly behind the RADEON 9800 XT – by no more than 9-10%. Overclocking helps the ASUS reach the level of the GeForce 6800 again.
The demo record on the Research map abounds in light sources and thus requires multi-pass rendering. It means graphics processors that support long shaders have an advantage here since processing more light sources in a single pass reduces the total number of passes required to construct the scene. This advantage shows up in the results: the N6600 GT/TD is more than 20% ahead of the RADEON 9800 XT.
The N6600 GT/TD loses its position in the “eye candy” mode: its performance falls to the level of the RADEON 9800 XT as soon as 1280x1024 resolution and goes lower still in 1600x1200. Overclocking amends the situation somewhat: at 560/1040MHz frequencies the N6600 GT/TD is successfully competing with the RADEON 9800 XT even in the highest resolutions.
The speed of the ASUS N6600 GT/TD is sufficiently high in this game, too. This graphics card surpasses its immediate rivals in 1600x1200, save for the RADEON X800 PRO. It seems like the core frequency is more important for Painkiller: Battle out of Hell than the number of pixel pipelines.
It’s almost the same in the “eye candy” mode with its full-screen anti-aliasing and anisotropic filtering, except the highest resolutions where the memory bandwidth comes into play. The graphics cards with a more efficient memory controller come out on top then.
The demo we recorded on the d_canals_12 map is almost all water rendered with the help of pixel shaders, so graphics cards with a high pixel shader performance are on top here. They are the 16-pipelined RADEON X800 PRO and GeForce 6800 GT. The ASUS N6600 GT/TD keeps on the same level with the GeForce 6800; the latter has 12 pipelines and a 256-bit memory bus, but the ASUS card makes up for that with its high operational frequencies.
The N6600 GT/TD is slower than the RADEON 9800 XT in the “eye candy” mode, but overclocking helps to close the gap. By the way, take note that the GeForce 6600 GT AGP can give out a playable frame rate in one of the most difficult scenes of Half-Life 2, in the most popular resolution today (that is, 1280x1024) and with enabled full-screen anti-aliasing and anisotropic filtering.
The scene recorded on the d3_c17_02 map depicts a combat in the streets of a city. This scene is CPU-dependent due to the numerous character models, but it’s clear as soon as 1280x1024 resolution that the RADEON 9800 XT is faster than the ASUS N6600 GT/TD.
The card from ASUS falls even further back behind its competitor in the “eye candy” mode. The gap is as big as 40% in 1600x1200 resolution. However strange it seems, but the GeForce 6600 GT AGP can yield about 60fps only in 1024x768 here.
Splinter Cell: Pandora Tomorrow is usually somewhat faster on graphics cards with GPUs from ATI Technologies. Even the overclocked N6600 GT/TD cannot come close to the results of the RADEON 9800 XT.
This game doesn’t support full-screen anti-aliasing correctly so we don’t publish the results for the “eye candy” mode. We also want to remind you that Splinter Cell doesn’t have any means to benchmark the performance of the graphics card, so we perform our tests manually, with the FRAPS utility. This method has an inherent inaccuracy in itself.
The N6600 GT/TD cannot claim the first place among eight-pipelined solutions in this 3D third-person shooter, either, although overclocking helps to reduce the gap between it and the RADEON 9800 XT from 15% to 5-6%. This game is tested manually, too, so the numbers should be regarded as approximations.
The excellent performance of the ASUS N6600 GT/TD in this flight simulator is largely due to the efficient OpenGL driver from NVIDIA. The ASUS card even caught up with the GeForce 6800 at overclocking.
The ASUS feels at its ease with enabled full-screen anti-aliasing and anisotropic filtering, too, although the narrowness of its memory bus is felt in 1600x1200. Overclocking helps this card to get the second place, after the GeForce 6800 GT.
The N6600 GT/TD performs better than the RADEON 9800 XT in Lock On, but overclocking doesn’t give any substantial gains, seemingly because of the very low frame rates all the participating graphics cards produce in this game.
Quite surprisingly, the N6600 GT/TD even outdid the GeForce 6800 GT here! Overclocking adds next to nothing to the ASUS card. We want to remind you about the capricious behavior of Lock On – the numbers should be regarded as nothing more but approximations.
The new Colin McRae Rally makes a good use of pixel shaders and, according to the diagrams, the RADEON X800 PRO with the RADEON 9800 XT go unrivalled through this test. The ASUS N6600 GT/TD sinks to the last place when it works at its default frequencies. At the overclocked frequencies it is faster than the GeForce 6800 and comes close to the GeForce 6800 GT in high resolutions.
It’s different with full-screen anti-aliasing and anisotropic filtering: although the RADEON X800 PRO with its advanced memory controller remains on top, the RADEON 9800 XT slows down considerably and has almost the same result as the overclocked N6600 GT/TD. At its default frequencies the N6600 GT/TD is almost as fast as the GeForce 6800.
The ASUS N6600 GT/TD is clearly faster than the RADEON 9800 XT in this test. Overclocking helps to add a few more fps. The speed is quite normal for a strategy game, even in 1600x1200.
We enable FSAA and anisotropic filtering to see the N6600 GT/TD breaking away from the RADEON 9800 XT at the default frequencies. This is a rare thing to witness: as a rule, NVIDIA’s cards generally work somewhat slower that ATI’s ones in the “eye candy” mode.
Perimeter is an exceedingly demanding game that can use up all the resources of a modern graphics card. In fact, none of the devices participating in this test, except the relatively expensive RADEON X800 PRO, can ensure a satisfactory speed here. The N6600 GT/TD boasts a more advanced architecture than the RADEON 9800 XT, so its result is better. The ASUS card enjoys a 30% advantage over the 8-pipelined solution from ATI in 1600x1200 resolution.
When we enable FSAA and anisotropic filtering, the 128-bit memory bus of the ASUS N6600 GT/TD becomes a bottleneck. The device is much slower than the RADEON 9800 XT in this mode.
The semi-synthetic benchmark from Square Enix reports the RADEON 9800 XT suits better than the ASUS N6600 GT/TD or any other GeForce 6600 GT for that matter, for playing Final Fantasy XI. This game seems to depend greatly on the bandwidth of the graphics memory subsystem. Our tests suggest that the GPU frequency gain doesn’t affect the performance of the card much.
Aquamark3 loads the graphics card with its geometrically complex scenes with a high overdraw coefficient. There are few pixel shaders here, not to mention version 2.0 ones. Thus, the performance of this test directly depends on the speed of the vertex processors and on the card’s ability to remove invisible surfaces. The GeForce 6600 GT is an example of such a card, so the ASUS N6600 GT/TD is far ahead of the RADEON 9800 XT. Overclocking results in a nice performance boost, but the ASUS still cannot overtake the GeForce 6800.
The new graphics card from ASUS isn’t that spectacular in 1600x1200, although its performance matches that of the RADEON 9800 XT. Overclocking is even more efficient here – the N6600 GT/TD almost overtakes the GeForce 6800 GT at 560/1040MHz frequencies.
Comparing the total 3DMark03 scores, we can see a big advantage of the ASUS N6600 GT/TD over the RADEON 9800 XT. The gap is almost 2,000 points. Let’s see what each of the subtests has to show us.
The N6600 GT/TD gains a clear victory in the first test. This test is simple, however, and uses just a small portion of the capabilities available in modern graphics cards.
The N6600 GT/TD also beats the RADEON 9800 XT in the “eye candy” mode, except in 1600x1200 resolution. Yet overclocking is so rewarding here that the ASUS card easily outperforms its ATI competitor even in highest resolutions.
The second test’s rendering techniques are suited better for graphics cards with NVIDIA’s NV4x architecture. That’s why the ASUS card is on the winning side again, while overclocking adds 6-7fps more to the frame rate, absolute speeds being about 30-60fps.
The RADEON 9800 XT cannot challenge the N6600 GT/TD even in the “eye candy” mode.
The third test is a copy of the second one in terms of the rendering techniques employed. The results are quite naturally the same.
The ASUS N6600 GT/TD shows the advantages of its modern graphics architecture over the older one in the “eye candy” mode, too.
The fourth of the 3DMark03 tests stands somewhat apart from the other three. This is probably the only test that should be applied to test modern graphics cards as it is the only one to use complex version 2.0 pixel shaders. Although the RADEON 9800 XT executes such shaders quite quickly, the ASUS N6600 GT/TD is even better at this task: up to 20% better in 1280x1024! The speed of the N6600 GT/TD is probably impeded by its 128-bit memory bus in higher resolutions, but it is never slower than its competitor anyway.
The ASUS card is almost as successful in the “eye candy” mode. It is no worse than the RADEON 9800 XT here and outperforms the 12-pipelined GeForce 6800 at overclocking. Thus, the total score of the ASUS N6600 GT/TD in 3DMark03 seems more than deserved.
At the default frequencies the N6600 GT/TD is rewarded a higher score than the RADEON 9800 XT in 3DMark05. Overclocking helps the ASUS card to beat the GeForce 6800. Let’s see where this result comes from.
The first test from the 3DMark05 suite looks like a regular 3D shooter. The N6600 GT/TD is already faster than the RADEON 9800 XT and gives way only to the two much more expensive cards: GeForce 6800 GT and RADEON X800 PRO. Overclocking gives a nice speed bonus to the ASUS product.
3DMark05 doesn’t support full-screen anti-aliasing in resolutions above 1024x768 on graphics cards equipped with less than 256 megabytes of memory. Anyway, the results of the N6600 GT/TD are very high even here. Overclocking increases the frame rate almost to the level of the GeForce 6800 GT.
It’s hard to tell the genre of the second test. It looks like an RPG, but the numbers we have here are nearly the same as we got in the first test: the ASUS N6600 GT/TD is confidently ahead of the RADEON 9800 XT as well as of the GeForce 6800.
It’s roughly the same in the “eye candy” mode with its anti-aliasing and anisotropic filtering.
The third test can be considered as an adventure game. It’s the most complex in the suite, using sophisticated pixel shaders that are difficult texturally as well as math1ematically. The NV43 graphics processor is more advanced than the ATI R360 and supports long pixel shaders, and the ASUS N6600 GT/TD proves it, beating the RADEON 9800 XT. Unlike in the previous two tests, the ASUS card cannot outperform the GeForce 6800 at the default frequencies – the latter has 12 pipelines all the same – but it does so at overclocking.
The N6600 GT/TD is just a tiny step ahead of the ATI RADEON 9800 XT in the “eye candy” mode, but it’s a win nonetheless. Overclocking helps to increase the gap even further.
Like in 3DMark03, we can’t but acknowledge that the ASUS N6600 GT/TD wins in 3DMark05, and quite deservedly.
The ASUS N6600 GT/TD’s winning features are cute looks, highest quality of manufacture and crystal-sharp 2D image. The cooling system, regrettably, is not as effective as we might wish: the HSI chip equipped with a simple passive heatsink used to become very hot during our tests. Another drawback of the N6600 GT/TD is the lack of such things as S-Video à RCA adapter or Molex splitter among the accessories. But as we said at the beginning of the review, we may have dealt with a presale version of the product, and off-the-shelf cards may be complemented with everything necessary.
We don’t want to talk long about the performance of this card. It is like any other GeForce 6600 GT AGP would have. That is, it’s high enough for playing most modern games with comfort. The N6600 GT/TD is generally faster than the RADEON 9800 XT, but sometimes finds itself behind it, especially when full-screen anti-aliasing and anisotropic filtering are enabled. Well, modern games are so difficult that it’s often not possible to use 4x FSAA on the GeForce 6600 GT AGP or RADEON 9800 XT. Like all graphics cards on NVIDIA’s processors, the ASUS N6600 GT/TD works well with OpenGL applications, so it’s going to suit perfect to people who play shooters from id Software and flight sims.
Today, products based on the NVIDIA GeForce 6600 GT AGP chip are in fact the most optimal choice on the mainstream graphics card market. Things may change with the arrival of the AGP version of the RADEON X800 but it’s unclear as when this is going to happen. Until that, the GeForce 6600 GT AGP remains the best choice for any owner of an AGP system who doesn’t want to spend more than $180-230 for the graphics card. Considering the traditionally high prices on ASUS products, we expect the price of the reviewed card to be closer to the second number of the range, but you spend this money to warrant a highest quality of manufacture and excellent image quality in all resolutions. Enthusiasts and overclockers might be interested in the Top Limited Edition of the card that is equipped with faster memory and is clocked at higher frequencies. Overall, the ASUS N6600 GT/TD is almost free from drawbacks and deserves our sincere applause.
Highs:
Lows: