by Yaroslav Lyssenko
07/10/2006 | 10:26 AM
The leading positions of the major supplier of PC hardware parts ASUS are ensured on one hand by huge production volumes and on the other by the effort of the company’s R&D department because products with the ASUS brand boast specific characteristics you cannot find in their analogs from other manufacturers.
Graphics cards manufactured by ASUS in the last years can be divided into several groups: one group are exact copies of the etalons developed by ATI and Nvidia; the second group comprises unique products with passive cooling systems or improved technical characteristics (these come out in the Silent and the Top product series); the third group includes ordinary products with nothing particularly remarkable about them, even though they bear the ASUS brand.
On transitioning to the R580-based graphics card series, ASUS seems to have discontinued its Top product series which used to feature increased clock rates, redesigned coolers and extra accessories.
Thus, the most powerful graphics card on an ATI GPU that ASUS is offering at the moment is ASUS EAX1900XTX/2DHTV/512M which is going to be the subject of this review.
ASUS just wouldn’t be it own self if they didn’t offer their highest-performance solution in a box smaller than before. So, here is the same box the size of a briefcase:
The design of the box was borrowed from the ASUS EAX1800XT TOP for some reason (for details see ASUS EAX1800XT TOP/2DHTV Graphics Card: Living on the Edge): a totally black color scheme and a picture of an armored soldier. The box has a handle so that you didn’t have any troubles carrying it home. Yes, there’s a downside to this large package, particularly a higher rate of warehouse storage, but it is well compensated by the effect it produces to attract the potential customer. The sheer size of the box is an indicator that this is a top league product!
There is some technical info about the card on the front of the box like the amount of graphics memory, two DVI-I connectors, support for HDTV and VIVO. The box also mentions the copy of Peter Jackson’s King Kong that you will find inside, too. You can unfold the front panel to learn more about the ASUS EAX1900XTX but you can’t see it through a window as the boxes of previous flagship products from ASUS permitted. Instead of a window, the potential buyer can read a detailed description of the ASUS Extreme Video and ASUS Extreme Gaming concepts and the technologies these concepts are embodied in:
Frankly speaking, we don’t think these technologies are going to be utilized by too many users who buy this graphics card because they require ASUS’ exclusive driver to work. ASUS’ driver often lags behind the current version of ATI’s official Catalyst driver and is much larger (about 100MB in total), while the use of the latest version of the official driver guarantees the user maximum performance and lack of problems in new games.
There are a few compartments inside the box the accessories are snugly fitted in. The graphics card is firmly fixed in a separate small polyurethane-foam box and is thus secured against damage during transportation. Here are the accessories we found in the box:
The software pack comes on 9 discs and, besides discs with drivers and a multilingual user manual, includes Peter Jackson’s King Kong , Project: Snowblind and an ASUS GamePack disc with three more games. Also in the software pack are programs for watching, authoring and editing video content, namely ASUS DVD XP, CyberLink Power Director 3 and Medi@Show.
Unfortunately, these are in fact all the accessories to the ASUS EAX1900XTX, although it belongs to the category of the most expensive and high-performing graphics cards. It’s good to have a gaming hit and four less popular titles in the package, yet we would want to see something special in a box with such a powerful graphics card, something that would put it apart from other companies’ products, especially since ASUS has always been renowned for its ability to surprise the buyer in some nice way.
For example, a gamepad, like the ASUS XitePad included with the EAX1800XT Top (for details see our article called ASUS EAX1800XT TOP/2DHTV Graphics Card: Living on the Edge), would be welcome here since the EAX1900XTX is targeted exclusively at PC gamers. We don’t know why ASUS doesn’t enclose the mentioned gamepad with its best R580-based solution as they did with their other product, the EAX1900 CrossFire.
Apart from this, the accessories to the ASUS EAX1900XTX are more than sufficient. The packaging itself is good, too, but they might have changed the design which is the same for as many as four graphics cards from ASUS: the EAX1800XL, EAX1800XT TOP, EAX1900 CrossFire and EAX1900XTX.
After we’ve seen the package of and accessories to the ASUS EAX1900XTX, it’s time to have a closer look at the card itself.
So, this is the good old reference design developed by ATI. The sticker with the manufacturer’s logo in the middle of the cooler’s casing is the only indication that this is a product from ASUS. On the reverse side of the PCB there is also a paper sticker that reads “ASUS EAX1900XTX/2DHTV/512M/А”.
It’s all the same on the graphics memory side: there are eight chips of GDDR3 memory from Samsung (K4J52324QC-BJ11) in 136-pin FBGA packaging on the PCB. The total amount of graphics memory is 512MB and it is clocked at 775 (1550) MHz exactly as described in the official Radeon X1900 XTX specification. As usual, the memory contacts the cooler’s base through rubber-like heat-conductive pads. The pads are too thick to be very efficient, but they do their job all right. The graphics processor is clocked at 650MHz in compliance with the official specification. To improve the contact with the copper base of the heatsink a thick dark-gray thermal paste with low thermal resistance is used, as usual.
The Rage Theater chip installed on the card endows it with VIVO functionality. It is impossible to use the video input and to attach an YPbPr-interfaced device to the card simultaneously because the appropriate adapters occupy one and the same connector. This is a common problem with almost all graphics cards supporting this functionality.
ASUS’ engineers must have been left pleased with the capabilities of the standard cooling system because they didn’t introduce any changes into it. The cooling system installed on the ASUS EAX1900XTX is exactly alike to the cooler we first saw on the Radeon X1800 XT: the air from inside the PC is driven through the heatsink’s copper ribs and is then exhausted out of the system case (for details see our article called The New RADEON: A Family Portrait). Thanks to a thick copper base and heat pipes, this cooler copes easily with the rather hot R580 chip, but the effective cooling comes at a cost. The high-speed fan and the resonating plastic casing of the system produce a very specific sound that is irritatingly distinctive against the general noise from a working PC. You will hear this sound only when the fan speed is increased above normal, though.
We are always interested in finding the overclocking limits of a high-performance product. The ASUS card was tested too and was found capable of working at 670MHz GPU and 800 (1600) MHz memory clock rates. This is a normal result for a Radeon X1900 graphics card manufactured by ATI’s guidelines.
Graphics cards are usually overclocked in our labs in an accessible, safe and cheap way, so we only helped the card’s own cooler by setting a 120mm fan to blow at the PCB. We also tried to replace the card’s thermal interface with a different one, but that only helped increase the GPU clock rate to 675MHz which isn’t impressive for an R580 chip, either.
And finally we decided to use RivaTuner v2.0 RC 16 and set the speed of the card’s stock cooler at the maximum value. To improve the overall stability and overclockability of the card we increased the GPU and memory voltages with ATITool 0.25 Beta 14 (from 1.175V to 1.475V voltage for the GPU and from 2.084V to 2.096V voltage for the memory).
As a result we managed to increase the GPU frequency to 710MHz but the system began to hang up after running tests for an hour and we had to lower this frequency to 705MHz. The memory chips were stable at 815 (1630) MHz. So, our results suggest that the R580 chip works almost at the limit of its capabilities. It is virtually impossible to use the card when its fan works at the maximum speed because of unbearable noise. The increase in the voltages of the graphics card didn’t improve its overclockability.
Water- or cryogen-based cooling may bring you better overclocking results, but it is unlikely to be used in an average PC.
The cooling system on the ASUS graphics card behaves in an expectable way. In other words, it works exactly like the cooling system of the ATI Radeon X1900 XTX reference card the ASUS EAX1900XTX is an exact copy of. When you turn the computer on, the fan works at the maximum speed, producing unbearable noise, but then its speed goes down in a step-like manner and the fan becomes almost silent. The fan speed management system seems to be ready to react to an increase in the GPU temperature, but the level of noise always remains in a comfortable range even though it has that irritating “plastic” twang in it. Some people may also find it discomforting that the card begins to “play” with the fan speed under high computational loads, increasing and decreasing it, and thus changing the tonality of the noise, all the time.
As we wrote in the previous section, we checked the overclocking potential of the ASUS EAX1900XTX at the maximum speed of its fan. Using a sound-level meter Velleman DVM1326 we can now provide you with an objective estimate of the amount of noise produced by a particular graphics card (the measurement methodology was described in our article called Brother Against Brother: Gigabyte GV-RX19X512VB Vs. GV-NX79X512DB).
The level of background noise in our test room was 36dBA. At a distance of 1 meter from the working testbed with a passively cooled graphics card installed our sound-level meter showed 40dBA. We measured the level of noise produced by the cards in three modes: 2D, 3D typical and 3D maximum (the maximum level of noise with the automatic fan-speed adjustment enabled). The noise is measured at a distance of 1m and 5cm from the working testbed assembled in a Chieftec LBX-01 system case with the side panel removed. Here are the results:
The speed of ASUS’ cooler was set manually, so the results are identical in all the modes. The decibel is a logarithmic rather than linear measurement unit, so an increase in the sound level by 3dB corresponds to a twofold increase in the sound intensity. However, because of the non-linearity of the human hearing it is assumed that the perception of a sound source having become two times louder corresponds to a sound level increase by 10dB.
We increased the GPU and memory voltages when overclocking the ASUS EAX1900XTX, so we decided to check the card’s cooler at the maximum speed and at 75% of the max speed. One look at the results is enough to make it clear that the reference cooler installed on Radeon X1900 XT/XTX is not at all suitable for extreme overclocking. When its fan speed is the highest or near highest, it can of course cool the card, but it produces a tremendous 70dBA of noise! You won’t be pleased to be near this graphics card for long!
So, if you are into extreme overclocking and want to experiment with an ASUS EAX1900XTX, the first thing you should do is to get rid of the reference cooler. It’s up to you to decide on the replacement, basing on your tastes and financial capabilities. You may want to use a high-quality air cooler from Zalman or Arctic Cooling, a water-based solution or even a phase-change cooling system.
This graphics card provides a very high image quality in 2D applications, including 1800x1440@75Hz and 1600x1200@85Hz modes. The ASUS EAX1900XTX/2DHTV/512M delivers a sharp picture without fuzziness or shadowing which you can sometimes see on cheap graphics cards with non-standard PCBs and inexpensive components.
We ran our tests in the following testbed:
We set up the ATI and Nvidia drivers in the same way as always:
We selected the highest graphics quality settings in each game, identical for graphics cards from ATI and Nvidia, except for the Pacific Fighters flight simulator that requires vertex texturing for its Shader Model 3.0 rendering mode. Radeon X1000 doesn’t support this feature therefore we ran the game in Shader Model 2.0 in this case. We did not edit the configuration files of the games. We also didn’t use the driver profiles optimized for given games. To measure the performance we either used the integrated tools of the games we tested in, or if there were none available, resorted to FRAPS utility. If it was possible, we measured minimal performance as well.
To load the video subsystem to the full extent and to minimize the influence of the CPU speed on the performance results we didn’t test the systems in the “pure speed” mode. We only ran the tests in “eye candy” mode with full-screen anti-aliasing and anisotropic filtering. The graphics subsystem potential is used in a more optimal manner in this case. Besides, the image quality is considerably higher than in those cases when no full-screen anti-aliasing and/or anisotropic filtering are enabled.
We enabled FSAA and AF from the game if possible. Otherwise we forced the necessary mode from the ATI Catalyst and Nvidia ForceWare graphics card driver. We didn’t include any results of the overclocked graphics cards, because we didn’t manage to raise the clock speeds high enough. We have also tested the following graphics cards besides the ASUS EAX1900XTX/2DHTV/512M:
I would like to stress specifically that today we are looking only at single-chip solutions, because the dual-chip GeForce 7950 GX2 doesn’t have any actual competitors from the pricing as well as the performance standpoint. Therefore, if you would like to squeeze the maximum performance from a single graphics card, then you should check out our article called Two for One: Nvidia's Dual-Chip GeForce 7950 GX2 Reviewed, because this solution is quite far ahead of Radeon X1900 XTX with few exceptions, while its price is about $100-$200 higher.
These games and applications were used as benchmarks:
First-Person 3D Shooters
Third-Person 3D Shooters
Even if you turn on full-screen antialiasing and anisotropic filtering, all modern graphics cards will still provide an exceptionally satisfying speed in Battlefield 2.
The ring-bus memory controller helps the ASUS EAX1900XTX increase its advantage over the opponents when performing full-screen antialiasing in high resolutions. In the resolution of 1600x1200 the reviewed card is 15% faster than the GeForce 7900 GTX. The average performance is over 100fps. This is far above the 60fps point which is the widely accepted limit of comfort for first-person shooters.
Products on chips from the Canada-headquartered ATI Technologies are usually slower than their opponents on Nvidia’s GPUs in OpenGL applications. The ASUS card is about 15% behind the GeForce 7900 GTX in all the resolutions tested. Nvidia’s solution has a more efficient OpenGL driver and is capable of processing two times the number of Z-values per clock cycle, which helps a lot when rendering stencil shadows.
Note that it’s only in 1600x1200 that the performance of the cards, except for the GeForce 7900 GTX, sinks below the 60fps mark.
This test seems to be unable to load 48 pixel processors as is proved by the coinciding results of the Radeon X1800 XT and the Radeon X1900 XT. The GeForce 7900 GTX is much faster than the ASUS card and even allows playing the game comfortably in 1280x1024 with enabled FSAA and anisotropic filtering.
In the closed tunnels of the Oblivion that feature a lot of light sources and complex shadows, pixel shaders processing speed and Z processing efficiency matter the most. Although ASUS EAX1900XTX and GeForce 7900 GTX retain certain parity here, you should pay attention to the difference in minimal fps values. Thanks to twice as many pixel processors onboard, the ASUS card proves a better balanced solution in this test application than the Nvidia based product.
Of course, the open areas of Oblivion game with complex vegetation models may be pretty difficult even for the latest graphics accelerators. Only if we look at the average fps rate, we will be able to claim parity between ASUS EAX1900XTX and GeForce 7900 GTX. However, Nvidia’s flagship product cannot ensure smooth gaming experience with the minimal fps rate of less than 20 fps. So, the best solution for TES IV game remains Radeon X1900 XTX represented in our today’s review by the ASUS graphics card.
The Radeon X1900 XTX from ASUS provides a frame rate of over 90fps in 1600x1200 with enabled full-screen antialiasing and anisotropic filtering, leaving the rest of the graphics cards behind in this test. Well, the participating devices all show satisfactory speeds, except for the GeForce 7900 GT which is slower due to its low memory and GPU clock rates.
Pixel shaders are used on the Research map to process the lighting model with many light sources, so it is logical that a high frequency and large number of pixel processors make a difference in the rendering speed for this scene. The graphics cards all deliver comfortable speeds here, though.
The ASUS EAX1900XTX is still in the lead, but the GeForce 7900 GTX is successfully competing with the Radeon X1900 XT. The slow GeForce 7900 GT now runs along with the Radeon X1800 XT which has 16 pixel processors, too.
The HDR support for ATI’s Radeon X1000 cards still has a beta status in the current version of Far Cry, but the performance of the ASUS EAX1900XTX is rather high in this mode and allows using 1280x1024 resolution. Unfortunately, Far Cry’s implementation of HDR is not efficient when it comes to the Radeon X1000 architecture.
The same thing can be observed on the Research level. The difference in the GPU frequency being a mere 25MHz, we are not sure why there is a 20% difference in speed between the Radeon X1900 XT and the ASUS EAX1900XTX.
The efficient memory subsystem makes the ASUS EAX1900XTX a leader in 1600x1200 due to the use of full-screen antialiasing, but the speed is too low for comfortable play, so the victory by a margin of 4fps is of no practical value. In the lower resolutions the ASUS is just a few fps behind the GeForce 7900 GTX.
ASUS EAX1900XTX wins only in 1024x768 resolution with a slight advantage over GeForce 7900 GTX. However, in general we can state that our hero performs equally fast with the Nvidia based rival. If we compare the results against those of Radeon X1800 XT, we will see that the numerous pixel shader processing units quite pretty demanded here.
Only the lowest resolution, however, can be considered acceptable for comfortable gaming, because the minimal fps rate may drop below 45 fps in especially complex scenes.
Half-Life 2 is already not the hard nut it used to be for the graphics hardware of the past. Today’s graphics cards easily reach the performance ceiling and have similar results in every resolution, even with enabled full-screen antialiasing and anisotropic filtering.
Looks like even the finalized Half-Life 2 engine cannot load all 48 pixel processors of ASUS EAX1900XTX to the full extent. If ASUS’ advantage in the lowest screen resolution is only 3-4fps over its GeForce 7900 GTX competitor, then in 1600x1200 the picture is just the opposite. Nvidia’s flagship product manages to get about the same few fps ahead of the rival.
ASUS EAX1900XTX copes beautifully with its task even in 1600x1200 in eye candy mode: its shows 76fps that ensures smooth and pleasant gaming experience.
One of the most promising games of this year – Prey – is coming out very soon. Although this game uses the original Doom III engine, the performance in the demo version of this game is very much different from what we have seen before in the id Software product. GeForce 7900 GTX is ahead of all in the test mode with enabled FSAA and anisotropic filtering. At the same time we can’t say that ASUS EAX1900XTX yields a lot to its direct competitor by losing only 5fps. ASUS solution maintains a steady second position leaving GeForce 7900 GT the same 5 fps behind, which is actually quite unexpected for a game based on the OpenGL id Software engine.
Even with enabled FSAA the leadership in Quake 4 belongs to GeForce 7900 GTX. ASUS EAX1900XTX cannot really take advantage of its 48 pixel processors here. Although in 1600x1200 the more advanced memory bus interface of the Radeon X1900 helps reduce the gap, GeForce 7900 GTX remains ahead.
ASUS EAX1900XTX graphics card ensures acceptable level of performance in all resolutions, showing over 60fps even in 1600x1200.
Serious Sam 2 uses a number of shaders with a lot of texture lookups which allows the GeForce 7900 GTX to stand out among its opponents. The ring-bus memory controller helps the ASUS card narrow the gap, but the game is practically unplayable on it in resolutions higher than 1280x1024.
The engine of this game makes use of a number of shaders, so the ASUS version of the Radeon X1900 XTX can show its best here. The ASUS card takes the lead as the display resolution grows, being greatly aided by its highly efficient memory controller. The GeForce 7900 GTX isn’t too far behind, though.
Just like Serious Sam 2, this game is optimized for the GeForce 6/7 architecture and the Radeon X1000 family have been generally slower here. You should also be aware that you can’t achieve the maximum graphics quality on Radeon cards here because none of the Radeon X1000 series chips supports vertex texturing. This feature is only available on GeForce 6 and 7.
The ASUS EAX1900XTX ensures a comfortable frame rate in 1280x1024 with enabled FSAA and anisotropic filtering.
Above in this review you’ve seen games optimized for Nvidia’s graphics solutions, but the latest title in the X3 space simulator series is quite the opposite.
There is almost no difference between the GeForce 7900 GTX, GeForce 7800 GTX 512 and GeForce 7900 GT in low resolutions despite the considerable difference in the cards’ technical characteristics. The ASUS EAX1900XTX, Radeon X1900 XT and Radeon X1800 XT rank up in the order of descending performance and are all much faster than their GeForce series competitors.
The game doesn’t use too many shaders, so the ASUS card provides a comfortable speed in every resolution of the “eye candy” mode.
The drop in the Radeon X1800 XT’s performance is due to the game’s using shaders that cannot be executed quickly enough on 16 pixel processors. In 1280x1024 the ASUS EAX1900XTX is 15% faster than the GeForce 7900 GTX.
Note that both the flagship graphics solutions are capable of running this game fast in 1600x1200 with enabled FSAA and anisotropic filtering.
The graphics cards must have a high scene fill-rate and be fast at rendering stencil shadows to succeed in this test. The engine of Dawn of War obviously prefers Nvidia’s GeForce 7 graphics cards. Here, the ASUS EAX1900XTX is no rival even to the cheaper GeForce 7900 GT, not to mention the GeForce 7900 GTX.
Although the ASUS card provides an acceptable average performance in all the resolutions, you should be aware that its speed may bottom out to below 20fps. This has a negative effect on playability, of course.
With an advantage of 700 points the ASUS card enjoys the first place in this benchmark. Nvidia’s flagship is a little better than the Radeon X1900 XT that has considerably lower frequencies than the Radeon X1900 XTX.
The Radeon X1900 XTX in ASUS’ implementation is in the lead in the first test, although its advantage over the Radeon X1900 XT and the GeForce 7900 GTX isn’t bigger than 5-10%.
The second test abounds in shadows which are processed faster on the GeForce 7900 GTX. This card takes the lead in 1024x768. In the higher resolutions the ASUS EAX1900XTX regains the lead thanks to its ring-bus memory controller.
High pixel shader performance and an efficient memory subsystem are both equally important in the third test. These conditions are favorable for the Radeon X1900 family. Note that the GeForce 7900 GTX is slower not only than the ASUS EAX1900XTX, but also than the Radeon X1900 XT!
So, the first place of the ASUS EAX1900XTX in 3DMark06 seems well-deserved. The Radeon X1900 XTX graphics card is a lucky combination of high pixel shader performance and an efficient graphics memory subsystem and seems to be better balanced in comparison with the competing solutions.
The GeForce 7900 GTX wins in the newest version of Futuremark’s benchmark, leaving the ASUS EAX1900XTX a little behind. Nvidia’s flagship product superbly handles tasks in which high shader performance should be combined with a high scene fill rate. It is such scenes 3DMark06 is comprised of.
The SM2.0 tests appreciate a high fill rate and a fast execution of version 2.0 shaders. So, the ASUS card has nothing to show its strength on and it takes the second place behind the GeForce 7900 GTX.
The second group of tests, using SM3.0 and HDR, is a playground for the ASUS EAX1900XTX. It’s here that the Radeon X1900 architecture shows its very best. The 48 pixel processors and the ring-bus memory controller are much more efficient at executing the huge amount of math1ematics-heavy version 3.0 pixel shaders and processing HDR lighting. Unfortunately, we can’t compare solutions from ATI and Nvidia in the FSAA modes because the latter do not support FSAA and HDR simultaneously.
In the first two resolutions the ASUS EAX1900XTX is about 15% slower than the GeForce 7900 GTX, but in 1600x1200 it is the memory subsystem performance that becomes more important. The Radeon X1900 XTX is so far unrivalled from this standpoint and leaves the GeForce 7900 GTX behind or equals it.
The second test isn’t that sensitive to the texturing speed as the first one, and the GeForce 7900 GTX gives way to the ASUS EAX1900XTX. In its turn, the ASUS is not much faster than the cheaper Radeon X1900 XT.
As you can see, the GeForce 7900 GTX wins the SM2.0 tests only by winning the first of them, which is large and very sensitive to the texturing speed.
We were left with mixed feelings after we tested the ASUS EAX1900XTX/2DHTV/512M. This card surely has a lot of advantages over its competitors and the card’s heart Radeon X1900 XTX very often delivers unrivalled performance, e.g. with enabled full-screen antialiasing or in scenes with a lot of pixel shaders. At the same time, GeForce 7900 GTX-based products will be a better choice and will provide a higher performance in scenes that require a high speed of texturing and in OpenGL applications.
Although the graphics cards from the Radeon X1900 and GeForce 7900 families are peers in today’s gaming applications, the ASUS EAX1900XTX may become a leader in upcoming games. The programmers put ever more math1ematic instructions into pixel shaders, so the 48 pixel processors of the R580 chip will be able to render complex scenes faster than the 24 pixel processors of the GeForce 7900 GTX. However, you should keep in mind that GeForce 7950 GX2 demonstrates much better results than Radeon X1900 XTX and has every chance to become an even more progressive solution.
The ASUS card is a copy of the reference design and is manufactured by Sapphire Technology, so it has a poor overclocking potential as is characteristic of products based on top-end chips.
Following its long-established marketing policy ASUS complements its highest-performance solutions with an impressive set of accessories the potential buyer is sure to appreciate. And so, the ASUS EAX1900XTX/2HDTV/512M is a very appealing choice for those users who not only need high performance but also want to find games and additional software in the graphics card’s package.