by Alexey Stepin , Yaroslav Lyssenko
02/05/2007 | 10:16 AM
The widespread Nvidia G71 chip features superb overclockability, besides everything else. Its frequency was 650MHz for the main subunits and 700MHz for the vertex shader processors on the GeForce 7900 GTX, but that graphics card is already out of production and is to be replaced with the GeForce 8800 GTS 320MB model, a solution featuring a new-generation GPU. However, the GeForce 7900 GTX has a younger brother, the GeForce 7950 GT, described in our earlier review called Nvidia GeForce 7950 GT Graphics Card Review and highly appreciated by overclockers.
Lower in the product hierarchy there is the GeForce 7900 GS model which is based on the G71, too, but has much more modest technical characteristics. Our tests showed that working at its default GPU and memory frequencies the GeForce 7900 GS is far slower than its opponent in the same price category, the ATI Radeon X1950 Pro. You can now find Nvidia’s solution at reduced prices in retail shops, yet this doesn’t help it much.
But considering the overclockability of the G71 chip, which can generally work at frequencies above 550MHz, the release of pre-overclocked versions of the GeForce 7900 GS was inevitable. After all, not all gamers can or want to overclock their graphics card by themselves. One such pre-overclocked card has been introduced by ASUS. The GPU frequency of the EN7900GS Top is raised far above the default and the manufacturer claims a 20% performance increase over the reference 7900 GS. The EN7900GS Top thus seems to be lifted up to the same level with the GeForce 7950 GT. Is it really so and can ASUS be offering two products in the same market niche?
We decided to check it all out by benchmarking the EN7900GS Top in comparison with ASUS’ EN7950GT, which is a version of Nvidia’s GeForce 7950 GT.
Both graphics cards are parceled in ASUS’ standard package, a huge box with a handle. The boxes are designed identically, the text with the model name being the single difference:
Besides that, the EN7900GS Top package has a sticker that tells you that this card’s speed is increased over the standard GeForce 7900 GS. ASUS claims a performance growth of 20% and we’ll check out this claim later on. The picture on the box shows a trooper from Ghost Recon Advanced Warfighter, this 3D shooter being included with the EN7950GT and EN7900GS Top.
Traditionally for ASUS products, the contents of the boxes are neatly put in separate compartments and the cards are firmly fixed in foam-rubber trays to avoid damage during transportation and storage. There is a flap on each box, yet there is still no window under it to give us a view of the card.
ASUS ships its EN7950GT and EN7900GS Top with the following accessories:
Strangely enough, ASUS saved on a second DVI-I → D-Sub adapter, although each card has two DVI connectors. This doesn’t quite agree with the packaging whose appearance and size implies that the EN7950GT and EN7900GS Top are to be regarded as top-end products. Well, LCD monitors are widespread nowadays, most of them featuring a DVI input which provides a higher image quality than the analog D-Sub. If the monitor doesn’t support DVI, the user can utilize the included adapter. It’s unlikely that the user would have two monitors with only D-Sub inputs, but anyway, this paltry economy on a cheap adapter looks strange on the part of such a respectable manufacturer as ASUS.
We have no complaints about the documentation enclosed with the card. The brief installation guide is written in 20 languages, and if you want a more detailed manual, you’ll find one on the included CD.
The disc with drivers contains the ASUS Enhanced driver that supports a number of ASUS’ exclusive technologies like GameFace Messenger, Splendid Video Enhancing Technology, and OnScreenDisplay (we described them in our ASUS EAX1900XTX review). It also contains additional software to support all those technologies. Unfortunately, the enhanced driver takes up about 140MB on your hard disk and is based on the out-dated version of ForceWare (91.47). The up-to-date version is numbered 93.71. If you don’t have any use for ASUS’ technologies, we guess it’s better to install the latest official ForceWare instead of ASUS’ special driver. You’ll have fewer performance and compatibility related issues then.
The CD case included with the graphics cards can store 16 discs. It is quite handy if you have to take so many discs with you on a trip. A full version of the 3D tactical shooter Tom Clancy’s Ghost Recon Advanced Warfighter is also included. This game is on our list of benchmarks, by the way. It is a rather demanding application, but the GeForce 7950 GT and GeForce 7900 GS are quite capable of delivering a good speed in 1280x1024, so this game is a proper choice.
Thus, the packaging and accessories of the EN7950GT and EN7900GS Top deserve our praises, yet we guess it’s no good that there is no second DVI-I → D-Sub adapter included. It cannot affect the cost of a top-end product much. We don’t think such paltry economy becomes the respectable ASUS.
Nvidia’s ban on developing and utilizing unique PCB designs for senior models of the GeForce series does not cover GeForce 7900 GT/GS, GeForce 7950 GT as well as less advanced GeForce 7 models. However, ASUS decided to stick to the reference design with its EN7950GT and EN7900GS Top.
ASUS EN7900GS Top
You can see it in the photographs that the graphics cards are both made by the reference design, the azure color of the PCB being the single difference. Each card is equipped with two DVI-I outputs and a universal 7-pin S-Video/YPbPr connector. Neither of them carries a VIVO chip. Thus, you can only tell the EN7950GT from the EN7900GS Top visually by the marking on the memory chips or by the stickers on the reverse side of the PCBs.
The ASUS EN7950GT carries 512MB of GDDR3 memory in eight HYB18H512321AF-14 chips from Infineon. These chips have a capacity of 512Mbit, 2.0V voltage and a rating frequency of 700 (1400) MHz. This is indeed the frequency the memory is clocked at by the card, providing a bandwidth of 44.8GB/s. Most of the GPU subunits are clocked at 550MHz and the vertex processors, at a 20MHz higher frequency. The frequencies are the same as those of the reference GeForce 7950 GT.
The ASUS EN7900GS Top uses Hynix HY5RS573225A FP-14 memory (256Mbit, 1.8V, 700 (1400) MHz). The chips are clocked at 720 (1440) MHz on the card, providing a bandwidth of 46GB/s instead of 44.8GB/s, yet this can hardly have a big effect on performance of the EN7900GS Top in games. It is the GPU frequency that ensures an increase in speed. It is 590MHz for the main GPU subunits and 610MHz for the vertex processors. This is far over the reference card’s 450/470MHz and may indeed provide the promised 20% performance boost. We’ll check this out soon.
The increased clock rates are the only point where the ASUS EN7900GS Top differs from the ordinary GeForce 7900 GS if you compare their specifications. They have the same amount of active GPU subunits: 20 pixel processors, 7 vertex processors, 20 TMUs, and 16 ROPs.
The EN7950GT and the EN7900GS Top are both equipped with the standard cooler we described in our reviews of Nvidia’s GeForce 7900 GT, 7900 GS and 7950 GT: a folded sheet of copper is glued to a small copper base. The cooler is equipped with a 45mm fan. The whole arrangement is covered with a plastic casing and secured on the PCB with four spring-loaded screws. Dark-gray thermal grease serves as a thermal interface. The memory chips are not cooled.
The reference cooler is popular due to the low heat dissipation of the G71 chip, but it may prove not that efficient when the G71 is clocked at 550MHz or higher. ASUS is confident the card won’t overheat, but we’d recommend you to make sure your system case is ventilated well before you try to install and use an EN7950GT and, especially, an EN7900GS Top.
We hadn’t expected much from the EN7950GT and EN7900GS Top as concerns overclocking, but the senior model did surprisingly well: we increased the GPU frequency from 550MHz to 660MHz and the card was perfectly stable (we even notched 675MHz at first, but had to roll back to 660MHz for the card would overheat and enter the throttling mode even while being cooled by an additional fan). The memory chips did well enough, too, speeding up from their default 700 (1400) MHz to 800 (1600) MHz. Thus, we reached the parameters of the GeForce 7900 GTX without “heavy” overclocking methods like volt-modding or liquid cooling.
Being already overclocked by the manufacturer, the EN7900GS Top performed modestly in that test, notching 620MHz and 820 (1640) MHz for the GPU and memory, respectively. The GPU frequency gain being very small, we decided not to benchmark the EN7900GS Top at the increased frequencies.
We want to remind you that the described graphics cards from ASUS are both equipped with the reference cooling system which is not too efficient. If you are going to overclock either of these cards, you should think about replacing the native cooler with something better. Otherwise the card may get damaged due to overheat.
During our comparative testing of the ASUS EN7950GT and EN7900GS Top we used the following hardware platforms:
The graphics card drivers were set up in such a way as to provide the highest possible quality of texture filtering.
We selected the highest possible graphics quality level in each game. We didn’t modify the games’ configuration files. Performance was measured with the games’ own tools or, if not available, manually with Fraps utility. We also measured the minimum speed of the cards where possible.
We tested the cards in three standard resolutions according to our testing methodology: 1280x1024, 1600x1200 and 1920x1200. We enabled FSAA and anisotropic filtering from the game’s menu. If this was not possible, we forced them using the appropriate driver settings of ATI Catalyst and Nvidia ForceWare. We didn’t run the tests with disabled FSAA, because the graphics accelerators reviewed today are powerful enough to ensure relatively high performance level with FSAA 4x.
We ran the tests with disabled FSAA only for those games that do not support FSAA due to the specifics of their engine or use HDR (FP16). The thing is that the GeForce 7 family cannot perform FSAA together with floating-point HDR.
Besides the ASUS EN7950GT and EN7900GS Top, we included the following graphics cards into this review:
We used the following games and benchmarks:
First-Person 3D Shooters
Third-Person 3D Shooters:
Having standard clock rates, the ASUS EN7950GT performs just like any other GeForce 7950 GT. Here, this is enough to compete with the Radeon X1950 Pro. When overclocked, the EN7950GT can challenge the Radeon X1900 XT.
As for the EN7900GS Top, it does provide a 20% performance boost. As a result, this graphics card, with fewer active GPU subunits, is almost as fast as the GeForce 7950 GT. That’s an achievement for a graphics card that had very moderate technical characteristics originally.
The EN7950 GT is as fast as the Radeon X1900 XT at the default frequencies and has no rivals at all at the increased clock rates. This must be due to the game running on a modified engine from Quake 3. It’s interesting that having a 20% advantage over the ordinary GeForce 7900 GS, the EN7900GS Top is also 20% behind the EN7950GT. It means that the number of TMUs is a crucial factor for this test and more important than the GPU clock rate as is indicated by the modest results of the Radeon X1950 Pro which has only 12 texture-mapping units.
The average performance of the benchmarked graphics cards is low in the 4x FSAA + 16x Aniso mode. The overclocked EN7950GT is the only one to provide a comfortable frame rate in 1280x1024.
The ASUS EN7900GS Top is 2-5fps slower than the EN7950GT throughout this test, their average frame rates ranging from 55 to 90fps. They see the back of the Radeon X1900 XT in front of them, but the overclocked EN7950GT is quite competitive against that Radeon, too, especially in resolutions of 1600x1200 and lower. The difference is 10% in favor of the AMD solution in 1920x1200.
The general picture is the same on the Research map, with minor deviations. The Radeon X1900 XT has a bigger lead in high resolutions due to the specifics of the map that puts a high load on the GPU’s computing subunits.
In the FP HDR mode there is almost no difference between the EN7950GT and EN7900GS Top in resolutions above 1280x1024. They must be limited by their memory subsystems.
This mode can’t be used for practical play in high resolutions, but both graphics cards from ASUS provide a high enough speed in 1280x1024.
The graphics cards from ASUS pass this test in the same way as the previous one. There is a negligible difference between them. From a practical point of view, there is no difference at all: both cards are equally suitable for playing F.E.A.R., but the EN7950GT is preferable if you are into overclocking. You can overclock it to the level of the Radeon X1900 XT (don’t forget to replace the native cooler with something more efficient).
The deferred rendering technique makes this game incompatible with FSAA, so there are only anisotropic filtering results here. We will only benchmark the cards in 1280x1024 resolution because of the high system requirements of the game.
There’s again a tiny difference between the EN7950GT and the EN7900GS Top. However, the overclocking of the latter card provides a smaller effect than in F.E.A.R.: you get an average performance growth of less than 10%. Both cards are slightly slower than the Radeon X1900 XT. The difference of 2fps isn’t crucial, though, and each of the three solutions can be considered suitable for playing this shooter.
The EN7950GT is getting farther away from the EN7900GS as the display resolution grows up despite the higher GPU frequency of the latter card. The average performance of the senior model is near 60fps in 1900x1200 whereas the junior model is lagging 10-12% behind.
The EN7900GS Top is for the first time faster than the senior model from ASUS, but not by much. The game engine probably doesn’t load all the pixel processors and TMUs, so the lack of them is easily made up for by the core frequency.
Both graphics cards have a considerable advantage over the Radeon X1950 Pro and are somewhat slower than the Radeon X1900 XT, not counting in the results of the overclocked EN7950GT. If you enable FSAA, the EN7900GS Top and EN7950GT provide a comfortable frame rate in 1280x1024.
The engine of this game, on the contrary, makes full use of all of GPU resources. Using shaders with multiple texture lookups, it is sensitive to the amount of TMUs. The game is also sensitive to the amount of graphics memory: graphics cards with 512MB of memory have much higher speeds than models with 256MB.
The large gap between the ASUS EN7950GT and the EN7900GS Top looks natural then. The former can be used to play the game with enabled FSAA at least in 1280x1024, but the latter cannot yield even 45fps on average. The cards from ASUS have similar minimum speeds, though.
The ASUS EN7950GT outperforms its junior brother beginning from 1600x1200 resolution, but the cards are both too slow for that resolution to be playable. Thus, the EN7950GT and EN7900GS Top can be viewed as equals when it comes to playing Hitman: Blood Money with enabled FSAA. The senior model is preferable if you are going to overclock your graphics card.
The competing graphics cards from ASUS deliver good performance in this test: the EN7900GS Top is as fast as the Radeon X1900 XT while the EN7950GT is somewhat faster than the AMD solution. The average frame rates are high enough for this game genre, but the speed fluctuates too wildly for comfortable play, occasionally plummeting to 4-10fps.
The current version of Gothic 3 doesn’t support FSAA, so we benchmarked the cards with anisotropic filtering only.
This game being a very demanding application, the ASUS cards both have modest results. However, the EN7950GT is preferable to the Radeon X1950 Pro as it has a higher minimum of speed. The Radeon X1900 XT is the only card that keeps the frame rate always above 25fps – the EN7950GT can’t do that even at the overclocked frequencies.
The game looks best with enabled HDR, so we benchmarked the cards in the HDR mode. Moreover, the Nvidia GeForce 7 series cannot support FP HDR along with full-screen antialiasing.
On installing the version 1.3 patch, the game runs faster, yet the best graphics card here, the overclocked ASUS EN7950GT, can’t notch even 30fps. The game seems to load heavily the graphics card’s texture-mapping units as is indicated by the extremely low results of the Radeon X1950 Pro and the relatively high results of the Radeon X1900 XT – they have 12 and 16 TMUs, respectively.
The Nvidia GeForce 7 architecture doesn’t allow to use FP HDR and full-screen antialiasing simultaneously, so we benchmarked the cards in TES IV: Oblivion using anisotropic filtering only. The game loses much of its visual appeal without HDR, although some gamers argue that point.
In closed environments ASUS’ products deliver high speeds in resolutions up to 1600x1200, and the overclocked EN7950GT is even somewhat faster than the Radeon X1900 XT, but a large part of the TES IV: Oblivion gameplay goes on outdoors where the requirements to the graphics subsystem are much harder. Let’s see what we have then.
Indeed, as soon as you go out, the Radeon X1900 XT with its 48 pixel processors breaks away from its opponents and the overclocking only improves the average, not minimum, speed of the EN7950GT. The average frame rate of the EN7950GT and EN7900GS Top is quite good for their class, and is even higher than that of the Radeon X1950 Pro, but their minimum speed is too low. It means you won’t avoid slowdowns and loss of control accuracy in graphically complex scenes.
Radeon X1000 series cards do not support vertex texturing, so they can’t use this technology to render the water surface with the highest possible quality in this game. This is only available on Nvidia GeForce 7 GPUs.
The EN7950GT and EN7900GS Top deliver similar frame rates, but the junior model provides a higher minimum of speed in 1600x1200. Both cards allow to play the game comfortably with enabled 4x FSAA and also give you the best-looking water surface.
It’s hard to choose between the Radeon X1950 Pro and the ASUS EN7950GT in X3:
When FSAA is enabled, 1280x1024 seems to be the highest resolution to play the game with comfort on the EN7950GT and EN7900GS Top. You can also try to play it in 1600x1200 as the speed is never lower than 30fps in that resolution.
The EN7950GT is far ahead of the EN7900GS Top. The biggest gap between them is in 1280x1024 and 1920x1200 resolutions, 15% and 20%, respectively. The difference is smaller in 1600x1200 for some reason, about 10%. Note also that the huge GPU frequency growth of the EN7900GS Top over the reference GeForce 7900 GS gives it a lead of only 8-10%. This may be due to high load on the TMUs and lack of graphics memory (256MB is too little for this game).
EN7900GS Top has not a single chance against EN7950GT, and even higher working frequencies do not help here. In 1280x1024 the lag is about 25%. Together with the low minimal fps rate it doesn’t give us any reason to hope for comfortable gaming experience. At the same time, the performance of EN7950GT is high enough for that, and overclocking provides additional speed reserves.
The ASUS EN7950GT scores 383 points more than the EN7900GS Top and 318 points less than the Radeon X1950 Pro. When overclocked, this graphics card scores over 10,000 points, yet cannot overtake the Radeon X1900 XT. Note also the impressive advantage of 1424 points the ASUS EN7900GS Top has over the reference GeForce 7900 GS.
It is in the first test, which demands a high fill rate from the graphics card, that the EN7950GT has the biggest lead over the EN7900GS Top. In the other tests the difference isn’t bigger than 10%. In the second test the overclocked EN7950GT cannot reach the level of the Radeon X1900 XT: the smaller number of TMUs doesn’t affect performance as much as in the other two tests whereas the numerous pixel processors of the AMD card help it cope better with math1ematics-heavy load.
As for the EN7900 GS Top, its performance is roughly that of the Radeon X1950 Pro: somewhat better in the first and third tests and worse in the second test. The outcome of the separate tests doesn’t quite agree with the overall scores because of the enabled FSAA. The benchmark doesn’t use FSAA by default.
The overclocked version of the GeForce 7900 GS offered by ASUS scores only 152 points less than the EN7950GT model. The cards both outperform the Radeon X1950 Pro (although they were both slower than it in 3DMark05).
Having 24 TMUs, the EN7950GT is competitive against the Radeon X1900 XT, scoring only 70 points less. The EN7900GS Top has another 69 points less. That’s a good result for an overclocked version of the GeForce 7900 GS, but what will we see in the SM3.0/HDR tests?
The EN7900GS Top is still a little slower than the EN7950 but is closer to the Radeon X1950 Pro – the difference is 87 points now. The Radeon X1900 XT leaves the others behind – its 48 pixel processors and technologies to accelerate the rendering of shadows with the CSM method come in handy in this test.
In the first SM2.0 test the EN7900GS Top occupies a position between the standard GeForce 7900 GS and the EN7950GT, outperforming the Radeon X1950 Pro a little. But in the second test the ASUS card is slower than both the senior model from ASUS and the AMD solution. Like in 3DMark05, the results of separate tests with enabled FSAA don’t agree well with the results the benchmark produces at its default settings.
Having tested ASUS’ EN7900GS Top and EN7950GT graphics cards, we can state that these two products are not equal to each other.
The EN7900GS Top indeed ensures an average performance growth of nearly 20% over the standard GeForce 7900 GS, yet this is not enough in most cases to reach the level of the GeForce 7950 GT. In some tests the EN7900GS Top is much slower which can be explained by its having half the amount of graphics memory as well as fewer active GPU subunits than in the EN7950GT. This can be seen in Call of Duty 2, Serious Sam 2 and Age of Empires 3: The War Chiefs. Thus, the GeForce 7900 GS can’t make a good replacement to the GeForce 7950 GT even if well overclocked. But for the Radeon X1950 Pro the ASUS EN7900GS Top is a dangerous opponent as it is considerably faster across a number of games.
As for the ASUS EN7950GT, it is an ordinary GeForce 7950 GT with all the consequences. It is competitive against the Radeon X1900 XT in some cases, yet the comparison is not quite correct due to the difference in the amount of graphics memory, and is at least as fast as the Radeon X1950 Pro (and even outperforms it greatly in such games as Neverwinter Nights 2, Call of Duty 2, Half-Life 2: Episode One, and some others). The EN7950GT has a high overclocking potential. We overclocked our sample of the card to the level of the GeForce 7900 GTX, although we aren’t sure all samples can conquer such frequencies.
The two products from ASUS we’ve described in this review lack originality. They are both based on the reference design and differ from the reference cards in the color of the PCB only. Both cards come with reference coolers and with not too many accessories (but the interesting game Tom Clancy’s Ghost Recon Advanced Warfighter is included with them). The EN7900GS Top may be a good choice if you don’t want to overclock your graphics card and if its price is considerably lower than that of the EN7950GT. Otherwise you should buy the EN7950GT – it is going to satisfy you for sure.
ASUS EN7900GS Top