by Alexey Stepin , Yaroslav Lyssenko
07/10/2008 | 02:51 PM
After a long period of confusion in its product nomenclature Nvidia has finally acknowledged that it cannot go on like this anymore. However, the GeForce 8800 GTS 512MB still occupies the place between the GeForce 9800 GTX and GeForce 9600 GT whereas the GeForce 9800 GTS seems to have been scrapped.
As we found out in our tests, the GeForce 9800 GTX delivers but barely higher 3D performance than the GeForce 8800 GTS 512MB but comes at a considerably higher price. Coupled with its other drawbacks such as the long 27-centimeter PCB and two power connectors, the market standing of the 9800 GTX doesn’t seem firm. Those 5-8 percent of advantage in speed over the GeForce 8800 GTS 512MB is surely not enough to make the GeForce 9800 GTX really attractive. So how can this product win the customer considering Nvidia’s prohibition to develop unique PCB designs for GeForce 9800 GTX?
There are indeed but a few ways a graphics card vendor can try. First of all, it is factory overclocking. It can help increase the performance gap between the GeForce 8800 GTS 512MB and GeForce 9800 GTX to such an extent as to justify the price difference between them. Nvidia permitting, many of its partners practice such overclocking. Zotac is among them. A part of PC Partner Group, this brand offers not only traditional solutions, one of which was reviewed by us earlier, but also provides a whole series of enthusiast-targeted pre-overclocked graphics cards under the common name of AMP! Edition. We will discuss the Zotac GeForce 9800 GTX AMP! today. The card may be the fastest GeForce 9800 GTX out there, so let’s check it out.
The graphics card is packed into a standard-size box colored warm yellow and orange hues typical of the brand.
The color scheme is nice but the picture on the face side of the box is not very friendly looking. Perhaps some people will find the predatory teeth-showing lizard quite an appealing symbol, though. As usual, the box shows the basic technical specs of the product save for the GPU and memory clock rates but the pre-overclocked frequencies of the card are indicated by the Overclocked and AMP! Edition captions.
The main cardboard box under the colorful wrapper contains pieces of foam rubber with cutouts for the card and accessories. The contents are thus protected against any hazards during storage and transportation. The accessories include the following:
There is nothing extraordinary about the accessories but you do get everything necessary to use the card in a high-performance gaming system or in a home multimedia center. The kit doesn’t include software for playing HD video formats, though. The lack of such software can be pardoned in cheaper solutions but the user can expect to find it included with a top-class product such as GeForce 9800 GTX.
There is a copy of Lost: The Video Game in the box (the game is also known as Lost: Via Domus). The game is based on a popular TV series and has already become a visiting card of Zotac products. It doesn’t boast exceptional visuals or an outstanding plot, though. It is just a nice bonus.
On the whole, the packaging and accessories of the Zotac GeForce 9800 GTX AMP! are good although the lack of a HD video player is a problem. It’s good the kit contains all the adapters necessary to establish a HDMI connection for both video and audio. It means you can connect your system to a large LCD panel with just a single cable.
Every GeForce 9800 GTX selling today uses the reference PCB design. They are all manufactured on contracted facilities and sent ready-made to Nvidia’s partners. Modifications are not yet allowed but the GeForce 9800 GTX is far simpler than the GeForce 8800 GTX/Ultra and we have no doubt this graphics card will appear with nonstandard PCB designs sooner or later. The Zotac card is a copy of the reference sample, differing from it with the stickers on the cooler’s casing and fan.
Despite its 256-bit memory bus that simplifies PCB wiring, the GeForce 9800 GTX is as long as 27 centimeters or as the GeForce 8800 GTX. Consequently, this card is incompatible with a number of system cases, not only with compact barebone or microATX but also with ATX cases that have a shortened design or a specific position of the HDD cage.
The power circuit is quite complex for a G92-based solution. The four-phase GPU voltage regulator is based on a Primarion PX3544 controller. The two-phase memory voltage regulator is based on an Anpec APW7066. The card has two PCI Express 1.0 plugs for connecting external power although this is not really necessary: the combined load on these connectors is but slightly above 60W, which is far below the load capacity of one connector (75W). However, you have to use both connectors for the card to start up and work. Zotac included appropriate adapters for people whose PSU has only one 6-pin power connector or no such connectors at all.
There are eight memory chips on board (Samsung K4J52324QE-BJ08, 512Mb, 16MB x 32). These are the fastest GDDR3 chips available today. They have a rated frequency of 1200 (2400) MHz at 1.9V voltage. The memory frequency is increased over that of the reference GeForce 9800 GTX: 1152 (2304) MHz against 1100 (2200) MHz. The memory bandwidth is increased from 70.4MBps to 73.6GBps as the consequence. It is not as high as the memory bandwidth of the GeForce 8800 GTX (86.4GBps), let alone GeForce 8800 Ultra (103.7GBps) but the G92’s memory controller is more efficient. It is the amount of graphics memory rather than its bandwidth that may sooner become a bottleneck. Unfortunately, the GeForce 9800 GTX carries only 512 megabytes of memory on board and this may be not enough for some applications especially as Nvidia’s current-generation GPUs are known to suffer from inefficient memory management. So, if you own a GeForce 8800 Ultra and play games at high resolutions, you should not hurry to replace your card with a GeForce 9800 GTX, even pre-overclocked.
The GPU installed on the Zotac GeForce 9800 GTX AMP! is marked as G92-420-A2 and has a standard configuration with 128 unified stream processors, 32 (64) TMUs, and 16 raster operators grouped into four sections with four 64-bit memory controllers. The main GPU domain is clocked at 756MHz while the shader domain is clocked at 1890MHz. Being far higher than the reference ones (675MHz and 1688MHz respectively), these frequencies should ensure a larger performance gap from the GeForce 8800 GTS 512MB.
As opposed to the GeForce 8800 GTS 512MB, the Zotac is equipped with two MIO connectors to support Triple-SLI configurations. Besides, the card offers two dual-link DVI-I ports (with support of HDMI/HDCP and resolutions up to 2560x1600 inclusive) and a standard 7-pin mini-DIN connector for analog video output in Composite, S-Video and YPbPr formats. The audio-over-HDMI feature is supported by connecting the graphics card to an internal S/PDIF output of the sound card using the included cable.
The card is cooled by a standard reference cooler from Nvidia whose origin goes back to the GeForce 8800 GTX. The cooler consists of an aluminum frame on which a large composite heatsink, made up of thin aluminum plates, is installed. Heat pipes connect the heatsink with the copper heat-spreader that contacts with the GPU die. The heat pipes are placed in such a way as to distribute the heat uniformly in the heatsink. There is a layer of dark-gray thermal paste between the cooler and the GPU. The memory chips and power elements transfer the heat to the aluminum frame via fabric pads socked in white thermal grease.
A Delta BFB1012L blower is installed at the back of the cooler. There are slits in the card’s mounting bracket for the hot air to be exhausted out of the system case. The whole arrangement is covered with a black plastic casing. The cooler is secured on the PCB with 14 spring-loaded screws and an X-shaped bracket. The casing is additionally fastened with three screws to prevent any misalignment.
Practice suggests that this cooling system design is optimal, combining high cooling performance with little noise. The cooler is going to cope easily even with the pre-overclocked version of GeForce 9800 GTX provided by Zotac.
Based on the reference design, the Zotac GeForce 9800 GTX AMP! is going to consume about the same amount of power as the reference GeForce 9800 GTX but we wanted to check out how the factory overclocking affected the card’s consumption. We performed our power consumption test on a special testbed configured like follows:
The 3D load was created by running the first SM3.0/HDR test from 3DMark06 in a loop at 1600x1200 with 4x FSAA and 16x AF. The Peak 2D load is emulated by means of the 2D Transparent Windows test from PCMark05. Here are the results:
Click to enlarge
The increased frequencies affect the card’s power draw. The peak power consumption of Zotac’s GeForce 9800 GTX is 5.2W higher in comparison with the reference card. That’s not much compared with the flagship cards of the previous generation, though. 113W is a modest value in comparison with 130W of the GeForce 8800 GTX, let alone 160W of the Radeon HD 2900 XT. As for the load distribution, the two power connectors are redundant even for the Zotac card: one such connector could easily sustain the load of 65W.
We measured the card’s noise 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. The card proved to be as noisy as the reference GeForce 9800 GTX:
So, the card is almost silent. You can’t hear it under normal conditions. Working on an open testbed, the GPU temperature was 66°C under load. This is 5°C higher than the temperature of the reference card but there is no danger of overheat. The cooler does its job well notwithstanding the factory overclocking. You should note that the GPU temperature depends on the room temperature and may differ in your particular conditions.
We hadn’t expected much from overclocking the card. Indeed, the GPU was stable at 771/1944MHz and the memory chips, at 1220 (2440) MHz. These frequencies are just slightly higher than the card’s default ones, so we didn’t benchmark the overclocked card.
For our performance tests of Zotac GeForce 9800 GTX AMP! we put together the following testbed:
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 synthetic 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.
Besides Zotac GeForce 9800 GTX AMP! we have also included the following graphics accelerators to participate in our test session:
The tests were performed in the following 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.
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.
The Zotac does not benefit much from the pre-overclocked frequencies, being about as fast as the ordinary GeForce 9800 GTX. The biggest difference between them is only 1.9fps, which is negligible compared with the average frame rates of 80-90fps. The ATI Radeon HD 3870 X2 is no worse for practical purposes but its minimum speed is lower and it consumes considerably more power.
BioShock doesn’t support FSAA when running in Windows Vista’s DirectX 10 environment. We benchmark graphics cards without FSAA in this game.
Both versions of GeForce 9800 GTX reach the performance ceiling set by the other testbed components, particularly by the CPU, at 1280x1024. The Zotac goes ahead of the reference card as the resolution grows up. The gap amounts to 7% at 1920x1200 and the Zotac is formally faster than the Radeon HD 3870 X2. The latter shows a higher minimum of speed, though.
The in-game benchmarking tool does not support 2560x1600 resolution so we had to limit our test to 1920x1200.
The Zotac card can’t compete with the dual-processor Radeon HD 3870 X2. The latter can yield an average frame rate of over 25fps as none of modern single-processor cards can. The overclocking effect is small in this game, hardly above 1%.
The factory overclocking has a stronger effect in this game. It adds 8% to the performance of the reference GeForce 9800 GTX at 1280x1024. With this addition, the Zotac is 26-27% faster than the GeForce 8800 GTS 512MB. Now it looks like a flagship G92-based solution indeed.
The performance increase is smaller at the higher resolutions, though. It is also insubstantial in absolute numbers to affect your gaming experience. The GeForce 8800 GTS 512MB provides the same level of comfort at 1920x1200 as the Zotac card does.
This game is tested at the High level of detail, excepting the Shaders option which is set at Very High. This way we try to achieve a compromise between image quality and speed.
The Zotac GeForce 9800 GTX AMP! is confidently in the lead but the gap from the reference card is a mere 4-7% depending on the display resolution. The average speed is good at 1280x1024 but the slowdowns to 17-18fps – even at the reduced game settings – are disappointing.
The much lower minimum speed of the GeForce 8800 series is due to the older version of the ForceWare driver. It is only recently that Nvidia has abandoned its practice of separating the GeForce 8 and GeForce 9 series on the driver level.
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 overclocking effect shows up in the classic way here: the pre-overclocked GeForce 9800 GTX is getting farther away as the resolution grows up. The gap is as large as 15% at 1920x1200. The average frame rate being over 45fps and the game having a frame rate limiter, this advantage only means a larger reserve of speed the gamer is unlikely to ever require.
The ATI Radeon HD 3870 X2 used to be an unchallenged leader in this test but the Zotac looks competitive against it. Your choice should depend on your personal preferences and the pricing of the two products because both deliver highest playing comfort in every display mode.
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 Zotac beats the Radeon HD 3870 X2 but the latter has a higher minimum of speed at 1280x1024. This victory has little practical value because the additional speed doesn’t affect the gamer’s experience.
None of modern single-chip graphics cards can deliver comfortable performance in Lost Planet at the highest graphics quality settings. The Zotac card is not an exception although it is formally the winner of the test. The Radeon HD 3870 X2 is slow because CrossFire technology is supported poorly for this game.
Like in many other tests, the factory overclocking adds 7-9% to the Zotac’s speed in comparison with the reference card. Thus, the Zotac is the fastest of the G92-based solutions but the others, including the GeForce 8800 GT 512MB, deliver a comfortable frame rate, too.
It is only at a resolution of 1600x1200 that the GeForce 9800 GTX AMP! benefits from having higher clock rates. Its advantage over the reference card amounts to 11% then. In the other modes there is almost no performance gain. The resolution of 1280x1024 remains the optimal one for graphics cards with 512 megabytes of memory because their performance can bottom out below 25fps at higher display modes.
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.
Overclocking the GeForce 9800 GTX can be rewarding in TES IV. The performance gain is up to 15% but the result is not really worth the increased power consumption and heat dissipation since the top-end cards are anyway fast enough here. Still, the Zotac GeForce 9800 GTX AMP! secures another win for itself.
The new add-on to Company of Heroes is tested in DirectX 10 mode only since it provides the highest quality of the visuals.
The average frame rate is increased while the minimum speed is as low as that of the reference card. As a result, the ATI Radeon HD 3870 X2 remains the only graphics card to ensure a really high level of gamer’s comfort in Opposing Fronts.
The expansion pack to C&C 3: Tiberium Wars having a frame rate limiter, you should consider the minimum speed of the cards in the first place.
The tested graphics cards are all equal in the gamer’s eyes due to the modest requirements of the game and the frame rate limit.
We can’t say that the overclocking of the GeForce 9800 GTX is rewarding in this game. The minimum speed is too low. The average performance gain is 6-10%, being the highest at the resolution of 1600x1200.
There is no advantage from the increased GPU and memory frequencies of the GeForce 9800 GTX card when it comes to the overall scores. The difference between the two versions of this graphics card is a mere 73 points, which is within the measurement error range.
It’s different in the individual tests: the Zotac doesn’t show much in the first test, but the overclocked frequencies give it an edge over the reference card starting with 1600x1200 resolution. The advantage is not high, amounting to 6-8%, which is not enough to compete with the Radeon HD 3870 X2 that wins two out of the three 3DMark05 tests by a large margin.
The ATI Radeon HD 3870 X2 wins even more confidently than in 3DMark05. And like in 3DMark05, the pre-overclocked version of GeForce 9800 GTX from Zotac is but slightly better than the reference card.
There are no speed benefits in the groups of tests, either. The Radeon HD 3870 X2 is not far ahead of Nvidia’s solutions in the first group, but leaves them behind in the second group that uses Shader Model 3.0 and HDR lighting.
The individual tests that we run with 4x FSAA show an effect from the overclocking of GeForce 9800 GTX. The performance gain is about 7%, which agrees with what we’ve seen in the gaming tests. The ATI Radeon HD 3870 X2 is inferior to Nvidia’s flagship single-chip solutions in the first test that depends on the texture processor performance, but leaves no chance to its opponents in the second test.
The Zotac card has a similar performance gain in the first SM3.0/HDR test. In the second test, the gap is only 4%. The Radeon HD 3870 X2 is far in the lead in both tests.
The Zotac GeForce 9800 GTX AMP! is indeed the fastest single-core graphics card based on the G92 processor. The pre-overclocked frequencies ensure this card an average advantage of 7-8% over the reference sample and increase the gap from the GeForce 8800 GTS 512MB.
On the other hand, the practical benefits are unclear. The Zotac doesn’t open up new opportunities for gamers in any of the games we use as benchmarks. If it delivered comfortable performance, the GeForce 8800 GTS 512MB could deliver comfortable performance, too. If the latter was too slow, the Zotac was too slow as well. This might have been expected because the only difference between these two cards is the clock rates of the GPU and memory but something more is needed for a breakthrough as the Radeon HD 3870 X2 showed in such games as Call of Juarez and Company of Heroes: Opposing Fronts.
Anyway, the Zotac GeForce 9800 GTX AMP! is a good product. It is considerably faster than the GeForce 8800 GTS 512MB, delivers highest performance in today’s games, is free from the drawbacks typical of multi-GPU solutions, and has good accessories. Our only gripe is about the lack of software for HD video playback in the kit. We can put up with that in cheap products but not in top-class ones like the Zotac.
The Zotac card is only going to be truly appealing if its retail price is lower than $230, which is the price of the GeForce 9800 GTX+. Otherwise, we’d recommend you to consider either ATI Radeon HD 4850 or Nvidia GeForce 8800 GTS 512. This is true for any other version of GeForce 9800 GTX, though.