by Alexey Stepin
06/01/2006 | 12:24 AM
June 22, 2005, Nvidia announced its landmark G70 graphics processor and claimed technological superiority over the previous leader ATI Technologies. ATI’s R520 made a tardy appearance on October 5 but didn’t put the company back in the lead being slower in some tests than the GeForce 7800 GTX.
ATI found itself with a large store of previous-generation chips (R423, R430 and R480) which had to be disposed of in some way or another. Otherwise the company would face severe financial loss. ATI solved the problem in an elegant, even though not very profitable, way by announcing performance-mainstream and mainstream graphics cards called Radeon X800 GTO and Radeon X800 GT, respectively. These cards were in fact the same as Radeon X850 XT but with fewer pipelines and lower clock rates, and they suited their purpose well, especially since there was some delay with shipments of Radeon X1600 XT. Radeon X800 GT and GTO are still quite appealing in terms of price, but their performance is rather too low by today’s standards.
But let’s get back to the R520. As we said, this chip didn’t become an unrivalled leader despite the number of progressive architectural innovations it embodied in silicon. The performance of the best solution on that GPU, Radeon X1800 XT 512MB, was overall higher than that of the GeForce 7800 GTX, but the Nvidia GeForce 7800 GTX 512 released November 2005 proved to be faster still (for details see our article called NVIDIA GeForce 7800 GTX 512: Faster, Higher, Stronger!). And although Nvidia couldn’t provide that card in mass quantities, ATI Technologies should have felt very disturbed at its lost technological superiority.
Well, you surely know that how this story ends: ATI hastened the development of the R580 chip and introduced it to the public on January 24, 2006 (for details see our article called The Fast and Furious: ATI Radeon X1900 XTX Review). The chip featured more general-purpose registers and three times the number of pixel processors compared with the R520, yet had the same number of texture-mapping units. This approach to GPU design might seem arguable, but the R580-based Radeon X1900 XTX and X1900 XT cards delivered excellent performance in games, leaving the GeForce 7800 GTX 512 behind in most cases. But what happened to the R520?
Not the luckiest claim for technological superiority, the R520 was found inadequate for the top-end market sector once R580-based graphics cards began to sell in mass quantities which they did right after the announcement of the new GPU. So, it was just like with ATI’s previous-generation chips and the company played the same trick again to put the already manufactured R520 chips to good use.
Some R520 chips are still being installed on Radeon X1800 XL while the rest of them are coming to us on the new mainstream graphics card which was announced at the same time with Nvidia’s mainstream GeForce 7600 GT. The ATI Radeon X1800 GTO card claims to be the best offer in a price range of $199-249. Let’s see what’s behind this claim.
The Radeon X1800 GTO must be compared with its immediate market rival GeForce 7600 GT.
Obviously, the Radeon X1800 GTO is made by disabling four pixel processors out of the R520’s sixteen. This approach to making mainstream graphics card can be traced back to Radeon X800 GT/GTO or even earlier to Radeon 9500/9800 SE. It means there may be an opportunity to transform an inexpensive Radeon X1800 GTO into a full-featured Radeon X1800 XL by turning on the disabled processors, but this is beyond the subject of this review (ATI may also be using R520 chips with defective subunits to make its Radeon X1800 GTO – the flexible modular architecture of desktop Radeon X1800/Radeon X1900 chips allows doing that).
The Radeon X1800 GTO looks technologically superior to the GeForce 7600 GT if you compare their specs. The cards have 12 pixel processors each, and Nvidia’s solution has a higher GPU clock rate, but it is equipped with a 128-bit memory bus whereas the Radeon X1800 GTO accesses its memory across a 256-bit bus. As a result, ATI’s solution enjoys a nearly 10GB/s advantage in memory bandwidth over its competitor and also features a more efficient ring-bus memory controller. Coupled with its 12 raster operators (ROPs), this should tell positively on the Radeon X1800 GTO’s performance in high resolutions with enabled full-screen antialiasing. We’ll check this out in our tests. Right now let’s take a closer look at the card’s design. The new mainstream graphics solution from ATI is represented by a PowerColor X1800 GTO graphics card.
The card is parceled into a normal-size box with a thin cardboard wrapping.
The package is designed alike to the PowerColor X1900 XT’s (for details see our PowerColor X1900 XT Graphics Card Review), with gray as the predominant color. A sketchy picture of the graphics card is on the box with its name in a large clear print and surrounded with numerous logotypes. Funnily, the card on the box has a cooler of the older type. The cooler actually installed on the Radeon X1800 GTO is different as you will see below.
A sticker on the box reports that this device has two DVI-I interfaces and 256 megabytes of memory with 256-bit access. There’s more detailed info about the card on the box’s side. The package looks neat and elegant and doesn’t annoy your eye with bright colors or stereotyped monsters.
Besides the graphics card, the box contains the following:
The accessories are modest, quite typically for PowerColor’s products. There are no specials here, yet you get everything necessary to use the card. Like with the PowerColor X1900 XT, we are somewhat disappointed at finding no games among the accessories. A good game in the software pack would make the PowerColor X1800 GTO definitely more appealing.
All Radeon X1800 GTO graphics cards are manufactured by the same design developed at ATI Technologies, so the following section refers not only to this PowerColor product but also to X1800 GTO graphics cards selling under other trademarks.
Developing a new PCB with a completely new wiring layout for the Radeon X1800 GTO wouldn’t make any sense, so it’s no wonder the card resembles the Radeon X1800 XL.
The cooler is designed in a different way and there’s no engineering connector in the top left of the PCB, but otherwise the Radeon X1800 GTO is exactly like the Radeon X1800 XL. They even have the same number of induction coils near the heatsink that covers the voltage regulator’s MOSFETs.
This PCB design isn’t without drawbacks, though. The PCB employed for Radeon X1800 is rather long, so you may have troubles trying to install your Radeon X1800 GTO into a small or barebone system case. Moreover, the Radeon X1800 GTO requires additional power while the GeForce 7600 GT doesn’t. The power consumption of the new graphics card from ATI isn’t going to be very high, but the 6-pin power connector is necessary since the power circuit hasn’t been modified to fully rely on the PCI Express x16 slot which can provide up to 75 watts of power (you’ll see below how much power the Radeon X1800 GTO needs in comparison with the Radeon X1800 XL).
There’s an ordinary R520 chip here which is the same as are installed on Radeon X1800 XL and Radeon X1800 XT. The resistors on the die package are located in the same manner on all Radeon X1800 chips, so it seems to be impossible to turn on the disabled hardware subunits by re-soldering. The subunits are probably disabled by means of melting the internal fuses, so BIOS updates are unlikely to have any effect. Still, there is a chance we can see modification-friendly cards, something like Radeon X1800 GTO2. The core clock rate of this card is 500MHz.
The card carries eight GDDR3 memory chips (Samsung K4J55323QG-BC20) in 136-pin FBGA packaging. Each chip has a capacity of 256Mbit, so the total of graphics memory is 256 megabytes. The chips are clocked at 500 (1000) MHz and work at 1.8V voltage. For comparison, the Radeon X1800 XL comes with faster, BC14 chips that are capable of working at frequencies up to 700 (1400) MHz. The use of slower chips with an access time of 2.0 nanoseconds helped cut the production cost of Radeon X1800 GTO to come extent, but also reduced its overclocking potential.
The card features a VIVO-supporting Rage Theater chip. Capturing analog video isn’t a very useful feature for today because such video mostly remained in the past, yet some users may find it helpful.
The cooler of the Radeon X1800 GTO differs somewhat from the one installed on the Radeon X1800 XL, but its concept (see our article called ATI All-In-Wonder X1800 XL: All I Want? for details) has remained the same. The graphics processor gives its heat to the cooler’s copper sole which is then evenly distributed in the ribbing by means of two U-shaped heat pipes. The ribbing is blown at by a small fan. The air stream then also cools the heatsink on the power regulator’s MOSFETs and leaves the card. The cooler is covered with a flat aluminum cap.
The cooler on the Radeon X1800 XL had a 4.2W ADDA fan (0.35A, 12V) which was criticized for being noisy. The new cooler has got a larger and quieter 3W fan (0.25A, 12V) which uses a 3-pin connection with a separate line for the speed sensor. The speed control system is the same as in the entire Radeon X1800/X1900 family. You can also control the speed of the fan manually by means of RivaTuner.
As usual, there’s a layer of very thick dark-gray thermal paste between the cooler and the GPU die. The memory chips give their heat to the aluminum casing of the cooler via thick rubber-like thermal pads.
The new cooler installed on the Radeon X1800 GTO will probably be slightly more efficient and less noisy than the one they used on Radeon X1800 XL. We don’t expect any miracles, though, because the heatsink has remained almost the same plus this system doesn’t exhaust the hot air out of the system case.
We measured the power consumption of the Radeon X1800 GTO on the following testbed:
We used a digital multimeter Velleman DVM850BL (0.5% measurement accuracy). To put a Peak 3D load on the card we ran the first SM 3.0 graphics test from 3DMark06 in a loop at 1600x1200 resolution with 16x anisotropic filtering. Then we created an extremely high 2D load by launching the 2D Transparent Windows test from Futuremark PCMark05. Here are the results:
The Radeon X1800 GTO consumes much less than the Radeon X1800 XL in which all the 16 pixel processors, TMUs and ROPs are enabled and working, but doesn’t reach the level of the simpler GeForce 7600 GT. After all, the R520 is a complex chip which was originally intended for top-performance solutions and its 48 watts is a good result. The card puts equal load on its internal and external +12V channels in 3D mode – each channel provides about 21W of power and the rest is consumed from the +3.3V rail. A high-wattage power supply isn’t required. You shouldn’t be afraid to install your Radeon X1800 GTO into barebone systems which usually come with low-wattage PSUs.
Next we tried to overclock the card and were quite successful at that. Using the stock cooler we managed to increase the core frequency of our Radeon X1800 GTO to 600MHz. The graphics memory was overclocked to 575 (1150) MHz which was an achievement for 2.0ns chips. At 600 (1200) MHz memory frequency the card would freeze after a while.
So, the Radeon X1800 GTO seems to be a good choice for an overclocker even if you don’t succeed in unblocking the disabled GPU subunits. You can also get better overclocking results than ours if you are into water or cryogen cooling and volt-modding, but there’s a higher risk of damaging the card through such experiments, too.
The fan speed control system of the Radeon X1800 GTO works exactly like on any other Radeon X1800/X1900 card. The fan rotates at its max speed after you turn the computer on, but then the speed is lowered to nearly silent level. After that the fan may speed up if the GPU temperature exceeds a certain value after your having worked in 3D applications for a while. The Radeon X1800 GTO becomes audible then, but fortunately it doesn’t produce the irritating high-frequency noise the Radeon X1800 XL used to emit. So, the noise characteristics of the Radeon X1800 GTO are acceptable, even though the card is not actually noiseless. You can also adjust the fan speed manually by means of RivaTuner. This method of reducing the noise should be quite safe if you don’t drop the speed unreasonably low.
The Radeon X1800 GTO doesn’t differ from the Radeon X1800 XL in design, so it delivers the same 2D image quality. We saw a sharp picture without fuzziness or shadowing in all display modes supported by our Dell P1130 and P1110 monitors, including 1800x1440.
We tested PowerColor Radeon X1800 GTO 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.
Since mainstream graphics cards do not always guarantee acceptable performance in contemporary games with enabled FSAA, we ran the tests not only with FSAA 4x + AF 16x, but also with FSAA disabled and only the highest level of anisotropic filtering activated.
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.
Besides Radeon X1800 GTO, we have also included the following graphics cards into our test session:
These games and applications were used as benchmarks:
First-Person 3D Shooters
Third-Person 3D Shooters
With anisotropic filtering alone or with FSAA plus anisotropic filtering, the PowerColor X1800 GTO is as fast as the GeForce 7600 GT, being a little ahead or behind it depending on the resolution. The cards both provide a comfortable average and minimum frame rate in all resolutions of the first mode and in resolutions up to 1280x1024 in the FSAA mode.
This game is tested using the Fraps utility, so there may be some inaccuracy in the results.
This OpenGL application makes use of UltraShadow II technology which accelerates stencil shadows rendering. The GeForce 7600 GT wins the “pure speed” mode, leaving the PowerColor X1800 GTO up to 50% behind.
But as soon as we add full-screen antialiasing to anisotropic filtering, the ATI solution is immediately as close as 15% in 1024x768 thanks to its much faster memory subsystem and 12 ROPs. There’s almost no gap between these solutions in higher resolutions, but the average performance of the PowerColor X1800 GTO and GeForce 7600 GT is rather too low there for comfortable play.
The PowerColor X1800 GTO is less than 5fps behind the GeForce 7600 GT in 1024x768. Unfortunately, graphics cards of this class cannot ensure a playable frame rate in this game in higher resolutions – you can only have it with more advanced devices like Radeon X1900 XT or GeForce 7900 GT.
Both the mentioned mainstream solutions yield an average frame rate of less than 50fps in 1024x768 in the “eye candy” mode. And then a strange thing happens: notwithstanding its 256-bit memory bus, the ATI solution is suddenly behind the GeForce 7600 GT! Moreover, the PowerColor X1800 GTO is exactly as fast as the Radeon X1800 XL in 1600x1200 although the two have a different number of execution units, TMUs and ROPs. So it’s not about the GPU, but probably about how the graphics memory is managed by the Catalyst driver.
The engine of Doom 3 uses the OpenGL API and is optimized for some special features of the GeForce 6/7 architecture. This is why the cards rank up in this test just as they did in The Chronicles of Riddick. The PowerColor X1800 GTO is slower than the GeForce 7600 GT in the “pure speed” mode and doesn’t give you enough speed even in 1280x1024 whereas the GeForce makes the resolution of 1600x1200 playable.
It’s different in the “eye candy” mode: the ATI solution easily overtakes the GeForce 7600 GT but 1024x768 is the only resolution where these cards yield more than 55fps.
High pixel shader performance is important for the Pier demo in which there’s a lot of water rendered by means of such shaders. Performance of graphics cards faster than the Radeon X1800 XL is usually limited in this scene by the speed of the central processor, but the CPU doesn’t interfere in the competition between the PowerColor X1800 GTO and the GeForce 7600 GT: the ATI solution is in the lead beginning from 1024x768 resolution.
The advantage of the PowerColor X1800 GTO over its opponent is bigger when we turn on 4x FSAA – it amounts to 20% in 1600x1200. This is enough for the card to provide a comfortable average speed of 55fps and higher. The GeForce 7600 GT with its 128-bit memory and 8 ROPs cannot contend here.
The requirements to the graphics subsystem are different on the Research map: it’s important to have SM3.0 support to render the lighting of the scene in one pass. Nvidia’s cards have always been better at that due to the specifics of the game unless you turn full-screen antialiasing on. Here, the PowerColor X1800 GTO is 10-17% slower than the GeForce 7600 GT depending on the resolution, yet it anyway provides an average frame rate of at least 70fps.
ATI’s solution easily overtakes its opponent after we turn on 4x FSAA. Both the cards deliver a good performance because Far Cry isn’t a very demanding application, although it is as beautiful as the most recent first-person 3D shooters are and even better!
HDR support for Radeon X1000 is implemented poorly in the current version of Far Cry . This is why the PowerColor X1800 GTO cannot contend with the GeForce 7600 GT here.
This is especially clear on the Research map where high pixel shader performance is important while the advantages of the 256-bit memory bus aren’t called for. Anyway, the resolution of 1024x768 is available on the PowerColor X1800 GTO in both cases.
F.E.A.R. demands that the graphics subsystem be capable of fast processing of pixel shaders this game abounds in. The PowerColor X1800 GTO has a lower pure speed than the GeForce 7600 GT, probably because of the higher core clock rate of the latter card. The difference in both minimum and average performance is 10%. Both these cards provide a comfortable frame rate in 1280x1024, yet you may want to limit yourself to 1024x768 to have a higher minimum speed.
The cards change places in the “eye candy” mode, but their minimum performance is only 25-26fps in 1024x768. It means you can play the game, but may come across such scenes where the performance goes down below comfortable level. This is why we don’t recommend turning on FSAA on these cards in F.E.A.R. if you use the highest graphics quality settings.
It’s characteristic of Half-Life 2 that the speed ceiling for ATI’s solutions goes higher than for Nvidia’s as you can see in 1024x768. In higher resolutions of the “pure speed” mode the average performance of the PowerColor X1800 GTO and GeForce 7600 GT is similar and over 70fps in 1600x1200. This is more than enough for normal play.
The game uses high-resolution textures, so the 256-bit memory bus of the PowerColor X1800 GTO comes in handy in the “eye candy” mode. ATI’s mainstream solution looks better here than the GeForce 7600 GT and provides a comfortable frame rate in any standard resolution, including 1600x1200. The opponent is lagging behind by 10-20%.
The tech demo Lost Coast runs on a greatly improved version of the Source engine which is much more demanding than the original game, using HDR and numerous advanced shader-based effects. None of the mainstream graphics cards can deliver a playable frame rate even in 1024x768, although the GeForce 7600 GT almost manages to do so.
Of course there’s no talking about playability when you turn on 4x FSAA. The maximum you can get from a modern mainstream card here is 40fps. Note that the PowerColor X1800 GTO outperforms the GeForce 7600 GT due to its efficient memory subsystem with a 256-bit bus and ring-bus controller.
This game is rather simple in terms of graphical complexity, so all the participating graphics cards provide a high frame rate in the “pure speed” mode. For some reason the PowerColor X1800 GTO enjoys a bigger advantage over the GeForce 7600 GT in lower resolutions, although we might have expected something completely different: the memory subsystem performance has a bigger impact on the overall performance of a graphics card in higher resolutions. However, there’s only a difference of 5% between the two in 1600x1200 but over 10% in 1024x768.
In the “eye candy” mode the PowerColor X1800 GTO is ahead of the GeForce 7600 GT too, but the gap between them grows wider as the display resolution gets higher. Besides, the average performance of these solutions is rather too low for comfortable play in 1600x1200 notwithstanding the graphical simplicity of Project: Snowblind .
The game was tested with the Fraps utility, but the measurement accuracy is quite high since we used one of the integrated videos recorded on the game engine.
Quake 4 runs on the Doom 3 engine, so GeForce 7 series graphics cards have a certain advantage in this test. However, the PowerColor X1800 GTO delivers an acceptable average performance in resolutions up to 1280x1024 and – if you overclock it – in 1600x1200, too.
In the “eye candy” mode the two mainstream graphics cards allow playing in 1024x768. In higher resolutions their performance is below 50fps. You can get higher results by lowering the level of detail, but the game will lose some of its appeal as a consequence.
Serious Sam 2 requires 512 megabytes of graphics memory on board the graphics card when you enable its maximum graphics quality settings. Otherwise the performance may be downright disappointing. The PowerColor X1800 GTO and GeForce 7600 GT do not have that much memory, so their speed at the highest settings is far from comfortable in the standard resolutions.
Note that the GeForce 7600 GT outperforms the PowerColor X1800 GTO but it’s not because of a difference in the number of texture-mapping units as is the case with GeForce 7900 GTX and Radeon X1900 XTX. The mainstream cards have 12 TMUs each, so the reason for the difference should be sought for in the engine of the game.
We used a demo supplied with the game, but measured the frame rate with the Fraps utility.
We didn’t test this game with enabled FSAA because HDR is disabled on Nvidia’s GeForce cards in this case, making the game much less appealing visually. Besides that, HDR is incompatible with the Bloom effect. TES IV: Oblivion doesn’t offer integrated benchmarking options, so we have to use Fraps and test manually. The results may be somewhat inaccurate as a consequence.
In dungeons and other closed environments it is important for the graphics card to quickly process pixel shaders, lighting and shadows. However, the PowerColor X1800 GTO is 15% to 25% slower than the GeForce 7600 GT. 500MHz is probably not enough to reveal the potential of the R520 chip. This is also indicated by the results of the Radeon X1800 XL which is much slower than the GeForce 7600 GT in 1280x1024.
As for playability, you can’t play TES IV: Oblivion on mainstream graphics cards like PowerColor X1800 GTO or GeForce 7600 GT if you use HDR and highest graphics quality settings. The average performance is too low even in closed scenes and is going to plummet as soon as you go out in the open.
The PowerColor X1800 GTO is better in open game scenes as the 128-bit memory bus of the GeForce 7600 GT cannot handle the much higher memory load even without full-screen antialiasing. The average speed if the $199-249 cards is no higher than 30fps. It means you have to lower the level of detail or disable HDR to achieve just a bare minimum of comfort in Oblivion .
The GeForce 7600 GT has the same pure speed as the PowerColor X1800 GTO in this test, but the latter delivers a better minimum speed due to its more advanced architecture that better copes with complex math1ematic computations. Anyway, the maximum resolution for comfortable play on either card is 1280x1024.
When we turn on full-screen antialiasing, the PowerColor X1800 GTO is ahead in 1024x768 already, which is expectable considering its advantage in graphics memory bandwidth (32GB/s against the 7600 GT’s 22.4GB/s). Mainstream graphics cards cannot provide a comfortable frame rate in higher resolutions of this test mode, but the ATI solution is in the lead anyway.
Since the gaming engine of Pacific Fighters is designed with Nvidia GeForce 6/7 graphics card families in mind, PowerColor Radeon X1800 GTO loses to GeForce 7600 GT in all resolutions. Nevertheless, it provides quite acceptable gaming performance in up to 1600x1200 with no FSAA. In case FSAA is enabled, you can use 1024x768 resolution, though the minimal fps rate is going to be quite low.
The PowerColor X1800 GTO does well in this space simulator test. Its performance in 1280x1024 equals that of the GeForce 7600 GT in 1024x768 and is quite high for normal play.
Moreover, although ATI’s solution allows using FSAA only in 1024x768, GeForce 7600 GT cannot boast even that: it offers only 42fps in this case which is insufficient for comfortable gaming in a space simulator.
Despite wide use of SM3.0 and HDR in this game, PowerColor X1800 GTO unfortunately yields to GeForce 7600 GT. Only overclocking helps improve this situation and only if we enable FSAA 4x. We cannot explain this phenomenon. Since there are a lot of shader effects, the game favors solutions with numerous pixel processors, however, Radeon X1800 GTO and GeForce 7600 GT feature the same number of pixel processors. Moreover, the difference in core frequencies is not big enough to cause ATI’s lagging behind here. These could be the Catalyst drivers, or some architectural peculiarities of the R520 processor.
Speaking of the mainstream solutions’ performance, we can state that it is not that high in SM3.0 mode with highest graphics quality settings. We can only detect relatively high average results in 1024x768 resolution.
Unlike the previous game, the defeat of PowerColor X1800 GTO in Dawn of War is quite logical, because of the game engine peculiarities: it uses a lot of stencil shadows and at the same time is limited by DirectX 8.0. Nevertheless, the performance of our main hero is high enough to offer comfortable gaming in all resolutions with enabled anisotropic filtering.
Unfortunately, Radeon X1800 GTO cannot handle enabled FSAA at all. Even though the average performance in this case is around 50fps, the minimal fps rate is 20rps, which will result into noticeable jumps in those scenes that have a lot of action. So, you may not be able to control your troops efficiently during the gameplay.
The total score in 3DMark05 shows that PowerColor X1800 GTO is 846 points ahead of GeForce 7600 GT, even though the test uses the default resolution of 1024x768 with disabled FSAA. If we take a closer look at the results of individual benchmarks, we will see where this victory actually comes from.
When FSAA is disabled, PowerColor falls slightly behind GeForce 7600 GT, although in 1600x1200 the lag is not more than 10%. We didn’t quite understand why it happened this way, because the scene of the first benchmark is large enough to benefit from 256-bit memory access. However, once the full-screen anti-aliasing is enabled, the rivals swap places, even though the performance difference between them remains the same 10%.
In Game 2 test that requires efficient work with lighting, shadows and fast vertex shader processing speed, PowerColor X1800 GTO takes the lead right away thanks to 8 fully-fledged vertex processors it features (while GeForce 7600 GT has only 5). With enabled FSAA it also benefits from the more efficient memory subsystem making the advantage between PowerColor X1800 GTO and its primary rival reach 20%-30%.
Game 3 test involves all resources of the graphics card: fast processing of complex pixel shaders, high fillrate and certainly 256-bit memory bus. As a result, PowerColor X1800 GTO is at least as fast as GeForce 7600 GT without FSAA and is 20% faster than the Nvidia solution with FSAA 4x and maximum level of anisotropic filtering.
The situation in 3DMark06 is completely different: the performance gap between GeForce 7600 GT and PowerColor X1800 GTO is only 44 points. Moreover, Radeon X1800 XL shows very similar results despite the 16 pixel processors it features. Now let’s take a closer look at the results of SM2.0 and SM3.0/HDR tests separately.
The total score of the SM2.0 tests looks juts like the numbers we have just seen. The only difference is a slightly bigger gap between PowerColor X1800 GTO and GeForce 7600 GT. Although the gap is completely eliminated once we perform some overclocking.
As for the SM3.0/HDR tests, the advantages here lie with the PowerColor solution thanks to Radeon X1000 architecture that was designed to process pixel shaders as efficiently as possible. Moreover, the high-performance memory subsystem of the Radeon X1800 GTO gets extremely demanded in the first test of this category.
GeForce 7600 GT boasts a pretty significant performance advantage in the first SM2.0 graphics benchmark. PowerColor X1800 GTO manages to catch up only after overclocking. The remarkable thing about the first test is 26 light sources, which may be the cause of Radeon X1800 GTO’s failure. Another possible reason may also lie in the Catalyst drivers.
When we enabled FSAA and AF simultaneously, PowerColor X1800 GTO regains its leadership thanks to the 256-bit memory bus and even gets far ahead of the competitor in 1280x1024. Graphics cards with 256MB of video memory cannot handle higher resolutions in 3DMark06 with enabled full-screen anti-aliasing.
Unlike Game 2 test in 3DMark05, the second SM2.0 graphics test in 3DMark06 shows that PowerColor X1800 GTO and GeForce 7600 GT go neck and neck with one another if there is no FSAA involved. Once FSAA is enabled, PowerColor solution gets far ahead leaving GeForce 7600 GT 20%-30% behind.
The first SM3.0/HDR graphics test allows PowerColor X1800 GTO to show its best with 3.0 shader processing and HDR support. However, it loses its advantage over GeForce 7600 GT in 1600x1200. Overclocking provides about 20% performance boost for PowerColor here.
The second SM3.0/HDR graphics test doesn’t really load the memory subsystem. The most important thing here is efficient work with HDR and dynamic shadows. PowerColor X1800 GTO doesn’t support Fetch4 function that speeds up CSM shadows processing and is implemented only in R580 GPU. Besides, its graphics chip works at lower clock speed than that of the GeForce 7600 GT. As a result, Nvidia solution gets ahead. So, Radeon X1800 GTO owes its victory in the SM3.0\HDR tests category solely to the first benchmark.
The response from ATI Technologies to Nvidia’s GeForce 7600 GT appeared quite a success and didn’t require too much struggle and effort: the unused resources of the R520 and the already existing PCB layout gave birth to a new mainstream solution featuring excellent list of specifications and high performance. And ATI didn’t have to develop anything from scratch.
The new Radeon X1800 GTO showed its real best in the same applications where Radeon X1800/X1900 based solution have always been strong, i.e. in games requiring fast pixel shader performance especially with enabled full-screen anti-aliasing. The latter effect is very demanding to the memory subsystem, and such feature as 256-bit memory bus width of the newcomer played a very important role here. GeForce 7600 GT with its 128bit memory bus simply couldn’t compete with the Radeon X1800 GTO here. As usual, ATI solution yielded to its opponent in OpenGL games or those gaming applications that were optimized for GeForce 6/7 architecture.
Besides, Radeon X1800 GTO turned out to be very modest when it came to power consumption: only 48W under full workload. It also boasted acceptable level of noise thanks to a new more efficient cooling solution than the one used on Radeon X1800 XL, and excellent overclocking potential. Especially I terms of GPU overclocking. The latter is actually not surprising at all, because R520 was initially designed to work at frequencies over 625MHz. Although a lot of chips didn’t go through the frequency sort-out procedure, they can still run at a much higher rate than the 500MHz nominal frequency set for Radeon X1800 GTO, which certainly ensures a significant performance boost.
Overclocking potential of Radeon X1800 GTO is slightly held back by the slower graphics memory with 2.0ns access time. However, since this memory communicates with the GPU via the 256bit memory bus, it still ensures higher bandwidth than the memory subsystem of GeForce 7600 GT. The more efficient Ringbus memory controller than the one from Nvidia also contributes to higher memory subsystem bandwidth.
Of course, Radeon X1800 is not absolutely flawless. In particular, it has a much more overloaded PCB than that of GeForce 7600 GT and requires additional power supply. Also remember that the production cost of Radeon X1800 GTO is quite high because of the more complex PCB layout that has been initially developed for more expensive solutions. less options for price reduction – this is the price ATI had to pay for releasing a new mainstream graphics card without much trouble. At the same time, Nvidia GeForce 7600 GT is based on a very simple PCB design and uses a special GPU, so its price may drop down quicker and hence Radeon X1800 GTO may lose its attraction for the end-user in the long run.
Radeon X1800 GTO is going to be replaced with the new ATI RV570 based solutions, so we wouldn’t really worry about the shortage of R520 chips. But in the meanwhile, this graphics card deserves the title of the best solution within $199-$249 price range.
Speaking of the particular Radeon X1800 GTO graphics card from PowerColor that we have reviewed today, I would like to say the following. This graphics card follows the reference design precisely and is a great choice for those who do not hunt for rich accessories bundle and are not willing to pay extra for included games.