by Alexey Stepin
04/19/2006 | 03:11 PM
A modern high-end graphics card is an extremely complex device not only because of the immense number of transistors the graphics processor consists of. Developing the printed circuit board for such a card is a challenging and resource-consuming task as it has to provide for stable operation of the GPU and graphics memory at frequencies above 500-600MHz. This gets even more complicated due to the 256-bit bus the GPU and memory communicate through. The power circuitry has to meet strict requirements, too, now that the power consumption level of today’s high-end card is as high as 80-120W.
So, developing an original PCB for a high-end graphics card is a risky and costly business for the graphics card manufacturer and if anything goes wrong due to some unaccounted-for factors, the card manufacturer’s reputation, and the GPU maker’s too, may suffer, not to mention the inevitable financial loss.
We’ve got the present situation as a result: all high-end graphics cards are now being made on facilities rented by ATI Technologies and Nvidia, the chief GPU makers in this world. The companies’ partners buy ready-made solutions and may equip them with an original cooler or change the clock rates and BIOS settings, but most often they only stick their own logo and supply original accessories. This is why there’s so much uniformity in the high-end graphics card sector, but on the other hand the user is guaranteed stability of operation because the GPU manufacturer should know better how to design PCBs for its own graphics processors.
The GeCube Radeon X1900 XTX graphics card that we are going to talk about today is yet another such product. Let’s see how it’s going to attract the potential customer.
This is the box the GeCube Radeon X1900 XTX comes to the retail market in.
The moderate-sized cardboard box is painted black and red colors; red-haired Ruby with a katana in hand and outlined in a flash of lightning is pictured on the front of the box. A sticker in the bottom right corner tells you there is a full version of the 3D tactical shooter Delta Force: Xtreme inside. The package looks all right, but lacks any catching details in our eyes. It may make it difficult for the card to attract a potential customer in a shop. The box contains the following, besides the graphics card:
There’s nothing extraordinary here, but you get everything you really need to install and use your GeCube Radeon X1900 XTX. The user manual is without any extras, but provides all the necessary information. The choice of the included game is somewhat questionable. Delta Force: Xtreme, although belongs to quite a remarkable series, doesn’t stand out among a lot of other similar projects. It is also a rather out-dated title, having been announced about one year ago, and cannot serve to demonstrate the capabilities of today’s graphics processors.
Curiously enough, the power adapter is like those included with Nvidia’s graphics cards, i.e. with two Molex connectors. All cards with ATI’s GPUs on board have previously been shipped with an adapter that had one Molex plug. It makes sense, considering the Radeon X1900 XTX can draw over 70 watts from the external connector. The connection resistance is two times smaller with the two connectors instead of one. This helps to avoid overheat and make the connection more reliable. Well, many modern power supplies are originally equipped with one or two cables for PCI Express graphics cards, so the mentioned adapter may not be necessary to you at all.
The rest of the accessories included with the GeCube Radeon X1900 XTX graphics card are quite ordinary and we have no complaints about them.
The graphics card doesn’t differ from the reference sample and is an exact replica of the card we described in our Radeon X1900 XTX review.
There’s only a paper label on the reverse side of the PCB that tells you this is indeed a GeCube product: GC-X1900XTXD-VIE3. They even left intact the picture of Ruby on the cooler’s casing and the ATI logo on the fan motor card.
Eight Samsung K4J52324QC-BJ11 GDDR3 chips in 136-pin FBGA packaging are installed on board the card. The total amount of graphics memory is 512MB and it is clocked at 775MHz exactly as described by the Radeon X1900 XTX specification. As usual, the memory contacts the cooler’s base via rubber-like heat-conductive pads. Its efficiency is rather low, particularly due to their excessive thickness, yet they cool the chips all right. The graphics processor works at 650MHz as described in the official specs, too. Very thick dark-gray thermal paste with low thermal resistance is employed as an interface between the GPU and the cooler.
The card carries a Rage Theater chip that endows it with VIVO functionality. You cannot use the video output and connect an YPbPr-interfaced device simultaneously because the appropriate adapters need the same connector. This is a drawback of almost all graphics cards with VIVO support, though.
The cooling system installed on the GeCube Radeon X1900 XTX was first examined by us in our Radeon X1800 XT review. Its origin can be traced back to the Radeon X850 XT cooler and its design implies that the hot air is exhausted out of the PC case. The thick copper foundation and the use of heat pipes help this cooler cope with the high heat dissipation of the R580 chip. Unfortunately, the level of noise isn’t always comfortable because the cooler’s plastic casing works as a resonator when the fan speed is increased. It’s even not the level of noise, but the “plastic” tone in its sound due to the resonance that is irritating.
The GeCube Radeon X1900 XTX is an exact copy of the reference Radeon X1900 XTX from ATI, and its cooling system behaves absolutely alike. The fan works at its max speed when the computer is turned on and is really hard to bear, but then slows down in steps to become almost silent. The speed of the fan would go up sometimes during our tests – the fan speed control circuit must have reacted to the GPU temperature getting higher, but the noise remained at a comfortable level even though with that irritating “plastic” quality.
The reviewed graphics card from GeCube is at the very top of the current Radeon X1900 series, so we didn’t expect it to be any good at overclocking. That was indeed so: the maximum frequencies the card was stable at were 675MHz GPU and 800 (1600) MHz memory. This review is about practical and inexpensive rather than extreme overclocking, so we only added a 120mm fan to blow alongside the card’s PCB. This is why we didn’t achieve any wonders, but you should be aware that the frequency potential of the GeCube Radeon X1900 XTX, and of any high-end graphics card for that matter, is almost exhausted even without any overclocking. The card is already working near its limit. Better overclocking results may be achieved by using water-based or cryogen cooling and by modifying the power circuit, but such experiments are rather expensive and involve not a small risk of damaging the card in the process.
The quality of 2D image provided by the card was expectably high in all display modes, including 1800x1440@75Hz and 1600x1200@85Hz. The GeCube Radeon X1900 XTX delivers a sharp picture without any fuzziness or shadowing which are sometimes observed with very inexpensive no-name products that use non-standard PCBs and low-cost components.
We tested the performance of our today’s hero on the following platform:
We set up the ATI and Nvidia drivers in the same way as always:
ATI Catalyst:
Nvidia ForceWare:
We select 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. 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. It is not only about more optimal use of the graphics subsystem potential. We get much higher image quality in this mode than in case no FSAA and/or no anisotropic filtering are used.
We turned on full-screen antialiasing and anisotropic filtering from the game’s own menu if possible. Otherwise we forced the necessary mode from the ATI Catalyst and Nvidia ForceWare graphics card driver. We didn’t test the card in overclocked mode, because the overall frequency increase during overclocking was not that significant. Besides GeCube Radeon X1900 XTX, we have also included the following graphics cards:
These games and applications were used as benchmarks:
First-Person 3D Shooters
Third-Person 3D Shooters
Simulators
Strategies
Semi-synthetic benchmarks
Synthetic benchmarks
The recently released GeForce 7900 GTX has shattered the previously unrivalled dominance of the Radeon X1900 XTX except where an efficient and fast graphics memory subsystem is crucial for the overall result (such cases are full-screen antialiasing in high resolutions or extremely high, above 4x, levels of antialiasing).
The hero of this review is still unrivalled in this test and it enjoys a 15% lead over the GeForce 7900 GTX in 1600x1200. The average performance is over 100fps in this case, so even the most demanding gamer won’t be left unsatisfied.
Even if the GeForce 7900 GTX didn’t exist, the GeCube Radeon X1900 XTX would still lose this round to the GeForce 7800 GTX 512. So, the GeCube is 15% slower than the GeForce 7900 GTX in every resolution – Nvidia’s card boasts a much more efficient OpenGL driver and can process two times more Z-values per clock cycle which helps a lot at rendering stencil shadows. The resolution of 1600x1200 is hardly available to owners of the GeCube Radeon X1900 XTX although the average performance is quite close to comfortable for this game genre.
The GeCube Radeon X1900 XTX fails to take the first place in Call of Duty 2 , either. The recently released GeForce 7900 GTX is noticeably faster and allows playing the game with comfort in 1280x1024 with enabled full-screen antialiasing and anisotropic filtering. A curious fact is that there is little profit from the 48 pixel processors: the results of the Radeon X1800 XT and the Radeon X1900 XT are nearly identical.
It’s like in The Chronicles of Riddick , except that the GeForce 7900 GTX is not so far ahead of the GeCube Radeon X1900 XTX. The gap amounts to 10% in 1600x1200 which is very little considering the overall predilection of the Doom 3 game engine towards the GeForce 6/7 architecture. The GeCube card is only 4.7fps slower than the GeForce 7900 GTX in that resolution, their absolute speeds being about 57-62fps. So, this resolution seems to be available for comfortable play on either of the two graphics cards.
The GeCube Radeon X1900 XTX yields over 90fps in 1600x1200 with enabled FSAA and anisotropic filtering, leaving the rest of the participating graphics cards behind. In lower resolutions the GeForce 7900 GT drops out of the common group due to its lower GPU and graphics memory clock rates.
The Radeon X1800 XT with only 16 pixel processors joins the outsiders on the Research map which uses version 3.0 shaders. The GeCube Radeon X1900 XTX is still in the lead, but is closely followed by the GeForce 7900 GTX which in its turn is successfully competing with the Radeon X1900 XT.
Although HDR support for ATI’s Radeon X1000 cards is not quite efficient in the current version of Far Cry , the performance of the GeCube Radeon X1900 XTX is rather high in this mode – you can play with comfort in 1280x1024.
It’s the same on the Research map. Note the huge difference – about 20% – between the Radeon X1900 XT and the GeCube Radeon X1900 XTX. We can’t explain this since the clock rates of the cards differ but very little. Can it be that the extra 25MHz of GPU frequency produce such a colossal effect here?
The GeCube Radeon X1900 XTX is the best in 1600x1200 resolution, but its performance is already too low for comfortable play. In the lower resolutions the GeCube is a little slower than the GeForce 7900 GTX, by less than 5%. As you remember, the cards are doing full-screen antialiasing here.
The GeCube enjoys a certain lead in resolutions higher than 1024x768, but Half-Life 2 is such an easy application for today’s graphics hardware that even the GeForce 7900 GT, the slowest card in this review, easily gives you more than 60fps in 1600x1200 with turned-on full-screen antialiasing.
It’s different in Half-Life 2: Lost Coast because this small tech demo features much more advanced special effects than employed in the original Half-Life 2 . This may seem like a favorable environment for the GeCube Radeon X1900 XTX to show its abilities, but its 16 TMUs slow it down. The leader is the GeForce 7900 GTX with its 24 TMUs.
The GeCube Radeon X1900 XTX is only capable of beating the GeForce 7900 GT her of all the participating graphics cards from Nvidia. But the 7900 GT belongs in a lower product category than the GeForce 7900 GTX and GeForce 7800 GTX 512 and costs far less. On the other hand, the GeCube ensures a frame rate of over 60fps even in 1600x1200, so even the most demanding players are going to be well satisfied.
High-end graphics cards, and the GeCube Radeon X1900 XTX is among them, often find themselves being limited in speed by the performance of the computer’s central processor, even when full-screen antialiasing is in use. Two graphics cards stand out in higher resolutions. They are the GeCube Radeon X1900 XTX and GeForce 7900 GTX, and the latter is a little bit faster even though it uses an ordinary graphics memory controller rather than the Radeon’s ring-bus controller. Both the leaders yield an average frame rate of over 100fps in popular resolutions, so you are guaranteed to have enough speed in Quake 4 with turned-on full-screen antialiasing.
The numerous shaders with multiple texture lookups handicap the GeCube Radeon X1900 XTX in this game. It is slower than the GeForce 7900 GTX, even though the gap is narrower in higher resolutions thanks to the Radeon X1900’s ring-bus memory controller. Well, you don’t get a comfortable frame rate even with a GeForce 7900 GTX in this game at high resolutions with FSAA. The GeCube Radeon X1900 XTX is limited to 1024x768. You’ll have to use this resolution and FSAA or higher resolutions without FSAA.
The GeCube is confidently ahead of the others in resolutions above 1024x768. It is a little slower than the GeForce 7900 GTX in 1024x768 when there’s a rather small load on the graphics memory and the higher scene fill rate brings victory to Nvidia’s card. Note that even in that resolution you get a bigger speed reserve with the GeCube Radeon X1900 XTX (although this is not so very important for a third-person shooter).
The engine of this game was originally oriented towards the GeForce 6 and 7 architectures and Nvidia’s cards have always been and remain superior to Radeon X1000 series products in this test. Even so, you can play this game in 1280x1024 with enabled FSAA and anisotropic filtering on the GeCube Radeon X1900 XTX. You should be aware that you can’t get the best-quality water surface in this game because none of Radeon X1000 series chips supports vertex texturing. This has been a prerogative of GeForce 6/7 series cards so far.
X3: Reunion continues the glorious space sim series from Egosoft. The game features more detailed and special-effects-richer visuals than we saw in X2: The Threat , although it still does not use complex version 3.0 shaders. Anyway, this is a heavy application for the graphics subsystem and is a good benchmark, too. Let’s see what speed modern high-end graphics cards can provide in the new space simulator. Can you get a high-quality picture by turning on full-screen antialiasing?
Nvidia’s cards are slower than ATI Radeon X1900 and even X1800-based products. There is also no difference between the GeForce 7900 GTX, GeForce 7800 GTX 512 and GeForce 7900 GT despite the considerable dissimilarities in the tech specs of these cards. It must be due to some features of the game engine. In higher resolutions the GeForce 7900 GT feels the lack of memory bandwidth and falls behind the senior GeForce 7 models.
The GeCube Radeon X1900 XTX, Radeon X1900 XT and Radeon X1800 XT rank up in the order of descending performance, but the latter isn’t far behind the R580-based products that feature 48 pixel processors. It means there are not as many math1ematical instructions in the game’s pixel shaders as to load fully all the extra units in the R580 chip. The GeCube Radeon X1900 XTX ensures a high enough frame rate in all resolutions, including 1600x1200.
The game makes use of Shader Model 3.0 and this explains the failure of the Radeon X1800 XT which has only 16 pixel processors against 24 and 48 in the GeForce 7 and Radeon X1900, respectively. The new flagship product from Nvidia again lacks efficiency when communicating with the graphics memory. It is yet competitive against the GeCube Radeon X1900 XTX in 1024x768, but loses its ground in 1280x1024 and finds itself about 15% behind. The gap is somewhat smaller in the higher resolution, but the GeCube anyway has better minimum and average frame rates. Both these solutions are quite capable of running the game at an acceptable speed in 1600x1200 with enabled full-screen antialiasing and anisotropic filtering.
Like all games that demand a high scene fill rate and fast processing of stencil shadows, Dawn of War prefers Nvidia’s GeForce 7 cards. Here, the GeCube Radeon X1900 XTX can’t beat even the GeForce 7900 GT, not to mention the GeForce 7900 GTX, in any resolution with turned-on full-screen antialiasing. The average performance of this solution is sufficient to play in any resolution, but the minimum of speed is below 20fps which may have a negative effect on your gaming experience.
Aquamark3 benefits from higher fillrate of the GeForce 7900 GTX, but cannot take advantage of the GeCube Radeon X1900 XTX potential, because there are hardly any pixel shaders 2.0, not to mention 3.0 in this application. Moreover, even in test modes with FSAA, the benchmark doesn’t load the memory subsystem that much and ATI graphics cards cannot benefit from their ring bus memory controller. So, GeForce 7900 GTX once again proves that it is extremely well fitted for older games and benchmarks.
The first prize here belongs to GeCube solution. The flagship Nvidia card yields about 700 points to the leader here, but at the same time it is slightly ahead of Radeon X1900 XT working at lower frequencies than Radeon X1900 XTX.
Throughout the entire Game 1 test GeCube Radeon X1900 XTX retains its leadership, although it is only 5-10% ahead of Radeon X1900 XT and GeForce 7900 GTX.
In Game 2 test, which is also quite extensive, GeForce 7900 GTX manages to win the leadership in 1024x768 resolution, however, in higher resolutions GeCube Radeon X1900 XTX regains the laurels thanks to the ring bus memory controller.
Game 3 test requires the maximum pixel shader processing speed as well as fast work with the memory, which is very favorable conditions for Radeon X1900. So, here GeForce 7900 GTX gets defeated not only by the GeCube Radeon X1900 XTX, but also by Radeon x1900 XT.
All in all, GeCube’s leadership is indisputable: Radeon X1900 XTX is a great combination of fast pixel shader processing and efficient work of the memory subsystem, so its victory in 3DMark05 is quite natural.
Although 3DMark06 is a more contemporary and more complex benchmark than 3DMark05, GeCube Radeon X1900 XTX yields to GeForce 7900 GTX with standard test settings. There is nothing really surprising about it: the fillrate of the Nvidia’s flagship product is much higher. As for the pixel shader processing speed, Nvidia’s solution ran mostly as fast as Radeon X1900 XTX 9and sometimes even faster). And as for FSAA, it is not enabled in 3DMark test suites by default. Besides, the total 3DMark06 score results from the SM2.0 and SM3.0/HDR test scores, which should be considered separately.
The SM2.0 tests cannot let Radeon X1000 architecture show its real best. The fillrate parameter is of primary importance here as well as shader 2.0 processing speed. As a result, GeForce 7900 GTX is a stable leader in this test, while GeCube Radeon X1900 XTX has to put up with the second prize.
The result in SM3.0/HDR is just the opposite, which is also quite logical keeping in mind the 48 pixel processors of the GeCube Radeon X1900 XTX solution. At the same time, Radeon X1900 XT that differs from our hero by slightly lower GPU and memory working frequencies, yields to GeForce 7900 GTX and sometimes even to GeForce 7800 GTX 512. Unfortunately, we cannot perform a totally fair comparison between ATI and Nvidia solutions in test modes with enabled FSAA, because Nvidia’s babies cannot use FSAA and HDR at the same time.
GeForce 7900 GTX beats GeCube Radeon X1900 XTX in the first two resolutions. The memory performance starts mattering more in 1600x1200, and this is an indisputable trump of the Radeon X1900 XTX. So, when it doesn’t outperform GeForce 7900 GTX, it at least retains convincing parity with the rival.
Unlike the first test, the second test is not so demanding to the texturing speed: the scene is not so big and the 24 TMUs of the GeForce 7900 GTX working at 650MHz frequency remain practically unused. GeCube Radeon X1900 XTX keeps its leading position in this test, although it is just a little bit – only 5% - ahead of the Radeon X1900 XT in 1600x1200 resolution.
So, looks like GeForce 7900 GTX owes its victory to the first benchmark that is more demanding to the fillrate.
Having tested GeCube Radeon X1900 XTX graphics adapter we have every right to say that it boasts all the advantages and drawbacks of an ATI R580 based graphics card. It is one of the today’s most powerful graphics accelerators that is especially efficient in high resolutions with enabled FSAA as well as in those applications that deal with simultaneous processing of numerous complex pixel shaders. Like any other Radeon X1900, GeCube Radeon X1900 XTX is often defeated by GeForce 7900 GTX, especially in those OpenGL games that require high fillrate.
However, just like any other Radeon X1900 solution, GeCube Radeon X1900 XTX has more promising long-term prospective than GeForce 7900. The main reason for us to think so is the overall tendency that we have been observing in the today’s gaming world towards more numerous and more complex instructions in the gaming pixel shaders. Therefore, Radeon X1900 with its 48 pixel processors looks much more confident in this environment.
As for the actual solution offered by GeCube, we cannot make any specific conclusion about it: it is an exact copy of the reference board. To be more exact, the card itself is manufactured together with all other Radeon X1900 products and doesn’t really differ from what most other vendors have to offer.
The accessories bundled with the card are not very numerous, which is slightly disappointing, because this product belongs to the high-performance graphics solutions. But at the same time it doesn’t have anything excessive that could have affected the price of the GeCube Radeon X1900 XTX. From this standpoint, our hero is a complete opposite to what Asustek Computer has to offer: they pack their top-notch graphics solutions with all sorts of accessories.
So, GeCube Radeon X1900 XTX is a pretty attractive buy for those computer users who need the ultimate performance and do not really care about the far not free-of-charge goodies they get together with the card.
Highs:
Lows: