by Anton Shilov , Alexey Stepin
04/27/2006 | 04:15 PM
Sapphire Technologies was originally a part of PC Partner, a huge holding from Hong Kong, China, but back in 2002, the company obtained its brand name along with very experienced personnel from ATI Technologies and started to penetrate the retail market of graphics cards.
Having production capacity of 1.8 million boards per month, Sapphire became the main add-in-card partner of ATI from the very beginning.
Being the No. 1 manufacturing ally of ATI Technologies is a honour, but not an easy task. Sapphire has to offer the full range of Radeon products from the world’s largest supplier of discrete graphics processors, something which obliges the company to make even not the most interesting products. Nevertheless, priveleges that Sapphire has allow the company to offer very advanced add-in cards too, something that allows Sapphire to differentiate itself from the other companies.
For years now Sapphire has been offering its Ultimate (see our review called SAPPHIRE RADEON 9800 XT ULTIMATE Edition Extreme Overclocking: Myths and Reality) and Toxic graphics cards, the former with some kind of advanced and silent cooling system, the latter with factory overclocking. In May, 2005, the company decided to add another cathegory of products: the Blizzard graphics card, which should be the absolutely the top offering of Sapphire.
Unfortunately, at the request of Blizzard entertainment the company had to scrap the Blizzard brand and use its Toxic trade-mark for its premium class product.
Toxic means something poisonous and dangerous, dangerous for other makers of graphics cards, perhaps. Meanwhile, the original name of the board we review today was Blizzard , means a severe snow storm with strong winds, which implies that the “Blizzard” graphics cards are very cold, something which means additional reliability and overclockability when it comes to computer components.
The original Radeon X850 XT Platinum Edition Blizzard was meant to use a technology called “liquid metal” from the company named NanoCoolers. The liquid-metal cooling technology was based on metal cooling loops that contain a patented fluid called liquid-metal. Since the metal is electrically conductive, an electromagnetic pump was used to propel the liquid within a loop rather quickly. The liquid-metal loops acted as typical heat-pipes: they rapidly and efficiently transfer heat from a heat-source to a radiator which is cooled-down either actively or passively. The technology promised extreme efficiency amid quiet operation (see this news story for details).
Unfortunately, the startup called NanoCoolers decided that it would switch its efforts to a different technology named Thin Film Thermoelectric, which left Sapphire without the advanced cooler for the Blizzard family. The company even tried to utilize cooling systems from Arctic Cooling, but quickly realized that with its volumes the latter just cannot supply enough coolers, as it also ships them to companies like Asus and HIS.
At the end, Sapphire Technologies teamed up with Thermaltake in order to develop a cooling system that should be quiet, efficient, provide additional performance headroom and create a hype among enthusiasts. So, nearly one year after Sapphire first introduced its Blizzard, it can ship one, developed using totally different technologies and based on a totally different graphics chip. However, just days before the boards were about to leave the warehouse, Blizzard Entertainment asked Sapphire to change the brand-name.
So, one year, no liquid-metal and no Blizzard. Will this new Toxic turn into a success for the company? Let us find it out!
The packaging of Sapphire’s Toxic Radeon X1900 XTX with liquid-cooling is very big and solid. When you see it, you realize for what you are going to pay your money. And if you are in a retail store, you will quickly locate the Sapphire’s Toxic on the shelves. There an issue with the shelf space for such a large box, however, given that the absolute majority of enthusiast-class hardware is selling through online stores these days, the size of the box is not too critical, even though the shipment cost may be slightly higher compared to typical boxes.
The packaging, in fact, includes two parts: one is made of cardboard and the second one is made of plastic. The cardboard one is pretty attractive and does not contain any game-related characters, e.g., monsters, aliens, warriors, etc. There is kind of psychedelic theme on the package along with the logotypes of Sapphire, ATI and the name of the series: Toxic. By looking at the box, you cannot actually figure out what kind of graphics processor is there, which further implies that there is only going to be one Toxic with liquid-cooling in the lineup.
After the cardboard cover is removed (which is a pretty tricky task, by the way), we see a box made of plastic that actually contains some basic information about the product. We can learn that the new Toxic with liquid-cooling is based on the Radeon X1900 XTX with boosted clock-speed, we can learn some additional facts about the liquid cooling used along with the board and some general information about Avivo, Shader Model 3.0, CrossFire and so on. Sapphire recommends to use 450W power supply with 30A current on 12V rail along with the board.
In order to make sure that the liquid-cooling and the graphics card are not damaged during the transportation, Sapphire packs them into a special transparent container that holds the board and the water-pump very firmly!
Certainly, Sapphire did not forget to include a rich set of accessories along with its top-of-the-range offering. The necessary cables, manuals and discs are packed into a separate box that can be found below the plastic container of the of the graphics card itself. Here is what we have found inside the box:
The product bundle is very similar to that of the Sapphire Radeon X1800 XL, which we considered as a generally good one. But what is nice for the Radeon X1800 XL, may not be enough for the Blizzard Radeon X1900 XTX.
For instance, the multilingual user manual is the same as the one supplied with all Sapphire add-in graphics cards. It does say how to install graphics board, but it does not teach end-users how to properly install the water block and what users should know about the liquid cooling in general and proper installation of the Blizzard in particular. Perhaps, it made sense to equip the Blizzard with a special user manual?
There is a good thing about all Sapphires’ high-end graphics card: Sapphire Select DVD, a dual-layer (8GB) disc with four popular games of different genres, from which user can get two and try every one for an hour. The particular DVD features games like Brothers in Arms: Road to Hill 30 , Tony Hawk’s Underground 2 , Richard Burns Rally and Prince of Persia: Warrior Within. For typical graphics cards, for which the cost matters, two games seem to be enough, but for the premium class, such as the Toxic Radeon X1900 XTX with liquid-cooling, it would be better to have four games instead of two, or, at least get two top games, such as F.E.A.R. or The Elder of Scrolls IV: Oblivion .
To sum up, Sapphires top-of-the-range offering looks exclusive outside, the Toxic liquid-cooled Radeon X1900 XTX is exclusive when it comes to the graphics card, but the Toxic with liquid cooling is not really exclusive when it comes to the product bundle, something we wish it was. Overall, the product is good, but having a special list of accessories for it would totally bring really mind-blowing experience for the user!
As we have already reported in our earlier articles, neither ATI Technologies, nor Nvidia Corp. allow their partners to tweak print circuit boards (PCBs) of their high-end graphics cards. Moreover, very few companies can actually afford this, as PCBs have become state-of-the-art technology nowadays and reworking them is not an easy task.
That said, the Toxic version of the Radeon X1900 XTX looks just the same as all the other Radeon X1900 XTX-based graphics cards on the market, except, of course, the cooling system.
Even the PCB revision 109-A51031-50 of the reference Radeon X1900 XTX board from ATI remained on the Sapphire Blizzard, hence, you should expect Blizzard to feature the same 136-pin BGA K4J52324QC-BJ11 memory chips from Samsung with 1.1ns access time and maximum clock-speed of 1800MHz.
It is interesting to note that the chip Sapphire uses on the Blizzard Radeon X1900 XTX was made on the 4th of week of this year, in late January, whereas our reference Radeon X1900 XTX board used graphics processor produced on the 45th week of 2005, in early November.
All in all, when it comes to PCB design, all the Radeon X1900 XTX graphics boards are just the same, even in terms of print circuit board colour. Maybe ATI should allow its partners to obtain PCBs with black or blue colour for their premium-class offerings?
Even though the PCB of the Toxic Radeon X1900 XTX with water-cooling does not differ from any other on the market, the actual heart of Toxic is a specially designed cooling system that is not available in the retail and that should ensure that the Blizzard is silent and overclockable.
One thing that we discover when we take a look at the product is that liquid-cooling only applies to the graphics processor. The memory is cooled-down using tiny radiators, which are attached to the devices using thermal glue. This does not seem to be a huge problem, as according to Samsung Electronics every GDDR3 K4J52324QC-BJ11 memory chip consumes up to 2.73W, not a lot isn’t it? Moreover, Samsung claims that the threshold temperature of its chips is 125° C, which will hardly be achieved under normal conditions. Still, some of computer enthusiasts may want to use a bit larger radiators on the memory, something we will discuss a little later.
The majority of water-blocks available on the market are made of copper. The copper is known for very quick absorption of thermal energy and has proved itself as the best material for water-blocks. For some reason, Thermaltake and Sapphire decided to use aluminum water-block for the Toxic Radeon X1900 XTX with liquid-cooling technology, which may mean that the cooling efficiency is not the highest possible, something which a buyer of a premium product may hope for in expectations for further overclocking.
The organization of the liquid cooling is pretty simple, but that’s the idea behind the Toxic: to provide a very simplistic liquid cooling. The coolant is circulated with a 12V mini pump developed by Thermaltake, is pumped into copper tubes surrounded by radiator which is cooled-down by airflow of a fan that can rotate at 2000rpm or 2500rpm depending on the setting. It should be noted that the fan exhausts heat outside the computer case.
We should point out that the cooling solution used is generally the TideWater mini from Thermaltake. While we do now know its maximum thermal performance, we reckon that the original TideWater could dissipate up to 120W from the GPU and the difference between it and the mini is the smaller size of the radiator used by the latter. As reported, the Radeon X1900 XTX has thermal design power of 90W, therefore, Thermaltake TideWater mini should be able to cool it down without much problems at default speeds, but it hardly has significant threshold for overclocking. However, we do not think we should blame Sapphire for this situation. The retail versions of the Toxic Radeon X1900 XTX with liquid-cooling technology take 2 slots. With larger water-pump it would take 3 spare slots, which is not an option for many computer enthusiasts.
While installing the water pump in a free add-in-card slot is generally a nice idea, this will not work for micro-ATX computer cases. Moreover, it will not work on the vast majority of ATI CrossFire Xpress 1600 (RD480) mainboards, as the main PEG x16 slot for graphics cards is the “lower one”, hence, either there is not a lot of space below, or the tubes for liquid will touch memory heat-spreaders, which is not good at all, as the rubber can be damaged by high temperatures.
All-in-all, if you want to get Sapphire’s Toxic Radeon X1900 XTX, make sure you have the lowest PCI slot available and that water tubes do not touch memory radiators. Also you should check from time to time whether the exhausting turbine is not filed with the dust, as if your PC tower is standing on the floor, there will be quite a lot of dust on the bottom of your case. Finally, you should make sure that the airflows inside your case can cool-down PCB of the Toxic graphics card: unlike the default cooling system, Thermaltake’s TideWater mini cools only the GPU and does not take any heat away from memory and PCB.
Unlike the typical coolers used on the Radeon X1800 XT and Radeon X1900 XT/XTX graphics cards, Sapphire’s Toxic liquid cooler cannot adjust its performance depending on the temperature. Users can only regulate the speed of the fan – low-speed is 2000rpm, high-speed is 2500rpm. The former setting is utterly quiet and provides enough thermal performance to cool down the board at default settings (provided that the computer case itself is properly cooled too). The latter mode can be heard, however, it is not as loud as that of the original Radeon X1900 XTX cooler working at 85% of its speed…
As we mentioned above, all the Radeon X1900-series graphics cards are generally produced at one facility, hence, they comply to quality and reliability standard that ATI Technologies defines. While there are some drawbacks with such approach, the indisputable advantage of this is very high production quality, which means that virtually all Radeon X1900-series boards should function without problems and provide exceptional quality of 2D image.
Our expectations regarding fine 2D quality were completely justified: the Toxic Radeon X1900 XTX demonstrated exceptional quality if image even in the highest resolutions that our Dell P1130 and P1110 monitors support – 1800x1440@75Hz.
Given that the Sapphire Toxic liquid-cooled Radeon X1900 XTX already comes pre-overclocked, we did not have hopes for any further potential to run at even higher speeds. Still, we attempted to add some more muscle to the board. Unfortunately, just as expected, the part did not show any significant speed boost – about 10MHz per core and something like 20MHz for the memory is not a truly good result.
However, we should point out that we had an early revision of the Sapphire Toxic liquid-cooled Radeon X1900 XTX, therefore, retail versions may overclock better, especially if end-user turns the turbine to the max and increases voltage settings a bit.
ATI Radeon X1900 XTX is widely known as the “champion” when it comes to power: this board consumes about 120W and if you are planning to acquire it, you have to ensure that you have a quality power supply unit.
We tested the performance of Sapphire Toxic liquid-cooled Radeon X1900 XTX on the following platform:
We set up the ATI and Nvidia drivers in the same way as always:
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 Sapphire Toxic liquid-cooled 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
The Sapphire Toxic X1900 XTX is only formally faster than the ordinary Radeon X1900 XTX, the difference between them amounting to less than 2fps. Well, the Sapphire card is the winner all the same. Its increased GPU and memory frequencies and its 48 pixel processors help it deliver over 100fps even in 1600x1200 with turned-on full-screen antialiasing and anisotropic filtering.
The Sapphire Toxic X1900 XTX is slower than the GeForce 7900 GTX in this OpenGL and stencil shadows using application, yet it ensures quite a playable frame rate nonetheless. The performance of the Sapphire card may bottom out below the 60fps mark by 3.4fps, but you can hardly notice this during actual play.
The Sapphire Toxic X1900 XTX and the GeForce 7900 GTX both allow using resolutions up to 1280x1024 at the “eye candy” settings. In the higher resolution, the results of both the cards are far below 60fps, so you cannot play the game normally. Like in the previous cases, the higher clock rates of the Sapphire card don’t provide any real performance boost relative to the standard Radeon X1900 XTX.
The senior members of the GeForce 7800 and 7900 families are slightly ahead of the Sapphire Toxic X1900 XTX in Doom 3 because the engine of this game is oriented at the GeForce 6/7 architecture. Anyway, even with a lower fill rate and lack of UltraShadow-like technologies the Sapphire card is still capable of providing a good frame rate in all the standard resolutions, including 1600x1200. It is the high-efficiency ring-bus memory controller than helps it nearly reach the level of the GeForce 7900 GTX in the highest resolution despite the considerably lower memory frequency.
The Sapphire Toxic X1900 XTX is still unable to break away from the standard Radeon X1900 XTX. The limiting influence of the CPU is felt strongly in low resolutions, but the two mentioned cards are in the lead in 1600x1200, stopping very short of the 100fps mark.
The GeForce 7900 GTX equals the Radeon X1900 XTX and Sapphire Toxic X1900 XTX in 1024x768, but the Radeons take the lead as soon as 1280x1024. In 1600x1200 Nvidia’s flagship product is even slower than the humble Radeon X1900 XT. As for the picture in general, Far Cry is not a hard application for modern graphics hardware even if we talk about high resolutions with FSAA.
Despite the poor implementation of HDR support for Radeon X1000 cards, the Sapphire Toxic X1900 XTX yields 56fps in 1600x1200 and is only 10% worse than the GeForce 7900 GTX. This should be enough for comfortable play, but we’d suggest you limit yourself to 1280x1024 to avoid slowdowns in the heaviest game scenes.
The same is true for the Research map, but the performance is somewhat higher since the action goes on indoors.
The Sapphire Toxic X1900 XTX is the single graphics card whose performance is near the comfortable speed of 60fps in 1600x1200 with enabled full-screen antialiasing and anisotropic filtering, although this card is slower than the competitors in the lower resolutions. This is a good illustration of the advantage of the ring-bus memory controller topology over the ordinary design. It should be noted that the Sapphire Toxic X1900 XTX has a worse minimum speed than the GeForce 7900 GTX, even though by no more than 5fps.
Similar to Far Cry , this game isn’t difficult at all for modern top-end graphics cards, particularly for the Sapphire Toxic X1900 XTX. The effect from overclocking is very small: the performance of the Sapphire hardly differs from that of the Radeon X1900 XTX. The difference is 1.5fps in 1600x1200, which is a negligible gap considering the absolute frame rates of about 100fps and higher.
The performance of the Sapphire Toxic X1900 XTX is high enough, but this card is slower than the GeForce 7900 GTX. The gap is about 15% in 1024x768 and less than 10% in the higher display resolutions.
This tech demo seems to be using pixel shaders with multiple texture lookups. This is the only explanation of the lower performance of the R580-based cards that we can offer. They have a much higher shader execution speed compared with Nvidia’s flagship and feel much more confident at running many math1ematics-heavy shaders simultaneously.
Project: Snowblind is a rather simple game and uses simple visual effects. The Radeon X1900 family’s skills at processing complex shaders are not called for here, yet the Sapphire Toxic X1900 XTX still turned in an average result of 75fps in the highest resolution of the “eye candy” mode. The minimum performance is no lower than 64fps, so smooth and comfortable game experience is guaranteed even in the hardest scenes.
The performance of top-end graphics cards is largely limited by the speed of the CPU in this game. That’s why there’s little difference between the cards. Yet it is clear the Sapphire Toxic X1900 XTX is competing with the GeForce 7900 GTX quite successfully, although the latter plays on home turf: the Quake 4 engine uses the OpenGL API and the stencil shadow technique. The average performance of the cards is never lower than 114fps in 1600x1200 with enabled full-screen antialiasing and anisotropic filtering.
The GeForce 7900 GTX seems to allow using the resolution of 1280x1024, but you may want to switch to 1024x768 to have some speed reserve and avoid slowdowns to below 15fps. But the Sapphire Toxic X1900 XTX and the Radeon X1900 XTX offer a bigger speed reserve in 1024x768 than the GeForce 7900 GTX does.
This game was not tested in the “eye candy” mode because its HDR support is automatically disabled even on ATI’s Radeon X1000 cards on turning on FSAA; the graphics quality without HDR is much poorer. The Bloom effect is not supported simultaneously with HDR, either.
The hero of this review is on top in Oblivion because its heart is an R580 chip with 48 pixel processors. This game abounds in mind-bogglingly complex shaders, so the Sapphire Toxic Radeon X1900 XTX turns in the best result. The GeForce 7900 GTX tries to challenge it in 1280x1024 but has a considerably lower minimum speed even there. Here, every resolution is available to the owner of a Toxic Radeon X1900 XTX!
Keep it in mind, however, that we’re dealing with closed scenes of the game like caves, dungeons, houses, etc. Even the most powerful cards slow down as you go out into the open.
None of today’s graphics cards can singly provide at least 50fps in the open spaces of the Tamriel Empire. Moreover, only the Radeon X1900 family cards can keep the frame rate above 25fps. The average performance of the GeForce 7900 GTX is almost the same, but its min speed may be much lower than 20fps in some hardest scenes even in 1024x768 resolution. This is too slow.
There are a lot of pixel shader-based effects in this game, and the 48 pixel processors help the Radeon X1900 keep its speed high in complex scenes as opposed to the GeForce 7900 GTX. The Radeons are also better in average performance in resolutions above 1024x768.
You don’t generally need as much speed in a third-person shooter as in a first-person one. An average frame rate of 45fps should suffice for comfortable play.
The Sapphire Toxic X1900 XTX and the Radeon X1900 XTX meet this requirement easily as they keep the game running at 45fps and higher even in 1600x1200. The G70- and G71-based solutions can’t repeat this feat – the speed reserve they offer is much smaller.
Pacific Fighters is optimized for Nvidia’s GeForce 6/7 cards, but the Sapphire Toxic X1900 XTX, just like the rest of the Radeon X1900 family, can ensure a high enough frame rate in the “eye candy” mode up to 1280x1024 resolution.
Alas, you can only get the highest rendering quality of the water surface if the graphics processor supports vertex texturing, but this feature is only available with Nvidia’s graphics cards.
The R580-based cards are victorious here since X3: Reunion makes wide use of pixel shaders when creating its visual effects. The Sapphire Toxic X1900 XTX has a negligible advantage 2fps over the Radeon X1900 XTX. Both these cards, and the Radeon X1900 XT too, are fast enough for playing in 1600x1200 with enabled full-screen antialiasing and anisotropic filtering. The GeForce 7900 GTX has a good result, too, but it’s 18% behind the Sapphire in terms of average performance.
Age of Empires 3 belongs to the new generation of games and actively uses SM3.0 effects at the maximum graphics quality settings. No wonder the Radeon X1900 cards feel more comfortable here as they carry as many as 48 pixel processors on board. The importance of these processors in this game is indicated by the low result of the Radeon X1800 XT that has a similar architecture but only 16 pixel processors.
You want an average performance of at least 55-60fps in a real-time strategy, just like in a first-person shooter, and the Sapphire card can give it to you in all the resolutions here. The results of the Radeon X1900 XTX and Radeon X1900 XT are not much lower. The GeForce 7900 GTX card is good in all resolutions, but doesn’t offer as high a speed reserve as the R580-based solution do.
We see the opposite in Dawn of War . Each character in this game uses a 512x512 texture while the landscape is generated by means of 1024x1024 maps. This means that the graphics card’s fill rate is very important here. Moreover, each war unit consists of about 2 thousand polygons, and there may be as many as 100 such units on the screen, each with a realistic shadow created by means of the stencil shadowing technique. As a matter of fact, Dan of War belongs rather to the previous generation of games as it doesn’t use complex pixel shaders, always keeping within the DirectX 8.0 framework.
With its UltraShadow II technology and 24 TMUs, the GeForce 7900 GTX manages to produce a high frame rate in all the resolutions whereas the Sapphire Toxic X1900 XTX is rather too slow for 1600x1200 (the min speed is below 20fps). The advantages of the Radeon X1900 XT architecture – 48 pixel processors and an accelerated rendering of shadows using the Cascaded Shadow Maps technique – are just not called for here.
Aquamark3 is a previous-generation graphics benchmark because it has a minimum of version 2.0 shaders, using instead version 1.1 and, less often, version 1.4 pixel shaders. Anyway, the Sapphire Toxic X1900 XTX isn’t much slower than the GeForce 7900 GTX – there’s a difference of only 10% between them in 1600x1200.
The Sapphire Toxic X1900 XTX wins this test, stopping less than 300 points short of the next landmark. 11,756 points in 3DMark05 is the best single-card result we’ve ever recorded in our labs.
The Sapphire card holds the first place in all resolutions, although its performance is almost the same as that of the Radeon X1900 XTX which works at the default GPU/memory frequencies, namely 650/775 (1550) MHz.
The second test is somewhat different. The scene itself is smaller, for example. The fill rate requirements aren’t that strict here. What’s important is the speed of execution of pixel and vertex shaders and the rendering of shadows and lighting. This doesn’t brings any changes into the ranks: the Sapphire Toxic X1900 XTX and the Radeon X1900 XTX still hold the first places except 1024x768 resolution where they are joined by the GeForce 7900 GTX.
It is in the third test that the Radeon X1900 can show its full strength. Its 48 pixel processors come in handy here. The GeForce 7900 GTX, currently the best offer from Nvidia, is noticeably slower than the Radeon X1900 XT and is quite far behind the Sapphire Toxic X1900 XTX and Radeon X1900 XTX.
So the overall 3DMark05 scores are confirmed by the results of individual tests even though they are all acquired in the “eye candy” mode. The new graphics card from Sapphire that features a liquid-cooling system and increased clock rates has no rivals in this benchmark, although it is not much better than the ordinary Radeon X1900 XTX.
The higher clock rates of the Sapphire Toxic X1900 XTX make it competitive against the GeForce 7900 GTX. Well, the standard Radeon X1900 XTX is a mere 142 points behind the Nvidia card, too. The overall 3DMark06 score is the sum of SM2.0 and SM3.0/HDR tests, so let’s check them out independently.
The Sapphire Toxic X1900 XTX has worse results in the SM2.0 tests in comparison with the GeForce 7900 GTX, but this is a consequence of the test’s demanding high texturing speed from the card. Note the big difference between the Radeon X1900 XTX and Radeon X1900 XT: each megahertz is highly important in 3DMark06. The 48 pixel processors on board Radeon X1900 hardly matters here – the Radeon X1800 XT is just a little slower than the Radeon X1900 XT.
It’s different in the SM3.0/HDR tests: the Sapphire Toxic X1900 XTX and the Radeon X1900 XTX have no rivals at all. The R580-based solutions show their best, mainly due to their 48 pixel processors which are not idle here as they were in the SM2.0 tests. This fact is indicated by the big difference between the results of the Radeon X1900 XT and the Radeon X1800 XT.
As we said above, the first SM2.0 test is the only of 3DMark06’s tests that requires a high fill rate and does not contain complex pixel shaders. This helps the GeForce 7900 GTX win two resolutions out of three, but the efficient ring-bus memory controller of the Sapphire Toxic X1900 XTX says its word in 1600x1200 where the memory load is exceedingly high.
The second SM2.0 graphics test is nothing else but an improved second game test from 3DMark05 and it is far less sensitive to the scene fill rate. The Sapphire Toxic X1900 XTX and Radeon X1900 XTX leave the GeForce 7900 GTX far behind here in all the resolutions. It’s only in 1024x768 with enabled full-screen antialiasing and anisotropic filtering that Nvidia’s card can match the Radeon XT1900 XT.
The results of the SM2.0 graphics tests show that the senior model of the GeForce 7 family only wins the first test which is sensitive to the scene fill rate and, accordingly, to the number of TMUs. Moreover, you should remember that the Radeon X1900 family feels better in FSAA modes due to their very effective memory controller, so the results of the separate SM2.0 tests somewhat contradict the overall scores we got in the “pure speed” mode.
Sapphire’s Toxic Radeon X1900 XTX with liquid-cooling is currently the fastest Radeon-based graphics card on the market. However, it is not only a fast board with water-cooling and with all the advantages and disadvantages of the Radeon X1900 XTX family, but this is the top-of-the-range offering from Sapphire Technologies and it should be considered as such.
The main advantages of the Toxic Radeon X1900 XTX with liquid-cooling compared to the rivaling solutions, primarily from the GeForce 7900 camp, is 48 pixel shader processors, which may become important for future games full of math1ematically intensive pixel shaders, support for HDR + antialiasing, a bit better performance in modes with 4x full-scene antialiasing and so on. The key benefits that the Sapphire’s product has over its brethren among the Radeon X1900-series is generally quieter cooling solution and increased clock-speeds.
Yet another advantage of the Toxic Radeon X1900 XTX with liquid-cooling is exceptional product bundle: it includes everything needed and a selection of games that users can choose themselves. We would be even more pleased if Sapphire supplied a special set of accessories with its premium boards, however.
Without any doubts, the premium offering from Sapphire is definitely a nice product, but not without some drawbacks which include a cooling system that hardly has enough performance for overclocking near the extreme levels, some design peculiarities that oblige end-users to have a relatively big computer case with proper airflows inside. Additionally, Nvidia GeForce-based graphics accelerators usually outperform those powered by ATI Radeon in OpenGL applications
So, let’s sum up all the pros and cons of the product that we have just reviewed.