Gtx 295 power consumption. ZOTAC GeForce GTX295 dual-chip video card review

The most powerful 3D accelerator to date is undoubtedly the dual-processor GeForce GTX 295. It was created in opposition to AMD's flagship, the dual-processor Radeon HD 4870 X2. The NVIDIA video card turned out to be much more powerful and efficient, surpassing the competitor in most parameters. Unfortunately, it appeared on the market a little late, when AMD's position as a market leader strengthened and was unable to turn the tide. However, the title of the formal leader among 3D accelerators belongs to GeForce, not Radeon.

Design features

Despite the fact that conceptually the GeForce GTX 295 and Radeon HD 4870 X2 are completely identical, they differ significantly in their constructive implementation. This is due to the peculiarities of the graphics controllers used in them.

The RV770 GPU on the Radeon HD 4800 series graphics cards supports 256-bit memory bus, while the NVIDIA GT200 GPU supports 512-bit memory bus. In addition, the RV770 has an integrated external interface controller, while the GT200 has it as a separate microcircuit. When creating a dual-processor graphics card, AMD cost one printed circuit board with two 256-bit buses to connect two 3D chips to the memory. It also had a PCI Express bus switch, two 1024 MB GDDR5 chipsets, and power supply circuits. Therefore, in terms of technological complexity, a dual-processor AMD video card differs slightly from a competitor's single-processor models, on which, in addition to the graphics controller itself, a second chip is installed, which is responsible for external interfaces.

As we have already said, architecturally, the GeForce GTX 295 does not differ from its competitor. From the point of view of the system, these are two video cards with the PCI Express interface, each with its own array of local memory and a set of output ports. Only they are connected not by means of system logic (motherboard chipset), but using an additional PCI Express switch. Video cards work in SLI mode, performing alternate rendering of frames and recording them into the output video buffer of one of them - the one to which the monitor is connected.

Structurally, the GeForce GTX 295 is made differently than the Radeon HD 4870 X2.

It actually consists of two printed circuit boards, only facing each other. One board contains the first GT200 graphics chip, the NVIO chip (responsible for external interfaces) and the nForce 200 bridge (PCI Express switch), on the other - the second GT200 with its own NVIO chip.

Failed to fit all components on one board. Moreover, the width of the memory buses of each of the video cards had to be reduced from 512 to 448 bits, respectively, and the memory size was reduced to 2 x 896 MB (128 MB for each 64-bit channel, of which each 3D accelerator has seven). The cooling system is located between the two boards; it consists of an aluminum chassis in contact with memory chips, transistors and auxiliary chips, copper inserts between the graphics cores, a set of radiator fins spreading out like a fan, and a fan with longitudinally spaced blades. The fan is located in one end of the card, the exhaust holes are in the other facing outward, as well as from the top.

The front side is covered with a metal casing covered with rubber, on which the logo of the video card manufacturer is applied.

The set of video outputs of the standard (reference) GeForce GTX 295 consists of two DVI and one HDMI. The developers have already abandoned the analog output on the TV, and the DisplayPort connector was considered unnecessary due to the small distribution of monitors with its support.

The top of the video card has one SLI connector for connecting to another NVIDIA video card, as well as an S / PDIF input for full HDMI (with video and sound transmission).

Despite the rather complex design, the reference design GeForce GTX 295 has tolerable dimensions, for example, it is comparable in size to a single-processor GeForce GTX 285. The installed video card blocks only one adjacent slot (two-slot cooling system), its length is about 27 cm. the power consumption of the GeForce GTX 295 at standard frequencies is 289 W, the power of the system power supply must be at least 700 W, two power connectors must be connected, one of them must be 8-pin.

Of course, a "sandwich" video card is quite difficult to manufacture, which negatively affects its cost and the final price. Fearing for the competitiveness of its products, NVIDIA has already significantly lowered prices for all its video cards, including the dual-processor one. And in order not to work at a loss, you need to look for ways to reduce the cost. The simplified single-board design of the GeForce GTX 295 has been prepared by several video card manufacturers, but such models have not yet become massive on the market. Most of the currently available video cards of this series are standard reference models, differing only in packaging, configuration and logo on the front panel.

This is the option that Innovision offers its customers.

There are not many differences compared to the reference video card - the clock frequencies of the graphics controllers (but not their shader units) and memory chips are slightly increased, which should not significantly affect the performance.
Clock frequency data (obtained using RivaTuner):

As for the rest, it is the standard (in design, functions, configuration and cost) GeForce GTX 295, the flagship of the current line of NVIDIA video cards.

Testing

In fact, the GeForce GTX 295 has one competitor - the already mentioned Radeon HD 4870 X2 video card. Can it theoretically show the same level of performance? I guess, yes.

Although each of the two 3D processors of an AMD graphics card has only a 256-bit memory bus, GDDR5 memory, unlike GDDR3 used on NVIDIA video cards, exchanges data at four times the base clock frequency (according to the Quadro Data Rate). Therefore, a 256-bit bus with GDDR5 memory is the same as a 512-bit bus with GDDR3 memory. In this parameter, the AMD video card even wins. Keep in mind, however, that the NVIDIA GT200 GPU has more processing power and efficiency thanks to its well-designed and optimized architecture.

However, a dual-processor AMD video card will cost the user less. The difference sometimes reaches 30% when it comes to "inexpensive" manufacturers like Palit. However, the Revolution 700 Deluxe model is not a basic reference version of the Radeon HD 4870 X2, but a very "advanced" video card, with an original cooler, factory overclocking and an expanded set of I / O ports.

Also for comparison, we provide data on the GeForce GTX 285 video card - the most powerful single-processor model from the current NVIDIA lineup. Our copy is also released by Palit, but this time it belongs to the base, not overclocked series, and has a moderate price (which is 60-70% lower than that of a dual-processor NVIDIA video card).
Summary data for three video cards:

Test bench composition:
- Intel Core 2 Duo E8300 processor overclocked to 3.82 GHz;
- 2GB DDR3-1067 memory from Apacer;
- motherboard Gigabyte EP35C-DS3R (Intel P35);
- FSP Everest 1010 1 kW power supply with 8-pin power connector.

The tests were carried out in the Windows Vista SP1 environment, the version of the drivers for NVIDIA video cards is 185.85, for AMD video cards - 9.6.

Performance. The benchmark results in demanding 3D games convincingly prove the significant advantage of the dual-GPU GeForce GTX 295.
Maximum performance (anti-aliasing and anisotropic filtering off):

Improved quality (AA 4x, AF 16, transparent texture anti-aliasing):

Yes, there are cases when the Radeon HD 4870 X2 does not lag behind (and even overtakes within 10%), but more often the GeForce's performance is 30-60% higher. As for the single-processor GeForce GTX 285, a lag of 60-70% is in the order of things for it. Which, by the way, clearly corresponds to the difference in prices between the 285 and 295 models.

Temperature and noise. The weak point of the GeForce GTX 295 is power consumption. Recall that even in a well-overclocked system, such a video card will consume at least half of the total power. Fortunately, only in 3D - in 2D mode, the video card halves the frequency of graphics chips, and completely drops the memory frequency to the minimum mark - 100 MHz. Under full load, the graphics chips warm up decently (according to our data - up to 80 degrees), but the noise level remains within the bounds of decency and is not noticeable against the background of the fans working in the case.

Part of the hot air is directed to the outside through the slots in the mounting plate, and part to the inside of the case through the open part of the casing. Therefore, you need to take care of proper ventilation of the case, otherwise the video card will quickly "warm up" all the components of the system.

conclusions

The GeForce GTX 295 is truly the most powerful graphics card on the market. The most powerful in terms of performance and energy consumption. It demonstrates record performance in 3D games (in Crysis at maximum settings and a resolution of 1680x1050 - more than 45 fps), but requires a power supply unit with a capacity of at least 700 watts to operate. Too high a price (we have more than $ 500), even without taking into account the cost of a high-quality case, makes it unaffordable for most users. Which is natural, since this is a "top" product of its generation.

The choice of manufacturer when buying a GeForce GTX 295 does not really matter. All video cards of the first series are manufactured in the same place. It is enough to focus on the cost of the product. The Inno3D GeForce GTX 295 video card has a standard design, basic equipment and almost nominal clock frequencies, but it does not cost more than its counterparts.

Based on materials from www. gigamark. com

Since the GeForce GTX 295 claims to be the fastest "single" video card, the following video cards will participate in this test - the GeForce GTX 280 and a bundle of two such video cards in SLI mode, as well as, of course, the Radeon HD4870X2. To evaluate the efficiency of the "internal" SLI video card GeForce GTX 295, we also tested the performance of one "half" of this card, disabling SLI in the driver settings. Test stand

In the diagrams, the results for the GeForce GTX 280 in singles and doubles are marked in dark green. The results of the work of the "half" of the GeForce GTX 295 and in the normal mode are marked in light green. The results for the Radeon HD4870X2 are traditionally shown in red. GeForce video cards were tested on drivers version 181.20, Radeon HD4870X2 - on Catalyst 8.12.

Testing in 3DMark Vantage Performance mode showed that the GeForce GTX 295 really pulls out the "leader's shirt" from the previous king of 3D graphics - the Radeon HD4870X2.

If we take the 3DMark score without taking into account the CPU score (GPU Score), then the advantage of the new product over the Radeon HD4870X2 is about 9%. Not so much, but it is quite noticeable and cannot be attributed to the measurement error. Especially when you consider that when fighting for the "crown" every percentage is on the account.

When tested in 3DMark Vantage, the graphics settings mode High increases the advantage of the GeForce GTX 295 to almost 30%, which is somewhat strange. It is quite possible that due to the use of a wider memory bus, the GeForce GTX 295 can get an additional advantage in the transition to the "heavy" graphics mode compared to the Radeon HD4870X2. However, the figure of 30% looks somewhat dubious.

However, the Extreme 3DMark Vantage test mode sums up the testing results in this application. Indeed, as the "severity" of the mode increases, the advantage of the GeForce GTX 295 over the Radeon HD4870X2 increases. In Extreme mode, we see a difference in the GPU Score of these video cards of about 12%.

When tested in the game Crysis at low resolution, we again see how the GeForce GTX 295 outperforms the Radeon HD4870X2 by 25% in performance.

With an increase in resolution, the difference in the results of the GeForce GTX 295 and Radeon HD4870X2 decreases slightly, but still remains noticeable - about 10% in favor of the new product.

The inclusion in the game Crysis of the heaviest graphics mode VeryHigh still turns out to be a very difficult test even for the most powerful video cards. In absolute terms, the results of GeForce GTX 295 and Radeon HD4870X2 noticeably sag even at low resolutions, but the advantage of about 10% still remains with the new product.

But the increase in screen resolution comes as a surprise. Among the "single" video cards, the victory remains with the Radeon HD4870X2, however, the GeForce GTX 295 is only slightly behind.

Nevertheless, in the game Far Cry 2 with the graphics quality settings Ultra GeForce GTX 295 in the dispute with the Radeon HD4870X2 takes a convincing revenge and its advantage is about 15-20%. Interestingly, in this test the results of the new product are very close to those of the GeForce GTX 280 in SLI mode, and this is for a "single" video card!

In Race Driver: Grid, the new GeForce GTX 295 initially outperforms the Radeon HD4870X2 at low resolutions, but then the balance of forces in this pair changes to the opposite, and the Radeon HD4870X2 comes out ahead. However, the difference in results by a few percent cannot be called significant, especially since all video cards show quite comfortable FPS in this game.

GeForce GTX 295 performance scalability

Now, based on the above results, we can find out how efficiently the two "halves" of the GeForce GTX 295 work in the SLI bundle. As a starting point, we will take the results shown by one half of the GeForce GTX 295 (SLI mode for the video card is disabled) and compare them with the performance of the same video card, but in the maximum performance mode (SLI mode for the video card is enabled). The diagram below shows the percentage increase in performance when the "internal" SLI mode is enabled for the GeForce GTX 295 video card.

As you would expect, in synthetic 3DMark Vantage tests, the scalability of the "internal" SLI fluctuates within 80% -90%, which is quite close to the theoretical maximum of 100%. In Crysis, the SLI efficiency is slightly lower, from 65% to 75%, but this is also a very good result. Far Cry 2 reacts very positively to the increase in the number of GPUs in the video system, here the gain from using the SLI bundle in high resolutions exceeds 90%! Good results were obtained in the game Race Driver: Grid - the SLI efficiency ranged from 73% to almost 90%. Next, we will look at scalability in a slightly different way. Namely, how the performance of the GeForce GTX 295 will change if it is installed in a motherboard with support for the PCI Express 1.0 bus, whose bandwidth is half that of the PCI Express v2.0 bus. To do this, we taped a part of the PCI Express slot of the GeForce GTX 295 video card with tape to make it work in PCI-E 2.0 8x mode.

In this screenshot you can see (we glued two images together so that all the parameters of this menu item in the NVIDIA control panel are visible) that after such a simple operation, the video card uses half the lines of the PCI Express 2.0 bus, while the resulting total bus bandwidth is equal to the interface bandwidth PCI Express 1.0 16x. The results of testing the performance of the GeForce GTX 295, obtained after the bus "cut", will be displayed as a percentage of the results obtained earlier at the speed of the PCI Express 2.0 16x bus.

As you can see from the diagram, the GeForce GTX 295 video card practically does not respond to a halving of the PCI Express speed. This means that by purchasing a GeForce GTX 295 video card, the owners of motherboards with support for the PCI Express v1.0 bus. practically will not lose in performance. Provided, of course, that the CPU is able to fully unleash the potential of this powerful video card.

Power Consumption Test

And, finally, we present data on the power consumed by video cards under various load conditions. In the table below you can see the results obtained for the power consumption of the GeForce GTX 295 and Radeon HD4870X2 video cards.

As you can see, in the "rest" mode, the new product has a very moderate power consumption, which is greatly facilitated by the drop of the GPU frequency to 300 MHz and the video memory frequency to 100 MHz. But the main purpose of a gaming video card is to please with high FPS at maximum graphics settings. As it turned out, here too the "appetite" of the GeForce GTX 295 is rather modest, especially in comparison with the Radeon HD4870X2. However, the maximum load mode created by the FurMar test in Xtreme burning mode showed the real ardor of the GeForce GTX 295, whose power consumption came very close to the 300W level. Note that the GPU temperature in this mode reached 93 and 96 degrees (the "leading" GPU heats up a little more). At the same time, the cooler of the GeForce GTX 295 video card was overclocked to a speed of 3400 rpm, which is about 80% of the maximum. Of course, the cooler noise was quite noticeable at this speed. At the same time, during testing in the game Far Cry 2, the cooler speed did not exceed 2400 rpm, while its noise was barely noticeable. Well, so that you can estimate the power requirements for the power supply, here are the total power consumption figures of our test stand based on the Core i7 965 Extreme processor overclocked to 3.84 GHz.

As you can see, the minimum requirements for the power supply indicated on the packaging of the Zotac GeForce GTX 295 video card are not so far from reality. However, it is well known that it is better not to save money on a power supply for top configurations, it will come out more expensive.

conclusions

The battle for the "crown" of the fastest 3D accelerator continues! Today we got acquainted with another "two-headed" video card from NVIDIA, which can rightfully be called the fastest video accelerator to date. The novelty has demonstrated excellent performance results in the most severe modes and is not distinguished by the "hottest" temper and noise (if not specially warmed up). Moreover, the new product can be used in Quad-SLI mode, which means we can expect even higher FPS and the number of "parrots" in 3DMark. But we will tell this story next time.

No one likes to be in second place, especially a company as ambitious as Nvidia. Most recently, the company introduced the GeForce GTX 260 and 280 graphics cards, and at that time AMD could only offer the unfortunate Radeon HD 3870. Not surprisingly, the GT200 graphics cards continued to lead nVidia. But no one expected AMD to strike back with the Radeon HD 4850 and 4870 graphics cards - by the way, none of the cards were faster than the flagship of NVIDIA - which could distract attention from the massive monolithic chip.

After gaining a lot of experience with its smaller and more flexible GPUs, AMD quickly introduced the Radeon HD 4870 X2 graphics card, containing two RV770 GPUs, 2GB of GDDR5 memory and a PCI Express (PCIe) bridge to link the chips. Suddenly things changed and AMD had the fastest split graphics card on the market. Six months after the announcement, a complete set of RV770-based video cards is already beating nVidia's positions in all markets conquered by the G92.

Even hardcore gamers, who can spend thousands of dollars to get the best gaming performance, find themselves at a loss. Looking to beat the 4870 X2 with nVidia's decision? Buy a pair of GeForce GTX 280s. Want to beat it with AMD's solution? Buy a pair of 4870 X2 and run it in 4-way CrossFireX mode. Need something faster? Add a third GTX 280. But you need an expensive motherboard and a very powerful processor to handle this kind of graphics performance. Is there a limit to this?

Therefore, limiting only one separate graphics card seems quite reasonable to determine the winner for today. Of course, with the existence of 4-way CrossFire and 4-way SLI configurations, today there is always where to spend more money, increase power consumption and get 10-20 percent additional performance ... But we decided to find the winner among single video cards.

Click on the picture to enlarge.

In 2008, nVidia was not too active against the Radeon threat, updating only the GeForce GTX 260 to more aggressively compete with the 1GB AMD HD 4870. According to the company, all GTX 260 graphics cards will be able to use the 216 shader configuration in the future. Cards with 192 shader processors will disappear as partners sell them off. Given the previous price level, Nvidia expects the new GTX 260 to beat AMD's current offerings.

But apart from that, nVidia wants to take the crown in performance of separate graphics cards, and the company is spending a lot of money on this. This article is devoted to a preview of the graphics card that nVidia plans to release during CES. The GeForce GTX 295 is nVidia's answer to the AMD Radeon HD 4870 X2 and uses the same design as the GeForce 9800 GX2.

GeForce GTX 295 in detail

Click on the picture to enlarge.

Initially, it was rumored that the GeForce GTX 295 will consist of two GT200s in a configuration that simulates a pair of GTX 260. In reality, the video card consists of two full-fledged GT200s, each with 240 processing cores, but with a memory configuration similar to the GTX 260.

The original GT200 contained 1.4 billion transistors and was manufactured at TSMC using a 65nm process. A version of the chip called the GeForce GTX 295 benefits from a reduced to 55nm process technology. In addition, according to Nvidia's Jason Paul, the company has tweaked the on-die latency to improve the performance per watt ratio, which should prove itself in power tests.

Like the GeForce GTX 280, each GTX 295 GPU uses 240 stream processors and 80 texture filtering / addressing units. But, like the GTX 260, at the same time the GPU 295 uses seven ROP / framebuffer blocks, which gives 28 ROPs in total, and a 448-bit interface is supplied to 896MB of DDR3 memory. The most important clock speeds also correspond to the GeForce GTX 260. The core clock, including texturing units and ROPs, is 576 MHz. Stream processors run at 1242 MHz. And the memory is at 999 MHz (1998 MHz effective frequency). As you can see, by architecture, each chip sits between the fastest and second fastest nVidia GPUs.

		GeForce GTX 280	GeForce GTX 260	GeForce 9800 GX2	Radeon HD 4870 X2
Technical process	55 nm TSMC	65 nm TSMC	65 nm TSMC	65 nm TSMC	55 nm TSMC
Stream Processors	480	240	216	256	1600
Core frequency	576 MHz	602 MHz	576 MHz	600 MHz	750 MHz
Shader frequency	1242 MHz	1296 MHz	1242 MHz	1500 MHz	750 MHz
Memory frequency (effective)	1998 MHz	2214 MHz	1998 MHz	2000 MHz	3600 MHz
Memory size	1792 MB	1 GB	896 MB	1 GB	2 GB
Memory bus width	448 bit x 2	512 bit	448 bit	256 bit x 2	256 bit x 2
ROP	56	32	28	32	32
Price	$ 499 (rec.)	~ $ 380 (from 12 thousand rubles)	~ $ 230 (from 8 thousand rubles)	From 9 thousand rubles	~ $ 500 (from 14 thousand rubles)

Two GPUs, one video card

At first glance, a GeForce GTX 295 looks like a 280 or 260. If you turn the card over and look at the back panel, you will see only one GPU. Like the 9800 GX2 and 7950 GX2 that came out earlier, the video card actually consists of two printed circuit boards, which are located in the form of a "sandwich", with a special heatsink with a fan in between, which is responsible for cooling. The two boards are connected by an SLI cable and are enclosed in a protective casing.

Click on the picture to enlarge.

It is only natural that the cooler design was specially adapted to cool two PCBs. You can see that both boards have holes through which air is sucked in. The board takes up two slots, so it is no wider than other single-chip nVidia offerings. In addition, the board is the same length as the GeForce GTX 280 (and AMD Radeon HD 4870 X2).

Nvidia is not yet ready to compete in power consumption with the GTX 295 with AMD's solution - so we were not allowed to provide a numerical value for power consumption. However, the card uses less power (idle and under load) than the Radeon HD 4870 X2. The following graph shows the power consumption of the system from the wall outlet.

Click on the picture to enlarge.

On paper, the GeForce GTX 295 uses a thermal design power (TDP) of 289 watts. The 4870 X2 has a thermal package of 286 W. When we measured the system's power draw from the wall, the GTX 295 consumed 10 watts less when idle than the AMD board. When running the Far Cry 2 benchmark in a 2560x1600 loop with anti-aliasing (AA) and anisotropic filtering (AF), the nVidia board consumed a whopping 50 watts less on average.

Of course, we should wait until early January when fan speeds are determined and power consumption is balanced. But we can already say that the transition to the 55nm process technology has a good effect on the GeForce GTX 295, despite the massive dimensions of the GPU GT200.

Test configuration

Our test configuration corresponds to a system for enthusiasts, in which it is not a shame to install a video card for $ 500 - it can be either a system with a top-end i7 965 processor or an overclocked i7 920. In any case, the platform coped well with the load of all four of our test video card configurations ... Understandably, if you are planning to upgrade to a 4-way SLI configuration (which the GTX 295 supports) or 4-way CrossFireX, then you need to pay close attention to CPU performance.

But today we are interested in the fastest single video card, prepared for 2009.

Hardware configuration
CPU	Intel Core i7 965 Extreme (3.2 GHz)
Motherboard	Asus Rampage II Extreme
Memory	6 GB DDR3-1333 7-7-7 (three channels)
Storage device	Seagate 250 GB Barracuda 7200.10 7200 rpm
Optical drive	Lite-On DH-4O1S BD-ROM
Power Supply	Cooler Master UCP 1100W
Video cards	Nvidia GeForce GTX 295 1.8 GB Nvidia GeForce GTX 280 1GB AMD Radeon HD 4870 X2 2 GB AMD Radeon HD 4870 512 MB

Again, we used the 64-bit version of Vista, which supports a full 6GB of system memory.

We used the following games for our tests.

Tests and settings

Tests and settings
	Very High Quality Settings, No AA / No AF, vsync off, 1900x1200 / 2560x1600, Patch 1.2.1, DirectX 10, 64-bit executable, benchmark tool
	Very High Quality Settings, 4x AA / 8x AF, vsync off, 1900x1200 / 2560x1600, Patch 1.2.1, DirectX 10, 64-bit executable, benchmark tool
	Highest Quality Settings, No AA / No AF, vsync off, 1920x1200 / 2560x1600, Patch 1.1, FRAPS / saved game
	Highest Quality Settings, 4x AA / 8x AF, vsync off, 1920x1200 / 2560x1600, Patch 1.1, FRAPS / saved game
	Highest Quality Settings, No AA / No AF forced in drivers, vsync off, 1920x1200 / 2560x1600, FRAPS / saved game
	Highest Quality Settings, 4x AA / 8x AF forced in drivers, vsync off, 1920x1200 / 2560x1600, FRAPS / saved game
	Ultra Quality Settings, No AA / No AF, vsync off, 1920x1200 / 2560x1600, Patch 1.0.0.15, FRAPS / saved game
	Ultra Quality Settings, 4x AA / 8x AF, vsync off, 1920x1200 / 2560x1600, Patch 1.0.0.15, FRAPS / saved game
	Very High Quality Settings, No AA / No AF, vsync off, 1920x1200 / 2560x1600, DirectX 10, Steam Version, in-game benchmark
	Very High Quality Settings, 4x AA / 8x AF, vsync off, 1920x1200 / 2560x1600, DirectX 10, Steam Version, in-game benchmark
	Highest Quality Settings, No AA / No AF, vsync off, 1920x1200 / 2560x1600, timedemo
	Highest Quality Settings, 4x AA / 8x AF, vsync off, 1920x1200 / 2560x1600, timedemo

Test results

We started our tests with Crysis, which supports 64-bit mode, which is one of the most demanding games on the market. In all cases except one, the GTX 295 was able to bypass the AMD Radeon HD 4870 X2. At 2560x1600 with anti-aliasing (AA) and anisotropic filtering (AF), nVidia admits there is a bug in the beta driver that negatively impacts performance.

Interestingly, the 512MB Radeon HD 4870 comes pretty close to the GTX 280 without AA and AF enabled. However, as soon as you turn on the options to improve the picture quality, half the memory size starts to slow down performance.

Once again, the GeForce GTX 295 demonstrates a solid lead, which is most evident at 2560x1600 with active AA and AF. If the performance results of the final version of the video card are the same (and at the price of $ 499 there is no need to wait for another), nVidia has a serious advantage over AMD graphics in this first-person shooter with the theme of World War II.

In this game, unlike Crysis, the GTX 280 was noticeably faster than the 512 MB Radeon HD 4870. However, the 1 GB version would certainly have performed better, although it would cost $ 40 more.

Dead Space - fresh "horror", transferred from the world of consoles to the PC. And, unfortunately, this negatively affected the quality of the graphics.

Although we set the indicated modes of full-screen anti-aliasing (which led to a corresponding drop in performance, that is, AA and AF were executed in the driver), we still noticed obvious "ladders" even at 1920x1200 and 2560x1600. Additionally, the Radeon HD 4870 X2 had strange but repeatable behavior at 1900x1200, where results were lower than 2560x1600.

This game clearly prefers Nvidia graphics cards, as even the GTX 280 outperforms the AMD flagship. However, this is not a game that we would add to our usual test suite, if only for the fact that it is dubiously affected by the options for improving the picture quality.

We discussed the results with Nvidia. Apparently, Dead Space has its own anti-aliasing algorithm, so enabling AA in the driver panel did not affect the game. You can enable AA in the game itself, but this does not affect the performance of the test in any way.

Fallout 3 is very entertaining, but performance seems to be limited by the platform. Regardless of enabling anti-aliasing and anisotropic filtering, the GeForce GTX 295 and Radeon HD 4870 X2 performed almost the same at 1920x1200 and 2560x1600. It is quite obvious that the systems here provide a lot of graphical power that has yet to be revealed. It wasn't until we turned on anti-aliasing and filtering (AA / AF) on the GTX 280 at 2560x1500 that we noticed a drop in performance. And the HD 4870 is also affected by the memory limitation.

AMD has been battling Far Cry 2 for the past month. It looks like the Catalyst 8.12 driver has improved the situation. Without anti-aliasing, the 4870 X2 wins at 1900x1200 and 2560x1600. Enabling the quality enhancement options results in nVidia taking the lead in both resolutions.

Since the game is based on the Valve Source engine, we were pretty sure Left 4 Dead would give AMD graphics cards an edge. But at every resolution, with AA / AF on or off, the GeForce GTX 295 came out ahead of the AMD Radeon HD 4870 X2. Considering the price difference between the GTX 280 and 4870 512 MB, the results for this pair are not surprising. However, these cards are still a good starting point for those looking to upgrade.

2009 at a glance

As for the tests carried out, I would like to say the following. To begin with, the review and results are preliminary. We will release a full review in 2009, when the final version of the GeForce GTX 295 hits the market and the boxes with this video card appear on sale.

Here we have presented the results of six games. Five of these are cited by nVidia as examples of the most anticipated games for the 2008 New Year season. Four of these five games participated in nVidia's "The Way It" s Meant to Be Played. Two games were already included in our traditional test suite. And we added Crysis to this mix. But the graphics card is still an engineering sample, and according to the data nVidia hasn't set final fan speeds yet.

Why did we decide to release a test like this? All of the games selected are in fact popular, and we understand how tired our readers are of seeing games as young as three years old are being used because they have become recognized performance metrics. Therefore, we decided to take a breath of fresh air instead of the annoying Supreme Commander or World in Conflict.

Of course, TWIMTBP is nothing more than nVidia's competitive edge. The relationship with the developers so that they can optimize their software for "hardware" is quite logical and understandable, if the relationship does not lead to infringement of performance with the competitor's video cards. Considering the number of times AMD graphics cards beat nVidia models in 2008, this developer relationship program doesn't pursue such vile goals. In fact, we were recently able to take a closer look at how nVidia contributed to the development of Far Cry 2 as part of TWIMTBP. nVidia spent three person-months (if I may say so) on-site consultations and performance tests, including regular driver updates. As a result, the GeForce GTX video cards did not have as many problems as the Radeon models (we described these problems during tests of games on a Core i7 system with configurations up to 4-way CrossFire and 3-way SLI).

Conclusion

At the beginning of the article, we asked ourselves, can nVidia be able to refute AMD's claim of the fastest stand-alone graphics card on the market? Overall, the GeForce GTX 295 was able to do just that, as hinted at in preliminary benchmarks based on some of the hottest games of the season.

According to nVidia, the GeForce GTX 295 will be released at CES in 2009, that is, in a few weeks. It will be sold at a suggested price of $ 499 - which roughly corresponds to the Radeon HD 4870 X2 - and will be available for sale at the time of the announcement (we cannot say about Russia). When we get our hands on the final retail sample of the video card, and not an early engineering sample, we can conduct thorough tests and more accurately assess the performance.

What do we know at the moment? The nVidia GeForce GTX 295 is a really fast graphics card. We know the company has moved to a 55nm process that has reduced power consumption - and the GTX 295 consumes less power when idle and under load than the AMD Radeon HD 4870 X2.

However, the GeForce GTX 295, like all other upcoming 55nm GPUs from nVidia, has other trump cards up its sleeve, not just performance numbers. This includes all the talk about CUDA, PhysX and the soon-to-be-released 3D stereo technology, which we already got to know at NVISION.

We did not pay enough attention to PhysX support in our article, as the list of supported games does not overlap with the list of the best games. We did not talk about CUDA either, because, apart from Badaboom, today there are too few applications for which it would be worth buying a video card with this technology. Of course, we were impressed with the stories of five or six developers about how CUDA can revolutionize development in their niche areas with Tesla cards, but transferring this advantage to desktop applications is problematic.

But in this respect, the situation is slowly but surely changing. EA should catch up, and in early 2009 a couple of PhysX-enabled games should be released, which will allow us to evaluate how this function justifies itself. We tried out Left 4 Dead (remember, it's not included in the TWIMTBP program) with the upcoming nVidia 3D glasses and the gameplay is absolutely amazing. But this topic is for another article, which will also be published soon.

Today, few people remember about the GTX 295 video card. Its characteristics are currently useless. She can hardly run even the most undemanding game. However, it can still be found for sale on forums and marketplaces. What is remarkable about it for customers?

Thorny path

The specs of the GTX 295 have been carefully designed. This approach of Nvidia was caused by unexpected "attacks" from the main competitor - ATI. The developers of the "red" decided to release a new architecture, which turned out to be in demand and popular among buyers. Nvidia had to solve this problem soon and bring a new leader to the market.

The first attempt to defeat the annoying competitor was the release of the G200 architecture paired with a video card. Novinka managed to show an excellent result in tests and acquire a lower power consumption level. Nvidia was on track to not only create an equal rival to the new Radeon HD 4870 X2, but also to take over the title of the fastest graphics accelerator.

Decision

Before the characteristics of the GTX 295 became known to the world, it was necessary to develop a strategy. Nvidia decided to build a dual-socket adapter based on a previously released architecture. Many experts were initially surprised by the manufacturer's decision. Previously, Nvidia announced that it is ready to develop only single-processor models with powerful hardware.

The company was prompted to such deviations from its principles by an ardent attack of a competitor. In addition, the development of a new architecture, which worked on the 55-nm process technology, largely predetermined further actions. Indeed, unlike 65nm, a pair of these chips produced much less heat. Although there were skeptics on this score who considered such a decision a failure.

Comparison

After the release of the Nvidia GTX 295, its performance was compared to the capabilities of the Radeon HD 4870 X2. It is for the sake of superiority over the ATI model that Nvidia has done such a difficult path.

Both models run on 55nm technology. Graphics cards look good, but still fall within their specific category. For example, X2 is good because it has high computing power. The Nvidia GTX 295 has much better texture and bitmap performance. The novelty has acquired a high frequency of execution units, support for hardware acceleration.

ATI's model turned out to be better due to the volume of local video memory. The GTX 295's performance is in line with the company's strategy. The developers wanted to create the fastest gaming graphics card, so they missed out on DirectX 10.1 support, integrated sound engine, decoder, etc.

Modifications

Like all video cards from Nvidia, this one received a reference model and modifications. The most interesting thing is that, in fact, none of the now well-known manufacturing companies took up the modification of the novelty. As a result, there were only a few improved versions on the market from EVGA, Zotac, Inno3D, XFX GF. In general, all these GTX 295 video cards received the same characteristics, but the difference was still noticeable.

EVGA

One of the most popular modifications was the EVGA GeForce GTX 295. The characteristics of the video card did not differ much from the parameters of the reference model. Still, some differences could be noticed.

Outwardly, the model seems very cumbersome. But there is nothing surprising in this, since the manufacturer had to place two large chips and the layout of a pair of 448-bit buses on the board. Both boards inside the video card run on a single cooling system.

The option, on the one hand, seems to be quite good, on the other, it raises many questions, since the 55-nm chip is still quite warm. In order for the adapter to be not too large and in the future could easily be connected to an ATX case, both motherboards had to be placed face to face. The developers have tried to keep the slots and connectors in such a way that the cooling system can remove hot air from the case.

EVGA GeForce GTX 295 cards operate with an independent four-phase power supply. However, only the top is connected to the power supply and is rated for loads up to 150W.

14 GDDR3 512 Mbit chips are responsible for the memory on the boards. The microcircuits operate with a voltage of 2.05 V and a frequency of 1000 MHz. Such indicators of the reference model in the EVGA GeForce GTX 295 were slightly improved: the frequency was raised to 1026 MHz.

Since we have a dual-processor video card, it houses two GPUs. Each works with 240 shaders and 80 texture processors. In the reference model, the chip operates at 576 MHz in the main domain and 1242 MHz in the shader domain. EVGA was able to overclock frequencies to 594 and 1296 MHz, respectively.

Pros and cons of EVGA GeForce GTX 295

The modified version is practically no different from the reference version, and this is perhaps the main disadvantage. Today companies indulge us with unprecedented factory overclocks and the ability to increase frequencies at home. In the case of this modification, no particular improvements were noticeable.

A good clone of the reference version appeared before the buyer. There is no need to expect a unique design, since dual-processor models are a priori difficult to change visually. But EVGA tried to independently increase the frequencies of the memory and processors. In 2009, users were convinced that this is the best graphics card for gaming.

Zotac

The modification from this manufacturer was released in a branded box, which was no different from those released earlier. The company does not change its signature black and gold combinations to this day. The box contained a wealth of information and drawings. The main characteristics of the Zotac GeForce GTX 295 were also written right there.

The design of the modification is no different from the previous model and the reference version. Before us are all the same two boards, assembled "face to face", which work with a single cooling system. The latter, in turn, has several copper heat collectors and heat pipes.

Due to the fact that there is not much space on the boards, it was decided to reduce the memory bus to 448 bits. This decision also affected the rasterization channels. But the developer managed to keep the number of unified processors - 240. The GPU runs at 576 MHz, and the shader pipelines - 1242 MHz. The figures do not differ from those presented by Nvidia in the reference version.

The video memory is filled with microcircuits with a total volume of 1792 MB. They operate with a voltage of 2.05 V at a frequency of 2 GHz.

To improve the 295, the specialists tried to overclock the processor and memory frequencies. Since the model can hardly be called cold, and with increasing voltage and frequencies, there is always a jump in temperature, it was necessary to provide stable cooling. Someone just opened the case for this so that there was an air flow inside, someone installed an additional cooler to increase the efficiency of CO.

The specialists managed to raise the main speed indicator to 670 MHz and the shader frequency to 1440 MHz. The growth averaged 15%, which is actually an excellent result. The video memory was overclocked to 2214 MHz.

Pros and cons of Zotac GeForce GTX 295

Unlike the EVGA modification, this model went on sale with benchmarks. The only difference is a modified plastic casing that covers the boards and a new cooling system. Otherwise, we have a reference model with the characteristics set by Nvidia.

The 295 from Zotac was not the most popular. Of course, buyers would love to see factory overclocking and surprises in the box. Instead, they received the usual configuration and standard parameters. Although the possibility of manual overclocking is a definite plus. And the overclocking result turned out to be pretty good.

Other modifications

Compared to its competitors, the modification from Inno3D looks good. The first thing worth highlighting is the well-thought-out cooling system. The board contains an aluminum plate with cutouts for the chips. On top of them were placed two radiators, between which there is one compact fan.

The amount of memory is 1792 MB, and the frequency at which it operates is 2016 MHz. The processor speed turned out to be the reference. But overclocking helped to raise this figure to 684 MHz. We managed to increase the memory frequency to 2376 MHz.

The XFX GF GTX295 576M 1792MB modification also did not receive any special parameters that could stand out from the rest. The only thing the developer provided in the package was Far Cry 2, which turned out to be undoubtedly a profitable marketing ploy.

conclusions

Many users believed that there were 295 1860 Mb. The characteristics of this model cannot be found on the Internet, since such a version did not exist on sale. GTX 295 in any modification was produced with a memory capacity of 1792 MB and a frequency of 2 GHz.

Despite the first attempts to launch a dual-processor model with two boards on the market, within a few months, variations appeared with a compact design and an efficient cooling system. And, nevertheless, this did not affect the characteristics of the model in any way. She immediately took her niche in the hi-end class, fulfilling its main task: to defeat the main competitor - Radeon HD 4870 X2.

Of all the modifications, the EVGA GeForce GTX 295 turned out to be the most successful. This version of the video card became the most popular, since it had factory overclocking already from the box. And if now many buyers are focusing on purchasing a version with overclocking, then this, undoubtedly, was a huge advantage.

And although in fact the factory overclocking did not give a huge increase in performance, as is the case with modern models, a slight increase still affected the overall system power. And if the EVGA GeForce GTX 295 fell into the hands of a specialist, then at home the processor frequency could be increased even more, and, accordingly, the gaming potential of the accelerator could be increased.

At the beginning of this year, as part of an international exhibition CES 2009 company NVIDIA presented two newest accelerators, which to this day are the most productive in their classes. Of course, we are talking about a single-chip GeForce GTX 285 and a dual-processor flagship GeForce GTX 295, which have become an excellent tool in the competition against ATI, whose four thousandth series of accelerators turned out to be extremely successful, having managed to regain the confidence of users. Nevertheless, the green giant was able to find the strength and respond to his eternal rival. Since the official release GeForce GTX 295 almost half a year has already passed, and the world wide web has long been filled with dozens of variegated reviews and tests published mainly in the middle of winter, because it was then that the overwhelming number of samples managed to get to test laboratories, therefore, almost everything is known about this video card. But I would like to return to this issue again, even if after such a long period of time. Cause? It is actually trivial - these are drivers that play an important role in the overall performance of a video card. ATI, whose latest software has become a real gift for overclockers and extreme lovers, was noted for its competent software optimization, which means that it makes sense to revise the results of previous tests. So, we have before us Gainward GTX 295 and our task is quite simple - to determine if there is still GeForce GTX 295 the fastest or still lost the palm Radeon HD 4870X2?

Introduction

Quite recently, the flagship from NVIDIA has undergone a number of design changes, as a result of which the video card has "lost weight" by exactly one plate, and the second actually accommodates all the necessary components. This step was justified, first of all, from an economic point of view, which in turn should help vendors to reduce production costs and offer a new product at a more favorable price - and this is already a significant trump card in the price war with AMD. The two-slot turbine has given way to a complex two-piece radiator with a single fan in the center, but the thickness of the accelerator remains the same and occupies the same two slots. At the time of this writing, "single-deck" versions of accelerators did not have time to reach retail stores, and moreover, according to our information, the vast majority of vendors are experiencing difficulties in having the necessary PCBs, which means that the first batches will not bring the expected price reduction. Sooner or later this situation will normalize and we hope we will have the opportunity to get acquainted with the new revision of the accelerator.

Design and packaging

Before starting to study the instance, I would like to briefly recall its main technical characteristics:

• Two graphics chips GT200b (55 nm);
• 240 stream processors per core;
• "Dual" 448-bit bus;
• 1792 MB of GDDR3 memory;
• Support for DirectX 10.0, Shader Model 4.0, CUDA, PhysX technologies.

NVIDIA engineers decided not to reinvent the wheel and used a well-developed scheme for constructing a dual-GPU video card, copying the design from the GeForce 9800 GX2. All microprocessors, memory chips, power elements and key controllers are soldered on the internal parts of the PCB, protected by a massive cooling system. We will be able to see the layout of the elements a little later, but first, let's examine the accelerator's appearance, packaging and delivery set.

The video card came to us in the traditional Gainward packaging - a medium-sized cardboard box with the branding for its own GeForce series - a winged spirit depicted against the backdrop of a mythical dilapidated city:

On the reverse side of the package you can see the advantages of new technologies from NVIDIA, a brief description of which is present in almost any vendor - apparently this is one of the requirements of the green giant that must be met.

Inside we find the accelerator itself, sealed in an antistatic package, an adapter from a DVI digital output to an analog D-Sub, an S / PDIF cable, a disk with drivers paired with the proprietary ExperTool overclocking utility, as well as a short instruction manual. The photo does not include two adapters from a standard PATA connector to a 6-pin and 8-pin PCIe connector, but they are also included in the bundle. An extremely modest set, but at the same time no frills, otherwise the already considerable cost of the product would have been even higher.

And here is the video card itself. As befits status things, it has a lot of weight, which is required by the installed cooling system. The metal casing is rubberized, although you can't immediately make out what kind of material was applied to the accelerator cover. Since we have a completely reference model in front of us, in order to stand out from the “crowd” of such devices, the manufacturer placed a branded sticker with its name on the left side of the card.

The back side does not contain critically heated elements and therefore is not protected by a stiffening plate and, by compatibility, a heat collector, which would definitely not interfere there. A yellow triangle indicates high heating of the board - another warning from Gainward, which should not be ignored. For example, in the assortment of the company EVGA there is a proprietary backplate for an additional heat sink, although it is sold separately. However, it can be easily installed on any two-plate GeForce GTX 295.

During operation, part of the heated air will remain inside the case (if, of course, you have a case), but also some of its mass will be removed outside of it, for which special slots are provided on the back plug. An HDMI connector is located next to them, and on the floor below there are two DVI ports, therefore, a total of the video signal can be output to three displays. There are also two LEDs signaling the normal or unstable operation of the video adapter.

Now let's take a look at our device. I would like to note right away that this must be done with extreme caution, since there is a possibility of damage to the fragile crystals of some chips and this must be remembered. To dismantle the cooling system, you only need one screwdriver and a hexagonal nozzle to remove the DVI port mount, and if you can not bother with the quality of the nozzle, then there may be options with a screwdriver. I would like to pay tribute to the assemblers - all the screws were tightened wonderfully, which caused a few problems. Prior to that, the author of the article used a simple Chinese screwdriver (about 30 cents), but its crosspiece was safely erased on the very first screw. Then a more expensive Polish instrument was purchased (as much as a whole one and a half dollars), which was enough for three screws - this time the tip crumbled. Surprisingly, the number three screwdriver (even more expensive) did not fail, saving some of my money. We will get acquainted with the cooling system later, but for now the following photo:

Two halves of the fastest video card to date. The assembly of all the elements is rather tight and one can only admire the NVIDIA engineers who managed to "squeeze" everything necessary onto one PCB.

Each graphics chip is labeled G200-400-B3, i.e. now we have a 55nm G200b with 240 shader processors.

The required amount of memory (1792 MB) is made up of 14 Hynix microcircuits with the HR5S5223CFR-N0C marking corresponding to the GDDR3 standard with an effective frequency of 2000 MHz.

Both boards are interconnected by two flexible cables of different lengths, so it will be difficult to confuse them, especially since each of them has a special key indicating the orientation in the connector. To the right of the connectors you can see the nForce 200 chip, which is used to organize the internal SLI mode.

To power the accelerator, there are two PCIe connectors for 8 and 6 pins located on different PCBs. According to the developers, during the maximum load, the accelerator can consume up to 289 W, while it is recommended to use a power supply unit with a power of more than 680 W. In practice, this number is an order of magnitude overestimated, and on average, 220-230 watts are enough for a card. For example, in our test bench, a 550 W power supply was installed and during active operation neither it nor the accelerator showed signs of emergency operation, therefore, NVIDIA is reinsured in this way. Now about the cooling system.

The G200b graphics chip contacts the heat spreader through a layer of rather thick gray thermal paste, while memory chips and other power elements transfer heat through special fiber pads. The weight of the radiator is quite impressive, in principle, like any top accelerator. Before assembling, do not forget to replace the thermal interface with a GPU, this is not critical for the rest of the components.

At the moment, there is not a single alternative air cooler on the market for the reference GeForce GTX 295, which is due to the complex design of each of the PCBs, and it is unlikely that manufacturers will be able to offer a new solution, especially if we take into account the fact that the production of dual-plate video cards is discontinued. Only the owners of CBOs can have a reason for joy - there is just a gigantic mass of compatible water blocks from almost all leading companies in this industry on the market.

This concludes the visual inspection and proceeds to the practical part, checking the performance level in popular test suites and modern video games. The opponent was chosen Radeon HD 4870X2(our copy was submitted by development from Gigabyte), introduced last summer and held the leading position until early 2009.

Whether the new ATI drivers helped outstripping the accelerator from NVIDIA or, on the contrary, threw it back, we now have to find out.

Test system configuration and software used

The following set of components and devices was used as the main test bench:

• Intel Core 2 Duo E8500 (E0) processor;
• Asus Maximus II Formula motherboard (BIOS 2104);
• RAM Mushkin Redline XP2-8000 (2 * 2 GB);
• Sumsung HD103UJ hard drive (1TB capacity);
• Be quiet! Power supply Straight Power BQT E5-550W;
• Cooler Master Cosmos S.

The processor was cooled by a closed-loop liquid cooling system. The room temperature did not exceed 24 degrees Celsius. The CPU frequency was boosted to 4.5 GHz and did not change throughout the test. Memory timings were set as 5-5-5-12.

Software:

A 32-bit operating system was taken as a basis Microsoft Windows Vista SP1... The latest driver packages were also installed: ForceWare 185.85 WHQL for video card NVIDIA and Catalyst 9.6 for the accelerator from ATI... All settings were used by default.

Benchmarks:

Games:

Crysis warhead
Far cry 2
World in Conflict
S.T.A.L.K.E.R .: Clear Sky
Devil may cry 4

The first thing I wanted to check was the level of heat dissipation, for which the utility was launched FurMark v1.6.5, able to warm up the video adapter as much as possible. The accelerator cooling system responds rather quickly to the temperature rise and after a couple of minutes the turbine noise becomes annoying. The heating level did not exceed 92 degrees Celsius. Well, quite an expected result - you have to pay for the record speed with decent noise and considerable temperature.

Let's move on to synthetic tests. Let's start with 3DMark06 from Futuremark... All package settings were used by default. First, the results of two accelerators in denomination:

Then with increased clock speeds:

We perform a similar procedure with 3DMark Vantage by determining the speed of video cards with factory settings:

... and in forced mode:

Let's summarize the obtained data in two small tables for clarity.

In 3DMark06, there is an obvious advantage of the Radeon HD 4870X2, which even outperforms its rival in nominal value, and increases its lead with increasing clock frequencies. The situation is completely different with 3DMark Vantage. The accelerator from ATI lagged behind the GeForce GTX 295 almost several times, losing ground almost without a fight. Overclocking also did not help much, which cannot be said about NVIDIA - the video card based on its chips showed a much better result.

Games. We have selected five popular video games, most of them very demanding on system resources and, in particular, on the PC graphics subsystem.

Crysis warhead

Undoubtedly, a gaming masterpiece, at maximum settings, capable of bringing even the most progressive computers to their knees. To determine the performance, the utility was used Crysis WARHEAD Benchmark Tool with picture quality settings options Enthusiast... All game applications took place in the environment DX 10.

The leadership of the video card from Gainward is difficult to dispute, in all test modes the NVIDIA flagship confidently bypassed the Radeon HD 4870X2, yielding only 1-2 “minimum” frames in 1280x1024 both with and without anti-aliasing / anisotropic filtering, which is very insignificant and not significant competitive advantage. We also note that the results of the overclocked GeForce GTX 295 - the increase in its performance is an order of magnitude better in any of the modes.

Far cry 2

This game has a built-in benchmark and we just have to determine the leader in this application. The quality settings were set to Ultra High, and the script scene was "run" three times, after which the result was averaged, and the system returned the value of the minimum, average and maximum FPS. In this case, we were interested in the smallest and largest readings, which we entered into the corresponding tables:

And the result was really interesting. The ATI veteran was able to outperform even the overclocked NVIDIA video card at its nominal frequencies. This suggests, perhaps, a premature conclusion: the top GeForce GTX 295 will not dominate in all gaming applications, and the Canadian developer has something to put in opposition to its eternal competitor. It is also worth paying attention to the number of frames per second at different resolutions and detail settings - they are almost equal and the difference is from tenths to 6 frames per second. Unfortunately, we were not able to test the accelerators on a monitor with a diagonal of more than 22 inches, which definitely led to a decrease in FPS and most likely we would have been able to determine the test winner for sure.

World in Conflict

The maximum possible settings and the integrated test of the video adapter performance should show the advantage of one of the dual-processor video cards.

In almost every resolution, the maximum number of FPS varied at the level of 61-63 frames per second, regardless of the installed accelerator, which cannot be said about its minimum values. The new product from Gainward was able to demonstrate 30 fps at best, while for the Radeon HD 4870X2 the worst achievement can be considered 52 fps. Note that ATI's video card feels much more comfortable at 1680x1050 than in 1280x1024 even with anti-aliasing and anisotropic filtering enabled, although theoretically it should be the other way around. Nevertheless, one more round can be given to Canadian developers, their drivers are really at their best.

S.T.A.L.K.E.R .: Clear Sky

A special benchmark is available on the official website of this cult game for free download, with which you can find out how quickly your video card will draw the endless landscapes of the virtual Chernobyl zone and even help you compare the result with data from the constantly emerging online database. The fourth scene was taken as a basis, called "Sun Shafts" that loads the video subsystem as much as possible. For the ATI accelerator, the mode was activated DirectX 10.1.

What do we see? And here we see the complete dominance of the GeForce GTX 295 in all resolutions and anti-aliasing modes. The minimum frame rate can drop to an uncomfortable 17 frames per second and this trend will continue if you increase the monitor diagonal, say to 24 inches with a resolution of 1920x1200 pixels. To avoid this, you can slightly weaken the quality settings, which will help the accelerators to produce a more acceptable amount of FPS.

Devil may cry 4

The last for today will be a three-dimensional third-person shooter Devil May Cry 4. In the benchmark we needed, the Super High mode was activated and the required anti-aliasing and filtering values ​​were set. It consists of four script scenes, upon completion of which a table is formed from the averaged FPS values ​​of each of them. The only drawback is the inability to find out the average and maximum number of frames per second. The fourth scene was taken as a basis, which was painfully saturated with various freaks and other evil spirits, for the drawing of which considerable resources of the tested video cards were required.

The speed of the GeForce GTX 295 is again at its best and does not allow the Radeon HD 4870X2 to snatch victory in any of the modes, doing this even with factory settings against the increased frequencies of the opponent in the person of ATI. It's nice to see that the "playability" is at the proper level and does not fall below 160 frames per second. Both cards responded well to overclocking, especially the accelerator based on NVIDIA chips.

Conclusion

At the beginning of our review, we asked a simple question - is the GeForce GTX 295 the fastest graphics card for desktop PCs today, or has ATI made an incredible leap forward and managed to get ahead of the leader? Oddly enough, it is still very difficult to give a clear answer. The dual-GPU flagship NVIDIA is very strong in synthetic 3DMark Vantage, but is inferior in 3DMark06, which is more popular among overclockers, in which the Radeon HD 4870X2 is the undisputed leader in this situation. Game tests also did not reveal the favorite of the competition, but most of them were won by the development of the green giant. We note the excellent level of software optimization for video cards, especially the latest versions of Catalyst drivers, which forced many enthusiasts to revise the test results, and extreme people to update the results on the bot. I would also like to say a few words about noise. Frankly speaking, its level is poor. Top accelerators with record speed and no less record heat generation make the turbines rotate at more than 2000 rpm, and this is already comparable to the sound of a small vacuum cleaner. In our subjective opinion, ATI's accelerator is somewhat quieter in 2D, but it becomes more noisy in tests and video games.

Benefits :

• Excellent overclocking potential;
• Excellent performance;
• Relatively low power consumption for a top accelerator;
• Good driver optimization.

Disadvantages:

• High heating level, but still lower than that of the ATI card;
• Noise under heavy load;
• Modest delivery set.

Yes, the cost of a new product from NVIDIA can scare away the most ardent adherents of the company, since not everyone can afford to give about five hundred US dollars. But if you need record speed and uncompromising performance, then this is your video card - it will justify every cent spent on it, but do not forget that the rest of the components (processor, motherboard, RAM) must also correspond to the proper level. We will continue to follow the development of the modern accelerator market and will try to promptly inform our readers about the vectors of its development and priorities. Stay tuned - the most interesting is yet to come! :)

Site administrationwebsite expresses gratitude to the following companies for the equipment provided for testing: