The Two Tigers: Asus EAH3870 TOP/G/HTDI/512M and EAH3850 TOP/G/HTDI/256M

Both leading graphics processor developers have become very tolerant to the fact that graphics card makers raise the clock speeds of their solutions above the nominal values in order to diversify their product range. Today we are going to tell you about two graphics cards from ATI Radeon HD 3800 family from Asustek Computer and will try to find out if the manufacturer can really benefit from increasing the nominal clock speeds of 3870 and 3850 models.

by Alexey Stepin , Yaroslav Lyssenko
01/29/2008 | 03:47 PM

Factory overclocking of graphics cards is a widespread practice. It is one of the ways for the maker to add originality to its product and set it apart from the competitors’ offers especially when it comes to graphics cards that have the reference PCB design and cooler. Such solutions may provide hefty performance benefits like with the GeForce 8800 GT 512MB or give little to no advantage like with the GeForce 8600 GT/GTS – it all depends on the degree of overclocking, the architecture of the GPU, and the parameters of the memory subsystem.

 

We have reviewed a number of such cards recently, all of them based on Nvidia’s GPUs, but AMD’s graphics department, the former ATI Technologies, currently offers two interesting products based on the new 55nm RV670 core, ATI Radeon HD 3870 and Radeon HD 3850. These cards showed good performance in most of modern games according to our tests. Being economical, quiet and inexpensive ($219 for the senior model and $179 for the junior one), they appeal to every gamer who cannot afford a top-end graphics card. Pre-overclocked versions of Radeon HD 3800 cards were sure to come out as well.

An interesting fact, the Radeon HD 3870 card had been originally expected to have a GPU clock rate of 825MHz and a memory clock rate of 1200 (2400) MHz but these parameters were lowered later, probably to keep its power consumption, heat dissipation and price within reasonable limits. Another fact is that notwithstanding only one explicitly declared frequency zone the previous graphics core from ATI had as many as 26 internal frequency zones, which largely accounted for the small performance gain the ATI Radeon HD 2900 XT 1GB GDDR4 provided over the ATI Radeon HD 2900 XT 512MB GDDR3. We don’t have reliable information about the internal structure of the R600 at such a deep level but it is logical to presume that the RV670 has multiple frequency zones as well. So the reduction of the official GPU frequency of the Radeon HD 3870 helped reduce the heat dissipation while sacrificing little in terms of performance and it also means that we can’t expect pre-overclocked versions of RV670-based cards to be very fast.

ASUS, the major graphics card supplier, has ventured to release pre-overclocked Radeon HD 3800 cards, but doesn’t promise much in terms of speed: a 4% performance growth over the reference card for the EAH3870 TOP/G/HTDI/512M and an 8% growth for the EAH3850 TOP/G/HTDI/256M. We’ve got these cards in our lab now and are about to check out if this promise is too pessimistic or maybe otherwise. Let’s take a look at the product package first.

ASUS EAH3870 TOP and EAH3850 TOP: Package and Accessories

The EAH3870 TOP and EAH3850 TOP come in identical boxes with identical accessories. The box is traditionally huge but offers a handle for your convenience.

 

The design is identical for both boxes. The main theme is related to the game included with the cards, Company of Heroes: Opposing Fronts. This is a top-class real-time strategy we use in our tests as a benchmark. The per-overclocked frequencies of the cards are alluded to by the Extra Performance sticker with a stylized performance chart. The EAH3870 TOP box additionally has a fiery caption “DDR4”.

The packaging is top quality with the contents carefully placed in individual compartments and the cards fixed in foam-rubber trays to be protected against any damage during transportation and storage. The ASUS EAH3870 TOP and EAH3850 TOP come with the following accessories:

There is a full selection of adapters, connectors and documentation including even such a trifle as an S-Video → Composite adapter. A full version of the user manual is provided on an individual disc. Proving its reputation of the supplier of top-class solutions, ASUS included a small but useful CD case. And of course we should mention the copy of Company of Heroes: Opposing Fronts because few manufacturers include bestselling games with their products.

The driver CD contains version 8.43 of ASUS’ exclusive driver based on Catalyst version 7.11. You may want to install the latest official Catalyst, which is version 8.1, instead. Among ASUS’ software the SmartDoctor program, an overclocking and monitoring tool, can be mentioned. The GamerOSD utility is similar to Fraps in functionality. It reports performance data and allows making screenshots or record gameplay videos as well as adjusting image parameters such as gamma, brightness and contrast. This is quite a useful tool considering that Fraps is not freeware.

So the two described graphics cards from ASUS get our praise for the packaging and accessories. This has become a tradition for ASUS’ products, by the way.

ASUS EAH3870 TOP: Design and Specifications

Although belonging to the TOP series, this model of the EAH3870 is a copy of the reference card with a standard cooler. The difference boils down to the stickers: “Company of Heroes: Opposing Fronts” on the cooler casing and “ASUS” on the fan:

 

As we noted in our earlier review, the PCB of the ATI Radeon HD 3870 is much simpler than the one of the Radeon HD 2900 XT. There is nothing wrong about it considering the 256-bit memory bus and the more economical RV670 chip of the newer card. The simple two-phase GPU power circuit is governed by a uPI uP6201AQ controller. A uPI uP6101BSA controller is responsible for the memory chips. Manufactured on 55nm tech process, the RV670 chip is indeed very economical and RV670-based cards are satisfied with a single 6-pin PCI Express 1.0 power connector.

The ASUS EAH3870 TOP carries the same memory as the reference card, Samsung K4U52324QE-BC08. These 512Mb chips work at a voltage of 1.8V and have a rated frequency of 1200 (2400) MHz. ASUS didn’t overclock them much, though. The memory is clocked at 1143 (2286) MHz, which is just a little higher than the reference memory frequency of 1125 (2250) MHz. So we can’t expect great benefits from this side. The card offers a total of 512 megabytes of graphics memory although the official website mentions it as EAH3870 TOP/G/HTDI/256М. This must be a type and the card’s name should instead have “512M” at the end.

The GPU frequency is increased far above the reference card’s 775MHz to 851MHz. This can hardly add more than 5-10% to the frame rate. The maximum we can expect is 15% but ASUS only promises a 4% performance growth. The GPU has a standard configuration with 64 universal shader blocks with 5 ALUs in each, 4 texture processors, and 4 rasterization processors. In other terms, the GPU can be said to have 320 streamed processors, 16 TMUs and 16 ROPs. As we noted in our reviews, the Radeon HD architecture is unrivalled in its computational capacity, and the number of ROPs is not a bottleneck in modern games. It is the amount of TMUs that can have a negative effect on performance which can be further aggravated by insufficiently deep application-specific optimizations in the driver.

Like any Radeon HD 3870 with the reference design, the ASUS card supports CrossFireX subsystems built out of two to four graphics cards. The card is equipped with two dual-link DVI-I ports and supports the audio-over-HDMI feature by means of an included adapter. It can also output analog video in Composite, S-Video and YPbPr format via its 7-pin mini-DIN connector.

The ASUS EAH3870 TOP is cooled with a reference cooler developed by AMD for this card specifically and described in our review of the new series. It resembles the cooler of the Radeon X1950 XTX but has a different and simple design actually. The cooler is based around a heatsink milled out of a whole chunk of copper. The monolithic heatsink minimizes thermal resistance for higher cooling performance without heat pipes.

The heatsink is cooled by a 75mm blower like the one employed in the cooler of the Radeon X1950 XTX. The hot air is exhausted out of the system case through the slits in the card’s mounting bracket. The load-bearing transistors of the power circuit and the memory chips are cooled with individual heatsinks, the memory heatsink being made from copper, too. Thanks to low heat dissipation of the RV670 even when clocked at 850MHz the fan works at a low speed and the cooler is almost silent. Its only drawback is the dual-slot form-factor, but it has already become standard for high-performance graphics cards and its clumsiness is made up for by its ability to exhaust the hot air out of the PC case which is not possible with single-slot coolers.

ASUS EAH3850 TOP: Design and Specifications

The ASUS EAH3850 TOP is a precise copy of the reference card from AMD only differing from it with the stickers on the cooler and fan.

 

The wiring of the power circuit is alike to that of the Radeon HD 3870 but there are almost no discrepancies in terms of the components employed. It is the same two-phase circuit with uP6201AQ and uP6101BSA chips responsible for power supply of the GPU and memory, respectively. Here, the external power supply connector is justifiable since the GPU and memory clock rates of the EAX3850 TOP are considerably higher than those of the reference card.

Like the reference Radeon HD 3850, the ASUS EAH3850 TOP is equipped with GDDR3 memory (Samsung K4J55323QI-BJ11) with a rated frequency of 900 (1800) MHz, but the chips are actually clocked at 950 (1900) which leaves almost no hope for further overclocking. Anyway, the memory bandwidth gain over the reference card is an impressive 7.7GB/s for a total memory bandwidth of 60.8GB/s. That’s far above 44.8GB/s of the Nvidia GeForce 8800 GT 256MB. The 256Mb chips provide a total of 256 megabytes of graphics memory.

The GPU of the EAH3850 TOP is pre-overclocked by the manufacturer from 670MHz to 730MHz. Coupled with the increased memory frequency, this is a promise of a higher performance growth than you can get from the EAH3870 TOP that virtually comes with only one, GPU, frequency pre-overclocked. The GPU configuration is alike to the senior model: 64 universal shader blocks with 5 ALUs in each. 4 large texture processors, and 4 rasterization processors. In fact, the card offers 320 streamed processors, 16 TMUs, and 16 ROPs.

The ASUS EAH3850 TOP can be used in a CrossFireX system together with one, two or three such cards. It offers the same connectors as the senior model: two dual-link DVI-I ports with support for resolutions up to 2560x1600 pixels, and a 7-pin connector for analog video output. Of course, the card supports, by means of an adapter, HDMI 1.1 and the audio-over-HDMI feature.

The card comes with the reference Radeon HD 3850 cooler which is somewhat more complex than the cooler of the Radeon HD 3870 because it called for more advanced cooling technologies to keep cooling performance high while remaining within compact dimensions.

There is a groove in the aluminum base a heat pipes lies in. It takes heat off the copper piece that has contact with the GPU die and distributes the heat uniformly in the heatsink consisting of aluminum plates. The air flow from the blower goes through the heatsink, makes a turn at an angle to the mounting bracket driven along channels in the plastic casing, and is thrown off sideways. The memory chips and the load-bearing transistors of the power circuit give their heat directly to the aluminum base through elastic thermal pads.

This is enough to cool the reference card especially as the heatsink has a larger dissipation area than in the notorious cooler of the Nvidia GeForce 8800 GT but will this design be efficient for a pre-overclocked version? You’ll learn in the next section of the review.

ASUS EAH3870 TOP and EAH3850 TOP: Noise, Temperature, Overclockability

Although the reviewed graphics cards come with reference coolers, we measured the level of noise produced by them to see if the fan management settings differ from those of the reference cards. We measured the noise with a digital sound-level meter Velleman DVM1326 using A-curve weighing. The level of ambient noise in our lab was 36dBA and the level of noise at a distance of 1 meter from the working testbed with a passively cooled graphics card inside was 43dBA.

There is no difference: the ASUS EAH3870 TOP is noiseless in every operation mode while the ASUS EAG3850 TOP was only audible under 3D load. The comfortable level of noise was achieved at a price, though: the temperature of the senior model’s GPU was as high as 84°C in idle mode and 92°C in 3D mode. The junior model had a GPU temperature of 64°C when idle and 76-77°C under load. These results shouldn’t worry you at all but they indicate that you’ll need a more effective cooler, like Zalman VF1000 LED, if you want to overclock these graphics cards further.

Our attempt at overclocking proved our apprehensions. The ASUS EAH3870 TOP refused to be overclocked at all hanging up every time we tried to increase its GPU frequency above the default. The ASUS EAH3850 TOP allowed to increase the GPU frequency from 730MHz to 780MHz and remained stable. 50MHz is quite an achievement considering the modest cooler. Perhaps the card could overclock even more if it had a better cooler. Against our expectations, the memory chips could be overclocked from 950 (1900) MHz to 1120 (2240) MHz. This should help achieve the performance of a Radeon HD 3870 in those applications where the amount of graphics memory is not a bottleneck.

Like the reference cards, the ASUS EAH3870 TOP and EAH3850 TOP are perfectly compatible with mainboards supporting PCI Express 1.0a.

Testbed and Methods

The comparative theoretical and gaming performance analysis of the Asus EAH3870 TOP and EAH3850 TOP was carried out on the following standard test platforms:

According to our traditional testing methodology, the drivers were set up to provide the highest possible quality of texture filtering and to minimize the influence of software optimizations that both – AMD/ATI and Nvidia – enabled by default. Also, to ensure maximum image quality, we enabled transparent texture filtering options: Adaptive Anti-Aliasing/Multi-sampling for ATI Catalyst and Antialiasing – Transparency: Multisampling for Nvidia ForceWare. As a result, our ATI and Nvidia driver settings looked as follows:

ATI Catalyst:

Nvidia ForceWare:

For our tests we used the following games and benchmarks:

First-Person 3D Shooters

Third-Person 3D Shooters

RPG

Strategies

Synthetic Benchmarks

We selected the highest possible level of detail in each game using standard tools provided by the game itself from the gaming menu to each user. The games configuration files weren’t modified in any way. The only exception was Enemy Territory: Quake Wars game where we disabled the built-in fps rate limitation locked at 30fps.

Games supporting DirectX 10 were tested in this particular mode. With a few exceptions, the tests were performed in the following most widely spread resolutions: 1280x1024/960, 1600x1200 and 1920x1200. If the game didn’t support 16:10 display format, we set the last resolution to 1920x1440. We used “eye candy” mode everywhere, where it was possible without disabling the HDR/Shader Model 3.0/Shader Model 4.0. Namely, we ran the tests with enabled anisotropic filtering 16x as well as MSAA 4x antialiasing. We enabled them from the game’s menu. If this was not possible, we forced them using the appropriate driver settings of ATI Catalyst and Nvidia ForceWare drivers

Performance was measured with the games’ own tools and the original demos were recorded if possible. Otherwise, the performance was measured manually with Fraps utility version 2.9.1. We measured not only the average speed, but also the minimum speed of the cards where possible.

The following solutions also took part in our today’s test session:

Performance in First-Person 3D Shooters

Battlefield 2142

This game doesn’t support display resolutions of 16:10 format, so we use a resolution of 1920x1440 pixels (4:3 format) instead of 1920x1200 for it.

The GPU clock rate increased from 775 to 850MHz doesn’t make the ASUS EAH3870 TOP much faster. It is only 4-6% ahead of the reference card, which cannot affect your playing comfort. The advantage of the ASUS EAH3850 TOP is outlined sharper, especially at resolutions above 1280x1024. Its performance is 10-13% higher than that of the reference card, yet this has little practical effect, too. The list of playable resolutions remains the same. Note also that the increased clock rates help the ASUS EAH3850 TOP compete successfully with the Nvidia GeForce 8800 GT 256MB.

BioShock

BioShock doesn’t support FSAA when running in Windows Vista’s DirectX 10 environment. That’s why we benchmarked the cards without FSAA.

The TOP series cards from ASUS are not any better than the respective reference card for playing BioShock. The senior model allows playing comfortably at resolutions up to 1920x1200 inclusive while the junior card can be used to play at 1600x1200 or 1680x1050. Both are somewhat slower than the Nvidia GeForce 8800 GT 256MB.

Call of Juarez

As usual, graphics cards with 256 megabytes of local graphics memory are slow in this game. The GeForce 8800 GT 256MB is an especially poor sight as it is three times slower than the 512MB version of the same card.

The small performance gain provided by the pre-overclocked EAH3870 TOP is not enough to overtake the Nvidia GeForce 8800 GT 512MB at 1280x1024: the ASUS card is 10% behind the leader. At the higher resolutions the Nvidia card suffers from driver-related problems with memory management and both versions of Radeon HD 3870 find themselves in the lead. That’s not a victory, however, as their speeds are too low for practical play.

Call of Duty 4: Modern Warfare

The ASUS cards produce a similar performance gain which is most conspicuous at resolutions below 1920x1200. As for actual play, the senior model is limited to 1280x1024 just as the reference Radeon HD 3870. The junior model is too slow even at the lowest tested resolution. The same is true for the Nvidia GeForce 8800 GT 256MB whose minimum speed is too low.

Crysis

The game being too hard at its Very High level of detail, we benchmarked the cards without FSAA to get a more playable speed.

The increased frequencies produce a bigger effect on the Radeon HD 3870 than on the Radeon HD 3850 in this test but there is no talking about acceptable speed in either case. You can’t play this game at the Very High level of detail on any existing graphics card. Perhaps the upcoming dual-processor solutions from AMD and Nvidia will change that.

Enemy Territory: Quake Wars

The frame rate is fixed at 30fps in this game as this is the rate at which the physical model is being updated at the server. Thus, this 30fps speed is the required minimum for playing the game.

Having higher performance due to the increased GPU and memory frequencies, the ASUS EAH3850 TOP offers more comfort for the resolution of 1600x1200. This 10-11% addition is indeed valuable as the frame rate is very near the allowable minimum. Otherwise, there are no changes and we can only note the small advantage of the ASUS EAH3870 TOP over the Nvidia GeForce 8800 GTS 512MB at 1600x1200 and 1920x1200. This must be due to memory management problems in the Nvidia ForceWare driver.

F.E.A.R. Perseus Mandate

Like Battlefield 2142, this game does not support resolutions of 16:10 format. So, we use 1920x1440 (4:3 format) instead of 1920x1200 in this test.

Notwithstanding its increased frequencies, the ASUS EAH3850 TOP has a very low minimum of speed for you to play normally at 1600x1200. The same resolution is the limit for the ASUS EAH3870 TOP although it delivers a more comfortable minimum speed than the reference card.

The Nvidia GeForce 8800 GT 256MB needs a special mention here. Being inferior to the Radeon HD 3870 in terms of clock rate and memory amount, it provides a considerably higher minimum speed at 1280x1024, which is the consequence of ATI’s saving on the amount of TMUs.

Half-Life 2: Episode Two

The ASUS cards are 7 to 14% better than the respective reference cards but this performance gain has no practical value for the gamer. Interestingly, the pre-overclocked EAH3850 TOP is as fast as the reference Radeon HD 3870 in this test except for 1920x1200.

Although the game obviously has modest graphics memory requirements, the Nvidia GeForce 8800 GT 256MB slows down suddenly after 1280x1024. Nvidia’s programmers have got something to work upon – the drivers call for improvements.

S.T.A.L.K.E.R.: Shadow of Chernobyl

The game doesn’t support FSAA when you enable the dynamic lighting model, but loses much of its visual appeal with the static model. This is the reason why we benchmarked the cards in S.T.A.L.K.E.R. using anisotropic filtering only.

The ASUS EAH3850 TOP and the corresponding reference card are out of the play right away as their performance is too low whereas the ASUS EAH3870 TOP allows playing at 1280x1024 even though being 12% slower than the Nvidia GeForce 8800 GT 256MB. This is due to the problem that has always plagued the ATI Radeon HD architecture – insufficient optimizations of the Catalyst driver for the particular game engine.

Unreal Tournament 3

Forcing FSAA from the graphics card’s driver doesn’t produce any effect as yet. Perhaps the FSAA support will be added by means of a patch.

The engine of this game was meticulously optimized for the existing hardware and every graphics card, including the Nvidia GeForce 8800 GT 256MB and the ATI Radeon HD 3850, feels at ease here at every display resolution. The pre-overclocked version of the latter card adds 16% to the frame rate to reach the desired 60fps mark. That’s the most significant practical effect from the overclocked frequencies of the ASUS cards we have spotted so far.

Performance in Third-Person 3D Shooters

Lost Planet: Extreme Condition

Both versions of Radeon HD 3850 feel a shortage of graphics memory right from the start while the ASUS EAH3870 TOP looks good against the more expensive GeForce 8800 GT 512MB, being only slower by 10-15% depending on resolution. Anyway, the gamer will have to lower the level of detail to play the game normally.

Tomb Raider: Legend

Overclocking is of little effect here and the ASUS cards don’t offer big advantages in comparison with the reference cards from AMD. The average frame rate of the EAH3870 TOP is below that of the GeForce 8800 GT 256MB, but Nvidia’s card has a terrible minimum speed at resolutions of 1600x1200 and higher, which makes the ATI Radeon HD 3870 and 3850 preferable for this game.

Performance in RPG

Hellgate: London

The ASUS EAH3870 TOP is slightly ahead of the GeForce 8800 GT 512MB but the Nvidia card seems preferable due to its higher minimum of speed even though 21fps is not really comfortable. The ASUS EAH3850 TOP outperforms the reference card by 10% at 1280x1024 but this has no practical value as its speed is anyway too low.

Gothic 3

The current version of the game doesn’t support FSAA, so we performed the test with anisotropic filtering only.

Having fewer TMUs than the Nvidia GeForce 8800 GT, both versions of Radeon HD 3870 have good results at 1280x1024 anyway. Their minimum speed guarantees the lack of any slowdowns that are sure to occur with a GeForce 8800 GT 256MB. Overclocking brings about a 3fps addition to the average frame rate, which has no effect on gameplay.

The cheaper ATI Radeon HD 3850 cannot cope with this test even at 1280x1024 and its overclocking gain is as small as with the Radeon HD 3870.

The Elder Scrolls IV: Oblivion

The game loses much of its visual appeal without HDR. Although some gamers argue that point, we think TES IV looks best with enabled FP HDR and test it in this mode.

The four RV670-based cards seem to hit against some ceiling that prevents them from having an average frame rate higher than 60fps. That’s not a consequence of the disabled vsync option. We don’t know the reason of that. Perhaps, there are some flaws in the current version of the Catalyst driver or in the software setup of our testbed.

It wouldn’t be correct to compare the ATI Radeon HD 3800 cards with Nvidia’s solutions due to the mentioned problem, but their performance is high for normal play even in open scenes. The speed is never lower than 25-26fps at 1920x1200 even if you play the game on a Radeon HD 3850 whereas the average frame rate is 44fps. These numbers have been a dream of every TES IV fan just recently.

Performance in Strategies

Company of Heroes: Opposing Fronts

The new add-on to Company of Heroes is tested in DirectX 10 mode only since it provides the highest quality of the visuals.

The ASUS EAH3850 TOP gains somewhat more than the senior model from the overclocked frequencies, yet both have problems when it comes to actual play. The same is true for the Nvidia GeForce 8800 GT 512MB, though. This card is far faster than the ASUS EAH3870 TOP in terms of average frame rate but its minimum speed is just as low, which means the same level of comfort for the gamer. The Nvidia GeForce 8800 GT 256MB fails the test due to driver-related problems.

Command & Conquer 3: Tiberium Wars

The game having a frame rate limiter, you should consider the minimum speed of the cards in the first place.

Here, there is no effect from overclocking Radeon HD 3800 series cards. The ASUS HD3870 TOP allows playing at all resolutions just as the reference card. The ASUS HD3850 TOP, like the reference ATI Radeon HD 3850, cannot provide a comfortable minimum of speed at 1920x1200.

World in Conflict

That’s one of the few cases when the overclocked frequencies of the Radeon HD 3870 bring about considerable benefits. The ASUS EAH3870 TOP provides a performance gain of 33-34% over the reference card at 1280x1024, 45% at 1600x1200, and 50% at 1920x1200!

The ASUS EAH3850 TOP doesn’t follow the suit and provides an ordinary 10-15% gain over the reference card. The amount of graphics memory must be the limiting factor for this test.

Performance in Synthetic Benchmarks

Futuremark 3DMark05

Judging by the overall scores, there is but a small effect from overclocking the Radeon HD 3850 and no effect at all from overclocking the Radeon HD 3870. Let’s discuss individual tests that were run at higher display resolutions with 4x FSAA.

The increased clock rates help the ASUS EAH3870 TOP match the performance of the Nvidia GeForce 8800 GT 256MB in the first and third test and almost overtake the 512MB version of the latter card in the second test. The ASUS EAH3850 TOP is ahead of the GeForce 8800 GT 256MB in the second test. The average performance gain from the increased frequencies of the RV670-based cards amounts to 6-12%.

Futuremark 3DMark06

As opposed to 3DMark05, the performance benefits of the ASUS cards are visible in 3DMark06 right from the start. The senior model is a mere 169 points behind the Nvidia GeForce 8800 GT 512MB. The junior model is ahead of the Nvidia GeForce 8800 GT 256MB and scores over 10,000 points.

The SM2.0 tests suggest that the Radeon HD 3850 benefits more than the Radeon HD 3870 from overclocking. The performance of the ASUS EAH3870 TOP in the SM3.0/HDR is noteworthy: it overcomes the 5,000 points mark and leaves the more expensive GeForce 8800 GT 512MB behind. Once again we can see that the ATI Radeon HD architecture is perfectly suited to intensive math1ematics-heavy work and can deliver better performance, with appropriate driver optimizations, than G80/G92-based solutions.

The ASUS cards and the reference ATI Radeon HD 3870/3850 can boast nothing in the first SM2.0 test, which is just what you could expect considering their 16 TMUs. The second test is not such a failure, though, with the ASUS EAH3870 TOP slightly outperforming the GeForce 8800 GT 256MB. The ASUS EAH3850 TOP is only 10% slower than the latter, which is a much smaller gap than in the first test.

The ASUS EAH3870 TOP is a mere 4-5% behind the Nvidia GeForce 8800 GT 512MB in the first SM3.0/HDR test, which is a very good thing for it. In the same test the ASUS EAH3850 TOP is 15% faster than the Nvidia GeForce 8800 GT 256MB. The second test shows the same picture, proving the efficiency of the ATI Radeon HD architecture at processing complex shaders again.

Conclusion

Our tests of the EAH3870 TOP and EAH3850 TOP graphics cards from ASUS have proved that factory overclocking of ATI Radeon HD 3800 produces but a very small effect and brings almost no benefits for real gaming applications. We don’t mean there is no performance growth at all: it varies from 3 to 5%, sometimes to 10-15%, but averaging 6-8% it almost does not affect your gameplay experience. Usually, if the ASUS cards are able to deliver comfortable performance, the same can also be achieved with the reference Radeon HD 3870 and 3850.

On the other hand, the ASUS EAH3870 TOP and EAH3850 TOP are indeed the fastest off-the-shelf Radeon HD 3800 featuring all the good traits of the series: good performance, high efficiency, and quiet operation. The junior model is obviously preferable to the Nvidia GeForce 8800 GT 256MB while the senior one, being somewhat slower than the Nvidia GeForce 8800 GT 512MB in games, is more available and can be a good choice for a gamer, too. Its retail price is higher than the recommended $199-219 because we’re talking about pre-overclocked versions. The price is also high due to the rich accessories including a full version of Company of Heroes: Opposing Fronts.

If you are not taken aback at the price – high-performance products from ASUS have never been cheap – the EAH3870 TOP and EAH3850 TOP may be interesting to you. But if you don’t need lots of accessories and free games, you may want to look for less expensive versions of ATI Radeon HD 3800 offered in plenty by other vendors.

Asus EAH3870 TOP Summary

Highs:

Lows:

Asus EAH3850 TOP Summary

Highs:

Lows: