by Sergey Lepilov
10/31/2009 | 01:43 PM
The recent release of the ATI Cypress graphics processor (RV870) and the first Cypress-based graphics card Radeon HD 5870 has been more than successful. The new product has proved to be as fast as the dual-processor Radeon HD 4870 X2 (and even faster in some applications) but at lower noise, power consumption and price. It is far superior to Nvidia’s single-processor solutions, too. The shortage of cards and some flaws in the Catalyst driver have been but temporary difficulties that usually accompany each newly released graphics card.
There is one more, hidden, danger for the commercial success of the Radeon HD 5870, though. We mean its junior counterpart called Radeon HD 5850. Based on the same GPU with a few subunits turned off, this graphics card is priced $100 lower. We don’t know if the disabled subunits can be turned back on to get a full-featured RV870, but we can check out the difference between the HD 5850 and 5870 in terms of performance and see if the gap can be bridged by overclocking. It is also going to be interesting to learn the difference in power consumption, noise and consumer properties of these models and to compare them with Nvidia’s solutions. We will carry out all these tests in today’s review.
The following diagram compares the official recommended prices of a few graphics cards. They are very close to the respective retail prices (except for the prices of Radeon HD 5850 and HD 5870).
As you can see, the GeForce GTX 285 is the only potential opponent to the Radeon HD 5850, being similar to the latter in price. The other cards are either much cheaper or much more expensive. A CrossFireX tandem of two Radeon HD 5850 would cost about $600 and there is no single graphics card or multi-GPU configuration with a similar price. The difference from the senior model is a substantial $100. So, the Radeon HD 5850 seems to have no opponents other than the GeForce GTX 285 but it is actually quite a dangerous rival to more expensive solutions as you will see shortly.
Now let’s take a look at the specs of these graphics cards:
Our ATI Radeon HD 5850 graphics card is a copy of the reference sample. It came to us in OEM packaging without any accessories. We will discuss our HD 5850 (left) in comparison with an HD 5870 (right).
The face sides of the cards are almost identical except that the Radeon HD 5870 is 38 millimeters longer, being actually the longest graphics card of today. Both devices are 100 millimeters wide and 37 millimeters thick, blocking the neighboring expansion slot on the mainboard.
The Radeon HD 5850 does not have a metallic cover on its reverse side as the HD 5870 has:
We guess the cover is no good for cooling and the Radeon HD 5850 feels better for not having it. There is a red plastic piece on the cooler casing of the cards which has slits on the right for exhausting hot air.
The card’s mounting bracket also has vent slits and a full set of modern interfaces: two DVI-I ports, HDMI and DisplayPort.
At the opposite end there are two 6-pin connectors for additional power (these are located at the top of the PCB on the Radeon HD 5870). According to its specs, the peak power draw of the Radeon HD 5850 is no higher than 170W (as opposed to the Radeon HD 5870’s 188 watts). A 550W or better power supply is recommended for computers with this graphics card (we guess this is a bit of an overstatement). The power draw plummets to a mere 27W in 2D mode.
Now let’s take a look at both cards with their coolers removed.
As opposed to the Radeon HD 5870 which has a 4-phase GPU voltage regulator, the Radeon HD 5850 has only three phases and a seat for an uninstalled fourth phase.
The RV870 chip of our Radeon HD 5850 is dated the 35th week of this year and has a simplified marking in comparison with the Radeon HD 5870’s GPU manufactured on the 33rd week of 2009.
The RV870 is manufactured on 40nm technology and has fewer shader processors (1440 against 1600) and texture processors (72 against 80) in the Radeon HD 5850 than in the HD 5870. The junior model’s GPU is clocked at a lower frequency (725MHz against the Radeon HD 5870’s 850MHz) and lower voltage (1.087V against 1.149V). In 2D mode the GPU frequency of each card is dropped to 157MHz and its voltage is reduced to 0.949V. Coupled with a reduced memory frequency, this helps achieve a substantial economy. The 334 sq. mm GPU die is turned by 45 degrees relative to the protective metallic frame. So if you are installing an alternative cooling system, you have to make sure its GPU-contacting piece is larger than 26 millimeters in both width and length. Otherwise, the GPU die won’t be covered fully.
Both cards are equipped with eight GDDR5 memory chips located on the face side of the PCB. The chips were manufactured by Samsung and are the same for both cards.
They are marked as K4G10325FE-HC04 and have an access time of 0.4 nanoseconds, which corresponds to a theoretical frequency of 5000MHz. However, the memory frequency of the Radeon HD 5850 is set at 4000MHz, giving some hope for good overclocking. The Radeon HD 5870 has a memory frequency of 4800MHz. The memory bus is 256 bits wide.
Here is a summary of the cards’ specs.
Now let’s check out the cards’ coolers and temperatures. Each is equipped with a dual-slot cooler of classic design: a blower is driving the air through a heat-pipe-based heatsink installed on the GPU and exhausting some of it out of the system case. Some of the air is left inside, though.
The cooling systems differ as we could make sure after dismantling them both.
So, there is a blower inside each cooler. It is installed on a metallic plate that serves as a heatsink for the power components and memory chips. The similarity ends here. The heatsink consisting of four 6mm heat pipes and aluminum ribs is soldered to the base plate on the Radeon HD 5870 whereas the heatsink of the Radeon HD 5850 is an individual element. It consists of a copper base about 4 millimeters thick, two 8mm heat pipes and aluminum plates that hang on the pipes and are additionally soldered to the base.
Besides, there is a flat heat pipe in the metallic plate right above the power components. It has no heatsink and helps distribute heat more uniformly in the plate. This solution doesn’t seem effective to us.
The coolers of both cards are equipped with 92x38mm blowers with PWM control.
The Radeon HD 5850’s cooler has a FirstD FD9238H12S fan whereas the Radeon HD 5870 comes with a fan from NTK (HK) Limited marked as FD9238H12S, too.
We guess the fans only differ with the stickers. Their speed is PWM-regulated from 1100 to 4800rpm (according to monitoring tools). We will compare their noise with those of other graphics cards shortly.
Right now we’ll show you how effective these coolers are. We tested them in a closed system case (its exact configuration is listed in the next section). The room temperature was 23.6-23.8°C during this test. The cards were loaded by FurMark 1.7.0 running in stability check mode at 1680x1050. The frequencies and temperatures were monitored with MSI Afterburner 1.2.0.
First let’s see how effective the reference coolers of Radeon HD 5850 and 5870 cards are in automatic fan management mode.
Bravo, AMD/ATI! Besides being effective, the coolers are never faster than 31% of their full speed even under FurMark, producing moderate noise. Our and other reviewers’ criticism of the reference coolers of Radeon HD 4870/4890 in 3D applications must have been heard by ATI engineers who have developed efficient and low-noise coolers. The GPU of the Radeon HD 5870 was no hotter than 76°C in FurMark and less than 65°C hot in 3DMark (at 28% cooler’s speed). The graphics cards were not audible in the quiet system case in 2D mode.
Next, let’s check out the coolers at 50% and 100% speed.
The coolers are much louder, of course, but the temperature of the Radeon HD 5850’s GPU lowers from 74 to 57°C at 50% speed and to 52°C at 100% speed. The temperature of the GPU of the Radeon HD 5870 lowers from 76°C to 60°C at 50% speed and to 53°C at 100% speed.
The distance between the mounting holes of the Radeon HD 5850 cooler is the same as on the Radeon HD 4870/4890 (75.5 millimeters), so you can try to install an alternative cooling system on this card. For example, a Thermalright T-Rad2 with two 92mm fans (1500rpm).
Unfortunately, a Thermalright VRM1(2) heatsink could not be installed on the voltage regulator elements of this card because the latter’s mounting holes are placed 3 millimeters wider than on the Radeon HD 4870/4890. Anyway, the alternative cooler easily coped with the GPU even without that additional heatsink.
The voltage regulator components were not hotter than 92°C then, which is lower than with the Radeon HD 4890 where they could be as hot as 110 or even 120°C. When the card was equipped with its reference cooler, these components were only 72-75°C hot.
Considering the low temperature and the reduced frequencies of the Radeon HD 5850 in comparison with the HD 5870, we had some hopes for good overclocking. We checked this out using the reference cooler, MSI Afterburner and AMD GPU Clock Tool. As a stability test we ran FurMark and the FireFly Forest scene from 3DMark06 with highest settings.
First, we tried to run the Radeon HD 5850 at the frequencies of its senior counterpart Radeon HD 5870:
Our Radeon HD 5850 found it easy to work at 850/4800MHz. Moreover, the GPU was stable at 910MHz with the default voltage of 1.087V! This was already an excellent result but we went further. After a couple of hours we found that the highest stable GPU frequency was 1010MHz at 1.21V voltage! We could not get higher by increasing the voltage more. Anyway, this is an impressive result. The memory chips overclocked less successfully, being stable at 4960MHz.
Interestingly, the overclocked card has a GPU temperature of only 67°C under load (this test was performed at 50% cooler’s speed).
Our Radeon HD 5870 was less successful at overclocking. We increased its memory frequency higher, to 5200MHz, but the maximum GPU frequency was only 905MHz. The GPU did not react to our increasing its voltage (up to 1.3V stepping 0.02V), so we had to stop at 905/5200MHz.
First, let me say a few words about the hardware that we used during this test session. All graphics cards were benchmarked in a closed system case with the following configuration:
To minimize the platform influence on the performance of the tested graphics cards I overclocked our 45 nm quad-core CPU with the multiplier set at 21x and “Load-Line Calibration” enabled to 4.1GHz. The processor Vcore was increased to 1.3825V in the mainboard BIOS:
The system memory worked at 1.56 GHz frequency with 7-7-7-14_1T timings and 1.64V voltage:
All other parameters available in the mainboard BIOS and connected with CPU or memory overclocking remained unchanged (set to Auto).
We are going to compare the performance of our Radeon HD 5850 and HD 5870 against that of two other graphics solutions on ATI chips: Diamond Radeon HD 4890 1 GB XOC (working at the reference frequencies for HD 4890) and HIS Radeon HD 4870 X2 2 x 1 GB:
From the competitor’s camp we will add the following three solutions: ZOTAC GeForce GTX 275 896 MB, LeadTek WinFast GeForce GTX 285 1024 MB and XFX GeForce GTX 295 2x896 MB.
All graphics accelerators except Radeon HD 5850 were tested in nominal modes. Radeon HD 5850 was tested at its default frequencies of 725 / 4000 MHz and at 850 / 4800 MHz that are the frequencies of Radeon HD 5870. This way we could check out the performance difference between the GPUs of these two graphics accelerators. Moreover, Radeon HD 5850 was also tested at maximum overclocked frequencies of 1010 / 4960 MHz in order to see if this graphics card will be able to outperform more expensive competition due to overclocking. We do understand clearly that other graphics cards can also be overclocked and hence run faster, but it wasn’t our goal to compare the performance of all solutions during overclocking today. That is why only our key hero got this “special treatment”.
Now let’s move on to software and benchmarking tools that we used. All tests were performed under the new Windows 7 Ultimate RTM x64 operating system with the following drivers:
The graphics cards were tested in two resolutions: 1680x1050 and1920x1200. We believe that it doesn’t make sense to test these graphics accelerators in 1280x1024, because the platform performance will be a limiting factor in most cases. Our monitor didn’t support resolutions above 1920x1200, but it is a minor issue, because very few gamers use higher screen resolutions anyway and the tested graphics cards cannot provide sufficient performance to ensure comfortable gaming experience in 2560x1600 resolution.
The tests were performed in two image quality modes: “High Quality” without any image quality enhancements and “HQ+ AF16x+AA4/8x” with enabled 16x anisotropic filtering and 4x full screen anti-aliasing (or 8x FSAA if the average framerate was high enough for comfortable gaming experience). We enabled anisotropic filtering and full-screen anti-aliasing from the game settings or configuration files. If the corresponding options were missing, we changed these settings in the Control Panel. Vertical sync was always off in driver control panels.
All games were reinstalled fresh under the new operating system and updated with the latest patches available at the time of tests. So, the complete list of test applications includes three popular synthetic benchmarking suites and 17 games of various genres, including such new titles as Wolfenstein, Batman: Arkham Asylum and Resident Evil 5. Here is the complete list of tests used with the settings (all games listed in their release order):
Here I’d like to add that if the game allowed recording the minimal fps readings, they were also added to the charts. We ran each game test or benchmark twice and took the best result for the diagrams, but only if the difference between them didn’t exceed 1%. If it did exceed 1%, we ran the tests at least one more time to achieve repeatability of results.
Now let’s talk about the actual benchmarks.
ATI Radeon cards are marked in red in the diagrams. Nvidia’s solutions are marked in green. The only exception is the Radeon HD 5850 overclocked to its maximum frequencies, which is marked in purple. First come the synthetic benchmarks.
Both synthetic benchmarks produce the same picture. The Radeon HD 5850 is better than the GeForce GTX 285, although the difference is not too big. When overclocked to the frequencies of the HD 5870, the HD 5850 gets very close to the latter and both are ahead of the dual-processor Radeon HD 4870 X2 in three out of the four test modes (the total of the two Futuremark applications). The Radeon HD 5850, overclocked to maximum frequencies, easily leaves the more expensive model from the same series behind.
The newest benchmark from Unigine Corp. supports DirectX 9, 10, 11, OpenGL, and hardware tessellation (only for DirectX 11 cards). It is very beautiful, especially on Radeon HD 58xx cards:
We used the following settings:
Anisotropic filtering and full-screen antialiasing were set according to the test mode. Here are the results:
Yes, the Radeon HD 58xx have low results. It is because they were tested in DirectX 11 mode and showed all the beauty of Unigine Heaven. If tested in DirectX 10 mode, the Radeon HD 58xx series are 20-22fps (at the low-quality settings) and 10-11fps (at the high-quality settings) faster than in DirectX 11 mode and only inferior to the dual-processor solutions then.
We guess we should show you the visual difference between the two latest versions of DirectX:
Looks impressive, isn’t it? The pavement in the left screenshot seems to be much easier to walk on than the right-screenshot pavement, though. :)
The default Radeon HD 5850 can only beat the competing GeForce GTX 285 in the low-quality mode. When we turn on FSAA and anisotropic filtering, the GeForce GTX 285 goes ahead although the Radeon HD 5850 is no worse in terms of bottom speed.
It is when we overclock the HD 5850 that we see the most exciting things. The Radeon HD 5850 is a mere 1-2fps behind the Radeon HD 5870 when both are working at the same frequencies. Thus, this game is indifferent to the Radeon HD 5850 having some of its GPU subunits disabled. Moreover, the more expensive Radeon HD 5870 is left behind when the HD 5850 is overclocked to 1010/4960MHz. The latter card takes third place then, right behind the dual-processor Radeon HD 4870 X2 and GeForce GTX 295. An impressive performance!
Crysis remains one of the hardest graphics card tests where even dual-processor solutions cannot deliver a really high frame rate. The Radeon HD 5850 easily beats the GeForce GTX 285 at the default frequencies, is a mere 1fps behind the Radeon HD 5870 at the latter’s frequencies, and is somewhat slower than the dual-processor GeForce GTX 295 when overclocked to the maximum. The dual-processor Radeon HD 4870 X2 is weaker than the GeForce GTX 295 here and even slower than the Radeon HD 5870. Comparing this with the results of the Radeon HD 4890 you can see that the new RV870-based solutions have brought about a new level of performance.
Unreal Tournament 3 is not a resource-consuming game, so it is the 8x FSAA + 16x AF results that have the biggest practical value. In this mode, notwithstanding the considerable advantage of the Radeon HD 5850, let alone Radeon HD 5870, over the previous-generation Radeon HD 4890, the new Radeons are no better than the GeForce GTX 285. The dual-processor GeForce GTX 295, which was challenged by the overclocked Radeon HD 5850 in the two previous games, is unrivalled in Unreal Tournament 3. The difference between the Radeon HD 5850 and HD 5870 is negligible in every test mode when they work at identical frequencies.
Lost Planet is an Nvidia-optimized game. But despite this fact, the Radeon HD 5850 at its default frequencies is no slower than the GeForce GTX 285 and even beats the latter in the high-quality mode and at 1920x1200. When clocked at the same frequencies, the Radeon HD 5850 equals the HD 5870 but both are inferior to the dual-processor Radeon HD 4870 X2. The Radeon HD 5850 eliminates this gap when overclocked to its highest clock rates, proving again its worth for overclockers. We should note that the first test of the integrated benchmark would occasionally stutter on the Radeon HD 58xx cards. Hopefully, this bug will be corrected in the officially certified version of the driver.
As opposed to Lost Planet, S.T.A.L.K.E.R. runs better on ATI’s solutions which enjoy an overwhelming advantage over their Nvidia opponents. The GeForce GTX 295 is the only card to put up a fight, but it too gives up before the onslaught of the Radeon HD 5850 overclocked to 1010/4960MHz. When clocked at the frequencies of the Radeon HD 5870, the HD 5850 is a mere 1-2fps slower than its senior counterpart, making us wonder if it makes sense to pay more for the latter.
Curiously enough, the Radeon HD 5850 is ahead of the GeForce GTX 285 in the low-quality image modes, but has no advantage when we turn on full-screen antialiasing and anisotropic filtering. Considering that the frame rate is comfortably high on every card, except for the Radeon HD 4890, in every mode, the FSAA + AF results are more important. And the Radeon HD 58xx, even though significantly better than the Radeon HD 4890 (and even better than the Radeon HD 4870 X2), have no advantage over the GeForce. We can also note the negligible difference between the Radeon HD 5850 and 5870 when both are working at the same frequencies. This difference is easily covered by overclocking the HD 5850 to its highest.
This game is no difficulty for today’s graphics cards. Even the slowest Radeon HD 4890 (we wouldn’t dream of calling it “the slowest” just a month ago) ensures a comfortable frame rate in the hardest graphics quality mode and high resolution. The Radeon HD 5850 is not successful against the GeForce GTX 285 but gets as fast as the Radeon HD 5870 if overclocked to the latter’s frequencies. When overclocked to the highest possible frequencies, our Radeon HD 5850 even beats the dual-processor Radeon HD 4870 X2!
It is clear that the graphics cards all hit the performance ceiling imposed by the speed of the platform (although our testbed is itself quite fast) in the easiest test mode. In the harder modes the Radeon HD 5850 equals the GeForce GTX 285 or even leaves it behind a little. The difference between same-frequency Radeon HD 5870 and HD 5850 is negligible. When overclocked to its maximum, the latter model beats every opponent excepting the GeForce GTX 295. We must note that the Radeon HD 58xx cards produced a fuzzy and darker image when we turned full-screen antialiasing on. That must be due to some flaws in the Catalyst driver.
Cryostasis: Sleep of Reason is one of PhysX-supporting games included into this test session. Alas, the Radeons are no match to the GeForce series here. But we can note that the Radeon HD 5850 and HD 5870 are close to the GeForce GTX 275 and GTX 285 and just a little slower than the dual-processor Radeon HD 4870 X2. The Radeon HD 5850 is identical to the HD 5870 when working at the same frequencies. Thus, the lack of 160 shader processors and 8 texture processors is not as significant as you might expect looking at the specs of the two models.
Although Nvidia took part in the development of Warhammer 40000: Dawn of War II, ATI’s new-generation solutions are quite competitive to the GeForce series. Take note of the weak support of dual-processor configurations in this game. Neither the Radeon HD 4870 X2 nor the GeForce GTX 295 can win here. It is the overclocked Radeon HD 5850 that takes top place. Take note that FSAA and AF do not affect the speed of the Radeon cards much although the resulting image quality improvement is obvious.
BattleForge, the only DirectX 11 game in this test session, was developed in collaboration with ATI, so the GeForce GTX 295 is the only opponent to ATI’s solutions here. The Radeon HD 5850 and HD 5870 deliver the same performance if their frequencies are equal. The latter card is the fastest choice for fans of this very beautiful strategy.
Do not wonder at the low performance of the GeForce series. Like BattleForge, Stormrise is an ATI-optimized game with support for DirectX 10.1. Nvidia has no chance here. The Radeon HD 5850 performs brilliantly, delivering at the default frequencies almost two times the performance of the GeForce GTX 285. It is almost as fast as the Radeon HD 5870 at the same frequencies and successfully competes with the dual-processor Radeon HD 4870 X2 at the highest frequencies.
For an unclear reason, we could not launch the DirectX 10 version of Tom Clancy's H.A.W.X. under Windows 7, so we had to test the cards in the DirectX 9 version of the game. Here are the results:
This test is not quite correct due to the above-mentioned reason, so we won’t compare ATI and Nvidia here. As for the Radeon HD 5850 and HD 5870 at the same frequencies, they deliver the same performance in this game, getting very close to the Radeon HD 4870 X2. When overclocked to the highest frequencies, the Radeon HD 5850 is as fast as ATI’s dual-processor giant.
You may remember that ATI’s solutions used to be generally faster than their Nvidia opponents in the 2006 Call of Juarez. The new Call of Juarez: Bound in Blood shows the same picture. The Radeons are only challenged by the dual-processor GeForce GTX 295 while the same-category GeForce GTX 285 falls far behind the Radeon HD 5850. When overclocked to 850/4800MHz, the Radeon HD 5850 gets as fast as the Radeon HD 5870.
Nvidia’s solutions have always been better than ATI’s in OpenGL applications. The new Wolfenstein proves the point once again:
It is only in the FSAA + AF modes that the Radeon HD 58xx series are ahead of their opponents. This is more valuable than the advantage of the GeForce series in the easier modes, considering the high frame rates.
Are you bewildered at the performance of the Radeon series? There is no wonder as it does not support PhysX technology which is used just everywhere in Batman: Arkham Asylum. The average and bottom frame rate of the Radeon cards is fixed at 12-13 and 21-22fps whereas Nvidia’s solutions crunch this application with ease. You suggest turning the physics off? No way! Batman: Arkham Asylum is quite a different game without it:
Feel the difference.
As you can see, one more PhysX-supporting game allows the current generation of Nvidia cards to beat ATI’s new solutions. It is only at the overclocked frequencies that the Radeon HD 5850 can compete with the GeForce GTX 285 but that’s hardly an achievement. Take note that the RV870-based solutions are no slower than the dual-processor Radeon HD 4870 in the FSAA + AF modes although the latter delivers a higher frame rate in the other modes.
Now let’s analyze the results using summary diagrams.
The first diagram compares the Radeon HD 5850 and the GeForce GTX 285, which come from the same price category, in the gaming tests. The recommended price of the Radeon HD 5850 is $299 as opposed to the GeForce GTX 285’s $339. However, it is likely that the retail price of the HD 5850 is going to be higher than recommended whereas the GeForce GTX 285 can only be bought for less than $300.
So, taking the performance of the GeForce GTX 285 as the reference point, here is how the Radeon HD 5850 compares to it.
The diagrams are indicative of the correlation between the graphics cards’ performance and the application-specific optimizations. For example, such games as Cryostasis: Sleep of Reason, Resident Evil 5 and, especially, Batman: Arkham Asylum run faster on the GeForce GTX 285 thanks to PhysX support whereas Warhammer 40000: Dawn of War 2, Call of Juarez: Bound in Blood, BattleForge and, especially, Stormrise run better on Radeons. The Radeon HD 5850 is simply faster than Nvidia’s solution in the rest of the games, except for Tom Clancy's H.A.W.X.
The following diagrams compare the Radeon HD 5850 with the Radeon HD 5870 where the latter is taken as the reference point. The speed of the Radeon HD 5850 is shown in three modes: at its default 725/4000MHz frequencies, at the frequencies of the HD 5870 (i.e. 850/4800MHz), and at the highest frequencies of 1010/4960MHz. The first pair of diagrams shows the results at two resolutions without image quality enhancement techniques.
And this pair of diagrams shows the cards’ performance with full-screen antialiasing and anisotropic filtering.
The default Radeon HD 5850 is an average 12.7% slower than the Radeon HD 5870 in the low-quality modes and 14.1% slower in the high-quality modes. When the frequencies of the Radeon HD 5850 are increased to the level of the HD 5870 (850/4800MHz), the gap is smaller at 2% and 2.6%. When overclocked to the highest frequencies, our sample of Radeon HD 5850 is 8.2% and 6.7% faster than the senior model of the series in the respective modes.
We carried out our power consumption measurements using special equipment. To create maximum load we launched FurMark 1.7.0 in stability check mode at 1920x1200 and Linpack x64 (LinX 0.6.3, 3500MB, 7 threads). These two programs load heavily the graphics card and CPU, respectively, so we can determine the peak power draw of the whole system. We also found out the most resource-consuming game – Lost Planet: Colonies – and measured the power consumption of the systems while running it, too.
You can see the results in the diagram:
Thus, the systems with new Radeon HD 58xx cards are among the most economical in this test. The Radeon HD 5850 is the best in this respect while the power draw of the senior Radeon HD 5870 is comparable to that of the Radeon HD 4890 and much lower than that of the dual-processor Radeon HD 4870 X2 which it often challenges in terms of performance. Note also the extremely low power draw of the Radeon HD 58xx systems in 2D mode.
The systems with dual-processor Radeon HD 4870 X2 and GeForce GTX 295 cards require the most power.
We also measured the level of noise using an electronic noise-level meter CENTER-321 in a closed room about 20 sq. meters large. The noise-level meter was set on a tripod at a distance of 20 centimeters from the graphics card which was installed in a system case with an open side panel.
The five 120mm system fans was set at 520rpm during this test and the background noise of the system case, if measured from a distance of 1 meter, was 32.9dBA. When the computer was shut down, the noise-level meter reported 29.9dBA. Subjectively, the comfortable level of noise is about 41dBA, which is marked with a red line in the diagram.
The measurement method is simple: the speed of the graphics card’s cooler was changing with RivaTuner or MSI Afterburner from 100 to 20% (25%) stepping 5%. At each speed the level of noise was measured and the results are used to build the diagram. The Radeon HD 5850 was as loud as the HD 5870 and the GeForce GTX 275, as GTX 285. Therefore, there are only four graphics cards in the diagram.
Although the Radeon HD 5870 has the highest level of noise among the four graphics cards, it is the quietest of all. This is no paradox. The cooling system of Radeon HD 58xx series cards does not speed up more than 31% or 2100rpm even under maximum load whereas the coolers of the other cards accelerate to 40% and higher and get louder as the consequence (this is especially true for the Radeon HD 4870 X2). Besides, the cooling system of the Radeon HD 58xx series is the quietest among the tested cards at a fan speed of 2000-2100rpm.
Summing up today’s test session, we have to admit that the Radeon HD 5850 is going to be the only real opponent to the Radeon HD 5870 until the arrival of the next generation of graphics cards from Nvidia. In fact, it is such a dangerous opponent that purchasing the top RV870-based model may not seem reasonable. To remind you, the difference in price between the Radeon HD 5870 and HD 5850 is $100 (or 25% of the price of the HD 5870) but the actual difference between the two cards is vanishing quickly as soon as you overclock your Radeon HD 5850 to the frequencies of the HD 5870. Can any Radeon HD 5850 be overclocked to 850/4800MHz? We guess it is almost 99.9% probability as long as all these cards use the reference design and are equipped with the same GPUs and memory chips. The option of increasing the GPU voltage from software (if your particular card cannot overclock at the default voltage) is already available.
As a matter of fact, we have not found any flaws in the Radeon HD 5850. The graphic card delivers high performance, has excellent overclocking potential, and comes with a very effective and low-noise cooler. Besides, the Radeon HD 5850 is highly power-efficient and 38 millimeters shorter than the Radeon HD 5870 and thus can fit into more system cases. Its DirectX 11 support is a nice bonus, too, even though there are few DirectX 11 games available as yet. There are still some problems with drivers (you can recall Lost Planet and Left 4 Dead) but that’s normal for new products from both ATI and Nvidia. The only thing this Radeon lacks is PhysX (and we don’t think the Radeon series will ever acquire it considering that Nvidia does not want to lose the advantage its solutions have in PhysX-supporting games), but anyway – bravo, ATI!
Our exploration of ATI’s new series will continue. We are going to publish a review of CrossFire configurations based on Radeon HD 5870 and 5850 and a special report on how much the Radeon HD 5870’s performance depends on the computer’s CPU. Stay tuned!