by Alexey Stepin , Yaroslav Lyssenko
03/03/2009 | 06:44 PM
A traditional home theater used to consist of a TV-set, a receiver with speakers, and a dedicated DVD player. Later on, players of Blu-ray and HD DVD discs came to market along with those new formats. For all the advantages of such systems, particularly their relative ease of use, specialized players have one fundamental drawback. They are not truly universal. Modern models often support a variety of formats besides the official commercial DVD and Blu-ray but there are just too many combinations of codecs, compression settings and media containers in the world of modern digital video. As the result, there is always a high chance of encountering a file the ordinary home receiver cannot play back, especially if it is an open noncommercial format, for example Matroska (.mkv) which is so popular among anime lovers.
The solution came quickly in the way of Home Theater PCs. An HTPC is a typical x86-compatible computer running some popular OS like Windows XP or Vista. In order to easier control the system and optimize it for large TV-sets, you can use the OS’s tools (Windows Media Center) or one of numerous multimedia shells from third-party developers. The fundamental advantage of an HTPC is its absolute universality. If you find it not to support some audio or video format, you can easily add this support by installing appropriate software. Even though an HTPC is somewhat more difficult to deal with, it is gaining popularity among home theater fans.
The next question is whether a HTPC, which is in fact an ordinary computer, perhaps in a nonstandard form-factor and with special software, can be used as a replacement for a modern game console? There are a lot of games available for the PC platform and some game genres, for example real-time strategies, are rare on consoles. Playing a game on a large screen with hi-fi sound can be much more fun than on an ordinary monitor, even if the latter is 24 inches wide.
So, do you have to buy a game console if you’ve already got a HTPC? It depends on the latter’s configuration. HTPCs are usually assembled in a flat desktop case that can be placed under a TV-set or into a hi-fi equipment stand, and such cases often have limitations regarding the installation of top-end graphics cards because of the dimensions or PSU wattage requirements or ventilation, or all these combined. Therefore we decided to limit this review to inexpensive graphics cards that are sure to suit a HTPC and have some gaming capabilities besides just offering hardware acceleration for HD video decoding.
Since low level of noise is an important factor for a HTPC, we have selected an inexpensive version of Radeon HD 4600 that claims to be absolutely silent. It is the PowerColor SCS3 HD4650 512MB DDR2 graphics card. Besides benchmarking the card in games at display resolutions typical for modern TV-sets, we will also check out its HD video decoding capabilities.
We received our sample of PowerColor SCS3 HD4650 nearly naked, wrapped into an antistatic pack and without any accessories, but this product comes to retail shops in a neat upright box of a small size and eye-pleasing design:
The packaging is not very informative but you can learn from it that the card is equipped with 512 megabytes of GDDR2 memory and Silent Cooling System 3. Despite the name of GDDR2, the memory is not like the type Nvidia used to install on early models of GeForce FX. It is ordinary DDR2 as we will explain shortly.
According to the PowerColor website, the PowerColor SCS3 HD4650 comes with the following accessories:
The accessories are scanty, but the product is very inexpensive. Unfortunately, a DVI-I → HDMI adapter is an optional accessory, so you may find that you won’t be able to connect a modern flat panel to this card right after purchase. Make sure the adapter supports both video and audio if you buy it separately from the card.
The additional expansion slot bracket with vent slits may help improve the cooling of your HTPC if you install it next to the card’s own mounting bracket. If the airflows are organized properly inside the system case, the hot air from the card’s cooler will be exhausted out of it.
So, the accessories promised by the manufacturer for the PowerColor SCS3 HD4650 are sufficient for its product category. The box design is not original, yet attractive.
The PowerColor SCS3 HD4650 has a simple and compact PCB because the RV730 has a 128-bit memory bus and modest power requirements. There is no need for a complex PCB with advanced power circuitry, especially when the graphics card is cheaper even than the Radeon HD 4670.
A large part of the PCB is covered with an intricate heatsink of the passive cooling system SCS3. It is connected to the heat-exchanger with two heat pipes and actually consists of thin aluminum plates put on those pipes.
The heatsink just hangs on the pipes as it has no other contact with the PCB. The heat-exchanger is made from aluminum (copper is more expensive and wouldn’t suit a product of this class). It is fastened to the PCB with four spring-loaded screws and has small rubber pads that prevent any misalignment that might damage the fragile graphics die. Thanks to its large area the cooler should be able to easily cope with the RV370 clocked at a reduced frequency – at least if there is any air flow at all in your system case. The cooler has a dual-slot design, so the PowerColor SCS3 HD4650 may not fit into some HTPC cases, particularly into compact models that have the PCI Express x16 slot turned around by 90 degrees.
Of course, such a low-power graphics card does not need external power supply. It gets all the juice it needs from the power section of the PCI Express x16 slot. The GPU and memory voltage regulators are very simple, with only one phase each, and governed by uPI Semiconcuctor uP6101 controllers.
As opposed to most modern graphics cards, the PowerColor SCS3 HD4650 employs DDR2 memory: four Samsung K4N51163QZ-HC20 chips on each side of the PCB. According to the specifications, these are gDDR2 chips, but they work at a voltage of 1.8V, like regular DDR2 memory, and have nothing to do with the hot and uneconomical GDDR2 type Nvidia once employed so unsuccessfully on the GeForce FX 5800.
These chips have a capacity of 512Mb (16Mb x 32), the HC20 suffix denoting a rated frequency of 500 (1000) MHz. The reference Radeon HD 4650 has a memory frequency of 400 (800) MHz, but the memory chips are clocked at 500 (1000) MHz on the PowerColor card. This increases memory bandwidth from 12.8GBps to 16GBps. Well, these numbers both look ridiculous in our time although the developers might have been proud of them at the time of GeForce FX 5800/5800 Ultra.
As usual, the only information we could gather from the GPU’s marking is the manufacturing date – the 33rd week of 2008 or mid-August. The die is installed at an angle of 45 degrees in its packaging, which is normal for the RV730. The lack of any protection on the packaging is alarming since the card is equipped with a rather massive cooler which is fastened with four screws only. The rubber pads on the heat-exchanger protect the die, but you should handle your PowerColor SCS3 HD4650 with care. Don’t take the card by its heatsink and do not apply serious effort to it.
The GPU has the maximum configuration possible for the RV730: 64 superscalar execution modules with 5 ALUs in each (for a total of 320 ALUs), 32 texture processors and 8 raster back-ends grouped into two large modules each of which is linked to two memory controllers. Each controller has two 32-bit interfaces, so the total width of the memory bus is 128 bits. Compared with the RV770, the architecture of the RV730’s functional subunits is not simplified, the consequence being that this inexpensive GPU incorporates as many as 514 million transistors. The GPU frequency is 600MHz just as in the official Radeon HD 4650 specs. This low clock rate may prove to be a bottleneck in some games, so we should expect this product to be considerably slower than the Radeon HD 4670 that has a clock rate of 750MHz.
An important feature of the RV730 is that it incorporates the UVD2 video-processor that offers hardware acceleration for both H.264 and VC-1. Coupled with good scaling and post-processing algorithms, it is the best solution in its class. The capabilities of UVD2 are complemented with an integrated audio core that supports audio-over-HDMI, which simplifies the use of the PowerColor SCS3 HD4650 in a home entertainment center. You only need to connect it with a single HDMI cable to your receiver or LCD panel.
The card’s interface connectors are not optimal for HTPC usage, though. It has two dual-link DVI-I ports but for a HTPC one of these ports would better be HDMI. This is not a serious problem, however, as you can easily correct it by means of the appropriate adapter included into the box. Besides, the PowerColor SCS3 HD4650 supports a traditional selection of analog interfaces for home appliances: Composite, S-Video and Component (YPbPr). The card has a universal 7-pin mini-DIN connector for that. There is no hardware support for CrossFireX: the PCB lacks appropriate connectors. Well, we don’t think anyone would want to use two such low-performance cards as Radeon HD 4650 in a tandem. It is simpler and more rewarding to buy an ordinary Radeon HD 4850 that will also be free from the typical drawbacks associated with the modern implementations of multi-GPU technologies.
Power consumption is an important parameter of HTPC systems because many of them are equipped with rather low-wattage power supplies and have nonstandard form-factor which prevents you from replacing the default PSU with a higher-wattage unit. Therefore we did some measurements with our PowerColor SCS3 HD4650 on a special testbed configured as follows:
The 3D load was created by means of the first SM3.0/HDR test from 3DMark06 running in a loop at 1600x1200 with 4x FSAA and 16x AF. The Peak 2D mode was emulated by means of the 2D Transparent Windows test from PCMark05 that simulates the user’s working with application windows. We got the following results:
Just as we could expect, the PowerColor SCS3 HD4650 needs far less power than the reference Radeon HD 4670 GDDR3, mainly because of its lower GPU clock rate, of course. As a result, its power consumption is no higher than 30W in 3D mode, which is higher than the consumption of the record-holder Radeon HD 4550, but the Radeon HD 4650 has more advanced capabilities. The card is absolutely silent in every mode because it lacks any fan.
It should be noted that, despite the passive cooler, the GPU temperature was under 40°C in idle mode and under 72°C in 3D applications, but we installed our PowerColor SCS3 HD4650 into a roomy and well-ventilated Chieftec LCX-01 system case. The temperatures may get higher in a cramped HTPC case. So, before installing the card into a HTPC case, you should make sure there is air flow around the graphics slot. Otherwise, the card may find itself overheating, especially if the PCI Express x16 slot is turned around by 90 degrees in your system case.
There is usually no point in overclocking graphics cards of this class, but we gave our PowerColor SCS3 HD4650 a try. Contrary to our expectations, the card’s GPU overclocked well enough.
The GPU was stable at a frequency of 700MHz. That’s a very good result for a cheap graphics card with passive cooling. The memory chip refused to work at frequencies higher than their rated 500 (1000) MHz. Since the memory was going to be a bottleneck in games, we didn’t benchmark the PowerColor SCS3 HD4650 with its GPU overclocked.
We are going to investigate the gaming performance of graphics cards that might be of interest to HTPC owners in the following testbed:
The graphics card drivers were configured in the same way as before: to provide the highest possible quality of texture filtering and to minimize the effect of default software optimizations. We enabled transparent texture filtering. As a result, our ATI and Nvidia driver settings looked as follows:
ATI Catalyst:
Nvidia GeForce:
We used the following synthetic benchmarks and games:
First-Person 3D Shooters
Third-Person 3D Shooters
RPG
Simulators
Strategies
Semi-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. The games configuration files weren’t modified in any way, because the ordinary user doesn’t have to know how to do it. We made a few exceptions for selected games if that was necessary. We are going to specifically dwell on each exception like that later on in our article.
Since we are talking about graphics cards that are often installed into HTPC type systems, we performed the tests only in two typical screen modes that shaped up basing on the specifications of contemporary LCD monitors: HD ready/720p and Full HD/1080?. In the former case we are talking about 1360x768 (because the physical resolution of most HD panels is 1366x768) or about 1280x720 (if the game doesn’t support the first resolution). In the latter case we used 1920x1080 resolution.
Besides PowerColor SCS3 HD4650 we have also included the following graphics accelerators to participate in our test session:
Since all our today’s testing participants do not belong to the high-performance category, we decided to give up full-screen antialiasing. We still enabled anisotropic filtering in all tests either through the game settings or forced it in the ATI Catalyst and Nvidia GeForce driver.
Performance was measured with the games’ own tools and the original demos were recorded if possible. We measured not only the average speed, but also the minimum speed of the cards where possible. Otherwise, the performance was measured manually with Fraps utility version 2.9.8. In the latter case we ran the test three times and took the average of the three for the performance charts.
Since we are going to talk about inexpensive graphics solutions used as part of HTPC systems, we couldn’t help checking out the multimedia capabilities of the new PowerColor SCS3 HD4650. So, we undertook a small additional testing aimed at decoding performance and playback quality of contemporary HD formats. We used the following platform:
Besides PowerColor SCS3 HD4650, we also included the following graphics cards:
We used the following tools to estimate the video playback quality in standard (SD) and high-definition (HD) resolutions:
The driver settings remained the same. However, according to the HQV HD suite requirements, the noise suppression and detail levels for Nvidia GeForce graphics cards were set to the maximums.
Keeping in mind that all tests are run under Windows Vista OS without disabling background services, the CPU utilization peaks shouldn’t be regarded as critical. It is much more important how much time it takes the CPU on average to complete the task. Note that the CPU utilization may vary. Therefore, 1-2% difference is not indicative of any advantage of a certain graphics accelerator over the competitor.
To estimate the CPU utilization during full-HD video playback (1920x1080) and full-HD video with enabled “picture-in-picture” (PiP) feature, we used the following movies:
We decided to give up tests with free online media content this time, because of an ongoing discussion as to whether this sort of performance analysis is justified and credible. As an excuse we can also say that a pretty popular Matroska (.mkv) container is currently not supported by existing commercial software.
The HQV benchmarks from Silicon Optix, the leading developer of video processors for home video players, are one of the few available methods of evaluating the playback quality of Blu-ray, DVD and HD DVD movies. They have one drawback, however. The tester’s perception is subjective while the notion of an ideal picture is rather vague.
We have been criticized for following the HQV HD test instructions too closely and giving out too low scores. However, our scores seem to have been quite correct because the playback quality of some GPUs we tested has improved dramatically over time.
ATI programmers forgot to enable video post-processing in Catalyst 9.1, so we had to perform the graphics quality test with Catalyst 8.12 while the GPU load was measured with Catalyst 8.12 and 9.1. HD video post-processing works again in Catalyst 9.2 and the ATI Radeon HD 4650 shows maximum results.
The three major developers of graphics solutions, ATI/AMD, Intel and Nvidia, are constantly optimizing video playback settings in their drivers, which should lead to playback quality improvements. Alas, ATI and Nvidia do not take the HQV test seriously as the lowering results indicate. So, considering the subjective nature of this test, you should not view the HQV and HQV HD results as the ultimate truth.
Although DVD has become obsolete, not all modern GPUs can play that old format at high enough quality.
None of the tested GPUs could achieve the maximum score of 130 points. The main problems are the inability of the GPUs to smooth out the jaggies in the Jaggies Pattern tests and transform filmed content (24fps, progressive scan) into DVD or 1080i HDTV (interlaced) with the 3:2 pulldown method (the so-called telecine). For some reason, the GPUs proved to be unable to render text over video, which is a serious problem.
Standard-resolution video is dying out. TV channels in the United States are already broadcasting in 720p whereas HTPC owners are going to use Full-HD TV-sets with a resolution of 1920x1080. As a result, it is far more important to ensure high-quality playback of high-definition video.
As you can see, many modern GPUs boast exceptional quality. In fact, nearly every mainstream and performance-mainstream GPU is very good at playing Blu-ray and HD DVD movies. The disappointing results only come from the integrated graphics cores Intel GMA 4500 and Nvidia GeForce 8300, and from the discrete Radeon HD 3400.
We want to note that no GPU could reproduce the Film Resolution Loss Test - Stadium with 100% correctness. The GPUs that received 10 points for that test had no moiré but had some flickering. If you feel like fault-finding, you should subtract these 10 points from the total score because, according to the HQV HD instructions, flickering means 0 points.
Another thing we must note is that Nvidia’s GeForce processors are still inferior at reducing noise to ATI’s Radeon HD chips.
Due to full hardware bitstream processing support, ATI Radeon HD graphics processors are stably better in this test even with "picture-in-picture" enabled. It is absolutely true for our today’s hero, PowerColor SCS3 HD4650.
Sicne our testbed has a pretty powerful CPU, those of you who can’t say the same about their processor, should be really careful about the graphics card choice for their HTPC. When our CPU was only 40% utilized, a slower processor may be loaded much heavier, which may lead to dropped frames in dynamic scenes, especially with "picture-in-picture" mode or Bonus View enabled.
Unlike VC-1, there are no evident leaders or outsiders during MPEG4 AVC/H.264 decoding. The average CPU utilization stays between 20-25 %, which is not too high. Of course, PowerColor SCS3 HD4650 performed very well in this test, just as we have expected.
We have to point out that it turned out harder for many GPUs, which support simultaneous decoding of two video streams, to decode a scene from The Day After Tomorrow movie with very high btirate than to playback the last Beowolf scene with enabled "picture-in-picture".
Considering how long the MPEG2 standard has been out there and that the integrated decoder for this format exists in all contemporary graphics processors, we were surprised to see pretty high CPU utilization readings in Alien vs. Predator movie. It could be some driver issue or the fact that hardware developers are not as interested in the old standard anymore as they used to be.
The results demonstrated by PowerColor SCS3 HD4650 were satisfactory enough: CPU utilizaiton in this test didn’t exceed 38%. We doubt that anyone will watch an HD movie and do some CPU-hungry work on the PC at the same time.
Here we would like to end our discussion of the multimedia capabilities and proceed to 3D performance of PowerColor SCS3 HD4650 graphcis card.
Even with FSAA turned off, the PowerColor SCS3 HD4650 is far from brilliant, being much inferior to the other tested cards. However, its performance is high enough at 1360x768 for you to play in the HD-ready mode at maximum graphics quality settings. Not bad for such a modest product!
Cheap cards are no good for playing Crysis Warhead. The best products of this category cannot yield an average frame rate of 20fps even. The PowerColor SCS3 HD4650 is slower than 10fps at 1360x768.
We disabled the integrated frame rate limiter in the game console for the sake of comparing the cards. The game’s built-in benchmarking options do not provide information about the bottom speed, so there is no such info in the diagrams.
The game having modest system requirements, the PowerColor SCS3 HD4650 delivers good performance, at least in the HD-ready mode, even though lagging behind the other cards tested for this review. Its average frame rate is 30fps in the Full-HD mode, which may seem to be enough since the in-game sped limiter is set at exactly 30fps. The frame rate may occasionally drop below comfortable level, however, so you should pick up something more advanced than the Radeon HD 4650 for Full-HD playing.
Far Cry 2 is optimized much better than Crysis Warhead, so the GeForce 9600 GT and Radeon HD 4670 can both be used together with HD-ready panels for playing this game even at the highest level of detail. Hamstringed by its slow memory and low GPU clock rate, the PowerColor is only competitive to the sluggish GeForce 9500 GT.
The game runs on the Source engine and has an integrated benchmark, but the latter does not report the bottom speed information.
The game’s system requirements are modest enough, and the PowerColor SCS3 HD4650 copes well at a resolution of 1360x768. At a resolution of 1920x1080 the card’s average frame rate is barely acceptable while the bottom speed must be even lower. Therefore this card is not the best choice for people who have got Full-HD panels.
To achieve a playable speed in this game we disabled FSAA and such resource-consuming options as Sun rays, Wet surfaces and Volumetric Smoke. We use the Enhanced full dynamic lighting (DX10) mode for our test and additionally enabled the DirectX 10.1 mode for the ATI cards.
The results are as depressing as in Crysis Warhead. Only top-end solutions can handle this game even if you turn off some of the most resource-consuming special effects. But we don’t think someone will install a GeForce GTX 295 or Radeon HD 4870 X2 into a HTPC system.
This is one of the least demanding games on our list, yet the playability of the PowerColor SCS3 HD4650 is still limited to the HD-ready mode, i.e. 1360x768, whereas the GeForce 9500 GT ensures a comfortable speed even at 1920x1080.
The PowerColor SCS3 HD4650 can only be used at 1360x768 and its bottom speed is somewhat below acceptable level even then. Its elder brother does just fine in the Full-HD mode (1920x1080) but cannot challenge the unrivalled GeForce 9600 GT.
The average frame rate of the PowerColor card is good in the HD-ready mode but its bottom speed is too low for performing the numerous acrobatic tricks the game abounds in smoothly. The GeForce 9600 GT is the only card that is close to delivering comfortable speed in the Full-HD mode, but you may have problems controlling your hero when the frame rate occasionally drops below 25fps.
The PowerColor SCS3 HD4650 is not fast enough for playing the game at the highest settings even without 4x FSAA. The Radeon HD 4670 is, on the contrary, capable of plunging you into the post-nuclear world of Fallout 3 on a HD-ready panel, but you need at least a GeForce 9600 GT to play the game normally on a Full-HD display device.
The Radeon HD 4670 and GeForce 9600 GT are the only cards capable of ensuring good speed at HTPC-typical resolutions. The latter is somewhat better of the two. Alas, the average frame rate of the PowerColor card does not reach even 20fps in the HD-ready mode.
The PowerColor SCS3 HD4650 looks good at 1360x768, but 1920x1080 is too much for it because of its low GPU frequency and memory bandwidth. The senior model of the series, Radeon HD 4670, is almost twice as fast as the PowerColor and copes with the Full-HD mode excellently.
Theoretically, the PowerColor SCS HD4650 is fast enough even in Full HD mode (1920x1080) but the integrated benchmark doesn’t report the bottom speed. The average frame rate being about 35fps, the bottom speed may drop below 25fps. You should select a better card for this mode, for example a Radeon HD 4670. In the HD-ready mode the PowerColor card feels at ease, being almost as fast as the GeForce 9600 GT. As a matter of fact, this game is known to prefer ATI’s GPU.
The game has a built-in frame rate limiter set at 30fps. It doesn’t work at a resolution of 1360x768 so we used 1280x720 instead.
Unfortunately, the junior model of the Radeon HD 4600 series is too slow for normal play at the maximum level of detail, especially at 1920x1080. The senior model looks better than the GeForce 9600 GT, particularly in the HD-ready mode. Don’t forget that the game doesn’t support the resolution of 1360x768.
It’s different in the other strategy game in our test program: the GeForce 9600 GT is in the lead whereas the Radeon HD 4670 cannot maintain a playable bottom speed. The Radeon HD 4650 is very slow, being inferior even to the GeForce 9500 GT. These cards cannot be used to play World in Conflict.
As opposed to the games, there is only one resolution, 1280x1024, here. 3DMark06 uses it by default.
The PowerColor SCS3 HD4650 has the lowest result in each category and is inferior even to the GeForce 9500 GT. As we have already seen in the gaming tests, the reduction of the GPU frequency and the use of slow memory have a fatal effect on the performance of the RV730-based products. The Radeon HD 4670 isn’t brilliant, either, being considerably slower than the GeForce 9600 GT.
We minimize the CPU’s influence by using the Extreme profile (1920x1200, 4x FSAA and anisotropic filtering). We also publish the results of the individual tests across all display resolutions to provide a full picture.
With 512 megabytes of memory on board the PowerColor SCS3 HD4650 could pass 3DMark Vantage tests in Extreme mode but its result is lower than 700 points due to its low clock rates, although the Radeon HD 4670 with GDDR3 memory scores over 1000 points.
So, the question is if you can spend an evening playing on a HTPC equipped with a rather low-performance entry-level or bottom-mainstream graphics card? Our test session doesn’t give a definite answer. Take a look at the summary diagrams that provide a full picture of performance:
The answer is positive when it comes to graphics cards like ATI Radeon HD 4670 GDDR3 or Nvidia GeForce 9600 GT, especially if you have a HD-ready panel only (it supports 720p mode which corresponds to a computer’s resolution of 1360/1366x768 or 1280x720 pixels). Of course, resource-consuming games as Crysis Warhead or S.T.A.L.K.E.R.: Clear Sky will be slow on such hardware, but the overall picture is quite optimistic: the average and bottom speeds are comfortable enough in most applications.
The same goes for 1080p mode and the GeForce 9600 GT card. The Radeon HD 4670 is more or less good in this mode, too. You can play at 1920x1080 on these cards in most of the tested games, but the Radeon HD 4670 cannot provide a comfortable bottom speed in such titles as Call of Duty: World at War, Prince of Persia and Fallout 3.
The entry-level solutions like the PowerColor SCS3 HD4650 512MB DDR2 are rather disappointing in 1080p mode even if you turn 4x FSAA off. On the other hand, these cards can deliver comfortable performance in such games as Call of Duty: World at War, Enemy Territory: Quake Wars, Left 4 Dead, Dead Space, Devil May Cry 4, Race Driver: GRID and X³: Terran Conflict. Not bad for entry-level products!
The specific graphics card we have tested today – PowerColor SCS3 HD4650 512MB DDR2 – can boast a number of multimedia-related advantages over Nvidia’s inexpensive solutions. Particularly, it offers hardware acceleration for VC-1 video, delivers excellent post-processing quality (as is indicated by the high results of the HQV and HQV HD tests), and features an integrated HDMI-compatible audio core that helps get rid of extra cables. Besides, the PowerColor SCS3 HD4650 512MB DDR2 is absolutely silent, which is an important property for a HTPC system that is meant mainly for watching movies and listening to music. The only limitation is that this card should not be installed into low-profile cases with poor ventilation, but a roomy model like Antec Fusion, Zalman HD or Chieftec Hi-Fi will make a good home for it.
Highs:
Lows: