While it may not be the most graphically demanding game on the market, the Sims 3 just happens to be one of the most popular. In this test we use FRAPS to monitor the framerates as the camera is panned through a house and an outdoor environment.
FarmVille may be one of those games which you either love or loathe but there is no denying its popularity considering at any one time there are over 20 million players tending to their farms and harvesting crops. While it may be fun, many players have expressed concerns over the fact that this simple web-based game demands such a massive amount of system resources. However, there is an initiative spearheaded by Adobe, NVIDIA and ATI to harness the parallel processing nature of modern GPUs to accelerate non-3D Flash-based applications.
With the slow transition from Flash 10 to Flash 11, there has been a number of Betas (10.1 Beta 1 and the newer 10.1 Beta 2) released that trumpet support for GPU acceleration in Flash-based environment. This GPU-based acceleration with ATI’s Stream and NVIDIA’s CUDA is mostly geared towards the streaming of high definition videos from services like YouTube and Hulu but it has the capability to make the transition into acceleration of Flash-based games as well. That’s where FarmVille comes into the equation. It is without a doubt one of the most demanding games on a modern CPU once your farm has grown to a certain size and can benefit quite a bit from GPU acceleration. Unfortunately, neither AMD nor NVIDIA make any mention of acceleration of Flash games in their press releases.
For this test we used an existing farm which had been built up to a high level and gradually manipulated objects, panned the camera, harvested crops and purchased items over a 10 minute test session. The CPU utilization was monitored and logged by Windows 7 Performance Monitor running a custom profile. In order to further tax our system and not introduce any additional variables into the equation, HyperThreading and Turbo Boost on the processor were disabled. Below, we see the differences between Flash 10 and Flash 10.1 Beta 2 with Hardware Acceleration enabled in the Flash control panel.
It goes without saying that at this time the majority of Flash games are not accelerated on the GPU but what was surprising is that the 10.1 Beta 2 seems to put even more load on the system than the older Flash 10 application.
That being said, there are some interesting things going on here so let’s begin with the star of today’s show: the Sapphire HD 5450 512MB. In a Flash 10.1 environment, it surprisingly looses out to the NVIDIA cards even though the actual amount of GPU load in all cases was less than 5%. On the other hand, it showed quite good performance when using Flash 10 where it beat out the NVIDIA cards by about 7%. This difference could be due to anything from driver overhead to programming incompatibilities but it should be interesting to see what happens in the coming months as the official released of Flash 11 draws close.
The GT 210 seems to have a few issues with Flash acceleration as well. Actually, Flash games don’t get a performance improvement in any way that we could see and once again Flash 10 seems to run more efficiently than the 10.1 betas.
Our huge surprise of the day comes with the GT 220 which for some reason showed about the same amount of CPU usage as the GT 210 in Flash 10 but actually saw a 3% improvement when using 10.1 Beta 2. This had us scratching our heads but according to GPU-Z’s load monitor, it did actually seem like the GT 220’s core was being utilized more often than the other GPUs. Whether it is a difference in driver writing or something else, at this time the GT 220 does seem to marginally help with Flash game acceleration.
Finally we come to the integrated graphics core on the i5 661 processor and it is pretty evident that it is only the CPU being utilized here. The 2% difference between Flash 10 and 10.1 is well within the margin of error especially considering the maximum CPU usage we saw in both cases was around 92% across all cores.
With YouTube’s move to high definition videos, many viewers suddenly found that they could not watch the newer, better looking videos without their computer getting bogged down. Granted, this is mostly an issue for netbooks and laptops but there are still plenty of lower-end and dual and single core desktop processors out there which just don’t have the horsepower necessary to process 720P or 1080P content. This means more often than not, that HD video you are intent on watching quickly turns into a slide show.
Adobe’s Flash 10 does offer some basic hardware acceleration for HD video streaming from YouTube, Hulu and other online sources but the 10.1 version takes things to the next level by offering Flash video processing which is accelerated by the GPU. This frees up CPU resources, allows for smoother playback on low-end systems and also provides an added benefit of (supposedly) higher picture quality.
In this test we watched a random selection of 1080P YouTube videos ranging from IMAX trailers to movie trailers and beyond while logging CPU usage in the Windows 7 Performance Monitor. In total, the selection of videos accounted for about 15 minutes of playback time per test.
Unlike our tests with the Flash-based FarmVille game, it is quite evident that the move to Flash 10.1 has a pretty dramatic effect on the overall performance of high-def YouTube videos. While a savings of 15% CPU usage might not be much, it could be the difference between smooth playback and a distractingly jerky video.
Both the GT 210 and the HD 5450 offer much the same performance which ends up being better than the IGP on the i5 661 but still trailing behind the GT 200.
A Question of Improved Picture Quality…or not.
In their documentation, ATI claims increased picture quality when it comes to Flash-based video acceleration. Did we see any improvement with picture quality? Absolutely not during full motion playback but there seemed to be a very slight difference when the video was paused and we had time to thoroughly examine the picture. The differences were so small however that we doubt that you would ever be able to tell the difference.
In order to test the high definition video decoding capabilities, we loaded up CyberLink PowerDVD Ultra 9.0.2320 and played high definition 1080P media with three different types of HD codecs, namely VC-1, H.264/AVC, and WMW HD. Hardware acceleration was enabled in PowerDVD to take advantage of the accelerated decoding capabilities of our GPUs.
Chapter 18 of the Batman Begins Blu-ray DVD was our source for VC-1.
Chapter 14 of the Transformers Blu-ray DVD was our source for H.264/AVC.
The entire clip of Magic of Flight from Microsoft.com was our source for WMW HD.
Remember, in these tests our CPU was running at 3.33Ghz which is quite fast for even the best dual cores currently on the market. As such, the CPU usage you experience on these tests will most likely be much higher. CPU usage was once again logged by the Windows 7 Performance Monitor.
Also note that we used a REFERENCE HD 5450 to run these tests and NOT the Sapphire HD 5450 due to the fact that the product we were sent is not equipped for HD playback through HDMI.
Here we see all of the discrete GPUs having roughly the same performance with the HD 5450 slightly edging out the GT 210 yet trailing the GT 220 by a narrow margin. Honestly, as long as you have a good CPU or dedicated GPU, you should be fine for VC-1 playback.
H.264 / AVC Playback
H.264 decoding is a bit harder on the system than VC-1 but nonetheless, all of the products we tests passed with flying colors. Things do change around a bit though with the HD 5450 now leading the way with the least amount of CPU usage.
WMV HD Playback
Here we see that the HD 5450 and ATI’s AVIVO HD seems to be much better at WMV HD decoding than NVIDIA’s PureVideo-equipped GT 200-series. The difference really is much more than marginal here but once again all of these GPUs should provide you with more than adequate WMV playback.
For all temperature testing, the cards were placed on an open test bench with a single 120mm 1200RPM fan placed ~8” away from the heatsink. The ambient temperature was kept at a constant 22°C (+/- 0.5°C). If the ambient temperatures rose above 23°C at any time throughout the test, all benchmarking was stopped. For this test we use the 3DMark Batch Size test at it highest triangle count with 0xAA and 0xAF and looped it for one hour to determine the peak load temperature as measured by GPU-Z.
For Idle tests, we let the system idle at the Windows 7 desktop for 15 minutes and recorded the peak temperature.
When it comes to acoustics, obviously this card is the quietest of the bunch due to its passive heatsink. This heatsink also acts like a double edged sword since it also contributes to the HD 5450 showing some of the highest temperatures of the bunch. However, we are sure that in a well-ventilated case, there won’t be any need for concern.
For this test we hooked up our power supply to a UPM power meter that will log the power consumption of the whole system twice every second. In order to stress the GPU as much as possible we once again use the Batch Render test in 3DMark06 and let it run for 30 minutes to determine the peak power consumption while letting the card sit at a stable Windows desktop for 30 minutes to determine the peak idle power consumption. We have also included several other tests as well.Please note that after extensive testing, we have found that simply plugging in a power meter to a wall outlet or UPS will NOT give you accurate power consumption numbers due to slight changes in the input voltage. Thus we use a Tripp-Lite 1800W line conditioner between the 120V outlet and the power meter.
Honestly, what can we say about the power consumption of this card? It is just phenomenal considering the performance difference between it and the NVIDIA GT 210. Make no mistake about it, the GT 210 is a very efficient card but it goes without mentioning that ATI’s 40nm manufacturing process
ATI has really been on a roll of late and the HD 5450 represents the next logical step towards absolute market domination. Within the next month, there will be a 5000-series product in nearly every price bracket which is a damn impressive feat for a company that was simply struggling to stay afloat not long ago.
When push comes to shove, reviewing a card like this one meant having to take a step back and looking at our typical methodology in a new light. It is obvious that the HD 5450 isn’t particularly good at playing games, nor will it win any awards for pushing sub-$75 performance to new heights considering the HD 4550 is priced lower and has near-identical specs. However, it is good at all the things customers shopping in this price bracket are looking for and will definitely prove to be a boon for system builders like Dell, HP and the lot since they can now slap a DX11 sticker onto their lower-end PCs.
Unfortunately, the term “DX11” is seriously just a stand-in act mixed in with marketing babble here since there is no way on the face of the planet that the HD 5450 will provide the necessary performance for viable DX11 gaming. Sure, you can turn off post processing, depth of field, motion blur, tessellation and all the other DX11 goodness but that totally defeats the purpose of using the advanced API in the first place.
When compared to the competition, the HD 5450 offers up a bit of a mixed bag but that’s not due to a lack of trying. The ~$60 512MB DDR3 version of the HD 5450 we tested in this review walks all over the GT 210 DDR2 in every one of our tests. However, from a pure performance perspective the $75 GT 220 512MB GDDR3 offers nearly double the performance while retailing for about 20% more. It seemed NVIDIA knew what was coming and has quickly cut the GT 220’s price. We’re sure the $70 HD 4650 DDR2 card would have offered a similarly massive gap as well if we had included it in the charts. That makes the HD 5450 a bit of a hard sell in our minds when you consider both the GT 220 and HD 4650 offer almost all of its HTPC functionality, have been released in low profile versions and are actually able to play games at reasonable settings. No, neither of those cards is quite as efficient as this new ATI product but the increase in power consumption yields a huge difference in overall game playability.
Something else we have to mention is Sapphire’s mind-boggling connector layout on this particular card. Granted, this “Eyefinity edition” is geared towards the professional world but we wanted to see one of two things: either a native HDMI connector supplemented by a DVI to VGA dongle or a DVI to HDMI adaptor. The simple addition of either of these simple and relatively inexpensive adaptors would have made this particular Sapphire card so much more versatile and would have even allowed them to simplify a horribly confusing HD 5450 lineup.
For those of us who view the computer hardware world through enthusiast-colored glasses, the HD 5450 will be the point of ridicule and jokes about its low performance. However, for the market it is targeted towards, it offers the absolute perfect mix of low-end gaming capabilities, HD decoding muscle and efficiency all wrapped into an extremely cost-conscious package. It beats the GT 210 512MB senseless while costing a similar amount but its overall performance is clouded by slightly higher priced cards from both ATI’s own lineup and NVIDIA’s side of the fence.
- Excellent HD video decoding performance
- GPU-accelerated Flash videos
- Extremely efficient
- Support for bitstreamed HD audio
- Price / performance versus GT 210
- Passive cooling
- You can get so much more with card that retails for $15 more
- Maintains nearly identical specifications as the HD 4550 while costing the same
- No HDMI connector on this version from Sapphire
Hardware Canucks is focused in-depth reviews of the the latest graphics cards, CPUs, storage, smartphones, notebooks and gaming peripherals. Join us as we unbox, review and benchmark the best computer hardware on the market.