Blu-ray Done Right: How Does Your Integrated GPU Stack Up?

Test Systems And Benchmarks

We'll be performing three types of Blu-ray-based benchmarks: playback image quality, CPU usage, and power usage.

As usual, we’ll use Silicon Optix’s HQV Blu-ray test disc in order to judge the high-definition playback quality of our various contenders. This benchmark consists of a series of tests that scrutinize image quality enhancement features, each of which is assigned a number of points.  When all of the tests are completed, the playback image quality of each graphics chipset is represented by the sum of all of the points it scored. The best possible total score is 100 points, and the worst possible score is zero.

To judge CPU usage, we will play back a Blu-ray disk for three minutes and record the CPU usage during this time. Since there are three different codecs (compression-decompression algorithms) used on Blu-ray disks (and each codec requires a different amount of processing power to decode) we must test each codec separately. We will use the Blu-ray disk 300 to test the VC-1 codec, the Blu-ray disk Tomb Raider to test the MPEG-2 codec, and the Blu-ray disk The Fifth Element to test the H.264 codec.

For the last CPU utilization test, we will check the capability of these graphic chipsets to accelerate picture-in-picture (PIP) video streams. To do this, we will use the Blu-ray disk Sunshine, which utilizes the H.264 codec and features PIP commentary during playback.

Finally, we will test total system power usage at the wall to see how much power H.264 playback requires compared to the PC's idle state.

Here are the particulars of the test systems we ran:

Swipe to scroll horizontally
Test System Configuration
CPUAMD: Phenom II X2 550 BE (3.1 GHz, 6.0 MB Cache)Intel: Core 2 Duo E7200 (2.53 GHz, 3.0 MB Cache)
MotherboardsAMD 785G: Asus M4A785TD-M EVOAMD 790GX: Asus M4A78T-EGeForce 8200: ECS GF8200AGeForce 9300: Asus P5N7A-VNIntel G45: Asus P5Q-EM
RAMDDR2: A-Data AD2800E002GOU (2 x 2 GB) DDR2-800 at DDR2-800 CAS 5-5-5-18 2TDDR3: G.Skill F3-10666CL7-2GBPK (2 x 2 GB)DDR3-1333 @ DDR3-1338 CAS 9-9-9-24-33-1T
GraphicsRadeon HD 4200 (785G) - 500 MHz GPURadeon HD 3200 (780G)  - 500 MHz GPU (simulated using underclocked 790GX)Radeon HD 3300 (790GX) - 700 MHz GPUGeForce 8200 - 500 MHz GPU, 1.2 GHz shaderGeForce 9400 - 540 MHz GPU, 1.4 GHz shader (simulated using overclocked 9300)Intel G45 - 800 MHz GPU
Hard DriveWestern Digital Caviar Black 640 GB 7,200 RPM, 32 MB Cache SATA 3.0 Gb/s
SoundIntegrated HD Audio
NetworkIntegrated Gigabit Networking
PowerThermaltake Toughpower 1,200 W, ATX12V v2.2, EPS12V
OSMicrosoft Windows Vista Home Premium x64 SP1
GraphicsGeForce 8200/9300: Nvidia GeForce 190.38 WHQLAMD 790GX: Catalyst 9.7AMD 785G: 8.632-090626a-083962E Beta
  • Proximon
    Great article. I think maybe the 4650 is a bit overkill, but that's just nitpicking.

    As long as you are talking about HTPC builds though, you might want to mention temps... aren't the 9300/9400 boards very hot?
  • epsiloneri
    Power draw is not interesting because of the electricity bill, it is the generated heat needed to be dissipated with the associated noise levels due to cooling that is critical for an HTPC.
  • HalfHuman
    i don't get it why a home theater would use a 1200w power source. at the same time i don't get why would someone evaluate the power efficency using this kind of power sorce. if you ask me i'd make this crazy ass power supplies illegal. a normal hometheater should not use more than 50w at idle and 100-150w at load. seems that this is what these actually consume. factor in the less than 5% load on the power supply and you get a masterfull 50-60% power efficency. i'd love to see some proper power supply test.
  • falchard
  • falchard
    BTW, I would like to see a "Can it play Crysis" article in the future that runs down every video card and IGP, then determines if it can possibly play Crysis and at what settings.
  • HalfHuman
    the 1200w power supply is green as in blue-green mould green.
    this is in fact an excellent power supply... if you use it. at 100watts load it has a "cool" 76% efficency. if the intel pc uses less than 82watts in load and 66watts in idle you can only imagine the efficency a power supply has at below 5% load. the site suggest around 65% so instead of having a proper power supply using 40watts or less when idle, you get this "green" efficient hummer who swollows 66w. i really like you articles guys but this kind of testing is not the way to go.
  • Efficiency isn't even tested below 20% load i believe But it should still be around 70-80% it is a Thermaltake Toughpower 1200w and all of them(3 listed on their site) are standard 80% eff rated or bronze. Ture a more modest Delta,Seasonic 250w or 300w would be much more appropriate for a htpc.
  • HalfHuman
    20% for this would be 240watts and efficency would still be reasonable.
    i posted some link but i see it's been removed. that review said something about 65% minimum.
  • drew_a
    Uh, guys... you might want to edit this article...
    "For the last CPU utilization test, we will check the capability of these graphic chipsets to accelerate picture-in-picture (PIP) video streams. To do this, we will use the Blu-ray dick Sunshine, which utilizes the H.264 codec and features PIP commentary during playback."

    on page 6
  • icepick314
    "If you are an audiophile, you should know that out of these remaining options, only the GeForce 9300/9400 can handle uncompressed eight-channel LPCM audio over HDMI 1.3."

    i did NOT know this...

    i thought only way to listen to uncompressed audio on blu-ray was using Asus Xonar HDAV 1.3 audio card to bitstream to your receiver...

    it's nice to know that IGP has enough power to handle 1080p while streaming HD audio codec....