AMD A8-3850 + 6670 Review
Preface
When AMD announced its APU series, people were amazed. It wasn’t bringing anything new to the market but simply redefining what we had, and still have. AMD APU’s forever banished that stigma about “onboard graphics”, but can it rival other CPU’s in raw processing power? We have the answers here.
Setup
Our test platform consisted of the following, – AMD A8-3850, ASRock A75M-HVS Motherboard, Cooler Master Elite 460W Power Supply, XFX AMD 6670 1GB GDDR5, Kingston Value 2x2GB DDR3 1333, Hitachi 500GB SATA 3.0Gb/s, ASUS DVD Burner / Writer, and lastly the NZXT Source 210 Elite. Out total was just a pinch under $525. [slideshow id=7] Let’s just start off by saying our setup of our AMD APU test platform was hell. Unfortunately for us and other testers/early adopters, documentation is very limited. Message boards and online communities seem to be the primary beacon of information and even there information was limited. After scouring AMD’s site we were left empty handed on the setup process for crossfire or what AMD likes to call “Dual Graphics”. After hours of trial, error, and lots of BSOD’s, we finally decided to visit the motherboards manual to find a nice little section titled “Setting up AMD’s Dual Graphics”. Surprisingly after following all the listed steps we still had bugs and issues, so after a quick 2 minute BIOS update, things were smooth sailing from there. Too our surprise, AMD Dual Graphics mode required the use of an onboard video slot for crossfire use, so we recommend choosing a motherboard that has at least HDMI and DVI if not also VGA.
Performance
The performance of APU’s CPU and crossfired GPU was surprising. There were some extremely odd results but nonetheless, all very impressive. Our tests included, Windows Performance Index, Quake Live, Battlefield Bad Company 2, F1 2011, RetouchArtist Speed Test, and lastly Unigine Heaven Benchmark 2.5 Basic. All test were conducted at resolution 1680×1050 unless noted otherwise.
Windows Performance Index –
This was more of just a test rather than a benchmark. We found our results to be quite interesting…
Higher is Better
[imagebrowser id=9] This all made sense after reviewing it since, when only using the XFX 6670 as the GPU, it takes load off the onboard GPU and memory hence the higher score. Granted we recommend you take this test with a grain of salt because we don’t really know what sort of tests they are running and they obviously don’t account for crossfire.
Quake Live –
Once again this was more of just a test versus a benchmark. The APU and 6670 crossfired were able to reach the games cap of 125 FPS on every map besides one. On Quarantine with the skybox enabled and everything maxed out at 1680 x 1050, our frames dropped to a still very impressive 113.
Battlefield Bad Company 2 –
This being one of our favorite games currently out, we had to give this a test run. We did not gather any specific results as frames were very jumpy at parts but we’d say a solid 70% of the time we were able to play between 28 – 33 frames per second. We only really noticed a spike when large explosions or massive amounts of enemies and gun fire were going off. This test was done on all medium graphics at 1680 x 1050.
F1 2011 –
So this is our first test we decided to actually gather some data for. It had its own built in benchmark and allowed for consistent testing. Anyone familiar with F1 2011 knows it was a graphical improvement over last years which isn’t a slouch to begin with. It includes DirectX 11 effects, like soft shadows and different blurring techniques. Here were our results.
| Resolution | Multi Sampling | DX Ver. | Settings | Skidmarks | Avg Frames | Min Frames | Total Samples |
| 1680×1050 | Off | 11 | Low | On | 46 | 40 | 5728 |
| 1680×1050 | Off | 11 | Medium | On | 39 | 32 | 4719 |
| 1680×1050 | Off | 9 | Medium | On | 39 | 33 | 4769 |
So as you can tell from our results, our APU and 6670 crossfire was not only able to run this game very comfortably on low at 1680 x 1050, but we were also able to bump it up to Medium graphics, with some being forced on high AND DirectX 11 enabled with a very playable average of 39 and low of 32. Lastly in our final test we used the same settings as Test #2 but disabled DirectX 11 and received almost identical results. This is most likely the case because of the lack of actual DirectX 11 effects.
Speedtest Retouch Artist ( Photoshop CS5 )
This benchmark uses a decent amount of CPU and RAM inside of Photoshop to apply a large list of actions to a very high resolution image. The test is available for free on their website, ( http://clubofone.com/speedtest/). Unfortunately this test, the APU falls flat on its face. We took the average of 3 tests per system, our AMD APU test bench, and our Intel test bench and the results were very displeasing.
Lower is Better ( Seconds )
[imagebrowser id=8] Granted the i5-2500K is clocked 1.2Ghz faster per core and has 2MB more L3 Cache, we found these results to still be fairly disappointing.
Unigine Heaven Benchmark 2.5 Basic
The test you’ve all been waiting for, and the results? – Odd. This benchmark is known for some of the most intensive graphics let alone DirectX 11 effects available today. DirectX 11 brought with it a feature called Tessellation, simply put this adds physical geometry to game objects. Rather than a floor painted with a flat rock texture, each and every rock has its own 3d shape. As you can imagine this introduces a great deal of new polys and puts a great deal of stress on the video card, but outputs beautiful visuals. Well now that you have a very basic understanding of tessellation, let’s start off by saying that the AMD A8-3850 in dual graphics mode with a 6670, handles it exceedingly well. Would it be considered playable? No, but given the fact that even if Intel’s HD 3000 integrated graphics could support DirectX 11, it’s raw graphical power would not even near that of AMD’s APU. That being said here are the results, [nggallery id=10] After taking a look at the results you’ll notice that with a quality bump in DirectX 9, the system performed increasingly worse. Versus its impressive DirectX 11 benchmarks, that yes did have lower texture settings but tessellation on medium that did not only noticeably better frame wise, but also looked a significant amount more realistic.
Conclusion
At the end we were left with one question, are AMD APU’s worth it? The only answer we could come up with was, it depends. Are you the internet and email person with the occasional light gaming? Yes, the APU handled multitasking very well while we installed a trial version of CS5, played Quake Live, and transferred data from our external drive. Do you work with some sort of editing software such as Adobe’s Suite and Autodesk products, or heavy gaming? Probably not, the platform was relatively cheap, was very low on power consumption, and very quiet. It did handle all our tests, some better than others but for a small price jump we feel you could either get a far better workstation / gaming rig. Where we see the APU platform is in laptops and netbooks. It does offer the best CPU and GPU performance combination on a single chip; but it is easily outpaced with a well budgeted AMD Phenom or Sandy Bridge system.