new diamond from FlanK3r - AMD FX-8120 first part
It has been more than a week since we had reviews new generation of processors. The performance has known me for some time before the reviews and I have a few days before the launch of the 8120 FX had at home. Processor I underwent standard testing methodology, which prevails more theoretical benchmarks in speed when I still managed to add a practical and decompressing video conversion program Handbrake. Therefore, take this part but rather a preview of what will come later to test 8150th flagship FX For it is the focus on practical things and some ideas of your suggestions in private messages definitely take over. Just ahead of you, please be patient, the FX is not the only thing all day on my work:-D ....
Processor FX 8120 I was surprised by its 95W versions. Let's look at a few facts about the processor FX-8120 95W:
stock frequency: 3100 MHz
frequency with TurboCore 2.0: from 3400 MHz up to 4000 MHz
frequency HT: 2600 MHz
frequency CPU-NB: 2200 MHz
dualchannel support RAM up to 1866 MHz DDR3
http://img812.imageshack.us/img812/2...uidteploty.png
At idle, however, the processor moves only on the frequency 1400 MHz with voltages below 1V, more precisely, 0.87V. As you can see, processor temperature range only 31 C [i] (cooled by Noctua NH-D14) [/ i]
If the processor is rated at 3100 MHz clock, the tension is also very low and only 1.128V:!: When Turbo is at 3400 MHz still grow at low voltage 1.25V, the maximum turbo at 4000 MHz, it is still acceptable 1.39V.
Turbo is jumping up and down from the idle states, through the basic beat up both states of Turbo, the same is true for the voltage and processor in fact is almost all in full "peace". During the work is the most common frequency 1400 MHz to 3100 MHz, the load is held between 3100 MHz to 3400 MHz. The second turbo only really flies up occasionally. Even if we look at the status of hwinfo,the software from there are not still too smart:).
http://img153.imageshack.us/img153/2058/hwinfok.png
How and what I tested?
Mezi testy jsem zařadil: superpí 1M, Hyperpi 32M, MaxxPi 4M, MaxxMem, MaxxFlops, AIDA Queen test, AIDA Hash, AIDA memory benchmark, Geekbench, Fritzchess benchmark, wprime 32M a 1024M, y-cruncher, winrar, 7-zip komprese/dekomprese, editing photos from car-salon action to *.jpg and 1024x768 resolution in Zoner Photo Studio,
http://img51.imageshack.us/img51/8291/genez022.th.png
rendering in Cinebench R10, Cinebench R11.5 and POV-RAY 3.7, work with the video files using the x264 benchmark 3.0 and 4.0, encoding the AVP trailer in AVS converter,
http://img835.imageshack.us/img835/4280/avssz.th.png
encoding of video "Inside Job" from *. avi to DVD using ConvertXtoDVD,
http://img198.imageshack.us/img198/3...verxdvd.th.png
In Media Espresso the same movie to mp4.DivX 640x480.
http://img801.imageshack.us/img801/2...pressop.th.jpg
Next 3D Mark Vantage (CPU skóre) megatasking = when I was in the background as the first benchmark is winrar and run the CINEBENCH R11.5 and MaxxPi 64M. In the results, however, counting only CINEBENCH R11.5 and finished calculation in MaxxPI (winrar is still running in the background).
With FX I tested also Handbrake and decompression in practice. With Thuban I have still not finish all benchmarks and in this last two test I compared it only with i7-2600k at 3.8 GHz.
I've included the test results with AMD x4 980 BE AMD x6 1100T BE, Intel Core i7-2600K (overclocked to 3800 MHz turbo "off"). FX 8120 processor and x4 980 BE was tested on top motherboard ASUS Crosshair Formula V
I tested in old Windows 7 without SP1
And now first look at the results with the default setting of the FX-8120 (I set the RAM 1866 MHz, since it's official native support for FX processors, why it is unnecessary to degrade some percentage of it).
AIDA
http://img225.imageshack.us/img225/2...amemory.th.jpg
Winrar benchmark
-here's excellent result for their IMC and the improved access to RAM
http://img263.imageshack.us/img263/7499/graf9.jpg
wprime 32M, superpí 1M and Hyperpi 32M
-less is better, the old benchmark superpi and replacement for all threads Hyperpí definitely does not quite fit with FX....
http://img696.imageshack.us/img696/7559/graf1s.jpg
geekbench, fritzchess benchmark and cinebench R10
-diagnostic multithreaded benchmarks, calculating chess moves and rendering looks better for FX
http://img405.imageshack.us/img405/568/graf2.jpg
AVS video encoding, y-cruncher result, wprime 1024M and rendering in POV-RAY
-Although AVS loads more fiber, but failed to efficiently. Y-Cruncher is the exact opposite-it is a very demanding calculation, wprime 1024M is also demanding on the processor and the same can be said for rendering in POV-Ray. Everything is in seconds, so less time is better. Here the total FX hold, it succeeds best in POV-Ray.
:!: in this graph I had wrong 2600k with FX 8120... :!:
http://img855.imageshack.us/img855/9854/graf3z.jpg
practice work with Zoner Photo studio, MaxxPi
-MaxxPi can effectively load cores, Zoner is quite commonly used photo-graphic editor. Zoner in practice seems surprisingly better at AMD CPUs
http://img830.imageshack.us/img830/9812/graf4.jpg
AIDA benchmarks and MaxFlops and MaxMem
- AIDA benchmarks used more for fun, and at MaxFlops maxmem is a theoretical throughput, more is better.
http://img36.imageshack.us/img36/3500/graf5.jpg
x264 video encoding and 7-zip compression/decompression
-x264 benchmark - I used no function AVX and XOP without. x264 is more powerful domain of FX processors even if the FX 8120 at the fundamental frequency. Compression and decompression in 7-zip is also a very good level.
http://img12.imageshack.us/img12/5944/graf6.jpg
winrar in practice, handbrake encoding and mediaespresso encoding, megatasking nr1
-handbrake and FX is the ideal combination as it would look like with the FX everywhere:), practical archive decompression experienced such good growth, there is FX 8120 very close to the Sandy Bridge 2600K at 3800 MHz! In megatasking however literally crushes x4 980 BE processor and not so far away from i7 2600 k.
http://img846.imageshack.us/img846/3849/graf7.jpg
Cinebench R11.5 and megatasking nr2 in R11.5
- again surprised megatasking (more is better) in R11.5.
http://img839.imageshack.us/img839/6080/graf8.jpg
The first part of minireviews is over, I wanted to please you even more simulations across Specview CAD applications, unfortunately, long after the test I just ran the table with no results: (...
http://img98.imageshack.us/img98/4464/specview.th.jpg
What I stressed? The clear benefit is better to work with RAM-decompression practical (and probably even compression) driven competition, practical encoding in Handbrake is excelent, and we are convinced that increased megatasking (1100T I have not tested it yet), since it is quite close to the core i7 2600 k at 3.8 GHz. The mistake is definitely the weaker performance of single thread and skipping in the modules, it can make a difference around 5%. Here, perhaps in the future AMD will work, because there is a fundamental stumbling block between mediocrity / and tarnish the good product. Had the increased performance of one core course would be increased by a significant proportion also multithread and FX would have looked better. Even though I can not say that this is a downright bad product. Somehow fit into today's average bid, and who wants to buy it, buy it.
Like most of all I see is this offer FX-8120, which, as you will see next time, offers a very nice OC potential and still acceptable price and consumption. I can not forget the motherboard ASUS Crosshair Formula V, which is whether indeed the FX processors tuned and witness the fact that it was in 80% of all reviews used just her.
PS:srry for english, it was quickly from translator. Later we will see OC and OC benchmarks with FX 8120 (next week). And in next month very,very interesting review in practice applications with FX 8150(stock and OC setings) compared to 2600K and 1100T OC
