Return of FX : GA-990FXA-UD7 - OC'd AMD FX 8150 / 6990 Performance Comparison Review

[polyzp]

PART I of V : Return of AMD FX, Enter Gigabyte








My System:


AMD FX 8150 @ 4.90 Ghz
Promilatech Genesis Cooler - 3 x 135mm Scythe Kama Flex 100CFM fans
Maingear T1000 TIM
8 Gb DDR3 Team Xtreem 2400 cl9 @ @ 1987 Mhz / CL9
Gigabyte GA-990FXA-UD7
XFX HD 6990 + Accelero Twin Turbo Cooler @ 990/1500 Mhz
RevoDrive 3 x2 - 240 Gb


The first thing that came to mind when I first laid eyes on my new Gigabyte 990FX UD7 board was that it was quite the beautiful board. Much respect for the simple PCIE 16x slot latches, the Crosshair V uses a more traditional type, and sleek heatsink design.

EDIT: It should be noted that I received a revision 1.1 board from Newegg Canada, and that it does have Load Line Calibration (LLC) controls in the bios.






The new Team Xtreem DDR3 ram has a cheap but cool looking reflecting sticker, but what is most appealing are the stats: 2400 Mhz @ Cl9, which is only surpassed by Corsair's grossly overpriced and discontinued Dominator GT dual channel flagship memory.







The memory height is not found online, and could be helpful to potential buyers.

Team Xtreem memory height : 52mm

Promlatech Genesis allowed height: 54mm

So this is awfully close to the tallest ram you could use with this cooler. I have seen Dominator GT ram at a height of 54mm barley fitting underneath the heat sink, found here, and a 53 mm Corsair Vengeance stick here.



Comparison Systems :




AMD FX 8150 @ 4.81 Ghz
Promilatech Genesis Cooler - 3 x 135mm Scythe Kama Flex 100CFM fans
Indigo Xtreme TIM
G. Skill 2200 cl7 @ x 2190 Mhz / CL10
ASUS Crosshair V
XFX HD 6990 + Accelero Twin Turbo Cooler @ 990/1500 Mhz






Intel Rig


Intel i7 3770k @ 4.7-4.9Ghz
Various 8 Gb DDR3
Various Z77 Motherboards
Various GPUs


Results:





Sources:


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26


Discussion:

We see that for the benchmarks used in Part I of my review, FX only falls signficantly behind when only a single thread is used. When only one core is being used we see a 25-40% benefit per Ghz for the 3770k over the FX 8150.

When all threads can be used, Bulldozer does a good job of making up for the lost single threaded performance. Scaling for FX outshines the i7 3770k by a significant margin. Most notably with techarp's x264 HD, where my FX 8150 @ 4.9 Ghz beats a 5.0 Ghz 2600k, and narrowly loses to a 4.9 Ghz 3770k. (Check source 19)

In TrueCrypt 7.1a we see a 4.9 Ghz FX 8150 performing slightly better than its 22nm 3770k intel counterpart at 4.7 Ghz.

In 7-Zip we see the FX 8150 jumping 2.7 % percent ahead of its 3770k counterpart at the same 4.9 Ghz clock for Compression, but falling behind 2.7% with decompression.

The temperature of a 3770k is also seen to sky-rocket up to 78C during a SuperPi 32m test, while my FX 8150 doest hit above 59C.

Next week well see how my GPU handles graphically intensive workloads on my new Gigabyte board.

Stay Tuned! I am free to answer any questions and also am willing to take any requests you guys might have!

http://AMDFX.blogspot.com

[polyzp]

Part II

[polyzp]

Part III

[polyzp]

Part IV

[polyzp]

Part V

[FlanK3r]

score for 32M at Asus C5F is strange...Can not be so big diference in 100 MHz CPU clock...

[polyzp]

I thought this too but according to my old screenshot thats what it scored. My score now seems alot more common for an overclocked FX 8150.

[demonkevy666]

I think my flares are taller >_< 60mm and also discontinued.

[BeepBeep2]

Something is seriously wrong with your 32M result...
Also, why do a board comparison with different clocks? ...then you have your benches of a 3770K at all different clock speeds from others but you didn't run them yourself...
Also try to take into account power consumption if you are going to really compare FX and 3770K.

[polyzp]

The gigabyte board actually allows a higher stable overclock than the Crosshair v so I decided to compare max oc to max oc, like most review sites would do.

Also, the 3770k results are just collected from numerous sources so of course hardware will be different, but generally for CPU intensive tests the most important factor is the cpu and its clock.

I will be doing a power consumption comparison in the sections to come!

[mAJORD]

mm.. I don't think "scaling" is a valid performance metric :S

and just a small correction. Single Threaded 'benefit' for IVB should be 30-65% over BD "Ghz/Ghz"