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:


[IMG]http://1.bp.*************/-yV4O0s7Bw9E/T6ELYY-QqmI/AAAAAAAAATk/pzU69yeI0jU/s1600/PART1-2.png[/IMG]

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.*************

[polyzp]

Part II

[polyzp]

Part III

[polyzp]

Part IV

[polyzp]

Part V

[XBEAST]

I feel sorry for you, I really do.

[---TK---]

you say the IB skyrocked to 78c during super pi, what cooler was used on the 3770k? you didnt mention SB temps during that test. I will post up a 32m super pi at 4.7ghz and my max temp. temps on cpu`s are not valid unless the same cooler and TIM are used as well as case and ambient temps


Uploaded with ImageShack.us

[polyzp]

Thanks for the info. Sandy bridge generally runs less hot than ivy at the same high overclock, and it looks like to get to 4.9 ghz on ivy voltage is pushed enough to put the chip in a very hot region. (especially around 1.5v in the data point). My chips at 1.45V and doesnt go above 59C on the hottest core, your chip at 1.41V gets 64C on its hottest core. Yet my chip is notoriously more power hungry while overclocked.

Another factor that allows them to only be compared with a grain of salt is how intel and amd chips measure temps differently. But the 28C difference just shows that its generally hotter, not that its exactly 28C hotter in all cases.

Ill be doing a power consumption comparison as well due to high demand.

[---TK---]

you keep harping on IB heat but I do not see a super pi 32m on IB at least as far as I can tell, surely it will beat my 7m 31.436. BD did 21 minutes+ and 16:50, thats pretty terrible.

[polyzp]

Originally Posted by ---TK---

you keep harping on IB heat but I do not see a super pi 32m on IB at least as far as I can tell, surely it will beat my 7m 31.436. BD did 21 minutes+ and 16:50, thats pretty terrible.

SuperPi is probably the most anti-FX benchmark i know. FX even falls far behind thuban with this 6 year old benchmark.

[---TK---]

heh, you tested super pi 32m, and even commented on IB "heat" in your 8150 test, plus you did not supply any IB, SB scores. IB can run that test like 2.5x before your test was done on the first pass.

[polyzp]

Nonetheless Superpi scores are a bad representation, by themselves, of the overall capabilities of each chip, performance wise. In each case, 78c and 59C are at the lower spectrums of temperatures for each chip. I will retest with 3DMark 11, where ********* has shown 3770k temps approaching 100C.

[Agent-A01]

another one?

[---TK---]

Yeah, wonder if he started 15 other threads at various forums.

[silis]

Originally Posted by polyzp

Part V

Here is one more for you, how great it is again.

http://www.guru3d.com/article/geforce-gtx-690-review/12

[IcE]

Originally Posted by silis

Here is one more for you, how great it is again.

http://www.guru3d.com/article/geforce-gtx-690-review/12

Okay, I lol'd

[---TK---]

thats godawful thanks for posting. anybody thinking about getting a 690 with the flagship 8150 please read

[polyzp]

Originally Posted by ---TK---

thats godawful thanks for posting. anybody thinking about getting a 690 with the flagship 8150 please read

A 690 is overkill for a stock FX 8150. Anyone buying that card for only 1080p gaming is way over their head. Lets see some triple monitor gaming with that same gtx 690 at a high overclock on the FX.

[---TK---]

Originally Posted by polyzp

A 690 is overkill for a stock FX 8150. Anyone buying that card for only 1080p gaming is way over their head. Lets see some triple monitor gaming with that same gtx 690 at a high overclock on the FX.

who said 1080p? that was just 1 of the resolutions in the link
it goes up to 2560. the link just shows how terrible the cpu is at stock.

[polyzp]

Originally Posted by ---TK---

who said 1080p? that was just 1 of the resolutions in the link
it goes up to 2560. the link just shows how terrible the cpu is at stock.

I am sick of review sites using resolutions lower than 1080p for high end gaming GPUs, because its not realistic.

A single 680 is more than enough for 1080p gaming,

two 680s is just overkill.

For such an expensive GPU, it would be better if they did 1080p at the minimum, 2560, then triple monitor.

[---TK---]

its called a cpu scaling test for a reason bro. the lower the resolution the worse a bad cpu performs.

[polyzp]

Originally Posted by ---TK---

its called a cpu scaling test for a reason bro. the lower the resolution the worse a bad cpu performs.

Real world performance is more relevant in my opinion. Most other review sites did eyefinity resolutions in their reviews of the gtx 690.

[Spets]

Originally Posted by XBEAST

I feel sorry for you, I really do.

This..

Also people running 3 monitors are going to choose a better CPU, just like reviewers do.

[---TK---]

Originally Posted by polyzp

Real world performance is more relevant in my opinion. Most other review sites did eyefinity resolutions in their reviews of the gtx 690.

sound kinda mad guru3d put an 8150 in the 690 review? just that one page tells people to stay away from that cpu no matter what the resolution. guru 3d do some impartial reviews.

[kapu]

Hilbert @ GTX690 Review.

Now I deliberately included an AMD FX 8150 processor as AMD's current offerings are just not well suited for high-end SLI.

FX Series are weak for REAL life usage.