3800 X2 vs Opteron 170 Graphic Benchmark Results

Results from same system, same ram sticks with same configurations.

PNY 6800GT kept at same speed on all test (416/1100)
All graphics setting set to minimum on Nvidia driver window, No AA/AF
Vsync Off.

2540 Mhz
2x512 MB Ram 1T 2.5/3/3/8 (OC @ 212 DDR)
HTT = 4
Futuremark 01

3800 X2 Opteron 170
26850 27894
26853 26600

Futuremark 03 (Overall score/CPU score)

3800 X2 Opteron 170
14046/1270 14172/1344
14027/1265 14171/1345

Futuremark 05 (Overall score/CPU score)

3800 X2 Opteron 170
6041/6404 5895/6700
6019/6657 5972/6984

Aquamark

3800 X2 Opteron 170
84323 85077
84133 85381

Everest memory bandwidth

3800 X2 Opteron 170
5808/2541/52.5 5510/2490/54.8
5812/2556/50.5 5572/2485/54.4
5833/2539/52.2 5713/2483/53.5

2540 Mhz
4x512 MB Ram 2T 2.5/3/3/8 (OC @ 212 DDR)
HTT = 4

Futuremark 01

3800 X2 Opteron 170
26549 26519
------ 27662

Futuremark 03 (Overall score/CPU score)

3800 X2 Opteron 170
12918/1193 14149/1276
-----/----- 14136/1274

Futuremark 05 (Overall score/CPU score)

3800 X2 Opteron 170
5962/5857 5972/6287
----/----- 5938/6101

Aquamark

3800 X2 Opteron 170
83166 84237
------ 84259

Everest memory bandwidth

3800 X2 Opteron 170
5690/1921/57.5 5607/1903/58.3
5605/1923/56.8 5609/1913/58.2
5694/1927/55.7 5626/1910/58.8

2340 Mhz
2x512 MB Ram 1T 2.5/3/3/8 (OC @ 234 DDR)
HTT = 4

Futurmark 03 (Overall score/CPU score)

3800 X2 Opteron 170
13891/1236 14010/1302

Futuremark 05 (Overall score/CPU score)

3800 X2 Opteron 170
prgrm crash prgrm crash

Aquamark

3800 X2 Opteron 170
82794 83562

Everest memory Bandwidth

3800 X2 Opteron 170
5918/2578/51.4 5662/2488/53.4
5910/2583/51.3 5833/2491/52.7

good job!!
imho the ram changes are more interesting than the changes in cpu. i would like to see a highly cpu limited game in this if you dont mind. also i would like to see some encodeing comparisons.

Intresting.

Coming from a X2-3800 to a Opty 170 these are some intresting results. But my opty goes a little further then my 3800, and hopfully a little more over the long haul.

Oh ya the 1T to 2T is overall a 1-3% change in prefromance over the long haul. More and more tests are coming up with this since its the big question as of lately.

Nice results look forward to more if your doing them

Nice work! I'm assuming that the better results coming from the 170 are due to the larger cache. I would guess it's about identical to the 4400 performance if it was clocked at the same speeds.

I'm not understanding how to read these results..

Sorry, I couldn't get the space bar to register on this post at all, that's why the numbers are crunched so closely together.

processor 1 processor 2

graphics score/cpu score
(no "/" indicates only a graphics score)

memory read/write/latency

the 3800x2 is sold and en-route to it's buyer.

great work man! rseven is right that the two should be pretty much the same, with the opty being a bit higher due to its 1 mb cache. however, let's not forget that the opterons can oc better. i'm sitting at 2.8ghz , lets just say christmas has come early this year.

aznkc730 said:

great work man! rseven is right that the two should be pretty much the same, with the opty being a bit higher due to its 1 mb cache. however, let's not forget that the opterons can oc better. i'm sitting at 2.8ghz , lets just say christmas has come early this year.

Click to expand...

X-mas came early here too!

nice work, it seems the X2 gets more memory bandwidth than the Opty. Christmas hasnt come yet

Thanks for the benchies Kimo

lol xmas wont come for me. only 2 places in NZ sell the opty 170, and they are both selling for $720+.. oh well, no milk and cookies for santa this year im afraid

Careface*

party animal said:

nice work, it seems the X2 gets more memory bandwidth than the Opty. Christmas hasnt come yet

Click to expand...

I was reading on XS that the Manchester's consistently got better memory scores then their Toledo counterparts due to their stronger memory controller. Since the Denmark is basically a Toledo I'm assuming the same applies here.

$720, that is rediculous. Are things normally that expensive in NZ, or because the opty is saught after.

Good work and thanks for those results. I'll add a few results when I get my Opty 165.

Christmas will be on time here

A few questions/observations:

1.Opty < X2 in FM01 test# 2 (CPU test I believe)
2. Opty beats the X2 in all other FM runs.
Why is this? Can you run FM01 agian?

3. The Opty < X2 in the Everest bandwidth section. Why is that?


1.

Super Nade said:

Good work and thanks for those results. I'll add a few results when I get my Opty 165.

Christmas will be on time here

A few questions/observations:

1.Opty < X2 in FM01 test# 2 (CPU test I believe)
2. Opty beats the X2 in all other FM runs.
Why is this? Can you run FM01 agian?

3. The Opty < X2 in the Everest bandwidth section. Why is that?

Click to expand...

Exactly what I was wondering too!
It does seem strange about the Everest bandwidth tests favoring the X2. Could this be because of higher cache latency with the 2x1mb cache on the Opteron?

Unfortunately I didn't have enough time to run all the tests on the 3800 X2 that I wanted with the 4 sticks. I had to ship the processor off soon. I kind of regret not staying up 2 more hours to get at least 3 more runs from the 3800 X2.

All test were done on a freshly installed Windows XP Pro OS with only the minimum drivers required for the OS to work. The newest Nvidia dual core drivers used were dated from last week. EACH and EVERY test was done on a fresh boot. I would run one test and reboot, then do the next test again. Usually done with a graphics test, reboot, then Everest memory bandwidth test, reboot, then graphics test again. This was an open case computer laying on its side with minimal heat held in the case. Oddly, although it's probably more subjective than anything, the system 'seemed' to score higher when warmer(???). Maybe it's just components having to deal with less change in temp variations.

Super Nade:

#1. I unfortuntely only ran 1 test for that on the 3800 x2. I really believe that the x2 would have scored lower again, had I run a second test. My memory (my mind that is) serves me that the those scores always seemed lower than my current 170 scores. Had I run a second FM 01 test on that one I believe you still would have seen scores lower on the x2 and consistent with the rest of the tests.

2: The 170 is a tad higher accross the board on all graphics benchamarks, except for FM 2005. Even as we speak now, my tests still show that FM 2005 favors the 3800x2 by a minimal margin. I don't know why this is. FM 2005 is the questionable program here, maybe it's affected negativly by 1mb cache on each core (maybe similar to something like have 2 gigs ram to run an OS sometimes does not provide better performance because the memory address is much larger to allocate a small amount of active memory??). BTW, all programs were run in dual thread mode).

3. EVERY TIME i've run Everest memory bandwidth it has shown lower scores than the X2. This even applies to varying cpu speeds and FSB. I'm at a complete loss about this. However, you may notice that the gap becomes smaller when using 4 sticks of ram.
I've had 3 processors on the exact same system. I can tell you that the dual cores lost in memory read performance against my 3500 clawhammer (fx53/55 core ada3500dep4as), BUT were better on memory write than my single 3500 Clawhammer.

Wow you are real patient to have run all those test, nice job.