Who's telling the truth here? (Q's on benchines)

[jahjahbinks]

I decided to put my OC AMD 4400+ (2706MHz) on test with my Pentium D 830 (3GHz, stock). And the results were rather surprisng. Besides HD (I got WD Raptor on the 830 and Barracuda 7200.7 on 4400+), everything else is the same (of course MB, CPU and RAM aren't the same :p: ), Graphic card is GeForce 6800 (I only have one so I swapped from one to the other, same driver is used on both pc's)

First I did PCMark05

830: 4713
4400+: 3064

Comment: Wow.... :slap: I didn't know this benchmark loved Intel and WD raptor. :confused:

Then I did 3DMark05:

830: 3771
4400+: 3813

Comment: .....so my OC'ed AMD took the lead by a very small margin, it could be that it's limited by my graphic card but I thought the difference should be more since A64 always dominated in game performance.

Finally just to make sure I wasn't dreaming when CPU-Z showed me that 2706 figure on my AMD machine, I decided to run SuperPi 8M, and results are:

830: 7 minutes 41 seconds
4400+: 5 minutes 48 seconds.

Comment: Simply :slapass: by AMD, period.

But I still wonder what went wrong with PCMark05. :confused:

[Major]

But I still wonder what went wrong with PCMark05. :confused:

Is your OC 100% dual primestable on the X2 ? super Pi32 x2 ?

It could just be Hyper Theading advantage ?

[afireinside]

Raptor is fast as hell, hence the PCM05 score. If you want to see more Intel spanking run 3d01. 3d05 is 98% graphics card intensive on non 7800 setups.

[Vapor]

Do you have the .exe driver?

[Magnj]

i thought i read that PCM05 was way Intel biased

[Vapor]

An X2 should get 2.5x its clocks without much tweaking (as long as all drivers are present).

[jahjahbinks]

Do you have the .exe driver?

You mean the nvidia forceware v78.01?

[Vapor]

No, the A64 driver.

[jahjahbinks]

Is your OC 100% dual primestable on the X2 ? super Pi32 x2 ?

It could just be Hyper Theading advantage ?

Yeah, running two instances of Prime95 for over 3 hours before I stopped it because of all the noise fans were making... :p: And I couldn't wait to run 3dmark

[Major]

My X2 4400+ (see sig) gets 5900 or so in PCMark05 @ 2.8ghz, I just looked and I've deleted the scores but I'll run it tomorrow at 2.7ghz so you can compare ;)

[Vapor]

http://www.amd.com/us-en/assets/content_type/utilities/amdcpu.exe

[Vapor]

I get 7188 (http://service.futuremark.com/compare?pcm05=73291) with a tweaked 2.75GHz run with my 7800. So I guess you should be getting around 2x-2.4x your CPU clocks with that 6800.

[jahjahbinks]

Also I am running 1:1 with 2.5-3-3-7 timing. I thought for A64, no matter what you do (tight timing on a divider, or loose timing with 1:1, see http://www.dfi-street.com/forum/showthread.php?t=23876), as long as u have the same CPU speed, it's the same.

[Vapor]

Also I am running 1:1 with 2.5-3-3-7 timing. I thought for A64, no matter what you do (tight timing on a divider, or loose timing with 1:1, see http://www.dfi-street.com/forum/showthread.php?t=23876), as long as u have the same CPU speed, it's the same.huh? Was that at me?

When I said tweaked I meant with software and driver settings, not hardware. At the time, I wish my Redline hadn't died, I could have run 276x10 @ 1.5-2-2-0.

[jahjahbinks]

huh? Was that at me?

When I said tweaked I meant with software and driver settings, not hardware. At the time, I wish my Redline hadn't died, I could have run 276x10 @ 1.5-2-2-0.

No, I was talking to myself to see what could went wrong. :p:

I installed your file, I thought it's a similar program like Intel processor id until it asked me to restart the computer when it's done, that's when I realized it's a driver, lol. Well, I am running PCmark05 again.

[jahjahbinks]

Wow, this time I got 3527!!!!!!, and oh boy, that's with a freaking ATI X300 card!!!

p.s. to keep the story short, I have been waiting for the R520 from ATI so I just bought the cheapst PCIe card I can find when I started building this AMD pc last week. Just felt lazy to swap that 6800 back in again.. :p:

[Vapor]

Still seems like a low score--can you tell us what the break down of scores are?

[jahjahbinks]

LOL, i just closed that window, okay, let me get it warmed up again.

Strange thing is that ATI x300 beat 6800.

[jahjahbinks]

HDD XP startup: 8.98 MB/s
Physics and 3D: 83.28 fps
Trans. Windows: 34.36 win/s
Pixel shader: 22.1 fps
web page render: 3.587 pages/s
file decryp: 46.412
graphic mem - 64 line: 345.8 fps
HDD general usage: 5.88 MB/s
Multi 1
Audio comprs: 2695.94
video encoding: 405.52
Multi 2
Text edit: 165.31
Image decomp: 31.07
Multi 3
File comprs: 5.46
File encry: 25.01
Virus scan: 28.91
Mem latency - 16MB: 11.22

[Vapor]

I've boldfaced the ones that are lower than they should be and will likely get helped by the 6800. Looks like you're scoring about right.

HDD XP startup: 8.98 MB/s
Physics and 3D: 83.28 fps
Trans. Windows: 34.36 win/s
Pixel shader: 22.1 fps
web page render: 3.587 pages/s
file decryp: 46.412
graphic mem - 64 line: 345.8 fps
HDD general usage: 5.88 MB/s
Multi 1
Audio comprs: 2695.94
video encoding: 405.52
Multi 2
Text edit: 165.31
Image decomp: 31.07
Multi 3
File comprs: 5.46
File encry: 25.01
Virus scan: 28.91
Mem latency - 16MB: 11.22

[mr_knowitall15]

pc mark 05 is higher because of the raptor. hard drive seems to have a BIG impact on pcmark05s scores. I remeber seeing a 14k score that was done with a stock speeds system, but ran from a ramdisk...btw those scores seem low...i ran my system totally at stock(except vid card was at ultra speeds) because for some reason, the clocks i use for EVERYTHING else wouldnt work in 05, and i got a 4584. surely a dual core should rape me in Pcmark...not barely squeeze by. also, the 4400 should be MUCH higher. from other reviews and stuff, AMDs dual cores are beating intels in everything. Someone said they thought hyper threading, but thats not it because only the 840 Extreme Edition has hyper threading enabled, the 820, 830, and normal 840 do NOT.

[jahjahbinks]

Okay, I am gonna wait for Major's result tomorrow.

[Vapor]

Okay, I am gonna wait for Major's result tomorrow.Your results are exactly where they should be if you weren't using a defragged Ratpor.

[jahjahbinks]

I ran it again on 830 so I can compare the two results, I put the numbers in blue, this time 830 gets 6800, and 4400+ gets X300, so that explains the big difference in video tests. Raptor on 830, Barracuda on 4400+

HDD XP startup: 8.98 MB/s 11.95
Physics and 3D: 83.28 fps 125.11
Trans. Windows: 34.36 win/s 303.03
Pixel shader: 22.1 fps 102.96
web page render: 3.587 pages/s 2.763
file decryp: 46.412 65.944
graphic mem - 64 line: 345.8 fps 1311.62
HDD general usage: 5.88 MB/s 8.87
Multi 1
Audio comprs: 2695.94 1666.997
video encoding: 405.52 281.41
Multi 2
Text edit: 165.31 98.59
Image decomp: 31.07 23.76
Multi 3
File comprs: 5.46 3.97
File encry: 25.01 33.29
Virus scan: 28.91 52.4
Mem latency - 16MB: 11.22 8.32

it seems multitread test 1 is where AMD dominated, and web page render. So conclusion here is that PCMark05 is heavily graphic dependent and hard drive dependent.

[Vapor]

Same graphics card? Looks like you're missing ATi drivers or still have nV drivers. Otherwise everything looks good enough (unless your X2 is using a defragged Raptor).

[jahjahbinks]

ok, i am happy. :woot:

I have edited my last post. :D

[Vapor]

Yes, you should be happy with those scores with the hardware config you have :D

[jahjahbinks]

Just curious, is my overclocked 4400 faster than a stock FX-55 at 2.6 gig?

[Vapor]

In every respect, yes. Your clockspeed is higher and you have two cores.

[jahjahbinks]

Sweet! My next goal is 2750. I heard that 2800 is very difficult to achieve by air cooling.

[Vapor]

Nearly impossible with stock cooling--a Big Typhoon should prove capable though (IMO).

[jahjahbinks]

It seems people can OC 3800+ to the same speed as 4400+ with just a little more or even no extra effort, I wonder where did the 200 MHz difference go. :)

[jahjahbinks]

Did some benchmarks in Sandra, OC 4400+ clearly beats 4800+ in all categories, woot!

[Major]

I'm glad you got it all figured out ! :toast:

Feel better now ? :D

Sorry I didn't get a chance to run it at 2.7ghz, it probably wouldn't have helped you too much since the configurations are so different. Here is my work rig (see sig)

http://www.xtremesystems.org/forums/attachment.php?attachmentid=37606&stc=1

[jahjahbinks]

That's fine, I just read a similar post in another forum regarding low PCMark05 score on AMD, the trick is to put (/usepmtimer ) in boot ini.file, his score went up from 2600 to 5800. But when I searched google for what this /userpmtimer does, I ended up seeing problems reported by people from using this when playing games, so I am not gonna run PCMark05 on my machine. :)

[Vapor]

the use /usepmtimer thing gets fixed when the driver is installed....the benchmark favors Intel (in single core) because of the multithreaded tests where the A64s die.....the benchmark favors AMD (in multi core) because they're faster at the singlethreaded stuff and are also faster at the multithreaded stuff :D