Duvies X2 4400+ OC adventure part#2

[Duvie]

Sandra screenies:

http://home.comcast.net/~duvie23/x2-2552-11232-sandra.JPG ***100% tested stable 3 days***

http://home.comcast.net/~duvie23/x2-2596-11236-sandra.JPG ***passed 12 hour stress testing***

http://home.comcast.net/~duvie23/x2-2629-11239-sandra.JPG ***limited testing...dont like vcore w/ heat...need an XP-90 and sub 50c***

http://home.comcast.net/~duvie23/x2-2662-11242-sandra.JPG ***limited testing...previously failed 32mb with not eact round but I was running 10x265 1:1 w/ cas 2.5-4-4-8***

speed testing at 2662mhz

http://home.comcast.net/~duvie23/x2-2662-11242-tmpgenccpu1.JPG

http://home.comcast.net/~duvie23/x2-2662-11242-tmpgenccpu2.JPG

2x faster then single cpu.....200% increase

http://home.comcast.net/~duvie23/cinebench2003-11242-2662.JPG

My best at 1.8ghz winchester at 2.66ghz was 376....87.2% increase...the other one was running 533ddr though and 7400's in sandra...maybe tweaking later will get that up more...

http://home.comcast.net/~duvie23/x2-2662-11242-povray37-bench-1cpu.JPG

http://home.comcast.net/~duvie23/x2-2662-11242-povray37-bench-xcpu.JPG

This newest beta version of Povray is fully multithreaded as seen in cpu task manager and subsequent 2x better score. This benchamrk doesn't draw picture but is the basic benchmark for comparison at their website.

http://home.comcast.net/~duvie23/x2-2662-11242-ADT2004-1CPU.JPG

http://home.comcast.net/~duvie23/x2-2662-11242-ADT2004-MULTI.JPG

As you can see the multi version of the test is multithreaded a bit. I had to take screen pic just at same time cause it wont post an end time. As you can see there is a 52+% gain...This was converting 11 frames to AVI 640x480...This will drastically cut down by 2 hour renderings for DVD quality 24fps clips...


Limited stability testing at 2662mhz

http://home.comcast.net/~duvie23/x2-2662-11242-superpi32-tmpgenc-1.JPG

http://home.comcast.net/~duvie23/x2-2662-11242-superpi32-tmpgenc-2.JPG

As you can see 1.536v and 56c....more then I like so it wont stay there....run S&M and the temp gets to 60c within about 14 minutes....Will wait for the 90c but unlikely I will run anything more then 2.6ghz and under 1.5v....

Update: I have not seen one benchmark it wont complete...no errors and no glitches...I wont prime test it yet cause the heat will be near 60...this was for fun and hopefully stability is not too far off of this mark...





Part #2 (versus the old stuff) more to come.....

test systems...

AMD64 3000+ (1.8ghz) stock settings, HTT 5x, 200HTT, 200 divider, cas 2,2,2,6 PC3200

AMD64 2.665ghz (9x296) HTT 3x, 296HTT, 183 divider, cas 2.5,3,3,8 ~533ddr

AMD X2 4400+ @2.662ghz (11x242) HTT 4x, 200divider, cas 2.5,4,4,8 ~484ddr

All other factors are similar...

Vid card = BFG 6800 OC 390/850 soft-modded to FX4000 Quadro (drivers 75.90)


1.8ghz versus 2.665ghz = +48.1%


SYNTHETIC APPLICATIONS:

Applications---------------------------------------3000+ (1.8ghz)------------2.665ghz---------X2 2.662ghz

SANDRA (mem scores)----------------------------5323---------------------7454--------------6668
SANDRA (cpu ALU)---------------------------------8,248-------------------12,133-------------22,460
SANDRA (multi integer)---------------------------17,195------------------25,355-------------50,609


GAMING APPLICATIONS:

3DMARK01SE-----------------------------------------19,762------------------24,700-------------23,328

3DMARK03---------------------------------------------10,465------------------11,265--------------

AQUAMARK3(Total)--------------------------------57,855------------------67,056-------------
AQUAMARK3(CPU)--------------------------------9,320--------------------12,968-------------

UT2003 (FLYBY)640X480-------------------------251.46------------------347.22-------------
UT2003 (Botmatch)640X480-----------------------93.66------------------129.93-------------



CAD APPLICATIONS:

CINEBENCH 2003(cpu 1x)-------------------------256----------------------376---------------704
CINEBENCH 2003(C4D Shading)----------------311----------------------452---------------
CINEBENCH 2003(Opengl hardware)----------3563---------------------5000-------------
CINEBENCH 2003(Opengl software)-----------1353--------------------1958--------------


POVRAY v3.5 (Chess.pov) seconds------------332---------------------226---------------226

MATH APPLICATIONS:

SUPERPI SSE2 (1mb)(sec)--------------------------44---------------------30-------------31
SUPERPI SSE2 (32mb)(min:sec)----------------39:09-----------------27:28----------


SCIENCEMARK (Molecular)(sec)----------------98.52-----------------63.30----------56.73
SCIENCEMARK (Primodia)(sec)----------------456.65----------------311.57---------243.98

 

[BlingBlingArsch]

id like to see u complete and extend ur set of benches. and:
SANDRA (mem scores)----------------------------5323---------------------7454--------------6668
how would u interpret that result? would dual-core benefit from ddr2?

 

[Duvie]

Perhaps....If I could get the new 216,233,and 250 dividers to work maybe I can hit 8000 but sandra and others are not recognizing it....

i could interpret it one way...In most apps I run memory bandiwdth is not important so it doesn't effect me...A gamer it would effect and a person who lives and dies buy superpi may not be happy...

 

[Duvie]

Just ordered my XP-90c with Zalman ZM-F2 92x25mm Adjustable Case Fan 38/53CFM 20/36dBA....and 1 more case fan to replace a 80mm exhaust fan that was starting to fail.....only running 3 case fans right now versus my 5 (not including PSU fan on either) back in my P4 days

See if I can try for a bit more.....

I do need more vcore options though...as normal the vcore settnig dont really work on the Neo....I can set 1.45 and 10% over vid and get the same vcore as if I set 1.55v and 10%....either way I top out at 1.504v actual...loading at 1.536v....