Printable View
Last benchie for tonight and still at 1.55v managed to pull off a perfect 3dmark06 run with Gecube X1900XTX stock hsf air @732/868 :)
3dmark06 = 7,170
http://service.futuremark.com/compare?3dm06=356684
http://www.fileshosts.com/DFI/NF4_SL...64_7170_tn.png
eva that 3dmark03 score is some strange to me, i get the same score with a single evga 7900gtx @stock and a fx62 stock.
how did you score that with ur x1900 oced an the fx60 oced
ATI single cards aren't that strong in 3dmark2003, Nvidia 7800/7900 series is stronger in 3dmark2003 :)
Also 3dmark2003 also gets a boost with FSB/HTT speed, but i'm at 16x213HTT due to cold bug heh
But i might go back to 3dmark2003 and redo the runs and see heh
Managed 738/868 for 3dmark05 and 3dmark2001 at 1.575v/2.195/2.023/1.564 but it froze in GT1 in 3dmark06, so 732/868 seems like the stable point on stock hsf so far.
3dmark05 = 14,264
http://service.futuremark.com/compare?3dm05=2083234
http://www.fileshosts.com/DFI/NF4_SL...8_14264_tn.png
3dmark2001 = 45,404
http://service.futuremark.com/compare?2k1=9000571
http://www.fileshosts.com/DFI/NF4_SL...8_45404_tn.png
fantastic score eva :D how are you folks all keeping you vltage regulator so cool?
My x1900xt master is maze4 watercooled and temps never pass 40*c load...but my v.reg chips sit at 60*c idle and load at anywhere up to 80*c!!!!!
120x38mm fan over voltage regular chip :D
First FX-60 benchmark beyond my cpu's cold bug/low HTT bug. Work around using toast app to keep cpu loaded when idle.
FX-60 @15x229HTT with 233 upclock and 2x 512MB OCZ PC3200 Gold UTT BH-5 @264mhz 2-2-2-5 :)
3dmark2001 = 46,083 :cool:
http://service.futuremark.com/compare?2k1=9000607
http://www.fileshosts.com/DFI/NF4_SL...8_46083_tn.png
3dmark05 = 14,335
http://service.futuremark.com/compare?3dm05=2083369
http://www.fileshosts.com/DFI/NF4_SL...8_14335_tn.png
3dmark06 = 7,175
http://service.futuremark.com/compare?3dm06=358008
http://www.fileshosts.com/DFI/NF4_SL...68_7175_tn.png
Nice work, Eva2000.. under stock HSF.. Wow!!
Just looking the first result you have give for 3D05.. my CPU was clocked around the 3080 if i remember well.. looking your score i see my score was really CPU limited... you was having better score in the first part or i was at 793/900mhz.. (perhaps 799/900.. don't rememeber ..)
YOu will got easy the 15K..
Just for show you at what point the first bench is limited CPU...( in my case.. )
http://img142.imageshack.us/img142/7...stitre23jh.jpg
I definetevely need better cooling (or change my CPU)...
thanks was that single core @3080mhz ? you would be cpu limited i'd say somewhat
Yes here my orb's: http://service.futuremark.com/compare?3dm05=2045383
totally cpu limited in my score.. i can put any speed on core and memory.. the first bench never move higher /or lower.. .. i have suspect this before.. but this time i m fully sure...
I can't wait to see your GPU under water.. with your FX at this speed, you should increase quickly the score..
Nice job eva, your card isn't that bad.. only a bit worse than mine on the core and I'm on water, and much better on mem (I only get ~820)
Anyway can you try something?? See if you can reduce VDDQ to the minimum 1.785vand keep your clocks stable, I've always left mine there and there's no good effect to raising it, as far as ocing is concerned
edit: I also keep the VDDCI at 1.440v, oc'd to 702/792@ 1.4/2.001/1.785/1.440 (yes I'm a low voltage freak)
Yup slapped on Dangerden Acetal Maze4 Low Profile water block for water cooling to see how it goes.
H20 cooling results
With ~23C room ambient, X1900XTX on water cooling
@500/594 2D stock volts = GPU is 31C idle and 36C load
@650/774 3D 1.425v = GPU is 33C idle and 40C load
GPU/VDDC/VDDQ/VDDCI = GPU/Mem
Note: PASSED means ATI TOOL continued to run at that stated speed only.
GPU core clock tests with mem set to always 774mhz
1.425v/2.086/2.086/1.486v = 732mhz on GPU core an improvement of 30mhz over stock hsf air
1.450v/2.086/2.086/1.486v = 733mhz on GPU core with load temps of 40C. Sounds like similar issue with stock hsf air, GPU volts or temps aren't the problem but maybe VDDCI volts ? An improvement of 13mhz over stock hsf air.
1.500v/2.086/2.086/1.486v = 759mhz on GPU core with load temps of 42.5C. Now we're getting somewhere :D
1.525v/2.086/2.086/1.486v = 766mhz on GPU core with load temps of 42.5C.
1.550v/2.086/2.086/1.486v = 769mhz on GPU core with load temps of 42.5C.
1.575v/2.086/2.086/1.486v = 769mhz on GPU core with load temps of 42.5C.
1.500v/2.086/2.086/1.564v = 760mhz on GPU core with load temps of 42.5C.
1.575v/2.195/2.023/1.564v = 769mhz on GPU core with load temps of 42.5C. Have i hit a limit here ? On stock hsf air ATI Tool said my max was 720mhz on GPU but i could bench all the way to 732-738mhz on GPU.
Video memory clock tests with GPU set to always 732mhz
1.575v/2.195/2.023/1.564v = FAILED - upto 886mhz before artifact scanning dropped it to 882mhz.
1.575v/2.225/2.023/1.564v = FAILED - upto 900mhz before ATI Tool locked up during artifact scanning. Load temp = 43C
1.575v/2.266/2.023/1.564v = FAILED - upto 904mhz before artifact scanning dropped it to 900mhz.
1.575v/2.296/2.023/1.564v = FAILED - upto 895mhz before artifact scanning dropped it to 891mhz.
1.575v/2.288/2.023/1.564v = FAILED - upto 922mhz then locked up with slight artifacts.
1.575v/2.280/2.023/1.564v = FAILED - upto 882mhz before artifact scanning dropped it to 841mhz :confused:
1.575v/2.305/2.023/1.564v = FAILED - locked up at 871mhz
1.575v/2.288/1.985/1.564v = FAILED - upto 909mhz before artifact scanning dropped it to 900mhz
1.575v/2.288/2.023/1.533v = FAILED - upto 873mhz then artifacts. Guess VDDCI voltage still is important for mem clocks on water cooling.
1.575v/2.288/2.023/1.575v = FAILED - upto 900mhz then locked up
1.575v/2.288/2.135/1.564v = FAILED - upto 909mhz before artifact scanning dropped it to 900mhz then down to 891mhz
1.575v/2.288/1.935/1.564v = FAILED - upto 886mhz then artifacts
1.575v/2.288/2.047/1.564v = FAILED - upto 909mhz before artifact scanning dropped it to 900mhz then down to 891mhz. Similar behaviour as when VDDQ was at 2.135v
1.575v/2.288/2.015/1.564v = FAILED - upto 913mhz before locking up
1.575v/2.288/2.012/1.564v = FAILED - upto 909mhz before artifact scanning dropped it to 900mhz then locking up
H20 cooling results
Yup slapped on Dangerden Acetal Maze4 Low Profile water block for water cooling to see how it goes. With ~23C room ambient, X1900XTX on water cooling
@500/594 2D stock volts = GPU is 31C idle and 36C load
@650/774 3D 1.425v = GPU is 33C idle and 40C load
GPU/VDDC/VDDQ/VDDCI = GPU/Mem
Note: PASSED means ATI TOOL continued to run at that stated speed only.
GPU core clock tests with mem set to always 774mhz
1.425v/2.086/2.086/1.486v = 732mhz on GPU core an improvement of 30mhz over stock hsf air
1.450v/2.086/2.086/1.486v = 733mhz on GPU core with load temps of 40C. Sounds like similar issue with stock hsf air, GPU volts or temps aren't the problem but maybe VDDCI volts ? An improvement of 13mhz over stock hsf air.
1.500v/2.086/2.086/1.486v = 759mhz on GPU core with load temps of 42.5C. Now we're getting somewhere :D
1.525v/2.086/2.086/1.486v = 766mhz on GPU core with load temps of 42.5C.
1.550v/2.086/2.086/1.486v = 769mhz on GPU core with load temps of 42.5C.
1.575v/2.086/2.086/1.486v = 769mhz on GPU core with load temps of 42.5C.
1.500v/2.086/2.086/1.564v = 760mhz on GPU core with load temps of 42.5C.
1.575v/2.195/2.023/1.564v = 769mhz on GPU core with load temps of 42.5C. Have i hit a limit here ? On stock hsf air ATI Tool said my max was 720mhz on GPU but i could bench all the way to 732-738mhz on GPU.
Video memory clock tests with GPU set to always 732mhz
1.575v/2.195/2.023/1.564v = FAILED - upto 886mhz before artifact scanning dropped it to 882mhz.
1.575v/2.225/2.023/1.564v = FAILED - upto 900mhz before ATI Tool locked up during artifact scanning. Load temp = 43C
1.575v/2.266/2.023/1.564v = FAILED - upto 904mhz before artifact scanning dropped it to 900mhz.
1.575v/2.296/2.023/1.564v = FAILED - upto 895mhz before artifact scanning dropped it to 891mhz.
1.575v/2.288/2.023/1.564v = FAILED - upto 922mhz then locked up with slight artifacts.
1.575v/2.280/2.023/1.564v = FAILED - upto 882mhz before artifact scanning dropped it to 841mhz :confused:
1.575v/2.305/2.023/1.564v = FAILED - locked up at 871mhz
1.575v/2.288/1.985/1.564v = FAILED - upto 909mhz before artifact scanning dropped it to 900mhz
1.575v/2.288/2.023/1.533v = FAILED - upto 873mhz then artifacts. Guess VDDCI voltage still is important for mem clocks on water cooling.
1.575v/2.288/2.023/1.575v = FAILED - upto 900mhz then locked up
1.575v/2.288/2.135/1.564v = FAILED - upto 909mhz before artifact scanning dropped it to 900mhz then down to 891mhz
1.575v/2.288/1.935/1.564v = FAILED - upto 886mhz then artifacts
1.575v/2.288/2.047/1.564v = FAILED - upto 909mhz before artifact scanning dropped it to 900mhz then down to 891mhz. Similar behaviour as when VDDQ was at 2.135v
1.575v/2.288/2.015/1.564v = FAILED - upto 913mhz before locking up
1.575v/2.288/2.012/1.564v = FAILED - upto 909mhz before artifact scanning dropped it to 900mhz then locking up
Hmmm not happy with GPU clocks, might need to do a remount of the Maze4 waterblock to see if i can improve it somewhat heh.
http://www.fileshosts.com/DFI/NF4_SL...43Cload_tn.png
Kind of mediocre score huh ? From that i'd probably need 800Mhz on GPU to hit 15k in 3dmark05 LOL
3dmark06 looks better though = 7,338 http://service.futuremark.com/compare?3dm06=358600
http://www.fileshosts.com/DFI/NF4_SL...95_7338_tn.png
.. it's really nice... Eva 2000... this promise a lot..
for 799/900.. here the voltage i use with my HIS.. 1.575/2.25/2.25/1.503
( try up the MVDDC.. and lower the VDDCI)
nice work as usual eva
what cat's are you using? In the other thread the beta 6.6's supposedly boost 05, 06 around 400-500 points.
thanks guys
Lane will try your suggestions, but from what i've seen so far VDDCI needs to be high for stability for my card... but maybe by lowering VDDCI, i will be able to increase VDDC volts hmmm
I'm remounting the waterblock today to see if i can get better contact with GPU core heh
Praxis, i'm using official 6.5 ... will try 6.6 betas alot later.. need to be able to run my scores and compare with other 6.5 and below cats to see if i'm on track first heh
================================================== =
Update: remounted Maze4 LP GPU waterblock to see if i could get better contact and higher clocks. Seems to have helped so far.
GPU/VDDC/VDDQ/VDDCI = GPU/Mem
Note: PASSED means ATI TOOL continued to run at that stated speed only.
GPU core clock tests with mem set to always 774mhz
1.425v/2.086/2.086/1.486v = 739mhz on GPU core an improvement of 37mhz over stock hsf air and 7mhz improvement over first GPU waterblock mount :cool:
1.500v/2.086/2.086/1.486v = 759mhz on GPU core with load temps of 42C. Same max as first mount.
1.550v/2.086/2.086/1.486v = 769mhz on GPU core with load temps of 42C. Same max as first mount.
Got a bit tired of ATI Tool, so thought i'd do 3dmark05 continuous loop tests instead :)
1.575v/2.086/2.086/1.486v = PASSED at 772mhz looping 3dmark05 artifact free! Guess ATI Tool doesn't tell the whole picture :)
I am so impressed of your documentation of your tests! The listing of voltage--speed is like a dictionary. ;)
I have noticed the same thing, ATI Tool might report errors while other applications don't. Also, after a few days of -try-and-error I also remounted the Maze4, it was almost like a new card after that. So, that is a tip: If you cannot clock as high as you could earlier, remount the cooler.
well seems ended up around 765mhz gpu core is max when mem is >882mhz, can do 772mhz gpu core when mem is around 774mhz heh
I settled on 765/891 for now. Bit disappointed as i was hoping for at least 785-795mhz GPU on water.
http://www.fileshosts.com/DFI/NF4_SL...1_14564_tn.png
Max load temp was 43.68C or there abouts, never hits 44C so far in 23-25C room ambients :)
Interesting finding, i changed to my other PCI-E power connector on my 1KW PCP&C SLI psu, and now at same 1.575v ATI Tool managed to get to 778mhz instead of 769mhz!
Oh and piccies
http://i4memory.com/reviewimages/vid...900XTX_001.jpg http://i4memory.com/reviewimages/vid...900XTX_014.jpg http://i4memory.com/reviewimages/vid...900XTX_015.jpg http://i4memory.com/reviewimages/vid...900XTX_017.jpg
http://i4memory.com/reviewimages/vid...900XTX_019.jpg http://i4memory.com/reviewimages/vid...900XTX_022.jpg http://i4memory.com/reviewimages/vid...900XTX_023.jpg http://i4memory.com/reviewimages/vid...900XTX_024.jpg
http://i4memory.com/reviewimages/vid..._Maze4_003.jpg http://i4memory.com/reviewimages/vid..._Maze4_005.jpg http://i4memory.com/reviewimages/vid..._Maze4_010.jpg http://i4memory.com/reviewimages/vid..._Maze4_012.jpg
http://i4memory.com/reviewimages/vid..._Maze4_014.jpg http://i4memory.com/reviewimages/vid..._Maze4_022.jpg http://i4memory.com/reviewimages/vid..._Maze4_023.jpg http://i4memory.com/reviewimages/vid..._Maze4_024.jpg http://i4memory.com/reviewimages/vid..._Maze4_025.jpg
edit: darn 778mhz gpu clock was a once off fluke it seems, still 765-769mhz is max sigh
Well decided to try 6.6 beta and gained 8.67% in SM2.0 and 3.99% in SM3.0/HDR and overall gained around 5% in 3dmark06
Jump from 760/895 = 7,338 to 765/895 = 7,717
3dmark06 = 7,717
http://service.futuremark.com/compare?3dm06=359755
http://www.fileshosts.com/DFI/NF4_SL...95_7717_tn.png
I have the X1900 Master card too, but why it's not well for oc? :confused:Quote:
Originally Posted by rob[GL]
I have ASUS EAX1900 CrossFire Edition with the water cooling system. My GPU/mem is about 700/780 on default voltage, then I try to rising the voltage by ATITool it's not any help for oc. :(
Please tell me that how the X1900 CF Edition oc as good as XT/XTX. :confused:
Quote:
Originally Posted by Deschutes
Sawasdee-khrap !! The XTCF have 1.2ns mem, and the XTX mostly have 1.1ns and generaly OC better than XT/CF cards:)
for the voltage. Did you remeber to disable the services ati smart and hotkey poller??? ati does not always do it itself:)
Even if the XT has 1.1ns rams the XTX will still OC better than the XT ?Quote:
Originally Posted by SIOUX
depends on each card it seems - but on average XTX will clock better than XT... take 10x XTX cards and 10x XT cards and the average clocks and see hehQuote:
Originally Posted by farksy
the memory on my XT overclocks well but the gpu sucks (i've a VF-900 on it but still :/ )
edit: 1.1ns rams