1090T on the Crosshair III

W.Feather

[H]ard DCOTM x4 & [H]DCOTY x1
Joined
Nov 10, 2009
Messages
11,715
Here we go, got some spare time, ill be editing this post with more info as I bump clocks around, and as I get the scores.........I have already halfed the time per core on wPrime compared to my 550 @3.1 GHz......

WARNING: Pictures are ~1920x1080.....I hate thumbs, so be patient with em, the info is all displayed in text right below, pics are below text.

NEW TEST AT 4.0-
wPrime- 32m: 9.912s, 1024m: 312.166s, not sure why this is lower-
wprimescore.png

CineBench- 6.93points -
cinebenchscore.png

SuperPI- Was a little faster....but i forgot to SS before i closed, and well dotn want to spend ~20 min running them again
POV-RAY- Cant figure for the life of me what i did to get the test last time....so no go here

Any other syou want let me know ill get em going after some more folding for the [H]orde (NOTE: The last batch of info i have was NOT F@H stable)


4.0138 GHz @1.4875v(according to CPUz and CoreTemp)

WPrime Score: 32M- 7.3sec/1024M- 222.813sec
CineBench Score: 6.75, beat the reference i7 960 and i7 860....pic shown below
PoV-Ray Score: 5604.24 PPS (982.63 PPS CPU time) over 262144 pixels ........ NOTE: I think this was for one thread only....
Super-Pi Score(all tests): .184sec/.312sec/.685sec/1.499sec/3.549sec/8.030sec/18.009sec/44.354sec/1min 43.056sec/3min 51.414sec/8min 39.376sec/18min 54.612sec
Prime Stable:Ill let ya guys know when I get 5 hours to test:p
Temp: Idle--35C, Load(4M of wPrime and 32M of wPrime in a row)- 50C....not too shabby for not much thermal paste.....Ambient- 83 F

Found it strange that Super-Pi did NOT put all cores at 100%, rather it cycled a load of ~50% between the cores, and adding up all the loads put it at about 100%......also just realized that BOINC jumped back on, not sure if it was during SuperPi.....lol

CPUz Validation @4.013GHz - http://valid.canardpc.com/show_oc.php?id=1188813
I think I have an old version of CPUz, doesnt show correct CPU, and shows speed down to 802 Mhz ocasionally, no Turbo and no CnQ, so no reason it should :p

pics for 4.0 GHz-
CineBench-
cinebench.png

PoV-
povtest.png

Super-Pi-
superpicg.png
 
Last edited:
wow. is it me or is that voltage really high? ive seen others clock at around 4ghz with around 1.4v. not sure id wanna run 1.487 as a 24/7 setup. since i have the crosshair 3 board i was very hopeful that i could breach the 4 ghz barrier and not have to jack up the voltage.....this is rather depressing.
 
Last edited:
naw, thats just temp, see what my threashold was for heat at that level.....i am able to run stock speed at 1.3v....probably 1.28 havent tried yet.....just messing around quickly and didnt want to do voltage adjust so i tried a bit higher than normal
 
good numbers.. and yeah you do have an older version of cpu-z.. i think its up to 1.54 or 1.55 or something..
 
bit of an interesting tidbit......went out for dinner, and so i threw on Prime to test an undervolt at stock....for me 1.25 volt is too low for Prime95 stability, it froze ~30 minutes in....jacked voltage to 1.4 and will probably run again later(stock on this board is 1.475)....no worries, i have intentions to run it up to 4.4 GHz or so, and I think I have the headroom for it, depends on cooling more than anything(my TruePower new 750 should give me more than enough power, and the Crosshair III is a proven overclocker).....right now my load temps for all six cores in BOINC is 38, idle of 25.....and my room is above 80F....so not too bad.....on a side note, ive only done multi bumps....i should probably do some northbridge overclocking should I not, to edge out some more performance....

remember, im just jumping the speed up for some testing(and so I can say I did it:p), ill be sitting at stock mainly, since I run BOINC 24/7(minus gaming...if I remember to pause it) and need it to be super stable
 
bit of an interesting tidbit......went out for dinner, and so i threw on Prime to test an undervolt at stock....for me 1.25 volt is too low for Prime95 stability, it froze ~30 minutes in....jacked voltage to 1.4 and will probably run again later(stock on this board is 1.475)....no worries, i have intentions to run it up to 4.4 GHz or so, and I think I have the headroom for it, depends on cooling more than anything(my TruePower new 750 should give me more than enough power, and the Crosshair III is a proven overclocker).....right now my load temps for all six cores in BOINC is 38, idle of 25.....and my room is above 80F....so not too bad.....on a side note, ive only done multi bumps....i should probably do some northbridge overclocking should I not, to edge out some more performance....

remember, im just jumping the speed up for some testing(and so I can say I did it:p), ill be sitting at stock mainly, since I run BOINC 24/7(minus gaming...if I remember to pause it) and need it to be super stable


dont worry about boinc.. its not that demanding on the processor.. to give you an idea.. running WCG my temps sit at 50C max usually around 46-48C on my 940 @ 3.6ghz 1.49v on F@H A3 WU's it will hover around 55-60C depending on ambient temp.. as long as the temps look fine theres nothing really to worry about for long term stability.. you could do NB overclocking though there isnt a huge difference above stock but it will look better in synthetic benchmarks i guess..
 
I hate you, thanks for making me want to get a 1090T :p

Great stuff there man, keep showing us what this big boy is capable of! :D
 
I hate you, thanks for making me want to get a 1090T :p

Great stuff there man, keep showing us what this big boy is capable of! :D

no problem, it is definatly worth it over an X2 for gaming alone....I jumped ~25-30 FPS on average with this in BFBC2(I am still GFX restricted :p ), will be testing on Merc's 2 tonight....other tests and overclocking will come eventually, just have lotsa stuff to do, of course nothing poped up until after I got the CPU
 
naw, thats just temp, see what my threashold was for heat at that level.....i am able to run stock speed at 1.3v....probably 1.28 havent tried yet.....just messing around quickly and didnt want to do voltage adjust so i tried a bit higher than normal

id be really interested in what voltage would be needed to hold a stable 4.0ghz on this board.....cuz if i have to crank it up to 1.487v im not sure ill bother upgrading to the 1090t as i dont want to end up frying the board by overvolting it as a 24/7 setup. isnt stock voltage for the 1090t 1.375?
 
id be really interested in what voltage would be needed to hold a stable 4.0ghz on this board.....cuz if i have to crank it up to 1.487v im not sure ill bother upgrading to the 1090t as i dont want to end up frying the board by overvolting it as a 24/7 setup. isnt stock voltage for the 1090t 1.375?

i dont think I need 1.487v to hold 4.0 Stable, I honestly think it was overkill, i think i probably had my ram clocked too high and so the stress crashed the PC.....this weekend/next week ill have more time to play around with stable settings(last week of high school is this week), that 1.487 was more of a saftey to ensure I reached 4.0 GHz with no issue, and to see what my temps would be(rather good i must say)


as for stock voltage, my board actually has 1.425 as the "auto" voltage at all stock settings, so i'm not 100% sure, I took it down to 1.4v just to be safe, and the temps are pretty darn good
 
sorry for the delay, hopefully I can do more testing monday/tuesday night.....right now I am backing up my data(280GB of movies is taking 3+hours to frikin backup) to another PC, and will be doing a clean install of win7 ulti x64 in the next day or so....I think some of my low scores/lockup issues(at higher speeds with Prime 95) are because of my current install of win7(ive never trusted just dropping in a CPU in a machine w/o reinstall)....so ill let you guys know soon what to expect....still have to do stock benchmarks, and more OCing.......

god i have alot of crap on my desktop...
 
Just to point out, your POV-Ray scores are from using all threads. I get a score just a little higher due to my higher-clocked NB. If/When you get the chance, you might want to push yours to see what kind of gains are to be had.

Also, all your scores can be higher is you disable CnQ. It really messes with benchmark scores due to the time it takes for every thread to get up to fullspeed. I haven't found a real-world situations where it has any noticeable negative effect though.

Instead of Super-Pi, use Wprime. Will scale the calculations to all the cores.
 
CnQ is always disabled on my machines :p (I BOINC 24/7 other than when testing/gaming...well if i remember to turn it off when I game)


hmm, it seemed from the feed out on the bottom of POV-Ray that it was using a single core

ill see what I get with a higher NB, i think first im going to try getting my mem clocks down to spec at speed(11-11-11-25-1T, @1600, 1.65 volts....should be 7-7-7 something at 1.65volts, @1600).....meh never really messed with it before, but hate to pay for something and not get what I paid for :)

NB is clocked at 2006.9Mhz right now(stock speed), ill mess around in a day or so..... Ive been itching to get a Hackintosh going, so i may give that a try first....

edit: this is interesting....i just set my voltage to 1.35 in the BIOS, and when in windows both Core Temp and CPU-Z(both updated this time), show it as jumping from 1.35 volts to 1.475v, comes with a .3Mhz change in speed as well.......another interesting tidbit, Core Temp 0.99.6 sees my CPU as the AMD Phenom II X6 1009T BE :p
 
Looking at your posted screens, CnQ is working. That's why your CPU-Z shots are showing at 800Mhz. To completely get rid of it, you'll have to go to Windows Power Management settings and choose high performance. I've found that disabling it in the BIOS isn't enough. The reason your voltage is jumping is from Turbo. There are still a few issues to iron out with X6 BIOS support. To completely get rid of Turbo and it's effects, I had to use AOD.
 
CnQ is always disabled on my machines :p (I BOINC 24/7 other than when testing/gaming...well if i remember to turn it off when I game)


hmm, it seemed from the feed out on the bottom of POV-Ray that it was using a single core

ill see what I get with a higher NB, i think first im going to try getting my mem clocks down to spec at speed(11-11-11-25-1T, @1600, 1.65 volts....should be 7-7-7 something at 1.65volts, @1600).....meh never really messed with it before, but hate to pay for something and not get what I paid for :)

NB is clocked at 2006.9Mhz right now(stock speed), ill mess around in a day or so..... Ive been itching to get a Hackintosh going, so i may give that a try first....

edit: this is interesting....i just set my voltage to 1.35 in the BIOS, and when in windows both Core Temp and CPU-Z(both updated this time), show it as jumping from 1.35 volts to 1.475v, comes with a .3Mhz change in speed as well.......another interesting tidbit, Core Temp 0.99.6 sees my CPU as the AMD Phenom II X6 1009T BE :p


with pov ray make sure your using the beta version installed over the stable version it will have the multi-core support..

yep the voltage jump is due to the turbo mode..
 
guess ill know for sure when the next bios is released :) ill keep you guys updated
 
Ive installed the latest bios today......get great temps on stock voltage, and less fluctuation due to turbo mode.....ill be giving a full range of tests, at different speeds in the next week, doing stock speeds for reference now


ill also be pushing for a more voltage smart level at 4.0 GHz level.....I was getting 36 C load via Core temp during a 2 hour long Prime 95 run(did to get temps, knew it was stablet at stock) for a stock run....

just a heads up on what i'm doing with this combo :p, hope I can get it done before my new tablet comes lol
 
When were running Super PI did you use all 6 cores or 1.1 core will give a higher score,at least on mine it does.
 
its been a while since I did those.....IIRC Super Pi is a single threaded application....as a side note, I have left Super Pi out of my latest batch of benchmarks.....currently I have: wPrime, PoV Ray, Cinebench

Had sissoft, but since I have VLC going when i'm doing these(i have a short attention span :p ), anyhow, it lags the hell and does small freezes for me(not sure why).

If there is interest in Super Pi scores i can re add that in, no problems there....just let me know(and shoot me a download link, cant find my last download, and a quick google found version 1.1....the One i had before was 1.5 or somethin)
 
DoH I messed up.let see if I can find right program.

Edit:It was Hyper-PI I was thinking of
 
Last edited:
I can go get that and do a run or two of that, no issues there. My main purpose is to test multi-threaded performance......Ill be shooting for 3.6 GHz at 1.35v tomorrow or this weekend sometimes(depends on if I feel like BFBC2 :) )
 
Just curious hyper pi is faster on one core than 2-6 cores of the thuban which I thought was interesting.
 
interesting, ill DL and add it to my results.......ill probably post my stock results today or tomorrow, and then go for either 3.5, or 4.0, can you tell i'm bored :)
 
interesting update.....

about an hour ago I start my attempt at 4.0......posted at 1.4 v, never made it to logon screen......1.4125, posted, made to log on, crashed during log in......1.4325, posted, made into windows, started up Cinebench, crashed 12 "blocks" in.....so i went hell, ill go to 1.45....posted, got to where i did at 1.4325.....this is at 200x20 setting mind you.......didnt have time to mess with getting 4.0 stable enough to bench, so i go and try and see if i can load in at 1.3v(stock speed)...nope, crashed bios bad enough I had to hard reset it(clear CMOS button did jack shit)......as i do that a fuckin screw from the backplate of the GFX drops into the dam powersupply, so i just spent half a dam hour getting it out....probably voided the dam warranty since I needed to unscrew the fan(though I never broke the seal :) ), to get to the fuckin thing with a pair of pliers....i'm done for this week giving it a go.....maybe next week ill feel more inclined

edit: i should probably add that i changed my PSU orientation so its fan down....go figure the one time i drop a screw, it lands in litterally the one spot on the PSU where it could fall in.
 
it got too hot here for me to continue testing, ~89F with AC fighting the heat....., will continue the testing in a month or two when it starts to cool down(hell ill have fun on the nights we get to freezing !)
 
Well another mini update.


Got my H50, and now have this clocked 4.0ghz, at 1.49v actual ( 1.5325 in bios ) , with temp of 43c load....while folding :) If this WU finishes without an issue i will consider it stable (its already quick prime stable) , and ill get some benches for a true comparison.
 
Well another mini update.


Got my H50, and now have this clocked 4.0ghz, at 1.49v actual ( 1.5325 in bios ) , with temp of 43c load....while folding :) If this WU finishes without an issue i will consider it stable (its already quick prime stable) , and ill get some benches for a true comparison.

Excellent! Hope the folding goes well on this setup.
Now I gotta O/C the stock i7.
 
:) say by bye to my tail gate man :)

got crapy WU's on the 1090T (a 6701), and pulling 8k now, i'm figuring ~9-12k depending on the WU for other units.....and hell the Q9450 isnt even OC'd now :p hopefully i can hold you off more than those 3.6 days ( so long as you keep stock )
 
Well guys, more stuff coming up ;) Uploading now, will update OP with new test scores

http://valid.canardpc.com/show_oc.php?id=1422166

NB Speed is 2.8, cant for the life of me to get it at 3.0, at 3.0 in bios, it reverts to 2.0 in windows, 2.8 is good i guess:(.....RAM is at 1600, CPU at 4.01, needed 1.49v (1.535v in bios, the Crosshair III has a shit load of vdroop on the CPU, took a while to figure that out), you wont get 4.0 on this board w/o a nice bit of voltage.

The CPU is under an H50, Ambient of 87....My temps on this maxed out at 45c, did 50 passes of IBT, all stable, no problems with results....temps maxed at 48c on CPU, and the NB got to 55c max.

For the record, this setup, minus NB speed, gets 12k PPD to 14k ppd(excluding 6702 units) for F@H.
 
ha my 16k time beats your x6.. though it means absolutely nothing since its not enough to saturate any modern processor but still funny..
 
ya....it was interesting since my 4.0 at lower volts got some better scores in the wPrime.....dont see it as very effective comparison
 
ya....it was interesting since my 4.0 at lower volts got some better scores in the wPrime.....dont see it as very effective comparison


what version of wprime are you using and are you manually setting it to 6 cores?

woops didnt notice you changed the screen shots up there..
 
didnt change, added

wprime204 is the folder, so i'm guessing 2.04, and dont know how to set it to do 6 threads only, all 6 cores were used
 
didnt change, added

wprime204 is the folder, so i'm guessing 2.04, and dont know how to set it to do 6 threads only, all 6 cores were used


yeah v1.55 you had to manually set the core count thats why i was asking but didnt see that you put the screenshot up there.. funny that your only a little under 3 seconds faster on 32m then me but completely massacre my 940 on 1024m by almost 70 seconds..
 
dont know how much faster that can be done :) But ya, it is funny that it doesnt scale very well i guess between the tests.
 
damn just run cinbench 11.5 im going to cry.. lol your time is half of mine.. 2 cores + 500mhz more makes a huge difference in that.. got 3.54 points..
 
....my 550 got higher than that at 3.6.....ill have to see if i can find those numbers, been a while sinc ei ran it


edit: nvm....it was just shy of 2 ;)
 
ill give it a go when i get home, dont think it would be a great idea to do that remotely ;)


atleast i know this reference is getting out there :D
 
Back
Top