THe number for the 140W version I got it of the box mine came in, and the 125W version I got the # of Newegg
9950BE 140W version - HD995Z:banana::banana::banana:HBOX
9950BE 125W version - HD995ZXAGHBOX
Printable View
You can test the ram by booting into memtest. But since you are using AOD I think you may have to set your bios to your AOD configuration for 3.2, thus booting to your overclocked settings. I admit I don't really know how AOD works at bootup, I have always thought it only calibrates at the windows bootup session not the bios boot session. I say this because my bios startup screen shows 2.6 not 3.2.
Personally, I'd monitor heat during the Prime95 run. In my case, I see a rise from 37C to 47.5C with Prime95. Since my prime version shows each core's run, I can visually see which core bombs first and then each in succession. With my 3.2 run, Core#3 always bombed in Task#1 while the rest ran just fine. In your case, you may see a quick cascade of cores failing then lockup then reboot.
Also, ACC adjustment helped my 3.2 run. While very, very time consuming, it helped me attain 3.2. BTW, I can confirm ACC does goose the voltage. In my case - the AOD voltage setting is 1.35 and it gooses to 1.39 under very heavy load (OCCT load).
[/QUOTE]Quote:
Also, ACC adjustment helped my 3.2 run. While very, very time consuming, it helped me attain 3.2. BTW, I can confirm ACC does goose the voltage. In my case - the AOD voltage setting is 1.35 and it gooses to 1.39 under very heavy load (OCCT load).
I noticed the same thing on my rig, same setup as yours.
i tested my ram with the windows version of memtest (4 instances with 1gb of ram dedicated to each instance) and got no errors.
to test the settings with memtest in dos mode i have to set everything up in bios, thatīs right. aod doesnīt change bios settings. iīm booting into windows at default settings. all i changed in bios was 1066 mem speed and the primary 5-5-5-15 timings in unganged mode.
when setting up unganged mode i can adjust timings for both dimm banks seperately, just noticed that today.
i noticed a higher temperature with prime95 too. with aod stability test highest temp is around 40°C and with prime95 its 43°C.
funny thing is, that i canīt see which core craps out...
i just get a bluescreen with irq_not_less_or_equal and it reboots.
to be honest, i wasnīt really paying attention to the monitor till it turned blue. so i might see a core crapping out before the bsod if i stare at the screen all the time.
acc adds about 0.04v under 100%load on my system too.
Thank you both for your answers
Currently at 3.3 witn 1,35 volts
http://img171.imageshack.us/img171/4...3300mhzqo5.jpg
Ram is not pushed at all, so I keep them at 1,9 for the time being.
CPU NB voltage at 1,35 is ok with stock cooling?
Ek supreme is at it's way, so don't worry about that!:)
+2 at ACC helped a lot with 15 multi.Still haven't done much research as I was for 4 days at an overclocing event, but I will try more settings when I have more spare time.
But for sure the first thing is to put everything inside the HAF and start running the water.Then I will start real business!:p::p:
I just installed my 79-T ,, I will have results of the 140watt phenom that clocked 3.5ghz on my TA770// SB600...Well see if theres more left to sqeeze and funny you say HAF.
I just ordered the HAF:D .My long awaited TJ09 which actually came in, i decided to return. After looking it over ,, i realized two tripples would fit nicer in the HAF...Plus i wont mind painting the Steel vs aluminun.
The HAF iii post in Liquid cooling when i get everything together at the end of the week(two tripples will be nice and snuggled INSIDE) ...I like to mod things.
P.s If this mobo dosnt make my 9950 happy,, i have dibs on the new 790fx/sb750 DFI:D
Well I believe that HAF will serve my whole system well.
I like the RED-BLACK combination and the SUPREME is black too.
My HAF arrived 2 days ago at GREECE and I was one of the first to get it.
Excellent choice for waterccoling systems and I believe that there is absolutely no need for adding any extra fans.Especially the 23cm fan on the side will keep chipset and ram quite cool.Maybe a fan regulator will be good as an extra for bench time and I plan to use my old good akasa front panel.Also with black plate in front!:)
hey guys...im running custom water with 3 rads so temps are no issue, im running 100% load on prime atm with 1.4v in bios on the cpu and at 3.2ghz. my max loads dont pass 49-50 deg .
my question is whats regarded as "safe" max volts for 125w phenoms.
cheers
Ninjagordy. :yepp:
damn prime95...
seems to be my vista ultimate 64 that doesnīt work well with prime.
iīm on winxp now and itīs priming at 3.2ghz/2.4ghz nb 1.3875v+2%acc.
crazy thing... and i almost thought i had bad ram.
is there a 64bit version of prime95? the mersenne.org site seems to be down.
Nbvoltage on 79-T is 1.325v this is the Chip set volts...Im testing 1.61v CPUvid and 1.55NBvid as i type @ 3.4ghz:D
@same voltage i used on the TA770 this 79-T has my chip 2c higher temp wise for CPU.. The Vcore is 100% stable with almost no vdroop for VIDs..The TA770 i had to give 1.64v to get 1.61v.. The 79-T overvolts a little 1.587v = 1.61v
So far ASUS 79-T with 140watt 9950 @ 3.4ghz//2000mhzNB @ 1.61vCPUvid 1.55vNBvid and NO ACC yet.. Im testing the limits before i try ACC...My ram is @ 1067mhz ganged .. Using origanal BIOS..After some more tweaks ill update BIos and see what she really has:D
Havnt tried HTT yet ...I like to see limit with out using first.
http://i168.photobucket.com/albums/u...hz161volts.jpg
^^^nice clocks bro, I tried to do 1.3v for 3.5 but it wouldnt go. However I can do 3.4Ghz with 1.5v. But with ACC on, can't wait to see your results with it on.
ive settled for 3.1cpu/2.16nb/860mem...i think on vista x64 and stock cooling, its great haha!
^^not bad, right now my best daily setting that is `00% stable is 3.2Ghz @ 1.4v
Hmmm for some reason the MOBO kepps the CPUVID @ 4.87v...I want 1.55v but i see no setting to Change CPUVID on this thing...I have Vcore set @ 1.61v and CPUVID still reads 1.48v in CORETEMP.
Even the old TA770 let me set 1.55vid for both CPU and NB? When i set ACC it dosnt Raise or lower the CPUVID either...Only CPUvolts which is totally different..VID is very important for me to achieve 3500//2600 in which both CPU/NB needs to be set @1.55vid..Then I add the Vcore..
looks like i have some ASUS reading to do ,, ill see if Polygon has a MOD BIOS yet hahaaa.
http://i168.photobucket.com/albums/u...ra/asusvid.jpg
Update: Update: Update:,,
3.5ghz @ 1.6v but the VID dropped down to 1.462v with ACC auto? im so confused right now haaaaha.. But its making progress..Just working on the CPU with out NB for now ...Then ill change the Ram then ill work on NB...My Bios is a little Buggy with 1066 so its taking me time to get used to it. For the Most part to do 3.5 on the TA it took 1.65+ So ACC is working for the Most part on Auto.
http://i168.photobucket.com/albums/u...sus3511mhz.jpg
hey you guys should try this little generic CPU benchmark that was found over at TPU. Its pretty interesting, and this way we can compare scores :)
The forced 32 bit version yields better scores, but feel free to try both
http://forums.techpowerup.com/showthread.php?t=73579
I'll start it off :)
http://img374.imageshack.us/img374/6...1020sv1.th.pnghttp://img374.imageshack.us/images/thpix.gif
you dont find that a little warm for water cooling?? Im running 1.48v (sb600 and vista) for 3.2ghz and only hit 50c using prime95 on air. room temp is 22c
On a side note I lost 100mhz going from XP to vista (3.3ghz to 3.2ghz) and had to increase voltage .03 from 1.45 at 3.2 in xp to 1.48v in vista at the same speed.
That bench was written by a dude from this forum,Particle i believe :).
You're right http://www.xtremesystems.org/forums/...d.php?t=204361 ;)
how did you get 1.6V in the BIOS??? I thought the motherboard as CPU over volt protection above 1.55V:confused:
KTE has posted some overclocking advice regarding memory on another forum HERE that I thought you all might want to see. Since the relationship between the IMC and CPU is so important, he gives some insight on memory latency settings that could be helpful for those of us hitting a wall at a very high clock. His example below is for someone struggling with a 9850 on an EVGA mobo. I'm testing the memory tweaks today to see if I can breakthru the wall. (BTW: MAL=MaxAsyncLatency)
Quote:
Hold on, take it slowly or you'll crash again.
Set your HT Ref. to 200MHz, your HT-Link to 10x, keep your MEM stock. Now set 1.375v (or closest to it), 13.5x CPU Multiplier and try booting. If it boots fine increase tRFC to 127ns and MaxAsyncLatency (in MEMSET) to 3 values above what you get at boot. Then check voltage in some tools like Everest, HWMonitor and what CPU-Z shows (they don't always show correct and your board especially will not read K10 voltage correctly). If things seem fine, run some light stability test and that will give you an indication if its extremely unstable or slightly unstable. Don't try 2k6, it's too heavy and will crash the system if its unstable. Try something like SuperPi 1M first on 1-core. If that passes, then try wPrime 32M (set to 4-threads). If that passes fine then run 2k6 CPU tests. If that does fine then you can move up to Linpack 3 trials. Now this is very heavy load, probably the max you'll ever see, so any instability and it may suddenly crash. If 3 trials run fine, then go for more. If you can get 30mins worth of runs of that stable, then you should be fully stable in CPU/IMC and MEM, definitely load stable anyway.
That's a start.
Check your tRFC is above 105ns for all DIMMs, vDIMM can be kept to stock and check the MAL setting I mentioned in Memset above. If things are still unstable, first move is to move the MAL up 2 notches at a time. Once you've moved up 6 values/thrice and you still don't gain any stability, start to move up "NB VID" in AOD. Stay below 1.375VID and in most cases you should certainly find stability.
Its just like running on Intel setups. Once 2GB DIMMs are in you have to bump MCH voltage up and loosen MCH Performance settings. Many times vDIMM also has to be increased to gain any stability. They're much heavier load on the MEM controller by nature and even worse if in a x64 environement.