Fail blend test twice in a row.. about 4 hours each time.

Hi, i have my Phenom II x2 560 oced to 4ghz @ 1.55volts (cpu z reads 1.536 when under full load in prime)

I let it run for 10 hours with prime 95 at small ffts and it was just fine and temps never went over 40c.

Now when i run blend test i have failed twice. Both after about 4hours. Is my ram to blame? i only overclocked with the multiplier so my NB, HT, and memory speeds are stock.

my ram is gskill ddr3 1066 RIP JAW 4gb. I have 2 of them but for now i only have the one stick in incase i fry them.


Phenom II x2 560 4ghz 1.536v under load, 1.55 idle.
4GB 1066 gskill "rip jaw" ram. 7 7 7 18 2t
Asus m4n75td mainboard
Raidmax 530W psu
ati 4350 video card
plenty of case fans keeping the system nice and cool

asus put my ram at 8 8 8 20 1t by default, i changed them to gskill's specs manually. Im running blend test again with the 8 8 8 20 1t settings to see if it makes a difference.

any help appreciated.
48 answers Last reply Best Answer
More about fail blend test hours time
  1. The loose timings might help it but I am feeling like those RAM modules might be botched from the factory as far as those CL7 timings are concerned...MEM mods should not have a problem running at mfg. spec....BTW....your Avatar is the bomb....

    Iron Maiden Rules
  2. Your thinking my ram might be defective O.o ? Im running blend right now with the looser timings, hopefully it passes.

    should i run memtest all night with ALL STOCK settings (stock cpu voltage and mult) along with stock ram settings? ASUS put my timings to 8 8 8 24... kind of concerns me. it has that MEMOK! feature but im still unsure of what that even does..

    and yeah, maiden is easily one of my favorite bands!

    edit: btw, my temps for the current blend test are maxing at 36, sitting at 33. i think the artic silver is breaking in
  3. 5 hours 15min, stable

    max temp of 36C freaking WOOT
  4. SWEEEETT !!!...........Great to hear you got 'er going strong....and 36c wow thats a FROSTY CPU....

    memok! ...its just doing its own, POST before, POST...if you don't know POST stands for...Power On Self Test...its making sure the timings and speeds are ok to boot with....if all is ok you will see solid red light for about 15 sec. on the MOBO after you press the power button..then your POST splash screen will appear...if the settings are not able to be booted from then your suppose to reset CMOS then pres this memok! button it flashes and flashes and flashes until it gets back to a bootable config.....there is a whole section in your boards owners manual about it...BTW....My first system crash with my 965 machine I ran memtest86 for 14 hours at my now stable settings which are in my sig.....1600 MHz@1.6v...7-9-8-24-40-1T...It was at like 1100%...Run an instance for each of your cores....example.....4 core open 4 instances of memtest....set ram to be tested at since I have 8 GB each one instance is allocated a certain amount : 1--->1024 MB 2--->1024 MB 3--->2048 MB 4--->2048 MB...then click start test on each will test all of your memory yet still leave you with 2 GB free to be able to interface with the computer while it the "read me" file in memtest after you download explains how and why to not allocate "All available RAM"
  5. ill be sure to read the manual. Kinda disappointing in the ram though. I lowered the mult. last night and brought the fsb to 228 and i thought i saw windows boot noticeably faster

    would this be a result of my ram speeding up or the HT and NB. I failed blend but i kind of expected that with stock ram, nb, ht voltages.

    .. i should prolly check my memory thread :P
  6. just saw your edit

    i was just booting to my flashdrive with memtest running from there.

    in 10 min ill be 8 hours stable. I wonder if a 1.55 voltage on the ram will let me run 7 7 7 18 timings. or maybe go from 1066 to 1333 with same timings with just a voltage bump?

    edited first post, i wasnt at 2t on the timings.. its 1t. not sure why i thought it was 2t
  7. probably not...1066 ram is limited might get to 1200 MHz at stock voltages...If you are going for 1333 on that RAM I would drop the timings back to 8-8-8-24....if you want to see if your booting faster check it by using a program called boot racer.....Yeah the performance increase you saw would be attributed to the NB being OCed to 2228 MHz...your NB and HT multiplier are set at x10 on those two...I personally just OC with the CPU multiplier...and turn my NB up to 2800 MHz which is what my CPU is limited to....and manually set my NB and HT voltage to 1.2v...I keep my HT at 2600 MHz....Thhe biggest increase in performance that you will see when OCing will be when you get the CPU stable and then you start turning up that NB...The NB is essentially your Memory controller so the faster it goes the faster it can transmit data to the CPU...Keep in mind that the NB is very voltage sensitive even more so than most components...your voltage window is 1.1v to 1.2v....never exceed 1.25v on the HT or NB..
  8. how can i monitor NB voltages when in windows?

    i failed large ffts not even an hour into it. :( and by failed i mean i got a BSOD and it also restarted the computer for me.. before when i was getting a bsod i think it was for a different reason because it use to do a "memory dump" and have me manually turn it off. this time it was different.. makes me think its a different issue

    ram is actually at 8 8 8 20, not 24.

    so right now im sitting on stock ram speeds with 8 8 8 20. cpu at 4ghz and it is acting great but the only issue is failing large ffts in less than an hour.

    im going to head to bed and keep the computer off tonight.. first time its actually been off sense i put it together.
  9. Hmmm...

    What exactly are all of your BIOS settings? NB volts? NB speed? HT speed and volts? Do you have load line calibration enabled (LLC), or PCI Spread Spectum? What were the temps at BSOD?
  10. NB volt - AUTO
    NB SPEED - 2000mhz
    HT speed - 2000mhz
    HT volt - AUTO
    LLC - im not sure, i was told to keep it enabled but ill have to go check for it
    ALL spread spectrum are disabled (by default, love asus), including pci
    temps at bsod were max of 37*c on the cpu. not sure about any other temps.

    ill reboot and check for llc. Im not sure what i should set the volt for stock speeds on the ht and nb. im assuming 1.1 for NB and 1.2 for ht.
  11. no llc settings that i can find..
  12. NB and HT both have the same voltage envelope so you can set them both the same....1.1v - 1.2v

    LLC is just a vdroop compensation device in BIOS may just be on certain boards that it allows for adjustment....

    Great that all the spread spectrum stuff is disabled...

    I would re-check that temp with another program or something just to be sure that 37c is the correct temp.. it just seems really low to me... I suggest using....speedfan or coretemp....Also, Asus had some sensor problems on some of their boards...I had to Flash the BIOS to a newer version and viola temps reading perfect.....Just a thought.
  13. i have the newest bios. first thing i did.

    i have speedfan as well as core temp.. i was only using core temps so ill check the max temp with another blend test. Ill go manually set my voltages to 1.2 and see what happens.

    as far as vdroop goes.. i think 1.53 is required for my cpu to stay stable at load.. not 1.55. I think i just had to set it at 1.55 because of the vdroop dropping the voltage.
  14. speed fan reports

    CPU 37C
    MB 25C
    core 34

    coretemp reports
  15. Wow...those are some really cool load temps they look like idle temps....Well I do not believe your BSOD was temp was it your NB or HT....I would say loose up the timings some more on the RAM....Your voltage looks good on the CPU and RAM....I got to think a bit more on this.....
  16. well im almost 2 hours into large ftts stress test and everything is good

    only thing i changed where the hot and nb voltages to 1.2v

    nothing about 40C

    remember, it is only two cores on a cpu that was prolly meant to be a quad??
  17. So it may have just been the NB and HT voltages making your OC...BSOD out
  18. Yep, lets hope.

    what about these settings. they are at AUTO right now

    cpu/nb voltage
    cpu vdda voltage

    im going to check the manual right now but as usual, it prolly wont be any more descriptive than the bios was
  19. Anything on auto I change to the default value...instead of leaving them on auto...So look at what its defaulted to then enter the corresponding value
  20. i tried to look for a default value but the bios didnt tell me. It told me some of the other values' defaults but not those two.

    prime stable after 3 hours so far. cool
  21. Those ought to be OK on auto anyways....Glad its priming stable....So, far
  22. 5 hours large fft stable

    off to work, shouldnt be gone to long.. we wont be busy. ill post progress when i get home.
  23. failed prime on large ffts after 6 hours and 50min. Not good enough

    assuming that it was "more" stable than before, the only thing i changed was nb and ht voltage from auto to 1.2

    should i increase them more? the nb can go from 1.2 to 1.3.. not 1.25

    also, does v dropp effect nb and ht voltages? maybe i should relax the timings even more and see what that does?
  24. Damn.....I wouldn't push the NB past 1.25v...the absolute maximum is 1.35v...USE EXTREME CAUTION when upping this value...loosening the timings might work, hard to say at this point...Depends on the board as to how vdroop will act...Some boards actually compensate for the droop through LLC...vdroop can affect the MOBO as a whole in some instances...Not usually though...NB and HT are not prone to vdroop...
  25. well lets play it safe and think about this

    is my MB temp sensor the NB temp? do we not push the nb voltage because it cant handle power or because it gets hot? the MB temp never went over 27c when i was watching it.

    with my ecs board the integrated graphics gave me a temp and i always assumed that was the chipset temp but i never messed with those voltages.
  26. Yes the NB is the Temp sensor point for the MOBO....Some, get really hot and need aftermarket coolers to be OCed...others just can't handle the clockspeed or voltage increases...Its all in the quality of the MOBO....Most people do not even realize what an integral part the NB plays in keeping the OC stable or that they can even OC it at all as not understanding what it does....It has been A number 1 in performance boost for me...fps increased in Crysis 1 DX10 by an average of 11 fps....Having my NB OCed to 2800 MHz@1.2v...Lets just say it made all the difference in actual visible system speed increase....Essentially, the NB allows "everyone" to talk so, if the NB is at a higher clockspeed "everyone" can talk faster..Damn I ramble a lot....I hope you can make sense of that were correct in assuming that the integrated graphics temp is the NB thats where the IGP is located
  27. cool cool !

    i get your post 100%

    I left the NB voltage at 1.2v and left HT at 1.2 (can ht be increased? is it needed?)

    i put ram voltage to 1.55 and im running prime large fft again to see what happens.

    what temp should i look for with the NB if i do put it to 1.3v? i shouldnt need to though because its only at 2ghz..
  28. Best answer
    Right, without added clockspeed the NB does not require a voltage increase...Just my "rule of thump" is to always put the NB to 1.2v and leave it...stock or OCed....HT hasn't really mattered much for me in the scheme of things...No noticable improvment in performance that I could see...I also just set the HT to 1.2v and leave it....I run NB-->2800 MHz@1.2v....HT-->2600 MHz@1.2v.....I know the HT OC helps out you just don't notice it...LOL...

    At the temps your reporting your thermals are the least of your worrys when overvolting the NB...I would suspect maybe a 5c increase....just a guess though...
  29. well i dont want to touch he nb voltage if i dont have to. especially at stock speeds.. without a fan right on the heatsink.

    i guess its time to wait again and see what prime does this time
  30. IMO...the NB isn't the gremlin we are chasing here...I am suspect to believe that the RAM is for sure the root of that large fft failure you had
  31. if that is so, that disappoints me. Its already at looser timings.. and now im trying more voltage.

    the system is easily faster than anything ive used before, but i still wish the ram would play nice at factory specs
  32. Reading back on this thread...I see that you have the DRAM command rate at 1T...If the OC fails this time set it to 2T...It looks to be the one thing that I overlooked....I read through it like (5) times...finally seen it...lightbulb....the command rate has gotta be the culprit....
  33. what is command rate? i read that gskill ddr3 "does not like 2t at all"

    cant remember where but thats what i read
  34. cpu/nb voltage ; + 0.025 minimum
    cpu vdda voltage : 1.35 maximum
  35. but what should they be by defualt is the question

    large ffts failed, switched to 2t. ram voltage back down to 1.5

    8 8 8 20 2t 1.5v

    large ffts going now, ill report back in the morning.
  36. failed two hours into it with 1t so i switched to 2t and still failed 2 hours in.

    im at a los for words

    edit: could i lower my cpu back to slow speeds.. like 2.5ghz and see if that makes a difference? im starting to wish this llc crap was hear to i could try a slightly higher voltage

    the never click up for cpu volts is 1.565 and im not sure thats a good idea but thats the only way to get 1.55 when under load.
  37. Don't raise vcore too much. Just try your ram at 9-10-9-30 and at default vdim. If you have feature advanced clock clibration disabled it
  38. vcore has been fine at 1.55

    right now its at 1.5 under load @ 3.8ghz

    prime testing large ftts

    ram 8 8 8 20 1t @ 1.5v

    why set the ram like that? if this prime fails, ill try that but i think im going to try and get 3.8 stable and move forward
  39. passed large FFTs for 10 hours

    3.8ghz @ 1.532volts (idle!, at load its less. could get away with slightly lower voltage im sure)

    ram is at 8 8 8 20 1t 1.5volts

    nb and ht 2000mhz @ 1.2 volts

    aveerything else is on auto if its a voltage
  40. its' look good and stable. keep it
  41. time to crank up the northbridge
  42. maximum Phenom II NB at 2800
  43. yep yep. hopefully i can do that. thanks man!

    i dont know who to do best answer for.. you both are the only two that have really helped me :/
  44. ok ... no problem we just help !
  45. Best answer selected by firefoxx04.
  46. i picked lowjac because he has less points lol

    thanks, both of you
  47. Hey firefoxx...I had to work and I as unable to get back to you way back where henydiah.. Took over...I had to work sorry, but I am glad that you got it stable all the way around ....Thanks Henydiah for taking it over and helping foxx out the rest of the way...Your welcome...This was the most interesting thread that I have participating in a while....BTW....Firefoxx lets see some screenies of the prime95 test and a canard cpu-z validation...So I can post your results in the AMD Overclock Club...I would be happy if you would come on over to the thread and join the fun....Its stickied to the top of the TH OC forum page
  48. you're welcome lowjack :)
Ask a new question

Read More

RAM Phenom Overclocking