Nice chip you got thereQuote:
Originally Posted by Asmola
I guess I didn't push my CPU hard enough
I haven't push NB Voltage up to 1.4V ( only 1.3V )
Printable View
Nice chip you got thereQuote:
Originally Posted by Asmola
I guess I didn't push my CPU hard enough
I haven't push NB Voltage up to 1.4V ( only 1.3V )
Asmola: i seen your superpi, congratulations! This is more better than 965 C3 ! Crazy...whats maximum validation? Broken 4600MHz?
Newegg sent me a 1014 EPMW
Guys, what's the right way of testing the CPU-NB of the processor? I'm always testing the following way:
Increase the voltage from 1,25 to about 1.5 - 1.55 in steps of 0.05 and increasing the HT Ref. frequency, while keeping the cpu, HT and ram close to (over or below) their normal frequencies.
After increasing the FSB with some MHz I'm running blend for a while and if no error occurs, increase the FSB again. After some point, I resatart and increase the CPU-NB multi and decrease the FSB again in order to maintain the other 3 frequencies in the same range.
With this method I have a table with voltage scaling of the CPU-NB. The next is testing it together with the RAM. After putting everything together (when tested separately) I'm correcting the frequency and voltage, while looking up the table and testing with Prime again.
Is it ok?
I'm asking this, because I see a big difference, compared to my best C3's...this 3100 with only 1.25V looks unreal, lol. If I need to increase it to 1.3, or even 1.35 after complex overclocking of the system, it's still much better than the C3's CPU-NB.
Yes, it's crazy, that cpu-nb doesn't need as much voltage as the Denebs.
Wow, I never knew what a POS my 720BE was until dropping in a 1090T (1095T in cpuz). At default settings it gets stable 2600NB. Gained 2000pts on 3D'06 cpu test over the OCed 720
Asus M4A89GTD PRO/USB3 with 1301 bios
this is similar to how I do alsoQuote:
Originally Posted by I.nfraR.ed
I test everything individually to see its highest stable clock keeping everything stock except the part being oced, then I add 2 that have been tested seperately together.......like memory and NB leaving the cpu at stock and retest to find stable clocks then finally add the cpu into the mix
But Note: I only did it this way this time because the memory and motherboard are new but the cpu is old. If everything was new Id start with the cpu clocks then the nb then combine them and then add the memory last with tested timings.
Im sure many might have a different method but for me its worked best that way.
Ok, I'm not finished with the testing yet, but as you can see I will max it out very soon.
Additionally I should test the range 1.1 - 1.2V, because it's missing.
The tests are done with everything close to default, except the CPU-NB and checking for some stability with 12-13 loops in Prime Blend.
I thought I could share it with you, even it's not finished yet.
http://img690.imageshack.us/img690/4484/11255200.png
Ofcourse, when everything is put together I will need to increase the voltage and/or decrease the frequency, but I see a big improvement here!
Holy crap, 3370 CPU-NB! Is that blend stable?
I'm not sure if it's fully stable, I ran only 12-13 loops of blend, it passed 1024K + 2-3 additional loops. I just moved to the next voltage and frequency.
How much time do you want to run it, I can try now? I think It's a little bit pointless to do a full stability test at this stage, maybe after I find the maxed out frequency.
And I thought that 3.1Ghz was awesome -_-.
12-13 loops of blend is good enough for me, i'm impressed!
I'm also impressed! Is it my CPU or it's "normal" for many of the chips? I don't want to mislead those who are still wondering if they should move to X6, if my cpu is some exception from the rule...Quote:
Originally Posted by Zeus
I just ran some Blend, because I had no screenshot.
Here's 3367MHz with 1.45V, and it's still running:
http://j.imagehost.org/t/0207/NB-3367.jpg
The cooling is normal water with normal ambients, no cold air.
thats very good but why dont you overclock the cpu together with the nb and see if it stays stable?try 4.2ghz-4.4ghz with NB at 3.3ghz Then you will know if its good and then keep going up or start looking for stability?Quote:
Originally Posted by I.nfraR.ed
Maybe you're right, but I wanted to see how far it can go, I always do. The tables I make help me while benching with cold air :)
My new baby is here!! (1090T) :up:
I have not time till tomorrow to torture it, but I've done some quick tests.... only playing with multiplier (AOD): x19 = 3800mhz (air cooling), Turbo disabled, default vcore (~1.30) and well........ it's LinX (default settings) stable :yepp:
This thing runs happily cinebench 11.5 and x264 benchmark :)
Is this LinX a good stability test?????
Yes, LinX is good for stability testing. Bump up that NB PetNorth!
Thanks Richie!!! I'll do it!!!!!!! but not till tomorrow :(
BTW this is the batch
CCBBE CB 1011APMW
9E21693C00189
Mobo Crosshair IV (WOW, what a mobo :D )
I'm begining to loving this combo!!! :rofl:
Burning in the TIM using WCG BIONIC.
Awesome my friend!Quote:
Originally Posted by I.nfraR.ed
Looking forward to see what happens once you start to bring high CPU, CPU-NB and mem clocks together. ;)
Here is a guy testing 1055T 1015GPMW:Quote:
Originally Posted by thijs
http://www.coolaler.com/showthread.php?t=237169
So keep going:up:Quote:
Originally Posted by I.nfraR.ed
we dont know if its normal yet not to many people testing yet.
So do what you are doing keep pushing the NB till you cant anymore then lower it by 20-50mhz and overclock the cpu as far as you can:yepp:
Good luck!
Thanks :)
3400 is the max I can get with 1.5V. Tried also 1.55V which is too high for 24/7 and Thuban, imo. That voltage didn't help for going higher, it was freezing after 5-6 loops.
Normally with Deneb I get error when the NB is too high, or the voltage too low, here I get freezes :)
If I can run the NB at 3200 - 3300 for daily use it would be great.
I'm wondering if that frequency is real, it might be some bug in the bios? But if it was so, I couldn't achieve that, and I wouldn't have differencies in the results.
http://img248.imageshack.us/img248/1066/cachemem4.png
Good first pass, even if I can not pass over in the micro Mhz seems not, I am at 25C ° te'mperatura environment.
I'm for spending now to see 5870 CF da, then spent Vantage, and then ride my i7 mims step 920 and tests at the same frequency coo to see go.
Greetings from Vzla.
http://www.pctunerup.com/up/results/...00624.341K.jpg
Some benches from this evening
3DMark06 - 21287
http://img375.imageshack.us/img375/7...volts45.th.png
It's currently impossible to make it work with increased voltage, so this is the maximum of the card on stock voltages.
Will try to find the reason.
wPrime 1.55 - 5.601
http://hwbot.org/signature.img?iid=3...se&iehack=.jpg
Should be better at these frequencies, imo.