Printable View
O found the major bug which is runing my set:
I choose manually de fsb and set it to 266 - result is 1.9 vdimm
No change in BIOS and just choose 300 fsb - result is 2.1 vdimm
WTF! My mems give tons of errors with 2.1vdimm. They do 450 1:1 with my e6400 with only 1.9vdimm. Plz fix this!!!
I couldn't even POST 400x8.. I didn't try anything lower than 400fsb with 8multi because I was so mad that THAT didn't even work =x (that would make my RAM going 800mhz @ 1:1, which it should be able to do stock, but I couldn't touch it..)Quote:
Originally Posted by geof
Again my tips for those hitting FSB walls are
ONE: Try putting PCIE Freq. at AUTO. It has given me nearly 100FSB more than with it locked at any number.
TWO: Try lowering the vdim if you have it at +0.2 or higher. I get my best results at 0.1V
All the other tihngs I could say have been said many times already, but not these two.
Using the @BIOS program, which file do you select to flash bios after extracting the file you downloaded from gigabyte's website?
just set manually vdimm +0.1 in bios, you'll get 1.9v ;)Quote:
Originally Posted by MonkSP
for 400 and more with 8 multi i have to increase mch and fsb +0.2 , maybe it can help you ...Quote:
Originally Posted by Ribbon
Quote:
Originally Posted by mitftw
the biggest one :)
not exe and not bat so there is 1 left, the name depends on MB type and bios rev.
the one whose file name is your mobo model and the extension is the new BIOSQuote:
Originally Posted by mitftw
example: 965pds3.f4
Are you saying the BIOS is auto overclocking vdimm when FSB increases?Quote:
Originally Posted by MonkSP
i didn't try but it's possible because F4 bios seems to increase vcore with fsb in auto ...Quote:
Originally Posted by BottomsUp
It did the same thing for me when i used the F4 bios, i switched back to the F4c, F4 wasn't as stable for some reason.Quote:
Originally Posted by geof
I think you get 100 mhz more because you upgraded to F4 and then you locked the pcie freq, that is what I understand from your previous reply.Quote:
Originally Posted by delas52
a higher fsb with only +0.1 V, sounds like your power suppy not supplying the voltages needed :p:
please give the info because I can`t find it on page 1 ;)Quote:
Originally Posted by korruptedone
Fresh from Newegg, shipped with F3 bios flashed to F4c. This is a totally different animal compared to my MSI 975X Plat. That being said I get the same SuperPi 1M times with the MSI with a 200mhz lower cpu clock and lower FSB. Need to tighten these timings but cannot seem to find that option in the bios. Might need alittle help.
http://home.comcast.net/~phippy1/gigabyte.bmp
RacerX....go into the main bios page and push control + F1 ..then go back down to the overclock tab and you will see the latencies settings.
Okay don't understand what you're saying, but my point is setting PCIE to auto is not bad (for me at least) and can sometimes help.Quote:
Originally Posted by Pjoeloe
Less volts also mean less heat, and also these motherboards are very strange as you know so logic doesn't always apply.
Well I have had nothing but problems with the F4 bios for the ds3. I can't seem to get any consistancy like I did before. Sometimes I can boot up just fine and sometimes I cannot. I have spent about 20 hours and 500 reboots trying to get a stable OC starting from scratch and clearning CMOS after flashing. My system runs pretty stable with the occassional freezing during xp boot screen but I cannot get it to be dual prime stable like I was before with F3.
The closest I can get stable but still not dual prime is:
400x8 @ 3.2ghz @ 1:1
+.3vdimm
+.1vfsb
+.1gmch
1.40625vcore
I am racking my brain here..... I guess I'll just go back to the F3 bios....
Right!Quote:
Originally Posted by BottomsUp
Vcore and Vdimm. When u just raise htt to 300, vcore switches to 1.41 and Vdimm 2.10.
You have to force vcore 1.325 and vdimm +1 to get back to 1.9
Hey I see some people not liking the F4 bios for the DS3? Did I see that there is a F4C BIOS as well?
Before the F4, I couldn't OC at all, not even 1 Mhz. Now I can OC fine, except at I can't dual Prime95 stable at 3.0 Ghz, but I can at 2.92 Ghz.
Specs are:
E6300 Allendale OC'ed to 2.92 Ghz (8 X 365)
2 gigs of Corsair XMS2 DDR2 6400 (currently running at 730 Mhz though using 2.0 multpipler)
Gigabyte DS3
Vcore: 1.345 - DIMM: 2.1v (+.3v) - GMCH: +.2v from stock- FSB +.2v from stock
So get this, my memory isn't stable at 400x8 OC though it SHOULD be because its meant for 800Mhz. No matter how much I up the voltage on the VCore or GMch, I still am not stable. IF I try to goto 2.3v on my RAM, I can't boot (with some funky system file error). Any ideas of why memory isn't stable when it should be a 800Mhz? I'm thinking I should disable some of those advanced features in BIOS like the virtual technology blah blah and USB legacy things. Anyone have this issue?
I am in a very similar situation. when i dual orthos test my box it fails when I do a mix of cpu/memory but if i do a cpu only test(small ftt) i am totally stable. I have awesome expensive HZ Gskill memory and I am currently at +.3 vdimm, +.1vfsb, +.1gmch 1.4vcore but if I raise vfsb or gmch I cannot post and if I lower or increase memory v I cannot post....Quote:
Originally Posted by wolfman140
Aye, tonight I'm going to try disabling those advanced BIOS features and see what happens.
I also read that the Northbridge temp created a wall for some, so putting a fan on it helped. My EasyTune says that my Northbridge is a safe 44C...But who knows if that's accurate?
Yup, I complete Super PI tests just fine, but have errors with Memtest and Prime95. So...It's definately RAM related I think and not VCore. Then again, the northbridge acts as a communicator between all of them so...if that sucker is too hot, I imagine RAM related issues can occur.
Have you tried the Small FFT tests in prime, to try and isolate the errors?Quote:
Originally Posted by wolfman140
That's a big negative. I'm a noob at this so I don't even know what the FFT test is...Care to briefly explain:)Quote:
Originally Posted by korruptedone
Well, explain to me exactly what you are doing to prime your chip, and we'll go from there.Quote:
Originally Posted by wolfman140
For a quick test, I run two instances of Memtest. If I get errors with that, it means I most definately will get errors with Prime95.Quote:
Originally Posted by korruptedone
For the Prime, I run two instances of Prime95 each on the different cpu, and I do a Torture Test, and set the usable RAM for each instance to half (naturally). That's pretty much it. If I'm unstable, I will fail that test in the first 4 min or so.
Which test are you running in torture test, though? There are 3. If you aren't changing it, then on the Dialog Box that says "Run a Torture Test", set it to Small FFTs. It sounds like you are using Blend. Small FFTs will stress the CPU and try to keep your RAM out of the equation, as it uses a data set that fits into the L2 cache of the CPU.Quote:
Originally Posted by wolfman140