Particle, I have found unganged to be faster in 16M. Don't know about 1M.
It's also easier to stabilize - to get ganged stable at 860 7-6-6-16 I had to mess with drive strenghts on M4A79T Dlx.
Printable View
Particle, I have found unganged to be faster in 16M. Don't know about 1M.
It's also easier to stabilize - to get ganged stable at 860 7-6-6-16 I had to mess with drive strenghts on M4A79T Dlx.
The cpu if not stable will do this, this is my findings.......In my phase and DI validations I was being conservative with voltage and running my ram cranked.........this is why My validations failed, the IMC was under some stress and I was using minimal voltage just what I needed to pass benches........Quote:
Originally Posted by rbuass
I also notice this with people trying to validate 1 core..........fortunately for me I don't bother with 1 core validations.......so mine pass if I'm stable....
CPU-z will only accept your highest validation and one email address per account.........In my case I used 4 email adresses for review purposes......however my main email address was the one I used for 6.4 gig validation..........and unless I break 6.4g it will not accept another submission under that email address......
http://www.abload.de/thumb/benchesrmyl.jpg http://www.abload.de/thumb/3dm06ilpo.jpg http://www.abload.de/thumb/32mbencheszxp8.jpg http://www.abload.de/thumb/vantage8xcm.jpg
no text ;) xp does not work somehow and no time for cf so far..
Glad there are overclockers that can duplicate my results :)
Very odd. At 1.45V, I can't get the NB completely stable even at 2400MHz. Causes crashes and RAM errors.
Particle I Use +.250 to +.275 ( thats how gigabyte does it ) For 2600-2800 NB speeds. That appears to be the avg I need for any chip I use except the 720's they tend to have slightly weaker NB performance.
I'm going to try and isolate it out for now and see if maybe it's simply RAM. That would figure, wouldn't it? heh
might be ram placement bud, try swapping dimms? some ram like front banks, some like back.Quote:
Originally Posted by Particle
but with 0.05v more voltage sadly..Quote:
Originally Posted by chew*
Yah it will definitely vary among chips and motherboards but my guess is none of them will need more than 1.55....Quote:
Originally Posted by Oese
From my experience its 1.475 to 1.5375v, 1.55v didn't appear necessary.
The good news however is No chip prior to the 955's that I used could pass prime No matter how cold I got my room......I think one time in the winter I had 5C ambients.........electrically the cpu's could just not get prime stable............now even at much warmer temps they can.....
BTW AOD temps are off, part of it's gigabyte, part of its an old version.......Ive taken ambient temps and aod temps and cpu was 4-5C cooler than ambients.
Looks like progress all round, 4.0 eh ? Well let's see where I'm at :rolleyes:
First, I have to say that using the 955 makes me appreciate just how smooth and easy my experience was with the 720. That poor little 3 core cripple always tried its damnedest to make my wishes come true. The 955, well, it's a bit twitchy and I'm starting to think the m3a79T deluxe is a little twitchy too.
After my hissy fit last night I got the 955 looking 3.8 stable at sensible volts by using ACC. -2%, +2%, auto it didn't matter and prime ran for 6 hours as I slept. So ACC helped, or maybe it didn't, it certainly shouldn't, but maybe it did ? Certainly the power consumption dropped a little when I enabled it.
I set out this morning with a plan to test the stability of all 48 basic ACC settings. Not a very good plan. So, I thought, if ACC is dropping the power consumption why not just boot to bios, set the values, reboot into the bios and look for the settings that give the lowest power reading ? Meh. Maybe not the best of ideas but it sure beats 48+ hours of watching prime.
Well, I have the 48 measurements here. In fact I have over 100 measurements. A very strange thing happens with my set up when ACC is enabled with one or more values not equal to 0. With various peripherals removed and ACC off or = 0,0,0,0 the measured power while at the bios cpu settings screen is 230w +/-1. Change a setting and reboot, the measured power is now around 218w +/-1. The sort of thing I was looking for, but reboot again without making changes and the reading jumps back to 230. Repeat as many times as you like, whatever non-zero values you like, the sequence is 230, 218, 230, 218 etc. Even if it is inconsequential I still find it curious.
Anyway. Much messing around later and I have changed over to my 700w Seasonic, 670w at 12v is making for a much more enjoyable experience. Watching it closely I think even at 3.8 the Tagan was nearing meltdown, 12v rail drooping and efficiency shot to hell.
After 24 hours messing around I'm now happy the 955 can at least match my darling 720. Nb doesn't like more than 2.4, even getting the fsb to 266 has been a chore. 4.0 ? I'll see what I can do, though it may be best if you don't wait up.
Yah little tip, Look at my prime screenie......NB adds heat , I removed memory and NB from the equasion......as I've tried on other cpu's.......All I did was clock the core, under clocked HT and NB a tad and undervolted NB a tad to remove some heat ;)
Really not sure if its going to be possible with NB speeds up as i didn't try.
Might be better suited on watercooling or phase to get both core clock and NB(IMC) clock both at high speeds....
Have anybody tested DDR2 ram overclocking with a Phenom II X4 940 vs 955 CPU? I mean if ram/nb overclocking is better or worse since there now also is a DDR3 controller on the cpu.
ok just to add to chews post here is a test i did with the stock 940 and the 955
i cant post links cause the drivers for the 4890's isnt approved
Remember this is 100% stock as is
http://www.myalbumbank.com/albums/us...k_Settings.JPG
Ah ha. Nice one.Quote:
Originally Posted by chew*
I must admit this cpu is a lot more fun now I'm using a power supply that's up to the job. Apologies to AMD for all the nasty thoughts, it wasn't your fault at all. So, a bit of brute force, some chew* magic and ...
http://www.i2net.me.uk/files/Overclo.../955PI@4.2.jpg
http://www.i2net.me.uk/files/Overclo...ay/955@4.1.jpg
I'll work on subtlety later :D
Question:
Why does 3d06 show my clock @ 4.0 when it was run @ 4.1 ? I checked it twice :( I ran one previously at 4.0 but that was many crashes ago.
boot speeds ;)
Now the good news is as you increase cooling capacity there is no need to go easy on the NB........
Ok, I'm about to give up. Even with everything at stock settings and the NB just set to 2200MHz and 1.45V, it starts to cause errors. My NB hates me.
I can tell you that whatever your memory was doing before you will need to retune/tweak it particle........Memory is THE KEY to overclocking all phenom II's, the happier you can keep the IMC the higher they clock with less effort.Quote:
Originally Posted by Particle
Remember my stepping thread?
Take a look at your cpu and see what changed ;)
last we were working on a CV memory controller revision
These chips are CY making them different even from the 720be memory controller.
You need to start from scratch, try swapping memory in dimms etc etc
as seen here. CY memory controller revision ;)Quote:
Originally Posted by namegt
I was having same issues until switching bank locations for my dimm's. If you already tried that I dunno then:confused:..Quote:
Originally Posted by Particle
I switched banks and am having trouble just POSTing. I'll continue to play with it and see what I come up with.
Ok, I've started over from scratch. With DIMMs in the alternate banks (the far set), I've been doing better. Ran six passes of threaded mode Cinebench R10 x64 without issue at 2400MHz. Thanks for the tip, charged & chew.
That's good news, I hope it keeps going good.:up:Quote:
Originally Posted by Particle
Something else that keeps throwing me is that often the M4A79T won't have video on startup but is otherwise running fine. Windows will boot and everything but no video. I'd been clearing CMOS, but I think pulling the power entirely has the same effect, I can boot with video again. Simply turning off the board doesn't work.
Any beta BIOSes I should know about? :)
Six runs at 2600 and boot at 2800. Testing with 2800 in a sec.
Quote:
Originally Posted by ecat
Hmm you super pi1m time is...slow hmmm maybe chang to some higher nb clocks?, are you runing 1600 nb?