it's possible that it's a board limitation, because something similar happens to my clawhammer.
the following is for 230x10.
at 1.45 +3% (a touch under 1.5v) is prime stable for about 15 min
1.5: instant fail
1.5 + 3% is primestable for 12+ hours
1.55 fails within 5-10 min.
i haven't played around with the higher voltages much because my chip is very insensitive to voltage (at normal room temp, i need 1.75v to get to 2.4ghz, and then the load temp is hitting 60c, which is too high for me)
3500 CBBID0451FPAW
http://img.photobucket.com/albums/v636/loc_oc/30.jpg
Absolute max on the XP120. Will try yr settings later Andy.
I got the same voltage problem with my MSI NEO2. The max i can go is 1.6v (10%+1.45) if i go above the 1.45 it dont respond, so i leave it at that, and use the 10% option to get the max of 1.6v...makes no sense
here is what i get with 1.6v
SIOUX
My Neo2 lets me have up to 1.65V when I set 1.50 for VID and +10%..
I still can't quite get over it but he's right, just tried a ex-sucky winchester (0502 / CBBID) 3200+ that wouldn't go pass 2200mhz with any volts / ram timings.
With Andy's settings its now priming @ 2500mhz and I dont think thats the ceiling yet, nice find man and thanks for sharing the info. :cool:
There must have been a reason why mobo makers decided to have the voltages based on a base value multiplied by a percent. This might have been it. They might not have been able to officially say it because of contractual obligations. (Like with Intel and PAT) I wish I had a winne for testing.
what your mobo ?Quote:
Originally Posted by grenadier
maybe your mobo not provide to adding Voltage by %
for DFI NF4 it can change Voltage by %
:)
Thanks for the tip, but I'm already using 1.5V x 110% and it's tracking fine (SmartGuardian displays 1.61V while it should be 1.65V but not a big deal). Chip is still maxed out at 2.4Ghz 1.65V for stability, damn CBBID 0505 :(
(I can run 3DMark2001 at 2.5GHz 1.7V, haven't tried higher yet)
What's the max safe 24/7 voltage you guys would recommend? At 1.65V temps are always under 40C...
1.7v is = to ~2.0v on air for an XP. That is why. 1.6v = ~1.875v for an XP. So you'd be quickly killing the chip with that kind of voltage.Quote:
Originally Posted by bachus_anonym
As for this issue, here's a link to someone who seems to be having the same problem: http://forums.extremeoverclocking.co...41#post1623641
It sounds like a board issue. Maybe the VID is not being applied correctly in bios, for whatever reason.
You know when I briefly had my 3200+ Winnie 0451, the board I had had no voltage control, so I was only able to adjust it via clockgen. However clockgen @ 1.55v only gave my chip 1.45v - same thing? :confused: I thought it was clockgen, but maybe not.
Interesting. Did AMD change the higher VID codes?
I am not familiar with the chaintech mobo, but I have a similar method of changing core voltage within my nforce4 a8n-sli and have discovered that even though it looks like there is only one way to change voltage, the BIOS applies the percentages "behind the scene", so to speak...Quote:
Originally Posted by grenadier
Anyway the way to check for this is to load 1.4v in BIOS then boot into windows (with stock settings) use clockgen and record the VID value. Reboot and change the 1.4v to the next highest and repeat.
IN the asus board sometimes the VID for a relatively high (for this board) voltage ~1.6 will have a LOW VID but a ~1.5 will have a higher VID value. This means there is something else the BIOS is using to adjust the voltages. I am assuming it is similar to the other nforce4 boards, but since the asus was meant for all the noobs out there, they probably decided against adding another option for simplicities sake.
Is it best to run with a 1.6v that has a 1.35 VID on these boards vs a 1.6v with a 1.45v VID? (I haven't really done any testing on this, but it would be interesting to know...)
anyone else try this out?
The reason you set voltage with percentages is because A64s have two variables for setting the voltage.
First is CPU VID, and it is used for changing the voltage for C&Q. It is a function of CPU and not of a motherboard. It ranges from 0.8 V up to 1.55 V on 130nm CPUs and up to 1.45 V on 90nm CPUs.
The second is an usual CPU Voltage selector, used for overclocking. And it has to raise the voltage in percents, because it depends on the CPU VID selected. This feature depends on the board itself.
Hopefully i explained it right with my poor english :rolleyes:
I have a MSI neo2 Platinum. Stock settings for CPU VID is 1.500 which gives me about 1.376-1.392 volts according to CPU-Z. The sec cpu voltage is done by % in the following...
+3.3 gives me 1.424-1.440
+5.0=1.472-1.488
+8.3=1.520-1.552
+10=1.586-1.600
There is no other choice like 110% ect, though I never tryed lowering the CPU VID below 1.500, I will have to try that. I set it to 1.55 and there was no diff in voltage or stability.
I have one of the new 3500s and she is stable as a rock with +3.3 voltage@2.5 ghz by way of 11x multi and 228FSB. Now if I try 238FSB she crashes during 3DMarks cpu tests no matter what voltage I use up to 1.600. I can post and do CPU-Z @2.7+ but as soon as I run a program...Crash.
Things to note; in bios for CPU VID I can set it for AUTO as well, wonder if that will help. Why it is set by default to 1.500 in bios even though I have a winnie? Yet bios shows 1.4 voltage with that setting as well as CPU-Z and speedfan ect...
I knew there was something up since my cpu runs stable @2.5 with only 1.42 core, yet crashes@2.6 no matter what voltage.
One other thing, the weird part is if I use stock settings (no 3.3% over cpu vid) I idle at around 38-40. With 3.3% over (~1.42) I'm iding right now at 34 and max at 49. Seems like I'm using less voltage overclocked or something :p:
I'm going to try this on my 3000+ 0449 that wont clock past 2250 under water with a DFI NF4 when I get home. Will post my results. Thanks a lot for the tip, Andy.
Make perfect sense to me. Your english is fine. :)Quote:
Originally Posted by Ref
I love it how all the foreign people will add "Please excuse my english" at the end, when they did a great job.Quote:
Originally Posted by andyOCZ
Then people who should know how to speak english correctly type as if they don't know english! LOL!
It seems that only non-native speakers care to check if what they say/type is correct, ironically.
AFAIK, the Chaintech VNF4 Ultra does not have any % voltage settings. You can set VID and that's it. So, basically, I only have 1.4, 1.45, and 1.6 as viable settings. Chaintech needs to step up and come up with a better bios that fixes this!
I belev yoo was referrin ta mee?Quote:
Originally Posted by Ackbar
Hey man, I'm tired :)
Not only can I finally determine that this proc's mem controller sucks, but I broke my previous SPi record held by an old P4 by 3 secs :). I think the score for 2700 is a bit low, but I blame the divided memory. VID set to 1.325 with +136% special control (~1.73v actual). To sum it up, this trick gained me close to 500MHz. Thanks again Andy! :toast:
1.7v is a bit high for Winchester don't you think? I use 1.5v max.
If the temps are fine, why not give the CPU the voltage it wants? :p:Quote:
Originally Posted by dX.
Awsome news!Quote:
Originally Posted by Gogeta
At 33c load it's loving 1.7v :DQuote:
Originally Posted by dX.
very intresting trick man
but i wonder:
does it only work when u clearly see in windows that my voltage changes in bios didnt work?
cause while testing cpu vid about 1-2 weeks ago, it really did work here (smart guardian reported higher volts and higher cpu temps when increasing volts in the bios (running 1.55x104% atm, tested till 1.55x110%)
The only thing I can imagine is that the mobo's increase the voltage in each of the cases BUT increasing the VID is better for some reason.
I think it's important to figure out exactly what we're on to here and then somebody should write an article on this. It could be very important!
Is there anybody with an OLDER Winchester who also received a nice gain?