Max voltage for Ballistix PC3200?

P

PRCop

Guest
Hello all,

What is the MAX VOLTAGE which should be fed to a pair of Crucial Ballistix 512MB PC3200 sticks?

The stock voltage is rated at 2.8v and I am able to run these sticks at 255MHz @ 2.5-3-3-6 without a problem. They ran flawlessly for over 3 hours in MemTest, so I just stopped the testing on them.

Thanks for any info.
 
Great article, my friend! Was reading that even before you replied! lol

The max I can go on my cpu right now (3700+ SD), is 255MHz x 11 @ 2.805GHz with 1.5v on air (ThermalTake Typhoon cooler) and my Ballistix set at 255MHz with 2.5-3-3-6 timings. I am able to run MemTest for hours without an error (test #5), ran Super Pi in 27 minutes 47 seconds plus over 4 hours of Prime95.

As you can see, my system is pretty damn stable but I am running my memory at 2T. I would really like to have the memory at 1T.

Have you done any testing with 1T and 2T? I think that I can't do anything about the timings, as they currently are at par with your findings.

As for cooling, I have several 120mm fans on the side panel and one of them blows directly at the ram sticks :p
 
heh, so the airflow should be good. just make sure you keep a close on on the ram, be sure that it doesn't start feeling anything hotter than moderately warm, and they should last a long time :D

as for the 1T/2T thing... i've seen some other asus boards have this issue with 2 sticks too. do you have the latest bios?
for me.. going from 1T to 2T gets me like 2-4mhz more, at the most. nothing to write home about, that's for sure. :p

and thanks for the comments on the review. sorry for going OT here, but do you have any advice on how to make it better? it seems to be a bit lacking to me, but being an oc'ing site, benches and whatnot aren't needed imo.
 
I am currently running the 1008 BIOS because I tried the latest and was having lots of problem overclocking the system. But then again, I was rushing the process and not being careful on the overclock, so perhaps I should give it another try at the latest BIOS.

Do you know if the 1T problem was fixed with the latest BIOS for the ASUS A8N-SLI?

Another thing, for the past few days, I have been getting an error on Prime95 after about 30 minutes. The error says: Rounding was 0.5, expected less than 0.4

What could be the cause of this error? I would like to try and fix it.

As for your review, I really think it's great. Perhaps it would be good to do another test run like you did but using this time perhaps an AMD Athlon 64 3700+ "San Diego" :p ;) Since the memory controller has been updated as far as I know. And if possible, using an A8N-SLI mobo.

Another thing my friend, what is the safest MAX TEMP I should run my cpu at? Right now, with AC turned off in the room, it will run at about 51C FULL LOAD.
 
the max temp for any a64 is around 55-60ºC, stay under 60ºC and your good to go. try to stay under 55ºC if you can as a safe guard.
 
Sounds good to me! I upped my cpu voltage to 1.5v and am running Prime95 right now. Let's see what happens. Right now, the cpu temp at full load is at 47C.

To continue with this line of discussion, what would be the safest MAX vcore for this same cpu?
 
Thanks buddy! Right now, full load is at 46C-47C with Prime95 with 1.5v.

I have my BIOS vcore setting at 1.5000v but ASUS Probe and CPU-Z both show vcore as 1.536v with an occasional dip to 1.520v. Is this normal?
 
PRCop said:
To continue with this line of discussion, what would be the safest MAX vcore for this same cpu?
Click to expand...
it's.. not an exact science in the least. just know that for every 10c increase in temp, the approximate lifespan of the cpu will be halfed. the problem is that you don't know how long it'll last in the first place
for voltage, the lifespan reduction is sorta a square of the voltage increase. i'm not entirely sure how to interpret this, but i think it's so that a 10% decrease = 100% increase (or go the other way, 10% increase = 1/2 life)
but again, who knows how long it'll last in the first place

ontop of all this, different cores have different voltage and temp tolerances. what i have stated is a rough approximation, and some cores are far more or less sensitive, just like how various cores scale better or worse with more voltage and such.

that aside, i try to fit max voltage along a curve of temperature, though it should have a third axis: needed lifespan.
if you only plan on keeping the cpu for a week, feel free to give it 1.8-1.9v :D
however, that wouldn't be practical in the least, so don't advise that anyone does it.
for 90nm revE, i tend to state something like this:
under 40c = 1.6-1.65v (typical water temps)
40-50c = 1.55-1.6v (typical high end air temps)
50-55c = 1.525-1.55v
55-60c = 1.5v

while you can go over without issues, it's just a matter of how gutsy you're feeling. also, damage from temps and voltages happen over a relatively long period of time, unless extremes are in question. i feel it would be safe to go .1-.15v over my curve for testing purposes, but it should be avoided :p


moving on. the rounding error in prime95. this is simply an idication that you pushed something too far, and the cpu is spitting out bad math (think 2+2=5). simply decrease cpu mhz, increase voltage or decrease temps to fix this. maybe a combination thereof

the voltage fluctuations are normal.. though it's strange that your mobo overvolts. most i have seen undervolt by .02-.03v most of the time. :cool:


and lastly, the review. i definitly need a wider range of testing equipment, but.. i'll get there eventually
note that my 3700+ is a revE cpu with the updated memory controller, just in single channel form :p
 
You guys are just AMAZING!!!! Following your advice, Prime95 is once again completely stable! The trick was to take vcore up to 1.5v in BIOS (1.520v-1.536v on CPU-Z) and now it's completely stable once again.

I think I will throw the towel and settle at where I currently stand, which is:

255 MHz x 11 = 2.805 GHz

I could try to go up to 260 MHz but I am betting I will have to up the vcore even further, and I just don't feel comfortable adding more volts for perhaps a 55 MHz gain.

Once again, thanks everyone for your help. I am very grateful. :)
 
i would see where it maxs out without bumping up the voltage, but i have a tendancy to go overboard when trying to overclock :p

glad to see you have everything sorted out. or at least it seems this way :D
 
I thought I would never say this, but I think my CPU maxes out before the memory calls it quits! lol

I just tried to up the HTT to 260 and left the memory at 2.5-3-3-10 @ 2.8v and it ran MemTest flawlessly. It ran over 5 passes of test #5 without a hiccup.

So then I decided to give the cpu a try at 260. I booted in to Windows XP and right away, Prime95 failed.

I went back and forth between the BIOS and Windows upping the vcore each time and once I got Prime95 to fail only after 6 minutes lol, the vcore was at 1.6v and I sure don't want to continue to increase it.

In other words, by running at 255, I am getting the most out of this cpu and with the ability to run my memory at 1:1 with decent timings of 2.5-3-3-10 @ 2.8v.

Speaking of which, maybe I could up the memory voltage and try to get the latencies a bit tighter. What you think, (cf)Eclipse? Perhaps with a higher voltage, I can get 2.5-2-2-10.
 
you would either need a really good stick or a lot of voltage to get down to 2.5-2-2 at 255mhz :D

i think i would try dropping the cpu multiplier one notch, and see if the ram is capable enough to bring the cpu back up to the same speeds :D
(read: if you're topping out 255x11, see if the ram will do 280mhz for 280x10)
 
lol Just when I thought my overclocking routine was over! hahahahaah

Good idea, buddy. So, in case my memory will do 280, what would be better performance wise:

1. 280 x 10 = 2.80GHz
2. 255 x 11 = 2.80GHz

#1 or #2?
 
assuming you can keep 2.5-3-3, the faster ram will be better :D
even if you have to drop to 3-3-3.. 280x10 may still be better, but a bit less (like 0.5% :p)
 
These Ballistix are so good. Even though I can't get to run 280 x 10 because the memory just can't keep up with 280, even at 3-4-4-10, I have opted to stay with 255 x 11.

It gives me the same cpu speed of 2.8GHz but what amazes me is that I set the timings to 2.5-3-2-6 @ 2.8v and the system is MemTest and Prime95 stable! Yup, 255MHz at 2.5-3-2-6 with 2.8v! Gotta love these modules!

Even ran Super Pi 32m without a problem. Did it in 27m 26s. Not bad at all.

Thanks for all your help and info on this overclock!

:)

I have included a couple of CPU-Z and ASUS Probe shots.

screenshot.jpg
 
very nice oc. see if you can crank some vdimm into that ram, slap a fan over em, and see if you can get 2.5-2-2. that would be SWEET!
 
I did try that. But my Asus A8N-SLI only allows me a max of 3.0v for VDIMM. I tried 3.0v and there were less errors in MemTest. Maybe 3.1v would do the trick but my BIOS won't allow me.

One of the 120mm fans on the sidepanel, blows directly on top of the memory. ;) So that should help the cooling.

Anyone have an A8N-SLI with the latest BIOS installed, that can perhaps verify for me if the vdimm setting got cranked up?

Even though I can live with 2.5-3-2-6, what I would really love is to have the command rate at 1T. Right now it's at 2T.
 
I start getting LOTS of errors when running the #5 test within MemTest.
 
hmm, thats really weird, did it do that at stock as well? cause if it did, then you might have an RMA at hand. it might be a mobo related issue.
 
oh no, I can do 1T up until about 217mhz-220mhz. After that, I have to switch to 2T. I read somewhere that it might be a a BIOS problem with the ASUS A8N-SLI Deluxe mobo. I feel like trying out the latest bios from ASUS.

Right now I have bios 1008, which is the most stable one.
 
ive heard of the 1t issue with the Asus board aswell, its known as a good for stock board, but at ocing it lames up. im sure asus will fix it soon.
 
I will see if the forums at Asus have any info on this.

By the way, I decided to give your suggestion of running 255mhz at 2.5-2-2-6 a try. I have it at the board's max of 3.0v vdimm and right now, MemTest is at it's 7th pass of Test #5 and NO PROBLEMS so far!

*fingers crossed*

If I can actually run 2.5-2-2-6 @ 255mhz AND get 1T to work at this frequency, I will flip out! :p
 
i usually run all of memtest cause well, i feel better knowing its 100% clean. and i dont use super pi for stability, i use super pi 32m looped twice.
 
Sounds good to me! Right now it's on it's 14th pass of test #5. I will start MemTest again and let it do a complete test run. If it passes, I will try Super Pi for 2 loops. Will let you know how it goes.
 
PRCop said:
If I can actually run 2.5-2-2-6 @ 255mhz AND get 1T to work at this frequency, I will flip out! :p
Click to expand...
honestly, i would be excited too, and i think you can do it, cause i don't gain more than 1-2mhz from going from 2.5-2-2 to 2.5-3-2..
2.5-2-3 is a different story though, and i get like 10-15mhz more than 2.5-2-2 at the same voltage :D


as for 1T, i blame the mobo. try swapping ram slots?
 
Well, guess what guys! Left system running overnight and it ran Prime95 without a problem! Before that, I ran MemTest's full range test running for about 2 hours and it gave no errors.

This means that yes, my system is completely stable at 255mhz x 11 with 2.5-2-2-6 timings with 3.0v of vdimm!!!!!!

Now, I need to try and get around the 1t command rate problem. Right now, I am getting about 5,960 points on Sandra 2005 memory benchmark with 2t. If I can get it to run at 1t, I will definitely be near the 7,000 mark.

I know that it's only a synthetic benchmark, but I really want to have the 1t enabled, because many have said that running it at 2t, does not do the A64 any justice.
 
with 1t, you should definately be near or even over 7k. i hope asus wants to fix that issue for ya, as you should definately be able to run 1t over 220.
 
I'm flashing my mobo to the latest BIOS which is 1013. Let's see if this fixes the problem.
 
PRCop said:
I'm flashing my mobo to the latest BIOS which is 1013. Let's see if this fixes the problem.
Click to expand...
good luck!

also, i think i mentioned this already, but have you messed with the dimm slots the ram is in?
 
messing with dimm slots could fix it for sure. some mobos are really picky, they can be jerks. i remember Asus always being jerks with ram.
 
I read it but haven't tried it yet! :p I finished flashing to 1013 and the system is working without a problem. The BIOS settings remained the same EXCEPT for a few memory parameters.

Now I have a couple new parameters to deal with. Before, when viewing CPU-Z, it would show my timings as 2.5-2-2-6 AND 22 (Trc) , 16 (DRAM Idle Timer).

The 2 new settings in the BIOS are for Trc and DRAM Idle Timer. Now they show as 11 (Trc) and 16 (DRAM Idle Timer). So by updating my BIOS, my timings are a bit tighter at (Trc).

Also, the command rate was defaulted to 1t. I switched it to 2t to make sure the system worked. Now I will go back to the BIOS and give 1t a try.
 
PRCop said:
The 2 new settings in the BIOS are for Trc and DRAM Idle Timer. Now they show as 11 (Trc) and 16 (DRAM Idle Timer). So by updating my BIOS, my timings are a bit tighter at (Trc).
Click to expand...
yay. trc is one of the more important secondary timings. that'll bring latency down a bit :cool:
 
Ok, I switched it to 1t and it booted up but right off the bat in MemTest, I started getting errors.

The timings are at 2.5-2-2-6-11-16-1t

I will try to move them to the other slots. Stay tuned.
 
PRCop said:
Ok, I switched it to 1t and it booted up but right off the bat in MemTest, I started getting errors.
Click to expand...
:(

i would try to avoid giving it more voltage.. but i would also check to see if 3.1v fixes the problem. how many errors are you getting?
 
Mobo won't allow more than 3.0v on vdimm. I tried switching the slots but it won't boot into MemTest. I relaxed the timings and still no show. I will put them back to their original slots and try relaxing the timings.

It gave me lots of errors on 2.5-2-2-6-11-16-1t. Maybe relaxing the timings a bit, will allow me to run 1t.
 
You must log in or register to reply here.
Back
Top