oc causing bsod?

[divided]

when i try to run 2.5 1:1 on the memory in my sig. i can prime for like 2 hours then bam! bsod and reboot. question: is it a power issue or timing issue? ive read several people saying when they fail prime they get a calculation error, but never heard of anyone getting a reboot. any help would be much appreciated.

[Fight Game]

There's no way anybody could answer this question. A BSOD can happen for a wide variety of reasons. If your system is completely stable at stock speeds, then that would rule out any driver issues. Most of the time it's a setting on the ram. I'd start here.

 

Also, just fyi, there's no benefit of running your ram 1:1 with the processor. Matter of fact, if you have to stop upping the HTT (FSB),. because the ram can't keep up, you'll be missing out on alot of performance that cpu could be giving you. Read the guides here in the stickies on overclocking. There you'll learn how very little the ram frequency and timings means in terms of performance. Of course it depends on what you do with your computer, but generally speaking, for probably 95+% of us, this is true. And even if you fall into the other 5% category, still doesn't mean you should run your ram 1:1 with the processor, because chances are, the cpu will be able to go further, and you'll be missing out on this performance.

[Sorrento]

Also, just fyi, there's no benefit of running your ram 1:1 with the processor.

Its between 5 and 10% actually

 

Indeed AMD64 Socket 939 changed the perspective about running CPU/RAM at synch or asynch modes since the memory controller is super eficient... yet a diference, even a tinny one, still exists.

 

But, like Fight Game says: if ram is limiting you, just run in asynchronic mode with a cpu/ram divider and forget about it

[Fight Game]

Well I guess it depends on which benchmark you look at, as far as the percentages. Also depends on how accurate, and precise the benchmark itself is. Also depends on how the changes in the numbers translate to real world use. For example, if you overclock 1 component 50%, and in a couple benchmarks the numbers go up 50%, it doesn't mean anything will actually be 50% faster in real world use. I doubt that even if you overclock EVERY component 50% and EVERY benchmark was 50% better, the overall system would even come close to 50% better in real world use. And as usual it always depends on what you do with your computer. Depends on many many things, but bottom line is, is that the % increase in benchmark's is never the same % increase in real world performance. For the most of us here we want fast loading times (for windows, programs, etc), smooth game play, and usually the ability to do more than one thing at one time. For me, I like fast loading times the most. If I overclock every component, say...25%, I doubt my load times will be 25% faster. Probably more like 15% if I'm lucky. So normally when I'm waiting 20 seconds (omg 20?), then all this overclocking will save me 3 seconds on each load time. Add up all these 3 seconds, and it might equal all the time I spent overclocking and getting stable! Don't get me wrong, I like overclocking, not just for saving the ~3 seconds on each load, but more just for the feeling of getting something extra for free.

 

Edit..hmm still wouldn't neccessarily be free would it...since my time can = money, and I spent alot of time oc'ing.

 

Also, I think overclocking on slower computers makes more of a difference in real world use, since they were so slow it was easy to notice small increases. These computers out nowadays are already so damn fast, it's impossible (or very very hard) to notice 10-20%, except in a benchmark of course.

[divided]

well i guess i had more of a generic question. i understand the benefits of running 1:1 and i understand dividers as i have done the benches and analyzed the results. so again my orignal question....DOES A BSOD/REBOOT IN PRIME NORMALLY HAPPEN CAUSE OF POWER ISSUES OR RAM TIMINGS? i thought imporper timings would cause a calulation error, not a reboot. but again i could be wrong. ive noticed though that when i loosen the timings i can run for 8 hours, but when i run it again the next day or perahps the next week, it bsod and reboots, which the EXACT settings that were stable for 12+ hours. it seems like anything over ~235 with this mem and it fails. with timings as loose as 2.5-4-4-8

[Fight Game]

could be either, or both. Only way to know what's happening here with your system, is to try things, and narrow down the possibilities. gl

[divided]

from what ive seen it looks as if i loosen the timings. the bsod with go away. i just never heard of timings causing a reboot. i was thinking it is the psu. but its respected so i didnt look to much into it. i thought my mem/as i ve seen others, get it at 260+ stable. i cant even get 235....ill keep testing......

[ICON57]

i have seen the bsod, caused by memtimings, and right after it blue screened, it rebooted, so yeah, bad mem tinmings can cause both to happen...imho

[divided]

thanks for that definitive answer icon.

[Sorrento]

Well I guess it depends on which benchmark you look at, as far as the percentages. Also depends on how accurate, and precise the benchmark itself is. Also depends on how the changes in the numbers translate to real world use. For example, if you overclock 1 component 50%, and in a couple benchmarks the numbers go up 50%, it doesn't mean anything will actually be 50% faster in real world use. I doubt that even if you overclock EVERY component 50% and EVERY benchmark was 50% better, the overall system would even come close to 50% better in real world use. And as usual it always depends on what you do with your computer. Depends on many many things, but bottom line is, is that the % increase in benchmark's is never the same % increase in real world performance. For the most of us here we want fast loading times (for windows, programs, etc), smooth game play, and usually the ability to do more than one thing at one time. For me, I like fast loading times the most. If I overclock every component, say...25%, I doubt my load times will be 25% faster. Probably more like 15% if I'm lucky. So normally when I'm waiting 20 seconds (omg 20?), then all this overclocking will save me 3 seconds on each load time. Add up all these 3 seconds, and it might equal all the time I spent overclocking and getting stable! Don't get me wrong, I like overclocking, not just for saving the ~3 seconds on each load, but more just for the feeling of getting something extra for free.

 

Edit..hmm still wouldn't neccessarily be free would it...since my time can = money, and I spent alot of time oc'ing.

 

Also, I think overclocking on slower computers makes more of a difference in real world use, since they were so slow it was easy to notice small increases. These computers out nowadays are already so damn fast, it's impossible (or very very hard) to notice 10-20%, except in a benchmark of course.

So, you are saying that overclocking is the fountain of youth?? Because after some time interval the amount of saved time with overclocked systems can really make a diference

 

In the real world there is hardly a diference between running asynchronic mode or synchronic at same cpu speeds with the AMD64, thats what I mean... yet, with very strong benchmarks these diferences can be seen.

 

For example, at 2.9 ghz (11 x 264) with DDR400 memory on the 3:4 divider SuperPi 1m runs in about 29 seconds... while the same cpu speed but with Platinum Rev2 ram working 1:1 makes the same SuperPi run in 28 seconds. With mod version 1.4 the diference is around 0.8 to 1.1 seconds. Other benchmarks could point to similar diferences... but still the truth is these diferences are sooo tinny in comparison to systems running a memory controller on the Northbridge, this is what makes the AMD64 so remarkable.

 

In overclocking I can say it is a lot better to run 1:1 than using a divider... something in the memory controller or the mobo's chipset doesn't like some frequencies and speeds doing this