I've observed rather unusual behavior of my motherboard, while trying to find stable settings at 472x7 with PL:8 vs PL:9 . My configuration:
But then I repeated the prime95 after I turned PC on in the morning next day (so it was powered off for a good while) - now primes were failing like crazy, within minutes. So I thought WTH. I backed off to PL:9, left pimes again, came back - all is fine. Out of curiosity, switched to PL:8, primes for few hours - all is fine again, no errors.
And this pattern always repeats (tried 6 times already) - when my PC is after a longer period of being turned off ("cold" as mentioned in subject), there's no way I can get it stable at "only" 1.67 and PL:8 (I tried plenty of combinations of different other settings, but they give no visible effect - for reference it was with vtt 1.36, pll 1.56, fsb strap 333 (400 of course tested as well, the same results), memory 1:1 @5-5-5-15, ct: stronger). But when the PC is "warmed up" - it's almost rock stable.
"Almost" - because I did tests with GoldMemory as well (I find it much superior to memtest, it finds errors far, far faster). In "cold" situation - it was literally stuffed (>200 page history when I got back home) with errors, nonetheless - practically all of them happened early (first hour). Yesterday I tried it with Gold running for 7 hours but in "warmed up" scenario - it had only 4 errors total. As you can see, it confirms prime results (failures within minutes vs. 20 hours with no errors at all).
Now I wonder - what can be actually responsible for this type of behavior ? Any more experienced overclockers that have seen something like that ? Some self adjustment done by the board after running for some time ? Low temperatures (!?) ?
BTW - at PL:9 and 1.51v the board passes tests regardless of when they were started.
- Old E6400 with TR Ultra 120
- Asus rampage formula, bios 403 (with replaced el-cheapo double sided sticking tape for some real thermal compound - got -6C this way on NB)
- Corsair 2x1gb 6400c4 (v2.x - promos)
- Corsair 2x1gb 6400c4 (v6.x - psc)
But then I repeated the prime95 after I turned PC on in the morning next day (so it was powered off for a good while) - now primes were failing like crazy, within minutes. So I thought WTH. I backed off to PL:9, left pimes again, came back - all is fine. Out of curiosity, switched to PL:8, primes for few hours - all is fine again, no errors.
And this pattern always repeats (tried 6 times already) - when my PC is after a longer period of being turned off ("cold" as mentioned in subject), there's no way I can get it stable at "only" 1.67 and PL:8 (I tried plenty of combinations of different other settings, but they give no visible effect - for reference it was with vtt 1.36, pll 1.56, fsb strap 333 (400 of course tested as well, the same results), memory 1:1 @5-5-5-15, ct: stronger). But when the PC is "warmed up" - it's almost rock stable.
"Almost" - because I did tests with GoldMemory as well (I find it much superior to memtest, it finds errors far, far faster). In "cold" situation - it was literally stuffed (>200 page history when I got back home) with errors, nonetheless - practically all of them happened early (first hour). Yesterday I tried it with Gold running for 7 hours but in "warmed up" scenario - it had only 4 errors total. As you can see, it confirms prime results (failures within minutes vs. 20 hours with no errors at all).
Now I wonder - what can be actually responsible for this type of behavior ? Any more experienced overclockers that have seen something like that ? Some self adjustment done by the board after running for some time ? Low temperatures (!?) ?
BTW - at PL:9 and 1.51v the board passes tests regardless of when they were started.
Facebook
Twitter