![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Well a few weeks back, I posted a thread about how my system would sometimes pass Prime 95 and Super Pi, and sometimes it would fail despite being on the same settings. I seem to have finally found the answer to my problem. Skip the next few paragraphs to the end if you don't want to read my long drawn out story.
Firstly, one of my main concerns was my powersupply wasn't putting out enough amperage on the 5v rail (since my A7N8X-E Deluxe doesn't use 12v to power the CPU), and so I replaced my old 350Watt with a new 535 Watt Enermax PSU.
After replacing the Enermax, I found that low and behold, everything worked fine. I passed Super Pi and Prime 95 for 8 hours. Tested some games, and thought everything was great (this was at 227 Mhzx 11 @ 1.8). Then I decided to see if I can squeeze anymroe out of it and upped the Multiplier to 11.5 Well it was a no go, so I backed it back down to 11x Mulitplier, and decided to continue to stress test it.
After about 10 minutes of Super Pi, it failed with a Not Exact Round error. I was going o crap... not again, same problem. Maybe it wasn't the PSU. So I rebooted, rechecked my settings, and decided to run the test again, this time with Prime 95. Well, Prime 95 failed in 5 minutes, but Super Pi passed. I was like ok? So I rebooted, upped the voltage to 1.825, and tried again with Super Pi and Prime95. This time SuperPi failed, but Prime 95 ran for 4 hours.
Now I was stumped. So it's not the PSU, and I was still getting weird errors. So I decided to clock it back to stock, and reformat (thinking it might be my windows installation). After a format, I ran Super Pi (nothing loaded, not even drivers). Super Pi passed fine. I Then clocked it back to 2.5 @ 1.8, (still nothing loaded), and it passed Super Pi fine. Great I thought, just my windows installation.
I messed around with my FSB again, trying to see if I can squeeze a few more mhz out of it, but to no success, I backed it back down to 2.5 and ran Super Pi. Failed again withint 10 minutes.
Now i was really scratching my head. I tried all the ram configurations, different voltages (Ram and CPU). Nothing worked.
Then something occured to me, the system was stable at 2.5ghz (running Super Pi and Prime95) whenever I clocked it to 2.5 after clocking it back down to stock first. So I went back to bios, reset it to stock, installed drivers and software, rebooted, overclocked it again and ran Super Pi and Prime95. It ran both overnight fine. I clock it back up, then back to 2.5, and it failes again within 10 minutes.
So... after all that testing and messing around, turns out if I go straight to the overclock from another overclocked setting, it isn't stable. However if I clock it to stock (or even underclock it) and clock it back to 2.5ghz (227x11 @ 1.8), it would be 100% stable, and as long as I didn't touch the bios after that, it would remain so. I'm still scratching my head as to why, but I'm glad it's stable and fixed.
Firstly, one of my main concerns was my powersupply wasn't putting out enough amperage on the 5v rail (since my A7N8X-E Deluxe doesn't use 12v to power the CPU), and so I replaced my old 350Watt with a new 535 Watt Enermax PSU.
After replacing the Enermax, I found that low and behold, everything worked fine. I passed Super Pi and Prime 95 for 8 hours. Tested some games, and thought everything was great (this was at 227 Mhzx 11 @ 1.8). Then I decided to see if I can squeeze anymroe out of it and upped the Multiplier to 11.5 Well it was a no go, so I backed it back down to 11x Mulitplier, and decided to continue to stress test it.
After about 10 minutes of Super Pi, it failed with a Not Exact Round error. I was going o crap... not again, same problem. Maybe it wasn't the PSU. So I rebooted, rechecked my settings, and decided to run the test again, this time with Prime 95. Well, Prime 95 failed in 5 minutes, but Super Pi passed. I was like ok? So I rebooted, upped the voltage to 1.825, and tried again with Super Pi and Prime95. This time SuperPi failed, but Prime 95 ran for 4 hours.
Now I was stumped. So it's not the PSU, and I was still getting weird errors. So I decided to clock it back to stock, and reformat (thinking it might be my windows installation). After a format, I ran Super Pi (nothing loaded, not even drivers). Super Pi passed fine. I Then clocked it back to 2.5 @ 1.8, (still nothing loaded), and it passed Super Pi fine. Great I thought, just my windows installation.
I messed around with my FSB again, trying to see if I can squeeze a few more mhz out of it, but to no success, I backed it back down to 2.5 and ran Super Pi. Failed again withint 10 minutes.
Now i was really scratching my head. I tried all the ram configurations, different voltages (Ram and CPU). Nothing worked.
Then something occured to me, the system was stable at 2.5ghz (running Super Pi and Prime95) whenever I clocked it to 2.5 after clocking it back down to stock first. So I went back to bios, reset it to stock, installed drivers and software, rebooted, overclocked it again and ran Super Pi and Prime95. It ran both overnight fine. I clock it back up, then back to 2.5, and it failes again within 10 minutes.
So... after all that testing and messing around, turns out if I go straight to the overclock from another overclocked setting, it isn't stable. However if I clock it to stock (or even underclock it) and clock it back to 2.5ghz (227x11 @ 1.8), it would be 100% stable, and as long as I didn't touch the bios after that, it would remain so. I'm still scratching my head as to why, but I'm glad it's stable and fixed.