So I signed up for this forum today just to make this post, so sorry for being newby or whatever.
About 2 weeks ago I finished putting together a new system with mostly new parts. The CPU is a brand new Phenom II x2 555 BE, and the motherboard is a new ASUS M4A89GTD PRO/USB3, and the RAM is a pair of DDR3 dual-channel Kingston somethings rated at 1600 MHz. It has air cooling with a huge radiator and quiet fan that seems to work really well (coolermaster V8). The HDD and the case and most of the other stuff is old and used (by me), except the PSU (which I bought in a new-looking box but it has a sticker on the inside that says "REFURBISHED", oh well it works fine). So anyway after I got it to POST I hooked up the screen and went into the BIOS. The stock settings were:
200 MHz base clock
"AUTO" CPU multiplier (which is actually 16x unless you enabled cool-n-quiet)
100 MHz PCIe bus (which runs the internal graphics and maybe the HDD, idk)
"AUTO" CPU/NB multiplier (which is 10x unless you change the FSB clock)
"AUTO" HyperTransport multiplier (same as CPUNB)
Also this motherboard had this feature that supposedly unlocks the dormant cores automatically, but I wasn't really counting on that.
So I boot up into windows XP fine from an old HD. Then, when I rebooted, I just flipped the unlock switch and it showed up with all 4 cores It was stable at stock speeds with linpack for hours (Intel Burn Test). Then I upgraded to windows 7 x64 (took hours to get it to install right) and it didn't make a difference with the stability. Then I experimented with the unlocked CPU multiplier, and it seemed like the max stable cpu speed was 3.9 GHz with a multiplier of 19.5, at like 1.45 volts (4.0 would work, but not under high stress with safe voltages) This is all with 4 cores.
So 3 days ago I decided to reset the cpu multiplier to auto and try playing with the base clock instead. I got it up to 244 MHz, which was 3.904 GHz on the CPU and it was completely stable (2 hours of maximum stress linpack). It didnt pass the stress test at 225. So I left it at 244 and started experimenting with benchmarks and the CPU/NB and the HyperTransport bus. It turned out my system performed best when I left both of the multipliers at 10x (instead of letting them auto down to something less when I had the FSB overclocked). So I left it like that. At this point all the voltages were still auto. I had to leave the CPU voltage at auto if I wanted to use cool n quiet, which I like for everyday unintense use (like writing this post). The CPU NB voltage was like 2.35, which worried me a bit, but I've never overclocked with an AMD processor before, so i figured if it was on auto it must be ok.
So yesterday afternoon I was using the computer and I was done so i decided to put on linpack just for fun while I was not using it. But literally seconds after I started it, I got a BSOD. So i thought, hmm that's weird, maybe I'll take it down to 243 MHz. So I did, and I left linpack running while I mowed the lawn and had a snack. It ran for about an hour with no errors. Considering my computer hadn't even had time to heat up before it crashed the last time, I figured it was just a fluke. I left it at 243. Then, later that same day, I rebooted it and ran linpack again just to be sure. It wasn't immediate this time, but within 2 minutes it had crashed again. Now I was pretty confused, because I had just ran it for an hour successfully. So I went down to 240. I booted it up and ran linpack, this time it crashed in 5 minutes. So I did some experimenting to find out what was wrong. I underclocked the RAM and the CPU/NB and the HyperTransport, but not the CPU, to see if the cpu was the problem. It was. Then I put everything back at stock settings, plus the core unlocking. No crashing. Then I raised the cpu multiplier to 19.5 again, and it crashed under stress. So then I left it at 3.9 GHz but I turned off the extra 2 cores. No crashing. Then I activated just the first 3 cores at the same speed, no crashing. Then I activated the 1st, 2nd, and 4th cores only, no crashing. Then I activated all 4 cores, and it crashed under stress. So I was thinking about trying it at 3.8 GHz, but meanwhile I restarted it with all 4 cores. The bios gave me a message, something like, "This CPU does not have any cores to unlock! Press F1 to go to BIOS, F2 to boot without the core unlocker" So I pressed F1, and the bios looked completely normal. It recognized my processor the way it normally did, seemingly. So I rebooted it and apparently now I had my 3.9 GHz quad core again. I rebooted again and set all the settings to stock in the bios and turned off the core unlocker because I didn't want to mess up my brand new CPU.
So obviously the whole core unlocking thing is not supposed to be reliable. But does anyone have any theories about what's going on? Something definitely happened to it recently, but I don't know what.
Another note: The cooling is working great. With 4 cores at 3.9 GHz at full load with linpack, the absolute maximum temp I ever saw was 51 C. I guess that's just enough to crash it at 4 GHz, but I'm satisfied. I thought about replacing the fan with a more powerful one, but I enjoy the quietness.
The system is still pretty fast at stock settings, so I might just leave it that way if it seems risky to do more testing.
Any ideas? Also thanks in advance!
About 2 weeks ago I finished putting together a new system with mostly new parts. The CPU is a brand new Phenom II x2 555 BE, and the motherboard is a new ASUS M4A89GTD PRO/USB3, and the RAM is a pair of DDR3 dual-channel Kingston somethings rated at 1600 MHz. It has air cooling with a huge radiator and quiet fan that seems to work really well (coolermaster V8). The HDD and the case and most of the other stuff is old and used (by me), except the PSU (which I bought in a new-looking box but it has a sticker on the inside that says "REFURBISHED", oh well it works fine). So anyway after I got it to POST I hooked up the screen and went into the BIOS. The stock settings were:
200 MHz base clock
"AUTO" CPU multiplier (which is actually 16x unless you enabled cool-n-quiet)
100 MHz PCIe bus (which runs the internal graphics and maybe the HDD, idk)
"AUTO" CPU/NB multiplier (which is 10x unless you change the FSB clock)
"AUTO" HyperTransport multiplier (same as CPUNB)
Also this motherboard had this feature that supposedly unlocks the dormant cores automatically, but I wasn't really counting on that.
So I boot up into windows XP fine from an old HD. Then, when I rebooted, I just flipped the unlock switch and it showed up with all 4 cores It was stable at stock speeds with linpack for hours (Intel Burn Test). Then I upgraded to windows 7 x64 (took hours to get it to install right) and it didn't make a difference with the stability. Then I experimented with the unlocked CPU multiplier, and it seemed like the max stable cpu speed was 3.9 GHz with a multiplier of 19.5, at like 1.45 volts (4.0 would work, but not under high stress with safe voltages) This is all with 4 cores.
So 3 days ago I decided to reset the cpu multiplier to auto and try playing with the base clock instead. I got it up to 244 MHz, which was 3.904 GHz on the CPU and it was completely stable (2 hours of maximum stress linpack). It didnt pass the stress test at 225. So I left it at 244 and started experimenting with benchmarks and the CPU/NB and the HyperTransport bus. It turned out my system performed best when I left both of the multipliers at 10x (instead of letting them auto down to something less when I had the FSB overclocked). So I left it like that. At this point all the voltages were still auto. I had to leave the CPU voltage at auto if I wanted to use cool n quiet, which I like for everyday unintense use (like writing this post). The CPU NB voltage was like 2.35, which worried me a bit, but I've never overclocked with an AMD processor before, so i figured if it was on auto it must be ok.
So yesterday afternoon I was using the computer and I was done so i decided to put on linpack just for fun while I was not using it. But literally seconds after I started it, I got a BSOD. So i thought, hmm that's weird, maybe I'll take it down to 243 MHz. So I did, and I left linpack running while I mowed the lawn and had a snack. It ran for about an hour with no errors. Considering my computer hadn't even had time to heat up before it crashed the last time, I figured it was just a fluke. I left it at 243. Then, later that same day, I rebooted it and ran linpack again just to be sure. It wasn't immediate this time, but within 2 minutes it had crashed again. Now I was pretty confused, because I had just ran it for an hour successfully. So I went down to 240. I booted it up and ran linpack, this time it crashed in 5 minutes. So I did some experimenting to find out what was wrong. I underclocked the RAM and the CPU/NB and the HyperTransport, but not the CPU, to see if the cpu was the problem. It was. Then I put everything back at stock settings, plus the core unlocking. No crashing. Then I raised the cpu multiplier to 19.5 again, and it crashed under stress. So then I left it at 3.9 GHz but I turned off the extra 2 cores. No crashing. Then I activated just the first 3 cores at the same speed, no crashing. Then I activated the 1st, 2nd, and 4th cores only, no crashing. Then I activated all 4 cores, and it crashed under stress. So I was thinking about trying it at 3.8 GHz, but meanwhile I restarted it with all 4 cores. The bios gave me a message, something like, "This CPU does not have any cores to unlock! Press F1 to go to BIOS, F2 to boot without the core unlocker" So I pressed F1, and the bios looked completely normal. It recognized my processor the way it normally did, seemingly. So I rebooted it and apparently now I had my 3.9 GHz quad core again. I rebooted again and set all the settings to stock in the bios and turned off the core unlocker because I didn't want to mess up my brand new CPU.
So obviously the whole core unlocking thing is not supposed to be reliable. But does anyone have any theories about what's going on? Something definitely happened to it recently, but I don't know what.
Another note: The cooling is working great. With 4 cores at 3.9 GHz at full load with linpack, the absolute maximum temp I ever saw was 51 C. I guess that's just enough to crash it at 4 GHz, but I'm satisfied. I thought about replacing the fan with a more powerful one, but I enjoy the quietness.
The system is still pretty fast at stock settings, so I might just leave it that way if it seems risky to do more testing.
Any ideas? Also thanks in advance!