The question is simple enough -- Can I get the CPU to it's maximum speed at a reasonable voltage and then either be able to run less voltage at a given speed than was previously possible, or reach a higher maximum overclock?
Ok, I'll just start by saying this is my first thread, so please be kind. I've got a little experience, but I've never really had the time to do much fine tuning. Well, now I do so let the fun begin.
My idea about this machine is that it's to be a workstation/coding computer to last about 1-2 years before I replace it with whoever wins the next generation battle, with the main target app being a compile of a large Visual Studio build.
So which processor do I get? I've also got a 165 Opty that can do 2.65 GHz on 1.392V on a LanParty SLI-DR, so I was thinking about getting another one. Well, now that the supply is tightening up it's up to 330 or so OEM and 350 retail, I was kind of thinking that might be overkill for what I'm trying here. Also I knew from overclocking the 165 that one of the cores was substantially better than the other, so I knew that a single core good stepping would take me much higher than 2.65, and for about 2/3 of the price. So I went with the 148, as most of the others are slower and almost the same price or unavailable.
My Tools:
MSI K8N Neo2 Platinum work handme down check out the damage[/L]
Opteron 148 (11 multi)
2*512 OCZ 3200 Plat Rev 2
Zalman CNPS9500 LED
Raptor 73
The board will sustain me until my urge to buy the ASUS A8N32 Deluxe overwhelms me, or it's at least available.
I'll start out with a goal of one set of Prime95 SFF Torture tests as a stable platform as I move up/down the clock ladder, then I'll try to zero in on the maximum clock that can complete at least 8 hours of straight Prime95 torture. Given the approx lifetime of this CPU before I'll likely want to upgrade, I'll also set 1.55 or thereabouts as the maximum long term voltage.
2915 MHz at 1.513V - no errors after 36 min
2959 MHz at 1.616V - error after 11 min
2937 MHz at 1.52V - error after 13 min
I'm using a divider to get the memclock down at or below 200 to remove RAM from the equation to find the max CPU clock first.
Well, backing down just two MHz on the FSB and I was able to pull back almost a tenth of a volt - I think I've found the limit of this CPU at aroud 1.50-1.53 V.
Ok, I'll just start by saying this is my first thread, so please be kind. I've got a little experience, but I've never really had the time to do much fine tuning. Well, now I do so let the fun begin.
My idea about this machine is that it's to be a workstation/coding computer to last about 1-2 years before I replace it with whoever wins the next generation battle, with the main target app being a compile of a large Visual Studio build.
So which processor do I get? I've also got a 165 Opty that can do 2.65 GHz on 1.392V on a LanParty SLI-DR, so I was thinking about getting another one. Well, now that the supply is tightening up it's up to 330 or so OEM and 350 retail, I was kind of thinking that might be overkill for what I'm trying here. Also I knew from overclocking the 165 that one of the cores was substantially better than the other, so I knew that a single core good stepping would take me much higher than 2.65, and for about 2/3 of the price. So I went with the 148, as most of the others are slower and almost the same price or unavailable.
My Tools:
MSI K8N Neo2 Platinum work handme down check out the damage[/L]
Opteron 148 (11 multi)
2*512 OCZ 3200 Plat Rev 2
Zalman CNPS9500 LED
Raptor 73
The board will sustain me until my urge to buy the ASUS A8N32 Deluxe overwhelms me, or it's at least available.
I'll start out with a goal of one set of Prime95 SFF Torture tests as a stable platform as I move up/down the clock ladder, then I'll try to zero in on the maximum clock that can complete at least 8 hours of straight Prime95 torture. Given the approx lifetime of this CPU before I'll likely want to upgrade, I'll also set 1.55 or thereabouts as the maximum long term voltage.
2915 MHz at 1.513V - no errors after 36 min
2959 MHz at 1.616V - error after 11 min
2937 MHz at 1.52V - error after 13 min
I'm using a divider to get the memclock down at or below 200 to remove RAM from the equation to find the max CPU clock first.
Well, backing down just two MHz on the FSB and I was able to pull back almost a tenth of a volt - I think I've found the limit of this CPU at aroud 1.50-1.53 V.
Facebook
Twitter