How to determine an upper limit for CPU and memory
Given a CPU and memory modules, a common question is how to determine an upper limit for them so as to guide the overclocking and fine tunning process. By just booting into Windows or OS, and running all kinds of testing and benchmarking programs such as Sandra, Everest, SuperPI, 3DMark01/03, Prime95, ..., if the runs fail, it may well be due to heat stress of the CPU, OS, drivers, chipset, bus contention, memory bandwidth and latency limitation, ... and hence the full potential or some better upper bound for the CPU and memory cannot be determined.
Experiment
Using the DFI LP UT Ultra-D motherboard, Winchester 3000+, the G. Skill PC4400 Le, memtest86 was used to determine an upper limit of the CPU and memory. Results are shown.
For each memtest run, the CPU frequency, memory frequency, memory timing, CPU and memory voltages are set in the bios. Then run memtest to check for pass and failure. Depending on pass or failure, frequency(ies), timing, voltage(s), particular setting(s) are adjusted accordingly to improve the objectives.
By doing these steps and gradually adjusting frequencies, voltages, timing, settings, ..., a least upper limit for the CPU and memory was found.
It was able to boot into memtest86 at 3006 GHz
, but failed.
But it passed memtest86 at
- 2970 MHz for the Winchester 3000+ at 1.62 V
- 330 MHz for the G. Skill 4400 LE at 2.8 V, 3-5-5-10 1T
Also memory is not yet limited, whereas the CPU is, I determined.
By doing the above test, the previous results (listed in the first page of this thread) were improved.
This is an upper bound, it does not mean in Windows, all kinds of programs will pass and be able to run at that level of frequency and voltage, but it gives a guideline for overclocking and testing for actual programs and benchmarks.
E.g.
Memtest86 boot at 3006 MHz 1.65 V, memory 334 MHz 3-5-5-10 1T 2.8 V
Memtest86 pass at 2970 MHz 1.62 V, memory 330 MHz 3-5-5-10 1T 2.8 V
Windows XP boot at 2.95 GHz 1.60 V
Sandra CPU run at 2.94 GHz 1.60 V
SuperPI run at 2.90 GHz 1.60 V, memory 3-5-5-10 1T 2.8 V
3DMark01 run at 2.85 GHz 1.55 V, memory 317 MHz 2.5-4-4-8 1T 2.8 V, SuperPI 1M 30 sec
Prime95 run stably at 2.73 GHz 1.50-1.52 V, memory 303 MHz 2.5-3-3-7 1T 2.8 V, 23 C idle, 38 C load (11+ hours, user aborted)
...
Rule of thumbs, Prime95 is usually around 200-300 MHz below post/bios/memtest boot, 200 MHz below Windows boot.
Screen shots are shown in the next few posts.
