A Show for Computer Repair Techs by Computer Repair Techs
Jeff Halash from TechNutPC.com Talks to Computer Technicians
Google+ Jeffery Halash
Try Instant Housecall Today for Free/Offer Code: Podnutz
Finding your max overclock
I would like to share my process for isolating a stable overclock. In overclocking forums this is a hotly debated process. Online you will hear about various stability testing methods ranging from 2hrs to 48hr testing. Below is my process for finding a max overclock and testing it. This is not an actual cpu specific overclock guide, as different CPU’s can benefit from certain settings, but this will hopefully help you feel good about the overclocking process. Feel free to share how you do it as well.
First of all, if safe temperatures and voltages are used along with a good motherboard and power supply; Overclocking can be a safe and reliable way to get sometimes over 20% performance gains. I accept no responsibility for any damages that may occur while you overclock your system, do it at your own risk.
Before any overclocking is done, I do a full system benchmark including passmark v8 (entire system), asus realbench (cpu and gpu), cinebench (cpu and gpu), crystaldiskmark (for storage) and unigen valley (for gpu). I make sure I save results as screenshots or you can write them down.
I first determine what I want, MAX performance or Silent operation and choose a max cpu voltage that reflects those goals. This greatly depends on your ambient room temperatures, cpu and cooler being used. I then enter the bios and set the voltage for the cpu (some cpu’s might benefit from other tweeks besides cpu voltage) and a cpu overclock setting that I think the cpu can hit with that said voltage.
Example: With a intel i7 5820k, I may decide that 1.27v is my max voltage while generally 1.30v is a safe 24/7 voltage. I then go in the bios to set voltages and my settings for 4.5ghz; then try to boot into windows.
If it boots into windows, I then load up cpu-z and make sure voltage and cpu speeds are reading what I set. I also fire up MSI afterburner to watch cpu temps and cpu load across every core/thread. I then fire up Cinebench r15 and make 2 passes. If both finish I go back into the bios and raise it 100mhz. Then I repeat. If it fails; usually a BSOD or low score, I then go into the bios and lower it by 100mhz and repeat. Generally what I find is that when Cinebench fails, my end result if I do not change any voltage, will be 200mhz lower….usually.
Once two passes are successful in cinebench; I then do a benchmark run on Asus realbench. Once that passes I do the Asus realbench Stress test with max memory installed selected for the test and run it for 1hr. If that passes, I lower the memory to 50% of what is installed and set the benchmark to run 8hrs. If that passes I validate the cpu and being stable for general use. Depending on the customer’s use, I have also run Aida64 for 1-2hrs to backup the results from realbench stress test.
Once all stress testing is passed, I rerun the benchmarks to confirm the performance gains
CPU-Z (system info and watching CPU clock and voltage)
Unigen Valley (benchmark the video card)
Crystaldiskmark (Storage speed benchmark) (crystaldiskinfo has some smart data info inside)
Passmark v8 ( I use a paid version but free versions work, limited in save features)
Asus Realbench (works on all systems, has benchmark and built in stress test, not synthetic, real workloads which make it a great program)
Maxon Cinebench (CPU benchmark and GPU but primarily used for CPU)
MSI Afterburner (data logging for work load and temps for CPU and GPU)
This is the thermal printer I use for my QR codes, the included software makes the QR codes for me
Read more here:: Podnutz Daily #473 – Anthony Remis from Remis Computer Solutions