Is My RAM Overclock Stable/Safe? Few Questions

Dear People of Tom's Hardware,

I have recently taken the time to overclock the RAM in my system (without changing any voltage settings, which is what I'll be doing from now on for all of my components) and I just have a few questions to ask.

My first question is about timings. Currently, I have 4 sticks of 4GB Corsair Vengeance Blue CAS 9 1600Mhz RAM overclocked to 2000Mhz running at CAS 10. My timings on the XMP were 9-9-9-24, my new timings with the OC are 10-11-11-26. I have a question regarding the TRAS setting. I am no expert in this field of overclocking, but I've done research and all of the articles I have read say that the TRAS should be about the sum of the other 3 timings. Is this true? Isn't it just safe to tighten ALL the timings as much as possible without errors?

Anyway, I have done 2 1/2 full passes in MemTest86 with no errors and 3 runs of Hyper PI without any error messages popping up. I have had no sign of slowdown or issues running games and everyday computing. I just wanted to know, if I have no errors, is my computer still safe? I've heard of people saying bad RAM timings and the like could lead to OS corruption, so I want to be sure.

Thanks, Davcoll
Reply to Davcoll
4 answers Last reply
More about ram overclock stable safe questions
  1. Memtest isn't really good for stressing, its great to determine if you have a bad module not so much for stressing. I would run Prime95 for at least an hour on blended if it holds up its pretty stable but if inclined do a 32M Hyper Pi since you have it.


    "Isn't it just safe to tighten ALL the timings as much as possible without errors?" Yes, but your milage with sub timings like tRAS won't give you much real world performance. Testing using GeeKbench or the likes to confirm actual improvements vs stock over the OC, and over say tRAS timings.
    Reply to JamesSneed
  2. It should be perfectly safe at the stock voltage, I am not sure how much father past 1.5v is safe for DDR3 so I wont comment on that.

    As for stability if it did two passes of memtest is should be fine, I have never used hyper pie, I usually use super pi 32 million, but if it is similar to super pi that is a good test for stability as well.

    I have heard the same thing for TRAS, it certainly doesn't make sense for DDR4 to add all the 3 numbers together, that would be way higher than what my XMP profile sets it too. So I would say no it does not have to equal the first three numbers, though I have read the same thing before. I usually just try to tighten the primary timings as much as possible like you said, then just leave the timings alone. I have played around with secondary timings before, it gets pretty confusing.
    Reply to Dunlop0078
  3. JamesSneed said:
    Memtest isn't really good for stressing, its great to determine if you have a bad module not so much for stressing. I would run Prime95 for at least an hour on blended if it holds up its pretty stable but if inclined do a 32M Hyper Pi since you have it.


    "Isn't it just safe to tighten ALL the timings as much as possible without errors?" Yes, but your milage with sub timings like tRAS won't give you much real world performance. Testing using GeeKbench or the likes to confirm actual improvements vs stock over the OC, and over say tRAS timings.


    Yeah, I mainly used MemTest for getting the best possible timings without errors. I also could boot into Windows @2133 Mhz RAM and relaxed timings, but Hyper Pi could still detect errors that MemTest couldn't. Backed it down to 2000 Mhz in BIOs then ran Pi 3 times again with 32M and 4 cores with no issues. After that, I did what I said above and that's where I'm at. I only used Cinebench to test my scores, and it was 12 more points than with my XMP overclock.

    Currently, I'm just using games and computing to test for stability. Prime pushes my computer far beyond my comfort zone. It's been a few days and everythings been pretty smooth.
    Reply to Davcoll
  4. Dunlop0078 said:
    It should be perfectly safe at the stock voltage, I am not sure how much father past 1.5v is safe for DDR3 so I wont comment on that.

    As for stability if it did two passes of memtest is should be fine, I have never used hyper pie, I usually use super pi 32 million, but if it is similar to super pi that is a good test for stability as well.

    I have heard the same thing for TRAS, it certainly doesn't make sense for DDR4 to add all the 3 numbers together, that would be way higher than what my XMP profile sets it too. So I would say no it does not have to equal the first three numbers, though I have read the same thing before. I usually just try to tighten the primary timings as much as possible like you said, then just leave the timings alone. I have played around with secondary timings before, it gets pretty confusing.


    Do you know if there is a significant increase in temperature for RAM when overclocking? I didn't change the voltages (still 1.5v) it's just that this is something I don't hear talked about a lot. Other than that, good to know that's not always the case. I always thought you could just tighten them as much as possible (as long as you have no errors/instability that is). I suggest taking a look at Hyper Pi if you have a dual/quad core CPU, it's great for testing memory if you make it run at 32M w/ x amount of CPU cores you have. Not only can it detect errors super quick, but I think it is more optimized than than the latest version of Super Pi out at the minute. (not 100% sure though).
    Reply to Davcoll
Ask a new question Answer

Read More

Overclocking RAM Tom's Hardware