Did I just gimp my ram?

[Runninwild]

Hey guys, I recently put together a system and have begun overclocking it. My cpu is a bit of a dud (i5 3570k) which maxes out at 4.4ghz at 1.32V aida64 11.5 hours stable. My r9 290 runs 1107/1400 +25mv folding@home stable.

Before I got my cpu cooler and video card I lightly played with my ram 2x8gb corsair vengeance 1600mhz and Ive had it running at 2000mhz 9-9-9-24 at 1.5v for the past few weeks completely stable. Anyways I began pushing the ram to test its limits today, first I set the voltage to 1.6v, then tried 2200mhz and passed 3 passes of memtest86, then I tried 2400mhz and it wouldnt post, I loosened the timing to 10-10-10-26 and I was able to run 1 pass of memtest86 without errors so I tried 2600mhz, and also got it to run 1 pass. I then just decided to try and test its limits at these timings and I tried 3000mhz where it wouldnt post. So I tried upping the voltage to 1.65v and it still didnt post. So I tried 2800mhz with no post, then 2666mhz with no post. I then dropped the voltage back down to 1.6v and I was going to do further testing at 2600mhz and to my surprise it wouldnt post, so I tried 2400mhz with no post. I then dropped the timing back to 9-9-9-24, the voltage to 1.5v and set it to 2000mhz and it still wouldnt post. At stock voltage and stock timings it will boot at 1600mhz. Did I just gimp my ram? I was surprised how well my sticks were doing as most reviews I read on the set couldnt get it to run much above 2000mhz with loose timings and high voltage. I thought 1.65V was fairly safe for short term testing? Anyone got any ideas? Most of my voltage and bios settings are on auto, is it possible to the motherboard has changed one of the settings on its own which is causing the instabilities?

 

[Runninwild]

Well after a bit of research it looks like I damaged the memory controller in my cpu. Turns out you're not really supposed to go above 1.5v with ivy bridge. Whoops which I had of known that before I started playing with it. I guess I'm lucky it's still capable of running stock 1600mhz without any immediate errors in memtest. Going to try 1866mhz and hopefully it can take it!

 

[EarthDog]

1.65V wwon't kill that ram...no. PLENTY of people have run 1.65V on Ivybridge with no illl affects. I would restart from sratch and try again. Perhaps you corrupted something in the OS... it happens at times with memory overclocking...

 

[Runninwild]

I've been using memtest86 from a boot disk. I reset the bios and I cant get it to boot with anything above 1800mhz anymore and memtest finds errors at that speed. I'm pretty sure I damaged the memory controller. The good news it it did 4 passes over 4 hours at 1600mhz with stock timing and didn't find any errors.

I have a g1620 I might throw in the system and play with a bit. Are the memory multiplers unlocked on those cpu's as well or only on k series chips? I'm going to put my cpu back to 4.4fix with the ram at 1600mhz and run aida64 while I'm at work. If passed and the ram is able to clock higher with the Pentium cpu I might just sell my 4750k and find another one.

 

[Runninwild]

I tried with my g1620, I could get it to post at 1800mhz but not at 2166 which is the next available option on that cpu even with 1.6v. I cant tell if its my memory or the memory controller at this point but something is definitely damaged. My cpu still stress tests ok at 4.4ghz (13 hours of aida64 extreme today) so I'm kind of thinking its the ram. I never really thought you could damage ddr3 with such little voltage for such a short period of time but regardless my memory will no longer overclock or the memory controller will no longer support anything above a 1600mhz speed without errors

 

[EarthDog]

If you switched CPUs, its not the memory controller.

 

[Runninwild]

Do some memory controllers limit the speed the ram can run? Like the g1620 is a pretty budget chip, is it even capable of supporting 2000mhz ram?

 

[wingman99]

The g1620 is a budget chip however when they make chips from a single wafer some chips are hi-grade like i5 3570k some test low grade and some sections don't work on the chip like cores and it becomes a g1620.

 

[Runninwild]

So the memory controller shouldnt be a limiting factor in overclocking the ram. Basically I damaged my ram with only 1.65V? I thought that was generally considered a safe 24/7 voltage for ddr3? Could trying to boot it at too high of a speed damage it? I was always under the impression if it couldnt sustain the speed it just wouldnt post or would be unstable, I didnt think it could cause damage. I always thought damage came from too much heat and voltage.

Finished stability testing the system 3.1ghz at stock voltage, 1600mhz on the ram at stock timings and voltage, 1107/1401 on the gpu with +25mv. My cpu sucks at overclocking, 4.4ghz needs over 1.33V. Ram did 14 hours of memtest86 without errors at stock speeds, 1800mhz brings errors almost instantly now I guess i got too greedy with my high performing ram! I should have just stuck with 2600mhz at 1.6v

 

[wingman99]

You just had bad luck with your memory, the memory controller will work usually at DDR3 2400 speed, if you acquired a new kit you probably would not have the same problem over again it might not overclock well however. Your CPU is not to bad mine needs 1.33v Vcore for 4.5GHz.

 

[Runninwild]

1.33v gets me 4.4 aida64 stable, prime fails overnight and burntest either fails or crashes the system within minutes. Also temps hover around 100c with my hyper212 evo at that speed. I really wasn't expecting this cooler to be so loud, it's by far the loudest fan in my case, including my r9 290 after I made my own fan curve.

At 4.1ghz and prime95 running with my gpu running folding overnight my cpu maxed around 100c, gpu maxed 21st 89c the system was definitely toasty, especially after I wired my case fans to 6v but nothing I run stresses it nearly that much. Playing battlefield 4 it doesn't even get warm enough to get the fans going loud enough for it to be noticeably loud. Something I couldn't say before I set my own fan speeds on the gpu.

By stock voltage I should say auto, I've seen it climb as high as 2.4v under stress tests at 3.1ghz. Manually setting it to 2.25v and its not very stable at all.

 

[wingman99]

With my clock at 4.5GHz at a vcore of 1.33v my peak temperature running prime95 is only 84c with my hyper 212. Do you have good case cooling? Remove your side cover and see what you have for peak temperature under load.

 

[Runninwild]

I'm on ivy bridge which runs hotter then modern i5's. I've reseated the hsf a couple times as well it's just how my chip runs, at stock voltage and speed it barely gets above 60c. Airflow in the case isn't really a factor, it takes Intel burn test seconds to get the temp that high. While gaming it barely gets above 75c. I'm ok with the temperatures, it is what it is. The chip won't even boot at 4.6ghz regardless and I think I've tried as high as 1.4v. I'm more chip limited then cooling I believe.

I wired my case fans to run on 6v instead of 12 so airflow isn't optimal but it's still more then sufficient. I have 5 120mm fans and even with them at full speed it didn't really help my oc or the temperatures a huge amount. 4.1ghz at stock voltage and almost silent verse 4.3ghz and super loud with 1.35v and hot temps is an easy choice.

 

[wingman99]

I just upgraded from i5 2500k using the same cooler hyper 212 and it ran cooler with that processor 79c at 4.5GHz 1.34v. They used paste on the ivy bridge integrated heat spreader your chip should run at 84c, have your tried running with the side cover off to see? However your temperature is fine you are chip limited not cooling.

 

[Runninwild]

Maybe I have a bad application of thermal paste under the heat spreader. I thought about delidding but for such a poor chip I don't think it's worthwhile.

 

[wingman99]

I would not worry a bad application of thermal paste is only 1-2c difference. I would not do any delidding it is not worth it.