Question CPU/Voltage and related topics
Question CPU/Voltage and related topics
You should check for any indicators or warning signs that might suggest you need to adjust other voltages beyond your main CPU voltage. There are specific symptoms you can look out for. Since your DRAM is set to auto and you’re considering leaving it unchanged, you might want to verify stability and performance. The XMP profile suggests a voltage of 1.67, while the advertised max is 1.65—this discrepancy could be a clue. You may want to lower the voltage slightly or manually tweak settings outside the XMP profile to ensure optimal operation. Adjusting other voltages will help maintain balance and prevent overheating, which could lead to issues. If you can identify the right settings, you’ll have a clear record of your configuration.
I've tested many DDR3 (Intel) motherboards at up to 1.65V for years without any problems. I recommend setting it manually to 1.65V, which works well for 2133MHz. Right now I'm running systems at 2400MHz, 2133MHz/2133MHz/MHz with just 1.65V and no further adjustments are needed.
However, Intel is not AMD, so I don't know much about your memory controller capabilities. What voltage does the XMP profile set for 2133MHz, which could be a problem?
Also, wow, you have a 1500W PSU—at least you won't run out of power for overclocking. 😄
Sorry it took a while to get back here. Your AIO is performing well and there are no stability problems.
My HDD raid array crashed recently after the major update, but I don’t think the update caused it.
I’m not sure what those voltages mean or how to read them on my MOBO. Can you clarify what they are and their purpose?
My temperatures are low and the system runs smoothly at 4.5Ghz with a voltage of 1.45V. The advertised boost is 4.3, but I’ve reached 4.9. I’m worried about damaging something expensive and want to bring it back up safely without risking further issues.
You might notice "soc voltage" instead of VTT with AMD. Unfortunately, I don't have any guidance on overclocking AMD processors. For DDR3, a voltage range of 1.65-1.67V is generally safe. More details about AMD FX series IMC quality and the voltage it can manage would be helpful.
Ddr3 can go upto 1.8v.
The 1.65v ram was original release ram, right after having used DDR2. It wasn't until somewhat later that DDR3 refinements allowed the voltages to be dropped to 1.5v and for a few models, as low as 1.35v (not including DDR3L). There's no difference between amd and Intel for these generations, you'd have to go back several years, back to when Intel ran low-density ram, and amd ran either low or high density ram, back in the days of DDR and pre lga775.
Motherboards are different. As are cpus. Most times xmp settings worked perfectly stable, but xmp IS considered an overclock, so there is no guarantee it will be stable, just that it's attainable. This can mean you may need to bump dram voltages, nb voltages, system agent voltages or any combination, especially if the cpu itself is overclocked.
XMP on my lga1155 worked fine at 1.5v to get the rated 1866MHz. To get stable ram with a cpu locked core OC of 4.6GHz (i7-3770K with 3.8GHz turbo/3.4GHz base) I needed to manually bump dram voltage to 1.55v.
A bump from 1.65v to 1.67v isn't an issue. AMD memory controllers weren't as strong as the Intel versions.
With an FX cpu, you should not be concerned with temps, the cores do Not contain thermal strips, so temp readings as such cannot exist. There's only 2 programs qualified to give temp advice, and thats AMD Overdrive and Coretemp, set for Thermal Margins.
Because there's no accurate way to read temps, Thermal Margins are used instead. These are a amount of thermal headroom left in the cpu. It's a complex algorithm comprised of loads, voltages, usage, core usage, bandwidth usage and a bunch of other readings, all rolled up into a number. With that, the number itself is not important, it's what the number represents that is. A TM of 40 doesn't mean you have 40°C left, it means you have @ 40°C of workload left. The higher the workload, the faster that number gets smaller. 40's is idle or low usage, 30's is light usage, 20's is gaming usage, 10's is heavy usage and under 10 is pushing your luck.
It's like when ppl count from 10 to zero. They never actually stick to a perfect second, you always get to 3, 2½, 2, 1½, 1, ½, ¼, ⅛... and it takes another 15 seconds or more just to count down from 3. Thermal margins work the same.
The core temp for many same gen Intels was around 100°C. For the FX, it's 62°C. (as measured from jury rigged engineering samples by amd techs/engineers). But there's no way for a consumer to guage that temp. So you'll often see ppl with idle temps of 8-18°C, and load temps pushing 70+°C, non of which are accurate. Or in some cases, even physically possible unless they live in Alaska or Siberia and the pc was sitting next to an open window in the winter time.
Using thermal margins will tell you if you are pushing any OC too high. As long as you stay within safe voltage boundaries for that cpu/motherboard, I've seen ppl push FX upto 5.4GHz, But that was on a FX 4100 with a 280mm AIO, so cooling wasn't an issue.