F5F Stay Refreshed Power Users Overclocking Needs assistance: memory performance is limited for overclocking

Needs assistance: memory performance is limited for overclocking

Needs assistance: memory performance is limited for overclocking

Pages (2): Previous 1 2
D
Darkcell
Junior Member
12
07-28-2017, 10:56 AM
#11
Heh, I'm not really keen on either option but I can't adjust the timing or boost the frequency.
You mentioned you could reach 1866 by relaxing the timing settings. Do you prefer 1866 with more relaxed settings (perhaps something like 10-10-10-30), or would you rather stick to tighter settings at 1600? You'll need to decide.
Also, keep the Command Rate at 2; a lower rate might limit RAM overclocking.
D
Darkcell
07-28-2017, 10:56 AM #11

Heh, I'm not really keen on either option but I can't adjust the timing or boost the frequency.
You mentioned you could reach 1866 by relaxing the timing settings. Do you prefer 1866 with more relaxed settings (perhaps something like 10-10-10-30), or would you rather stick to tighter settings at 1600? You'll need to decide.
Also, keep the Command Rate at 2; a lower rate might limit RAM overclocking.

G
152
07-29-2017, 11:12 PM
#12
I've seen some solid performance with two different memory sets for Tradesman. I haven't heard much about the limits you mentioned. Micron's memory quality is quite good.
Micron is suitable enough for Intel to form a limited partnership with them. It doesn't mean they overclock, but it does indicate they're competent in their field.
G
GlennTheMaster
07-29-2017, 11:12 PM #12

I've seen some solid performance with two different memory sets for Tradesman. I haven't heard much about the limits you mentioned. Micron's memory quality is quite good.
Micron is suitable enough for Intel to form a limited partnership with them. It doesn't mean they overclock, but it does indicate they're competent in their field.

_
_BadoTommeh_
Member
50
07-30-2017, 05:33 AM
#13
Leaps-from-Shadows :
frenchy70 :
I'm not really keen on either option, but I can't adjust the timing or boost the frequency. 😕
You mentioned you could reach 1866 by relaxing the timing. Do you prefer 1866 with a looser setup (like 10-10-10-30) or the tighter 1600? You need to choose.
Also, make sure your Command Rate is at 2. A rate of 1 might limit RAM overclocking.
Yeah, I started at 1866 with a 9-10-9-30 command rate, but it failed HyperPi. I didn’t try lowering it to CL10 because I thought it wouldn’t help much (9/1600 is roughly 10/1866).
I also didn’t realize how crucial the command rate was for getting an overclock—would running at 2 be slower than at 1?
_
_BadoTommeh_
07-30-2017, 05:33 AM #13

Leaps-from-Shadows :
frenchy70 :
I'm not really keen on either option, but I can't adjust the timing or boost the frequency. 😕
You mentioned you could reach 1866 by relaxing the timing. Do you prefer 1866 with a looser setup (like 10-10-10-30) or the tighter 1600? You need to choose.
Also, make sure your Command Rate is at 2. A rate of 1 might limit RAM overclocking.
Yeah, I started at 1866 with a 9-10-9-30 command rate, but it failed HyperPi. I didn’t try lowering it to CL10 because I thought it wouldn’t help much (9/1600 is roughly 10/1866).
I also didn’t realize how crucial the command rate was for getting an overclock—would running at 2 be slower than at 1?

T
taconiebre
Senior Member
506
07-30-2017, 09:52 AM
#14
Try the CR at 2 and observe the result.
T
taconiebre
07-30-2017, 09:52 AM #14

Try the CR at 2 and observe the result.

F
64
07-30-2017, 04:42 PM
#15
Success !! It turned out I just needed to adjust the timings through trial and error. A big thank you to everyone who replied, as it inspired me to take a few extra steps that eventually led to a stable memory overclock.

I began by researching the settings for Digital and Analog IO voltage, as well as Sys Agent voltage—since in my BIOS I could only modify offsets, which wasn’t very useful when I read about setting them to 1.0, 1.15 or 1.25. I couldn’t even see what voltages were stock because even the CPUID HW Monitor only displayed the Sys Agent offset value.

Next, I raised the Dram volts to 1.65 and adjusted the three offset voltages accordingly. I set the frequency to 2000 and left timings on auto; it booted without issues, showing speeds of 11-13-13-35 2T.

Then I manually tweaked the timings to 10-11-11-32 1T, which also worked fine. I experimented with lowering everything, but most attempts failed to get a boot. When I could boot, HyperPi at 32MHz would crash instantly (though I didn’t try reducing CL to 9). I even increased the Sys Agent voltage to +0.25V without success.

Returning to 10-11-11-32 1T and testing with HyperPi worked perfectly. I decided to reset all offsets by switching them back to auto, and setting Dram volts to auto (1.5V) succeeded again—so it was really about finding the right timing!

I continued this process while keeping the CPU overclock at 4.3 GHz with adaptive voltages capped at 1.2V. Although I was eager to push for higher speeds and tighter timings, I realized that with the extra headroom in voltage and temperature, a decent overclock was enough.

Thanks again.
F
FuturzLeCanape
07-30-2017, 04:42 PM #15

Success !! It turned out I just needed to adjust the timings through trial and error. A big thank you to everyone who replied, as it inspired me to take a few extra steps that eventually led to a stable memory overclock.

I began by researching the settings for Digital and Analog IO voltage, as well as Sys Agent voltage—since in my BIOS I could only modify offsets, which wasn’t very useful when I read about setting them to 1.0, 1.15 or 1.25. I couldn’t even see what voltages were stock because even the CPUID HW Monitor only displayed the Sys Agent offset value.

Next, I raised the Dram volts to 1.65 and adjusted the three offset voltages accordingly. I set the frequency to 2000 and left timings on auto; it booted without issues, showing speeds of 11-13-13-35 2T.

Then I manually tweaked the timings to 10-11-11-32 1T, which also worked fine. I experimented with lowering everything, but most attempts failed to get a boot. When I could boot, HyperPi at 32MHz would crash instantly (though I didn’t try reducing CL to 9). I even increased the Sys Agent voltage to +0.25V without success.

Returning to 10-11-11-32 1T and testing with HyperPi worked perfectly. I decided to reset all offsets by switching them back to auto, and setting Dram volts to auto (1.5V) succeeded again—so it was really about finding the right timing!

I continued this process while keeping the CPU overclock at 4.3 GHz with adaptive voltages capped at 1.2V. Although I was eager to push for higher speeds and tighter timings, I realized that with the extra headroom in voltage and temperature, a decent overclock was enough.

Thanks again.

Pages (2): Previous 1 2