Trouble pushing an Intel G2358 to a stable 4.6Ghz

[Fisher]

Mobo: MSI Z97 Gaming 3

RAM: Kingston HyperX DDR3-1866 8GB (2x4GB)

CPU: Intel G3258 Anniversary

Cooling: Noctua D14

 

I seem to get a stable 4.5Ghz with the following BIOS settings:

 

CPU Voltage

1.265

CPU Frequency

4500

CPU Multiplier

45

CPU Ring Ratio

41

CPU Ring Voltage

1.2

System Agent Voltage

+ .160

I/O Analog Voltage

+ .160

I/O Digital Voltage

+ .160

BLCK

100

 

The temperature stays very reasonable, not even cracking 70.  Ring multiplier cannot go any higher and still stay stable @ 1.2v.

 

When trying to push the multiplier to 46, I run into instability issues.

 

The following settings are unstable:

CPU Voltage

1.325

CPU Frequency

4600

CPU Multiplier

46

CPU Ring Ratio

41

CPU Ring Voltage

1.2

System Agent Voltage

+ .160

I/O Analog Voltage

+ .160

I/O Digital Voltage

+ .160

BLCK

100

And:

CPU Voltage

1.325

CPU Frequency

4600

CPU Multiplier

46

CPU Ring Ratio

41

CPU Ring Voltage

1.2

System Agent Voltage

+ .180

I/O Analog Voltage

+ .180

I/O Digital Voltage

+ .180

BLCK

100

 

I stepped up core voltage incrementally from 1.265 to 1.325 with no success, then dropped it back to 1.265, increased SA/IOA/IOD voltages incrementally and began stepping up core voltage once more until I hit 1.325v core.  Then back down to 1.265, rinse/repeat all the way to SA/IOA/IOD offset of +.180v

 

I feel like I shouldn't have to push the core voltage even as high as 1.325 to achieve a stable 4.6Ghz.  Now I'm starting to wonder if I've missed something.  Temperatures stay below 76 degrees throughout all of this.  Does the community have any suggestions?  Please note that I am a first time overclocker and have only read guides particular to my model of CPU.

 

[ccokeman]

Try dropping the Ring ratio down one more step and retesting. I find that one of my 4770K chips will do another 100Mhz higher using less vcore with the same memory speed when I drop the ring ratio. Its a very fine line that you have to brute force it with voltage or bite the bullet and drop the cache ratio.

[El_Capitan]

Also, even though your memory is rated at 1866Mhz (not sure what the timings are), but set it to 1600MHz and loosen your timings. If they're 9-9-9-24-2T, set them to 10-10-10-30-2T. After Ivy Bridge, "CPU clock is king". You'll hear that term being used quite a bit.

[ccokeman]

Also, even though your memory is rated at 1866Mhz (not sure what the timings are), but set it to 1600MHz and loosen your timings. If they're 9-9-9-24-2T, set them to 10-10-10-30-2T. After Ivy Bridge, "CPU clock is king". You'll hear that term being used quite a bit.

 

Clock speed is king but cache ratio helps  

[El_Capitan]

 

Also, even though your memory is rated at 1866Mhz (not sure what the timings are), but set it to 1600MHz and loosen your timings. If they're 9-9-9-24-2T, set them to 10-10-10-30-2T. After Ivy Bridge, "CPU clock is king". You'll hear that term being used quite a bit.

 

Clock speed is king but cache ratio helps  

 

Yes it does, and so does making sure the memory isn't a factor limiting his CPU overclock.

[MedievalNerd]

From what I understand when pushing your chip to it's limits, the ring & core clock will start a tug of war. You should ideally reach your target Clock OC first, and then work your way up with your ring ratio.

 

Not sure how relevant this is to that CPU, but if you have more advanced settings like Transmitter CLK-De-Emphasis and things like that, you might need to tweak those to try and push the ring to it's limit.

 

For me the jump from 4.5 to 4.6 takes a whole 0.6 extra voltage. From 1.312 to 1.376 more precisely.

Edited November 14, 2015 by MedievalNerd

[jdm_freek]

im at 4.8 but with 1.32v