VCCIO & VCCSA Voltage in [Auto]

[Braegnok]

Hello, all my temps are fine, and I'm running CPU in default settings except for my memory.

On memory I have it set to manual , running @ 4600MHz 1.5V 19-26-26-46 and system is prime95 stable.

My CPU VCCIO is set to Auto and stays @ 1.312V

And CPU VCCSA is also set to Auto and stays @ 1.360V

Do these voltages seem OK, or should I manually set them a bit lower. Perhaps 1.225V on both and see if system remains stable, or just not worry about it.

Imjur link screenshot. https://imgur.com/a/xXCsDOB

Thanks in advance for any advice.

[Arne Saknussemm]

I would try and rein those in apart from anything for the extra heat

[hortons71]

Hello, all my temps are fine, and I'm running CPU in default settings except for my memory.

On memory I have it set to manual , running @ 4600MHz 1.5V 19-26-26-46 and system is prime95 stable.

My CPU VCCIO is set to Auto and stays @ 1.312V

And CPU VCCSA is also set to Auto and stays @ 1.360V

Do these voltages seem OK, or should I manually set them a bit lower. Perhaps 1.225V on both and see if system remains stable, or just not worry about it.

Imjur link screenshot. https://imgur.com/a/xXCsDOB

Thanks in advance for any advice.

My G-skill 4600 runs well with default VCCIO and VCCSA with default xmp2 profile, no extra voltage needed except 1.45v for DRAM. But the performance is very poor. It's more of a bandwidth problem, cpu has trouble keeping up when you trim prime / sec / tert timings. I usually aim for Read and Write performance at 92-94% of theoretical max at the given clock, and Copy not far behind without losing too much latency.

Every CPU behaves differently. VCCIO increases slightly at load and if it becomes too high I get Blue Screen or system lock. Too much VCCSA and Pcie start to be affected, freezes in games and sometimes Aura software can be affected and go from solid red to rainbow. So the highest I can go is 1.3v on both in Bios.
I have my RTL settings in auto, and when all is well they show 71 71 the same, and the PC always starts with the first try. But when you get too close to the edge they start to separate 1-2. This is a good indication of whether you are on the right track or have gone too far. You usually can't take someone else's settings and hope it works for you, you need to calculate some and then fine tune the rest with trial and error.

You should also keep in mind that high North Bridge clock can affect your RTL settings.

[ninezerofive]

It's always best to manually adjust VCCIO and VCCSA. It's a fine tuning process. I had about the same settings on Auto, but could lower it to 1.1~ both.
You are running them aggressively, try to set one of them to 1.25 manually, run stress test, if it passes, do the same with other one. Rinse and repeat.
Will help out with temperatures as well.

[Braegnok]

Thanks everyone for the advice! Going to try and lower both VCCIO, VCCSA down a bit.

[bass junkie xl]

i heard and see people say that vccio /vcssa over 1.3 v - 1.35 v is a no no for 24/7 1.4 vccio / vcssa can kil la cpu in a month or so

im running 32 gb team group xtreem 4133 c18 @ 4200 17 17 17 32 vccio @ 1.18 v / vcssa @ 12.v in bios witch dropped my 9900 ks temps a bit to

i wouldnt go for more then 1.26v for both 1.3+ for benches

[Braegnok]

Thanks vary much for the helpful advice everyone.

I was able to lower VCCSA from 1.312V down to 1.18125V

And lower VCCIO from 1.360V down to 1.19375V

Using AIDA64 System Stability Test and then running 3D Mark Time Spy,.. Everything is solid at the new manual set voltages.

[XYchromosone]

VCCSA @ 1.36v is high... definitely go manual and use lower voltages.

Update: I didn't realize you are running Team Group Xtreem 4133 c18. This runs at a higher voltage than 1.35v when OCing. I've seen the specs and it runs at 1.4v @ 4133MHz

[bass junkie xl]

i would run hci mem test for 1000 % coervage time spy and adia 64 are walk in the parks .

[Braegnok]

i would run hci mem test for 1000 % coervage time spy and adia 64 are walk in the parks .

Thanks for the advice bass junkie, good information.

Whenever I adjust my memory I always boot into USB flash drive with Memtest86 Passmark Ver. as it's the only program I'v found that doesn't require a installed OS. I like it because I don't want to worry about corrupting Windows, and it is much quicker to do stability tests of new timings from booting to a flash drive. Usually will give errors within 1-3 minutes, make more adjustments and do it again.

Once I pass Memtest86, than I move into Windows based software AIDA64, and then Time Spy to see if my scores improved.