Z390-I VCCIO and System Agent...What do you guys set your voltages at?

[tml4191]

Hardware:
- Z390-I motherboard
- (16gb x 2) 32gb Corsair Vengeance RGB 3000Mhz C15 (CMR32GX4M2C3000C15)
- 9900K CPU

I put VCCIO and System Agent voltages to Auto, which set VCCIO to 1.328 V and System Agent to 1.232 V.

I asked around on YouTube and Reddit, and nearly everyone said I could lower it.

So I lowered the VCCIO to 1.232 V in bios, which was the same as the System Agent voltage, and it jumped to 1.264 V, according to HWiNFO64 and HWMonitor.

Corsair also mentioned to leave it on auto for stability reasons, but I have yet to run into any problems at 1.264 V, even after succesfully passing OCCT medium packet, AIDA64 and MEMTest64.

How far should I lower the VCCIO and System Agent before calling it a day? I just don't want to have to CMOS reset. Some people even said go as low as 1.1xx.

What do you guys run your VCCIO and System Agent voltages at?

[eclypse]

I might have to check out that mem testing program.

My 9900k seems to want alot of vccio voltage. 1.34v. Basically I need to do xmp and run it high or I get blue screens and I'll get kicked to the desktop in bf5.

I have 4x 8gb g.skill trident z rgb 4266mhz.

Mb is a new asus extreme XI.

That's at 5.2ghz using 1.36v in the bios LLC 7 and 1.34v on load.

I've also tried raising the dram voltage a hair at 4.1v in the bios and under load will be like 1.43v.

That seemed to of fixed the issue but sometime durring reboot or cold boot I'll get a memory problem and the bios will force me to go into the bios to set the ram again.

So no idea.. bios issue? Cold boot mem voltage issue?

If I leave vccio at auto it's at 1.4+v in the bios.

I've tried losing the vccio voltage to around 1.31 and the blue screens will happen.

[Silent_Scone]

BF5 uses AVX instruction sets, so is fairly good at exposing instability as far as gaming workloads are concerned. It depends what the stop code you are receiving as to where the instability is coming from.

[Kelutrel]

I have a 9700K OC@5GHz and a Maximus XI Hero with BIOS 0903, 16Gb of DDR4 3200Mhz at 1.35v, and I have set my VCCIO and VCCSA at 1.1v without any problem. If I choose AUTO, the BIOS sets those values in the 1.3-1.4v range. I prefer to stay at 1.1v

[eclypse]

Silent Scone@ASUS wrote:
BF5 uses AVX instruction sets, so is fairly good at exposing instability as far as gaming workloads are concerned. It depends what the stop code you are receiving as to where the instability is coming from.

Did not know that.. cpu dont get hot durring bf5. When it would error it would just kick me to the desktop. No message or anything just closed the game. Been a long while since that's happend and feel it was related to dram voltage.. maybe not supplying enough ram voltage or the new pair of dimms I added need more voltage at there rated speed of 4266 then the original set.. who knows.

[Silent_Scone]

Doesn't necessarily need to be accounted for with heat, but the CPU is possibly unstable when running lighter AVX loads such as this. As I say, would need to know what the stop code is to know whether it relates to the memory or CPU.

4266 can require some manual tweaking, too.

[eclypse]

Silent Scone@ASUS wrote:
Doesn't necessarily need to be accounted for with heat, but the CPU is possibly unstable when running lighter AVX loads such as this. As I say, would need to know what the stop code is to know whether it relates to the memory or CPU.

4266 can require some manual tweaking, too.

Be tuff with no code.. no blue screen.

Going to go play around with it some more.

[Silent_Scone]

eclypse wrote:
Be tuff with no code.. no blue screen.

Going to go play around with it some more.

Sounds like it might be memory, then.

[Mappi75]

What i recommend:

use different test programs:

hci memtest pro and karhu RAM Test

normaly i would only use hci, because it founds every time errors and karhu runs12+ hours stable.

Now i tested a 64gb 3733 cl17 Kit and hci runs 24 hours.

With Karhu it shows me in a short time an error with "cpu cache enabled & stress FPU.