Vccio and sa voltages on 3000mhz ram

[Leoplate25]

Hi... I have an i5 9600k at stock and i am not OCing the cpu for the moment. Ram is Corsair Vengeance White led at 3000mhz xmp enabled but vccio and sa are way too high, 1.312 and 1.248v. Can i set these voltages to 1 or 1.1? I have an Asus maximus hero xi wifi. Thanks.

Rest of the system has a corsair h100i v2, a bitfenix whisper m 750w and a evga gtx 1070 ti sc.

 

[kastriot]

Read this:

https://www.reddit.com/r/overclocking/comments/ac0ybb

 

[TheMadDutchDude]

Yeah, that is very high. Scale back to 1.1v and you’ll be fine.

 

[Leoplate25]

Yeah, that is very high. Scale back to 1.1v and you’ll be fine.

Hi. Thanks for your answer. I ran Prime95 blend and memtest free and they ran without problems. Thanks i will keep thede settings and see what happens in daily usage.

Read this:

https://www.reddit.com/r/overclocking/comments/ac0ybb

Thanks!!!

 

[TheMadDutchDude]

I got the message from a vendor (not mentioning who!) when I reviewed an X99 board that it was normal as it helps stabilize the higher XMP profiles. However, what they aren't aware of is that anything above 1.25v (roughly) is considered unsafe for long term usage. I asked them why this was and if they had any plans to change their voltage ranges for IO/SA, but they were pretty clear that it won't be changing. It's kind of annoying from a reviewer standpoint, but you can't win them all. I tried...

 

[Leoplate25]

I got the message from a vendor (not mentioning who!) when I reviewed an X99 board that it was normal as it helps stabilize the higher XMP profiles. However, what they aren't aware of is that anything above 1.25v (roughly) is considered unsafe for long term usage. I asked them why this was and if they had any plans to change their voltage ranges for IO/SA, but they were pretty clear that it won't be changing. It's kind of annoying from a reviewer standpoint, but you can't win them all. I tried...

ASUS Maximus Hero XI and other Maximus boards has a "memory training" feature and if i turn XMP on with defaults VCCIO and SA voltages, the board turns on, off and then on again. Lowering these voltages, the "feature/problem" (or whatever you wanna call it) dissapeared, it starts normally, even on cold boots. Why? Someone tell me it was a PSU protection instead of memory training. I think it is because of the very high VCCIO voltage (over 1.3). When i manually set 1.3, the colour of the number turned purple (danger?). I don't know why or how, but now i am running VCCIO and SA at 1.1v both with XMP enabled (3000mhz) and stock clocks for the i5-9600k. As i said, i ran Prime95 and Memtest without any issues. I played Far Cry New Dawn, browse the internet and everything went and is ok.

 

[TheMadDutchDude]

All motherboards have memory training, for what it's worth.

Yeah, I believe that 1.25v is considered a 'warning' and is in yellow, and as you creep up to 1.35v, it gradually heads in the direction of 'DANGER!!' red. You can honestly try lowering it even further, but I've found good stability at 1.1v for the vast majority of memory testing that I do around those frequencies. It's good up until about 3600 to 3866, and after that will require more.

That's the thing. The XMP voltage is so massively over the top that it can actually cause you instabilities. Take my 2700X for example. No matter what profile I apply, it wants to feed by IMC 1.2v (the maximum safe voltage on AMD) by default. That is actually unstable for me and caused so many headaches that it's not even funny. I've now got my rig at tighter timings, higher frequencies and at 1.02v instead of 1.2v. That's close to 20% difference, which causes issues in and of itself.

 

[Leoplate25]

All motherboards have memory training, for what it's worth.

Yeah, I believe that 1.25v is considered a 'warning' and is in yellow, and as you creep up to 1.35v, it gradually heads in the direction of 'DANGER!!' red. You can honestly try lowering it even further, but I've found good stability at 1.1v for the vast majority of memory testing that I do around those frequencies. It's good up until about 3600 to 3866, and after that will require more.

That's the thing. The XMP voltage is so massively over the top that it can actually cause you instabilities. Take my 2700X for example. No matter what profile I apply, it wants to feed by IMC 1.2v (the maximum safe voltage on AMD) by default. That is actually unstable for me and caused so many headaches that it's not even funny. I've now got my rig at tighter timings, higher frequencies and at 1.02v instead of 1.2v. That's close to 20% difference, which causes issues in and of itself.

I had issues too. And now that you mention it i think it was because of the ram and vccio and sa voltages. I am gonna keep vccio and sa at 1.1v and i will se what happens in daily usage. The think i don't like is the heat produced by the cpu at stock settings, i know it's ok, but for example, running Prime 95 v26.6 SmallFTTs the CPU, at 4.3ghz reaches 58 degrees C, but if i disable turbo boost (3.7ghz) the cpu only reaches 39 degrees C. Why do i have 20 degrees more for only 600mhz? And the voltage isn't high. It's 0.98v at 3.7ghz and 1.1v at 4.3ghz.

P.S.: Voltage numbers turn yellow (warning, caution or whatever) when i set vccio or sa at 1.15, but if i set them at 1.1v they aren't yellow anymore. Passing 1.3v they turn purple (DANGER! Haha!).

 

[TheMadDutchDude]

That’s normal and your temps are perfectly fine. They’re good up until 105c now. So nothing to worry about.

 

[Leoplate25]

That’s normal and your temps are perfectly fine. They’re good up until 105c now. So nothing to worry about.

Well, thanks for your time and your info and answers! Really appreciate it! Take care!!!

 

[John Naylor]

As for safe DD4 voltages ... according to Intel

• 1.2V or lower = Best for DDR4
• 1.35V = okay voltage for overclocking kits( 3200+)
• 1.5V =absolute max voltage allowed for Intel XMP 2.0 profiles and max suggested voltage

if ya can find it on here, it's "certified by Intel" at stated voltages ... plenty over 1.35
https://www.intel.com/content/www/us/en/gaming/xmp-for-core-processors.html

As for testing, with modern CPUs / instruction sets, I find 24 hour synthetic stable OCs failing under application based multitasking benchmarks (RoG Real Bench). in < 1 hour. Much safer too as you won't get anywhere near the core voltages and temps you see w/ synthetics.

http://dlcdnmkt.asus.com/rog/RealBench_v2.56.zip

 

[erocker]

Voltages while high aren't too bad. I have the previous generation of your board and I was getting 1.4v VCCSA with a 3600Mhz kit. Asus just doesn't seem to know what to do with these voltages. I basically had to do stability testing starting with both voltages at 1.2v. Wound up at 1.15/1.2v respectively. For RAM voltage I like to set it over by .01v since it seems every Asus board I've used runs my RAM just under 1.35v due to LLC and things like that.

 

[TheMadDutchDude]

I would really like to be able to do the pre-testing for them so that they aren’t hurting IMCs for no reason.

At least yours runs under... my Crosshair is always over, no matter what LLC I apply.

 

[Leoplate25]

That’s normal and your temps are perfectly fine. They’re good up until 105c now. So nothing to worry about.

Hi, man, me again, haha! I lower IO and SA to stock settings and i bought and ran Ram Test and i didn't have any issues. Do you recommend to run SA at 1.050v and IO at 0.950v? Thanks!

Edit: Corsair Vengeance LED White 3000mhz 2x8gb XMP enabled.

 

[Vayra86]

Hi, man, me again, haha! I lower IO and SA to stock settings and i bought and ran Ram Test and i didn't have any issues. Do you recommend to run SA at 1.050v and IO at 0.950v? Thanks!

Edit: Corsair Vengeance LED White 3000mhz 2x8gb XMP enabled.

My experience with io and sa lower than 1.1v wasnt great with 3000mhz vengeance sticks. Will not pass 15 mins of occt linpack. Bump to 1.1 and all good.

 

[TheMadDutchDude]

What he said! I wouldn’t go to the bare minimum unless you’ve got hours to kill.

 

[Leoplate25]

My experience with io and sa lower than 1.1v wasnt great with 3000mhz vengeance sticks. Will not pass 15 mins of occt linpack. Bump to 1.1 and all good.

What he said! I wouldn’t go to the bare minimum unless you’ve got hours to kill.

Hi and thank you both! I ran Ram Test at 1.1v (IO and SA) for an hour (it covers near 99% of the ram tested) and i ran HCI MemTest too (400% coverage) and i had no issues. What you guys say? I forgot to mention that i ran SmallFFTs and Blend test for an hour each (no errors).

 

[Vayra86]

Hi and thank you both! I ran Ram Test at 1.1v (IO and SA) for an hour (it covers near 99% of the ram tested) and i ran HCI MemTest too (400% coverage) and i had no issues. What you guys say? I forgot to mention that i ran SmallFFTs and Blend test for an hour each (no errors).

That's good, then I'd say just roll with it. If you can pass torture testing, the only other way to find out is just your normal use cases. The main reason to do those tests is to make sure you can keep stability at higher temps.

As for safe DD4 voltages ... according to Intel

• 1.2V or lower = Best for DDR4
• 1.35V = okay voltage for overclocking kits( 3200+)
• 1.5V =absolute max voltage allowed for Intel XMP 2.0 profiles and max suggested voltage

if ya can find it on here, it's "certified by Intel" at stated voltages ... plenty over 1.35
https://www.intel.com/content/www/us/en/gaming/xmp-for-core-processors.html

As for testing, with modern CPUs / instruction sets, I find 24 hour synthetic stable OCs failing under application based multitasking benchmarks (RoG Real Bench). in < 1 hour. Much safer too as you won't get anywhere near the core voltages and temps you see w/ synthetics.

http://dlcdnmkt.asus.com/rog/RealBench_v2.56.zip

Voltages won't get near it in real world? I beg to differ. I've had Overwatch load up and bluescreen my system because of 100 C - max voltage applied. Synthetics simply max out the CPU, and in any high performance power plan, that means that is what the CPU could do under any sort of load for as long as is required.

Synthetics have a purpose. Stress testing, and stability testing under maximum stress. You're right, no use doing that for 24 hours unless you really need long term load stability (server/workstation etc.). For gaming, a somewhat lighter version of that does apply - it also means long term high performance is required.

 

[Leoplate25]

That's good, then I'd say just roll with it. If you can pass torture testing, the only other way to find out is just your normal use cases. The main reason to do those tests is to make sure you can keep stability at higher temps.

Hi, thanks! Then i good to go? What do you know about the last intel vulnerability? Something about zombieload? Is something i need to worry about?