F5F Stay Refreshed Hardware Desktop Secure VCCSA and VCCIO (Mem OC) support Rocket Lake

Secure VCCSA and VCCIO (Mem OC) support Rocket Lake

Secure VCCSA and VCCIO (Mem OC) support Rocket Lake

H
HERVVEHH
Member
69
03-18-2021, 07:38 PM
#1
I've found information suggesting 1.25V is a secure setting for VCCIO during continuous operation. Some references indicate 1.35V as the upper limit, but Rocket Lake appears to handle higher voltages better. This includes Vcore, SA, and IO parameters. The chipset divides VCCIO into two channels: VCCIO and VCCIO2. Motherboards also tend to apply more voltage automatically. For instance, my MSI B560 board delivers 1.25V SA and 1.31 IO when XMP is active at 3200Mhz. When I raise the frequency to 3600Mhz, SA climbs to 1.3V and IO to 1.41V. Many users also note higher voltages during auto mode. Though it's been nearly a year since the product launch, I suspect there were numerous reports about failing RKL CPUs if auto-voltages were truly risky. Most people simply enable XMP without checking voltage levels. My 32GB 3200Mhz CL16 RAM can reach 3600-36V if I set VCCSA to 1.23V and VCCIO to 1.32V. Lowering VCCIO or memory settings during cold starts often causes issues. Is this approach safe for long-term use?
H
HERVVEHH
03-18-2021, 07:38 PM #1

I've found information suggesting 1.25V is a secure setting for VCCIO during continuous operation. Some references indicate 1.35V as the upper limit, but Rocket Lake appears to handle higher voltages better. This includes Vcore, SA, and IO parameters. The chipset divides VCCIO into two channels: VCCIO and VCCIO2. Motherboards also tend to apply more voltage automatically. For instance, my MSI B560 board delivers 1.25V SA and 1.31 IO when XMP is active at 3200Mhz. When I raise the frequency to 3600Mhz, SA climbs to 1.3V and IO to 1.41V. Many users also note higher voltages during auto mode. Though it's been nearly a year since the product launch, I suspect there were numerous reports about failing RKL CPUs if auto-voltages were truly risky. Most people simply enable XMP without checking voltage levels. My 32GB 3200Mhz CL16 RAM can reach 3600-36V if I set VCCSA to 1.23V and VCCIO to 1.32V. Lowering VCCIO or memory settings during cold starts often causes issues. Is this approach safe for long-term use?

K
Kandy_
Junior Member
37
03-24-2021, 07:32 PM
#2
Seek the safest voltage possible since lower values are often misleading and too cautious for serious overclocking. Auto settings usually spike significantly, so manually adjust the voltage is best. The RAM speed depends on the specific chips, so inspect your memory modules in that burner to confirm the type.
K
Kandy_
03-24-2021, 07:32 PM #2

Seek the safest voltage possible since lower values are often misleading and too cautious for serious overclocking. Auto settings usually spike significantly, so manually adjust the voltage is best. The RAM speed depends on the specific chips, so inspect your memory modules in that burner to confirm the type.

C
ChainZoX
Junior Member
17
03-26-2021, 03:24 PM
#3
Igor's Lab indicated the safe range is between 1.45 and 1.5V for SA and IO on RKL. Another site mentioned a lower limit of 1.15V, but I’m struggling to boot at 3200Mhz below 1.2V. Some users seem to increase IO and SA beyond 1.4V on overclock.net, which might be acceptable. Also, my RAM is a Samsung D-die; it handles overclocks well when I push enough VCCIO, though memory training remains challenging.
C
ChainZoX
03-26-2021, 03:24 PM #3

Igor's Lab indicated the safe range is between 1.45 and 1.5V for SA and IO on RKL. Another site mentioned a lower limit of 1.15V, but I’m struggling to boot at 3200Mhz below 1.2V. Some users seem to increase IO and SA beyond 1.4V on overclock.net, which might be acceptable. Also, my RAM is a Samsung D-die; it handles overclocks well when I push enough VCCIO, though memory training remains challenging.

S
Stifelplays
Member
112
04-07-2021, 12:45 PM
#4
It looks like a high voltage requirement for such a low memory clock speed, even with 32gb storage doesn't help much. Usually I manage under 1.2v with the chip while running 16gb at 4000mhz CL16; the die handles the Dimm voltage fine. Although the board automatically sets these values around 1.4v+, it's not essential in my setup. Because of the memory die type and count, getting the desired performance needs a bit more effort from the memory controller. Long-term, it's safer to keep it lower.
S
Stifelplays
04-07-2021, 12:45 PM #4

It looks like a high voltage requirement for such a low memory clock speed, even with 32gb storage doesn't help much. Usually I manage under 1.2v with the chip while running 16gb at 4000mhz CL16; the die handles the Dimm voltage fine. Although the board automatically sets these values around 1.4v+, it's not essential in my setup. Because of the memory die type and count, getting the desired performance needs a bit more effort from the memory controller. Long-term, it's safer to keep it lower.

X
xFunnyman
Member
192
04-07-2021, 01:27 PM
#5
Rocket Lake and Gear 1 behave differently compared to other configurations. Many Rocket Lake CPUs struggle to exceed 3600Mhz even with high voltages, and achieving 3733Mhz booting was unsuccessful under those conditions. Some users report not reaching 3600Mhz at all on this platform. The situation seems influenced by both voltage settings and chip selection. Gear mode adds another layer of uncertainty regarding voltage accuracy. Traditional guidelines no longer seem reliable here.
X
xFunnyman
04-07-2021, 01:27 PM #5

Rocket Lake and Gear 1 behave differently compared to other configurations. Many Rocket Lake CPUs struggle to exceed 3600Mhz even with high voltages, and achieving 3733Mhz booting was unsuccessful under those conditions. Some users report not reaching 3600Mhz at all on this platform. The situation seems influenced by both voltage settings and chip selection. Gear mode adds another layer of uncertainty regarding voltage accuracy. Traditional guidelines no longer seem reliable here.

D
davachio
Member
204
04-13-2021, 04:59 PM
#6
D
davachio
04-13-2021, 04:59 PM #6

T
tlr13579
Member
178
04-13-2021, 05:51 PM
#7
Gear 2 mode requires a minimum of 4400Mhz RAM to work properly with models like the CL14 in Gear 1 mode. It doesn’t make sense unless you use a CPU paired with extremely high-speed RAM. However, on the overclock.nets forum, some users are testing CPUs with very high voltage settings for SA and IO. I checked through the posts and didn’t find any mention of IMC damage or failed CPUs. Based on what I saw, using 1.32V for IO and 1.24V for SA seems safe.
T
tlr13579
04-13-2021, 05:51 PM #7

Gear 2 mode requires a minimum of 4400Mhz RAM to work properly with models like the CL14 in Gear 1 mode. It doesn’t make sense unless you use a CPU paired with extremely high-speed RAM. However, on the overclock.nets forum, some users are testing CPUs with very high voltage settings for SA and IO. I checked through the posts and didn’t find any mention of IMC damage or failed CPUs. Based on what I saw, using 1.32V for IO and 1.24V for SA seems safe.

A
audi497mks
Senior Member
601
04-13-2021, 11:42 PM
#8
That's correct. Use only what's necessary.
A
audi497mks
04-13-2021, 11:42 PM #8

That's correct. Use only what's necessary.

C
cursino_8
Member
226
04-14-2021, 06:15 AM
#9
That other tech site seems completely off the mark, or maybe you misinterpreted the voltage readings from Igors' lab. From what I heard, around 1.45V is the safe upper limit, and about 1.8V marks when CPUs begin failing outright. Being conservative with voltages will likely lead to weak overclocks. My estimate is probably around 1.5V as a safe range, with a -0.3V buffer being ideal to avoid damage. A small negative offset like -0.05 should be better than nothing.
C
cursino_8
04-14-2021, 06:15 AM #9

That other tech site seems completely off the mark, or maybe you misinterpreted the voltage readings from Igors' lab. From what I heard, around 1.45V is the safe upper limit, and about 1.8V marks when CPUs begin failing outright. Being conservative with voltages will likely lead to weak overclocks. My estimate is probably around 1.5V as a safe range, with a -0.3V buffer being ideal to avoid damage. A small negative offset like -0.05 should be better than nothing.

A
AmazinglyCool
Senior Member
695
04-14-2021, 03:09 PM
#10
I matched what Somerandomtechyboi shared—I bought a few parts at a great rate and got to work. I felt confident about the voltage requirements. The VCCS reached up to 1.5 volts. Considering the goals, it seems clear they aimed for a durable design. It deserves recognition for handling G1/T1 workloads, though credit isn’t fully given for its performance in G1. A higher VCCO2 would be safer for G1. Around 1.32 volts should suffice. You might see odd events in the event viewer due to the IMC, even after passing stress tests. System: 11900KF 3200 b-die 16GB X2 @1.35v (14-14-34). After completing TM5, I reduced power consumption by about 25W below normal temp to run Prime95 Large FFTs. This chip isn’t the top choice for everyday use, but it can handle heavy loads—over 300W with proper cooling.
A
AmazinglyCool
04-14-2021, 03:09 PM #10

I matched what Somerandomtechyboi shared—I bought a few parts at a great rate and got to work. I felt confident about the voltage requirements. The VCCS reached up to 1.5 volts. Considering the goals, it seems clear they aimed for a durable design. It deserves recognition for handling G1/T1 workloads, though credit isn’t fully given for its performance in G1. A higher VCCO2 would be safer for G1. Around 1.32 volts should suffice. You might see odd events in the event viewer due to the IMC, even after passing stress tests. System: 11900KF 3200 b-die 16GB X2 @1.35v (14-14-34). After completing TM5, I reduced power consumption by about 25W below normal temp to run Prime95 Large FFTs. This chip isn’t the top choice for everyday use, but it can handle heavy loads—over 300W with proper cooling.