Should I upgrade my motherboard for Overclocking?

[Willy]

Just recently , I've just upgraded my PC to a RX 5700 XT, Ryzen 7 3700x and to an AM4 B450 Asus TUF Motherboard. I'm contemplating if it's worth returning this Motherboard for something Higher End or if it's even worth going through the hassle.

[r.linder]

โพสต์ดั้งเดิมโดย Willy:

โพสต์ดั้งเดิมโดย Escorve:

MSI afterburner is not the tool to be using for a CPU OC anyway. You're probably at stock then.

The tools to use for a CPU OC is BIOS, and in the case of Ryzen, Ryzen Master can be used as well.

Alright. I'll try Ryzen Master and see if I can this Voltage down. Any tips?

1. Set voltage to ~1.325v (lower is better)
2. Start at a reasonable frequency (i.e. 4.15~4.2 GHz; only around 21% of 3700Xs can run 4.15 GHz at 1.262v) and run a stress test like Prime95 (small fft) -- if it crashes, it's not stable. Adjust core frequencies until it's stable, pretty simple. If you can run Prime95 for at least an hour without a system crash, it's most likely going to remain stable.

[Willy]

โพสต์ดั้งเดิมโดย Escorve:

โพสต์ดั้งเดิมโดย Willy:

Alright. I'll try Ryzen Master and see if I can this Voltage down. Any tips?

1. Set voltage to ~1.325v (lower is better)
2. Start at a reasonable frequency (i.e. 4.15~4.2 GHz; only around 21% of 3700Xs can run 4.15 GHz at 1.262v) and run a stress test like Prime95 (small fft) -- if it crashes, it's not stable. Adjust core frequencies until it's stable, pretty simple.

Alright, thanks for the help! I greatly appreciate it!

[Willy]

โพสต์ดั้งเดิมโดย Escorve:

โพสต์ดั้งเดิมโดย Willy:

Alright. I'll try Ryzen Master and see if I can this Voltage down. Any tips?

1. Set voltage to ~1.325v (lower is better)
2. Start at a reasonable frequency (i.e. 4.15~4.2 GHz; only around 21% of 3700Xs can run 4.15 GHz at 1.262v) and run a stress test like Prime95 (small fft) -- if it crashes, it's not stable. Adjust core frequencies until it's stable, pretty simple. If you can run Prime95 for at least an hour without a system crash, it's most likely going to remain stable.

I put voltage down to 1.3, it's at 3.6 ghz atm.

[r.linder]

โพสต์ดั้งเดิมโดย Willy:

โพสต์ดั้งเดิมโดย Escorve:

1. Set voltage to ~1.325v (lower is better)
2. Start at a reasonable frequency (i.e. 4.15~4.2 GHz; only around 21% of 3700Xs can run 4.15 GHz at 1.262v) and run a stress test like Prime95 (small fft) -- if it crashes, it's not stable. Adjust core frequencies until it's stable, pretty simple. If you can run Prime95 for at least an hour without a system crash, it's most likely going to remain stable.

I put voltage down to 1.3, it's at 3.6 ghz atm.

You need to actually control the core clocks, not leave them at auto. Ryzen 3000's boost algorithim is awful for making good use of voltage.

[Willy]

โพสต์ดั้งเดิมโดย Escorve:

โพสต์ดั้งเดิมโดย Willy:

I put voltage down to 1.3, it's at 3.6 ghz atm.

You need to actually control the core voltages, not leave them at auto. Ryzen 3000's boost algorithim is awful for making good use of voltage.

I'm running the benchmark as we speak at 1.28 Volts at 3.9 Ghz.

[r.linder]

โพสต์ดั้งเดิมโดย Willy:

โพสต์ดั้งเดิมโดย Escorve:

You need to actually control the core voltages, not leave them at auto. Ryzen 3000's boost algorithim is awful for making good use of voltage.

I'm running the benchmark as we speak at 1.28 Volts at 3.9 Ghz.

You should be able to squeeze at least 4.1~4.15 all core at 1.3v. Maybe even 4.2 GHz, though 50 MHz all core won't make much of a difference anyway.

You could try 4.15v at 1.3v, and if it works, lower the voltage a little bit and test it again, if it passes Prime95, then lower voltage again and repeat until it stops being stable and go back to the last voltage where it was stable. If you can get 4.15 GHz running below 1.262v without crashing after at least an hour, you've basically won the silicon lottery of 3700Xs.

[Willy]

โพสต์ดั้งเดิมโดย Escorve:

โพสต์ดั้งเดิมโดย Willy:

I'm running the benchmark as we speak at 1.28 Volts at 3.9 Ghz.

You should be able to squeeze at least 4.1~4.15 all core at 1.3v. Maybe even 4.2 GHz, though 50 MHz all core won't make much of a difference anyway.

That's probably the whole reason of it being hot, now it's running idle at 35C. Yesterday it was at 45C. Thankfully I didn't play much or had it on for a long time.

[r.linder]

โพสต์ดั้งเดิมโดย Willy:

โพสต์ดั้งเดิมโดย Escorve:

You should be able to squeeze at least 4.1~4.15 all core at 1.3v. Maybe even 4.2 GHz, though 50 MHz all core won't make much of a difference anyway.

That's probably the whole reason of it being hot, now it's running idle at 35C. Yesterday it was at 45C. Thankfully I didn't play much or had it on for a long time.

As long as it's not 85+ degrees constantly, it's really not a problem. More than 85 degrees in an all-core, sustained high current load would be unsafe (which would include Prime95), the goal for those kinds of loads that slam the CPU is to keep below 80.

Idle temperatures on Ryzen 3000 are often pretty bad (compared to Intel and Ryzen 1000/2000) because boosting is using too much voltage, and it boosts a lot at idle unless you tell it not to with a custom power plan.

[xSOSxHawkens]

Escorve - pbo and voltage are not the issue you are thinking. Peeps have been barking up that tree incorectly. Its all about Amps and burning out the chips power delivery. That is what is killing chips like HWUB saw. That is also why some PBO works for most without harm, and why many boards (such as mine) run stock with a core voltage as high as 1.45 or 1.48 (mine boots at all defaults stock at 1.48v vcore). But it scales down dramatically with loading.

But regardless of the vcore FIT adjusts it as needed to stick within the Amp draw limit of the chip. Anytime you use manual voltage it over rides current limits. With 8 or 12 cores its *super* easy to hit *way* past Amp draw limits at 1.28v even. That is how peeps with 1.25 and 1.3v all cores are still seeimg burn outs. Safe voltages, but too much current through the cpus power circutry.

Hit up r/AMD. Prety big topic there...


@OP. Little point in the upgrade. Stick with ur b450 board and call it good!

[Willy]

โพสต์ดั้งเดิมโดย xSOSxHawkens:

Escorve - pbo and voltage are not the issue you are thinking. Peeps have been barking up that tree incorectly. Its all about Amps and burning out the chips power delivery. That is what is killing chips like HWUB saw. That is also why some PBO works for most without harm, and why many boards (such as mine) run stock with a core voltage as high as 1.45 or 1.48 (mine boots at all defaults stock at 1.48v vcore). But it scales down dramatically with loading.

But regardless of the vcore FIT adjusts it as needed to stick within the Amp draw limit of the chip. Anytime you use manual voltage it over rides current limits. With 8 or 12 cores its *super* easy to hit *way* past Amp draw limits at 1.28v even. That is how peeps with 1.25 and 1.3v all cores are still seeimg burn outs. Safe voltages, but too much current through the cpus power circutry.

Hit up r/AMD. Prety big topic there...


@OP. Little point in the upgrade. Stick with ur b450 board and call it good!

All I see on reddit is constant argument with everyone lmao.

[Willy]

โพสต์ดั้งเดิมโดย xSOSxHawkens:

Escorve - pbo and voltage are not the issue you are thinking. Peeps have been barking up that tree incorectly. Its all about Amps and burning out the chips power delivery. That is what is killing chips like HWUB saw. That is also why some PBO works for most without harm, and why many boards (such as mine) run stock with a core voltage as high as 1.45 or 1.48 (mine boots at all defaults stock at 1.48v vcore). But it scales down dramatically with loading.

But regardless of the vcore FIT adjusts it as needed to stick within the Amp draw limit of the chip. Anytime you use manual voltage it over rides current limits. With 8 or 12 cores its *super* easy to hit *way* past Amp draw limits at 1.28v even. That is how peeps with 1.25 and 1.3v all cores are still seeimg burn outs. Safe voltages, but too much current through the cpus power circutry.

Hit up r/AMD. Prety big topic there...


@OP. Little point in the upgrade. Stick with ur b450 board and call it good!

Also, what you put in ryzen master is completely different from what the SOC is reading. 1.18 was found to be a safe number to put as translate to about 1.25 on the SOC. This is why people are killing their chips, is because they input 1.25 but the actual voltage on the SOC is 1.4.

[xSOSxHawkens]

Yeh... BIOS all the way. Ryzen master is decent for monitoring. But actual changes should be done in BIOS/UEFI if they are supposed to matter.

[r.linder]

โพสต์ดั้งเดิมโดย Willy:

โพสต์ดั้งเดิมโดย xSOSxHawkens:

Escorve - pbo and voltage are not the issue you are thinking. Peeps have been barking up that tree incorectly. Its all about Amps and burning out the chips power delivery. That is what is killing chips like HWUB saw. That is also why some PBO works for most without harm, and why many boards (such as mine) run stock with a core voltage as high as 1.45 or 1.48 (mine boots at all defaults stock at 1.48v vcore). But it scales down dramatically with loading.

But regardless of the vcore FIT adjusts it as needed to stick within the Amp draw limit of the chip. Anytime you use manual voltage it over rides current limits. With 8 or 12 cores its *super* easy to hit *way* past Amp draw limits at 1.28v even. That is how peeps with 1.25 and 1.3v all cores are still seeimg burn outs. Safe voltages, but too much current through the cpus power circutry.

Hit up r/AMD. Prety big topic there...


@OP. Little point in the upgrade. Stick with ur b450 board and call it good!

Also, what you put in ryzen master is completely different from what the SOC is reading. 1.18 was found to be a safe number to put as translate to about 1.25 on the SOC. This is why people are killing their chips, is because they input 1.25 but the actual voltage on the SOC is 1.4.

The recommended max safe voltage for SOC is 1.125v though... and AMD recommends up to 1.2v.

In most cases, you don't need more than 1.125v even with some pretty heavy RAM overclocks.

[Willy]

โพสต์ดั้งเดิมโดย Escorve:

โพสต์ดั้งเดิมโดย Willy:

Also, what you put in ryzen master is completely different from what the SOC is reading. 1.18 was found to be a safe number to put as translate to about 1.25 on the SOC. This is why people are killing their chips, is because they input 1.25 but the actual voltage on the SOC is 1.4.

The recommended max safe voltage for SOC is 1.125v though... and AMD recommends up to 1.2v.

In most cases, you don't need more than 1.125v even with some pretty heavy RAM overclocks.

Currently I'm hitting 1.3v (BIOS) at 4.1 Ghz.

[r.linder]

โพสต์ดั้งเดิมโดย Willy:

โพสต์ดั้งเดิมโดย Escorve:

The recommended max safe voltage for SOC is 1.125v though... and AMD recommends up to 1.2v.

In most cases, you don't need more than 1.125v even with some pretty heavy RAM overclocks.

Currently I'm hitting 1.3v (BIOS) at 4.1 Ghz.

You might be able to squeeze in at least another 50 MHz, if you want to get the most out of that voltage. If you're lucky, you'll be able to do at least 4.2 GHz, but 3700X silicon is not as good as the 3800X's, so it's not too likely.