**THE OFFICIAL AMD CHIPSET DRIVERS THREAD**

[VortexA1]

Just changed mine from 99% to 5%

There must be a valid reason AMD have the minimum at 99%..?

 

[welshrat]

There must be a valid reason AMD have the minimum at 99%..?

I would like to know this as well as I changed mine last night and I havent noticed any issues other than lower idle temps. doesn't boost less or feel any less responsive. Very odd

 

[Sin_Chase]

I would like to know this as well as I changed mine last night and I havent noticed any issues other than lower idle temps. doesn't boost less or feel any less responsive. Very odd

Zen 2 uses an advanced sleep state when cores are not in use and become power gated off.

99% minimum allows the core to enter the sleep state when idling and therefore offer power savings but remain responsive when further required and awoken.

 

[VortexA1]

Zen 2 uses an advanced sleep state when cores are not in use and become power gated off.

99% minimum allows the core to enter the sleep state when idling and therefore offer power savings but remain responsive when further required and awoken.

That sounds ideal then. Surely it's best to leave it at that then.

 

[Nghtmare]

I left mine on 90%, therefore I allow it to lower the clock to lower itself under small load and go down to 0.9xxv instead of always keeping it above 1v.

 

[Yaayuh!]

Same here, defaults to 90%, I’ve left it there.

 

[opethdisciple]

New set of chipset drivers released yesterday:

Revision Number
1.9.27.1033

 

[ltron]

There must be a valid reason AMD have the minimum at 99%..?

Because if I remember correctly the hardware power management on the CPU takes over almost instantly at idle if min is set to 99%, if you change this to a lower value then Windows has more control around how power management and clock speeds behave and is applying a generic software profile which is less responsive and slower to ramp up clocks. However, doing this in hardware instead is faster in terms of going in and out of sleep states and ramping up clock speeds and hence more responsive.

 

[gpuerrilla]

if you change this to a lower value then Windows has more control around how power management and clock speeds behave. This should be faster in terms of going in and out of sleep states and ramping up clock speeds and hence more responsive in hardware vs software.

Are you saying windows is better (which I would guess isnt), just the way you closed off there isnt clear, but I think your saying otherwise.

 

[Illuminist]

New set of chipset drivers released yesterday:

Revision Number
1.9.27.1033

Notice anything better or worse?

 

[gpuerrilla]

New set of chipset drivers released yesterday:

Revision Number
1.9.27.1033

Thanks. Installed them now.

 

[ltron]

Are you saying windows is better (which I would guess isnt), just the way you closed off there isnt clear, but I think your saying otherwise.

No, I'm saying it's better to let the hardware handle it not Windows. Sorry, shouldn't post late at night and I've edited my post.

 

[Illuminist]

Thanks. Installed them now.

Notice anything better or worse?

 

[gpuerrilla]

Notice anything better or worse?

I would say when casually using the machine its now between 5-20w lower, I have a power meter at the wall. Not sure what other improvements there are yet.

 

[Panos]

Notice anything better or worse?

The 3900X doesn't go over 1.494v from what I see. Used to hit 1.5 until this update.

 

[humbug]

I would say when casually using the machine its now between 5-20w lower, I have a power meter at the wall. Not sure what other improvements there are yet.

This, My CPU is running a little cooler. this without 1003 ABBA, i'm still waiting for that.

 

[gpuerrilla]

This, My CPU is running a little cooler. this without 1003 ABBA, i'm still waiting for that.

Yeah I have not updated the latest BIOS either so more headroom left, should be really solid soon.

 

[JJimmyJ]

Really confused with this power plan malarkey.
So, doing some reading it looks like the hardware downclocks faster than any software can monitor, so hwi etc will not show downclocking when ryzen plan is in play. But many state that the hardware downclock only happens if min state set to 99%. The new ryzen plan based on new drivers sets min at 90%.
I tried using ryzen master to monitor and it just doesn't downclock.
So anyone know what the crack is?

 

[Lolik]

Really confused with this power plan malarkey.
So, doing some reading it looks like the hardware downclocks faster than any software can monitor, so hwi etc will not show downclocking when ryzen plan is in play. But many state that the hardware downclock only happens if min state set to 99%. The new ryzen plan based on new drivers sets min at 90%.
I tried using ryzen master to monitor and it just doesn't downclock.
So anyone know what the crack is?

Dunno if you still on Ryzen+ , but just before Ryzen 3000 was released , updated chipset drivers and they did made my 2600x (allcore oc) unstable and noticed more aggressive power delivery aswell, for ryzen 3000 it is just gets better with every bios and chipset upgrade (watching cores going into parking state is amazing), probably try doing some "safe" option and using windows balanced plan and play a bit in bios with digi vrm.

 

[humbug]

1004 next month. "over 100 improvements"

https://www.tomshardware.co.uk/amd-ryzen-cpus-new-microcode-bios-agesa-1004,news-61813.html