Generic CPU discussion

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 5057
Credit: 19256557623
RAC: 6975042

Don't be afraid to really

Don't be afraid to really "reef" down on the cpu cooler mounting screws.  The backplate socket nuts can take it.

My cooler manufacturer tells us to tighten as much as hand pressure can apply to get the best contact between the coldplate and the IHS for best temps.

The monitoring programs poll every second usually, but you are basically limited by the on-die temp reporting algorithm to what it displays and its polling rate.  AMD has never divulged that low level of information. There are dozens of on-die temp sensors in the Zen architecture all oriented to let any single core boost to its max clock at any time depending on load and power availability.

The TIM applied to the AIO coldplates is not bad, but it is not the best that can be purchased.  The silk screening application of the TIM is normally very even though I have seen bad applications plus the TIM patch does not cover 100% of the coldplate so is not covering 100% of the IHS.  You need 100% of the coldplate and IHS coverage for best heat transfer.

You might consider a better paste like Arctic MX-4 or MX-5 or similar.  I've used that myself for years or am now using Kingpin KPx paste which is a tad bit better for more cost.

[Edit]

No personal experience with that AIO cooler but from the images at least a better design than the old round coldplates that are common with the Asetek cooler base hardware. 

Can't tell if that coldplate is large enough to encompass the full IHS area.

Always best to use a "full-coverage" coldplate for best heat transfer from the IHS of the Zen, Threadripper and Epyc cpus.

 

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6595
Credit: 340075776
RAC: 319346

Keith Myers wrote:Don't be

Keith Myers wrote:

Don't be afraid to really "reef" down on the cpu cooler mounting screws.  The backplate socket nuts can take it.

My cooler manufacturer tells us to tighten as much as hand pressure can apply to get the best contact between the coldplate and the IHS for best temps.

Right. Consider it reefed. Three consecutive 'sensors' :

Tdie:         +72.6°C  (high = +95.0°C)
Tctl:         +72.6°C  
Tccd1:        +65.8°C  
Tccd2:        +66.0°C

and :

Tdie:         +73.8°C  (high = +95.0°C)
Tctl:         +73.8°C  
Tccd1:        +64.5°C  
Tccd2:        +63.5°C

and again :

Tdie:         +72.4°C  (high = +95.0°C)
Tctl:         +72.4°C  
Tccd1:        +65.5°C  
Tccd2:        +65.5°C 

... epic ! Thanks very much for that tip.

Quote:
The monitoring programs poll every second usually, but you are basically limited by the on-die temp reporting algorithm to what it displays and its polling rate.  AMD has never divulged that low level of information. There are dozens of on-die temp sensors in the Zen architecture all oriented to let any single core boost to its max clock at any time depending on load and power availability.

Aha. Which is why zenmonitor3 reports different clock speeds & power per core, moment to moment. I'd wondered about that.

Quote:

The TIM applied to the AIO coldplates is not bad, but it is not the best that can be purchased.  The silk screening application of the TIM is normally very even though I have seen bad applications plus the TIM patch does not cover 100% of the coldplate so is not covering 100% of the IHS.  You need 100% of the coldplate and IHS coverage for best heat transfer.

You might consider a better paste like Arctic MX-4 or MX-5 or similar.  I've used that myself for years or am now using Kingpin KPx paste which is a tad bit better for more cost.

[Edit]

No personal experience with that AIO cooler but from the images at least a better design than the old round coldplates that are common with the Asetek cooler base hardware. 

Can't tell if that coldplate is large enough to encompass the full IHS area.

Always best to use a "full-coverage" coldplate for best heat transfer from the IHS of the Zen, Threadripper and Epyc cpus.

I will keep all of that in mind. Again thanks.

Meanwhile the RAC/credit continues to climb like an F-16 on afterburner and no invalids yet ! ;-)

Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter ...

... and my other CPU is a Ryzen 5950X :-) Blaise Pascal

Tom M
Tom M
Joined: 2 Feb 06
Posts: 6876
Credit: 9787363334
RAC: 3272213

Mike, Do you have two

Mike,

Do you have two identical machines?  Or has the server just issued you two machine id's? Or does this have something to do with a Flatpak?

And exactly what is a "flatpak" in this context?  A virtual machine? If yes, that implies you are running two VM's?

Tom M

A Proud member of the O.F.A.  (Old Farts Association).

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 5057
Credit: 19256557623
RAC: 6975042

Mike Hewson wrote: Right.

Mike Hewson wrote:

Right. Consider it reefed. Three consecutive 'sensors' :

Tdie:         +72.6°C  (high = +95.0°C)
Tctl:         +72.6°C  
Tccd1:        +65.8°C  
Tccd2:        +66.0°C

and :

Tdie:         +73.8°C  (high = +95.0°C)
Tctl:         +73.8°C  
Tccd1:        +64.5°C  
Tccd2:        +63.5°C

and again :

Tdie:         +72.4°C  (high = +95.0°C)
Tctl:         +72.4°C  
Tccd1:        +65.5°C  
Tccd2:        +65.5°C 

... epic ! Thanks very much for that tip.

 

Ahhh! Much better and more normal as what is expected. I'd skip repasting and just run the heck out of your new afterburner equipped PC.

 

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 5057
Credit: 19256557623
RAC: 6975042

Tom M wrote:Mike,Do you

Tom M wrote:

Mike,

Do you have two identical machines?  Or has the server just issued you two machine id's? Or does this have something to do with a Flatpak?

And exactly what is a "flatpak" in this context?  A virtual machine? If yes, that implies you are running two VM's?

Tom M

I'd say the server hiccupped and issued a new hostID for some reason.  You can always merge both hosts.

Flatpak is one of the application packaging options that is being used in Linux distros.

The others being Snap used by Ubuntu and almost universally hated by longtime Linux users and the other is AppImage.

They all have pros and cons but basically operate in the same manner in bundling up all necessary dependencies for an application into the distribution bundle.  So the packages are much larger in general compared to just doing an apt install of the application because the necessary dependencies are grabbed outside of just the application binaries themselves during the install.

The benefit is that application upgrades or updates are self contained and the underlying OS updates can't break the application.

https://phoenixnap.com/kb/flatpak-vs-snap-vs-appimage

 

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6595
Credit: 340075776
RAC: 319346

Hi Tom.pop-os ( 12941316

Hi Tom.

pop-os ( 12941316 ) and fireball-xl5 ( 12941363 ) are two configurations of the same hardware, just with different storage setups. I'd been experimenting/mulling over [SSD + SATA] vs [SSD x2] and have finally settled on having two M.2 SSD's. That puts a 240GB SSD, the RTX 3060 and a 1TB SSD all running off the PCIe bus in fireball-xl5. I had chosen that motherboard for that possibility : there being two M.2 positions that are either side of the 1st/closest PCIe traditional slot. The whole point is that a good M.2 SSD mounted on PCIe can be read/write faster than SATA, or SAS SSDs for that matter.

A flatpak is a Linux software package type as Keith explains. Pop!_OS is Ubuntu derived, open source, has long term support and comes with Vulkan video drivers as default ( gaming ). Plus it comes with BOINC client/manager/gui with a really easy install.

Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter ...

... and my other CPU is a Ryzen 5950X :-) Blaise Pascal

Tom M
Tom M
Joined: 2 Feb 06
Posts: 6876
Credit: 9787363334
RAC: 3272213

Can anyone make case for

Can anyone make case for buying  a 3950x over a Ryzen 5950x because it is that much cheaper?

A Proud member of the O.F.A.  (Old Farts Association).

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 5057
Credit: 19256557623
RAC: 6975042

Tom M wrote: Can anyone make

Tom M wrote:

Can anyone make case for buying  a 3950x over a Ryzen 5950x because it is that much cheaper?

Uhhh . . .  . absolutely no.  I assume you must be talking about used 3950X prices on eBay or something because everywhere I looked for new, unused 3950X cpus, they were 3X as expensive as a brand new 5950X.

The 5950X smokes the 3950X in both single and multi-thread speed.  So the supposed price premium you are alluding to is well worth it.

 

GWGeorge007
GWGeorge007
Joined: 8 Jan 18
Posts: 3196
Credit: 5214306723
RAC: 4341350

DITTO!!  What he said!!

DITTO!!  What he said!!

George

Proud member of the Old Farts Association

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6595
Credit: 340075776
RAC: 319346

Where I bought it, on the

Where I bought it, on the day, the 5950X was the same price as the 3950X ! Both were AUD $800 including 10% GST, which works out to about USD $500 ( AUD $100 is/was worth USD $69 ) :

800 x (10/11) X 0.69 = 501.82

Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter ...

... and my other CPU is a Ryzen 5950X :-) Blaise Pascal

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.