Trying to understand Ryzen 3000 series boost speed variations

[lorry]

You guys could go the graphite pad route instead of LM. I'm still using the Innovation Cooling IC Thermal Pad I posted about over in the thread "show off your tech purchase."

What's your latest tech purchase?

It's not a bad thing at all, doing really well I think!! :D Have you tested the previous CPU with a few tests to see if there's much of an increase? ;)

www.techpowerup.com

The only minor issues I ran into was use something like tweezers to move the pad into the heatspreader don't touch it. Started to curl after removed from the plastic clamshell package, and it's slippery. Had a heck of time getting it to stay put while mounting the heatsink.

what are the thermals like when you use them?

And, it was suggested putting a drop of distilled water on the cpu just to temporarily hold it in place somewhat, ever heard that?

 

[Waldorf]

Yeah, copper on the block and 3770k with HS.

Those pads are a PITA and i replaced it with LM after 2 days,
as i had similar performance with silver/diamond based TP.
Especially when dealing with a rig that isnt layed flat (HS parallel to ground),
i have no problem installing any TP/LM.
Not happening with the pad.

 

[sneekypeet]

Not to dig up old news, but I would never buy anything IC related, point blank. The owner showed his true colors here, and is not a company I am willing to support, nor should any of you

 

[lorry]

There is also the Thermal Grizzly carbonaut pad remember, but thats really only for low end use I think?

Thermal Grizzly Carbonaut Thermal Pad - 38 × 38 × 0.2 mm

Innovative and efficient thermal pad, Reusable & extremely durable, Good thermal conductivity of 62.5 W / (m · K), 0.2 mm thickness - 38 × 38 mm, For Intel Socket 2066 and AM4 Socket CPU's

www.overclockers.co.uk

and this one ?

Coollaboratory Liquid MetalPad - 3 x CPU Thermal Pads

The innovation in the cooling of processors from High-End PC systems, notebooks and game consoles with a heat conduction pad!

www.overclockers.co.uk

 

[biffzinker]

Not to dig up old news, but I would never buy anything IC related, point blank. The owner showed his true colors here, and is not a company I am willing to support, nor should any of you

I thought it was another company responsible. The name seemed familiar but I didn't realize it was that company.

 

[sneekypeet]

I thought it was another company responsible. The name seemed familiar but I didn't realize it was that company.

https://www.innovationcooling.com/products/ic-diamond/

I may have come off a bit strong there too. Was not meant to chastise, more of a public service announcement.

 

[lorry]

IF all these things are meant to be SO much better, How come hardly anyone uses them then?

case of snake oil perchance?

 

[Waldorf]

LM is the cheapest of them all, and lots of fakes being sold on amazon that aren't from the brand that is printed on the box (e.g. kryonaut),
that i dont care about who and/or what is behind it, as i rather have a proper working product.
It does improve temps by at least a couple of degrees (C), does "depend" on things like cpu/type of cooler etc involved,
but i've always seen a noticeable gain over good TP like arctics and the like, no matter the cooling used.

edit it to make it simpler to read.

 

[lorry]

LM is the cheapest of them all, and lots of fakes being sold on amazon hat aren't from the printed name (e.g. kryonaut), that i dont care about who and/or what is behind it, as i rather have a proper working product.
It does improve temps by at least a couple of degrees (C), does depend on things like cpu/type of cooler etc involved, but always seen a noticeable gain over good TP like arctics and the like.

which liquid metal though? this IC one or the coolerlab one?

------------------------------------------------------------------------------------------------------------------------------------------

A thought, why haven't Intel and AMD come with torque settings for CPU coolers?
That would take all the guesswork and worry about over tensioning coolers, and give optimum sealing between cpu and cooler?

 

[Waldorf]

Coollab one.
I haven't had a LC not having screws that limit the amount you can tighten them, usually done with the length of the screw thread.

 

[lorry]

Knowing if a screw is threaded 4 mm or 5 mm in is a heck of a lot harder to gauge that having a torque wrench spin because you have reached the correct torque poundage though

 

[Waldorf]

You screw them in until it wont go any further.
As long as you dont force it, all thats needed.

 

[biffzinker]

https://www.innovationcooling.com/products/ic-diamond/

I may have come off a bit strong there too. Was not meant to chastise, more of a public service announcement.

It's okay, I didn't take in a manner that would offend me. Had I picked up on my suspicion of that company's name on the product I was purchasing from a third party. I would of looked somewhere else for a graphite pad I was interested in trying out.

You screw them in until it wont go any further.

On AMD's stock coolers there is a compression spring that prevents over tightening the four screws into the backplate.

 

[lorry]

Solved the pump question!
I was correct



Hey @Zach_01 or anyone else of course when I finally make the case swap and get things water cooled and then hopefully lower temps, a couple of questions.
Do you feel there is much more to be got out of this memory, or is it near enough at its limit?
I'm sure that you guys know way better than myself but with the hopefully better cpu temps that will result from water cooling, any idea how much of a performance gain there might be?

 

[PolRoger]

I can see the CPU is hitting almost all stock PBO limits of PPT:142/TDC:95/EDC:140 and that is because of your sub-70°C temp. Not often to see this. Because of relatively low temps the silicon manager is pushing CPU to the registered limits. Low temps give more headroom for all the limits but they stop at their factory sets.
Have you consider releasing them? It will give you potential more clock. And this is still within the stock silicon parameters.

May I suggest a few settings to try?

1. Just limits release (this may increase clocks marginally)
PPT: 147W
TDC:100A
EDC: 145A
PBOscalar: Auto

2. EDC capping (this will give more clocking)
PPT: 145W
TDC:100A
EDC: 130A (feel free to reduce it more)
PBOscalar: Auto

3. EDC capping and clock/voltage preservation (this certainly will give better clocking)
PPT: 145W
TDC:100A
EDC: 130A (feel free to reduce it more)
PBOscalar: X2 (Auto is X1, X2 mean longer and sustained clock and voltage. May increase temp a few degrees = 3~4C)

Nothing of this will have any dangerous outcome for the CPU, as you still relay on the internal silicon manager to adjust clock and voltage, known also as FIT (Silicon FITness controller).
This has nothing to do with manual/static OC which is disabling FIT.

#1
PPT: 147W
TDC:100A
EDC: 145A
PBOscalar: Auto

#3
PPT: 145W
TDC:100A
EDC: 130A (feel free to reduce it more)
PBOscalar: X2 (Auto is X1, X2 mean longer and sustained clock and voltage. May increase temp a few degrees = 3~4C)

 

[lorry]

certainly beats my best so far, but then I am still on air atm

 

[Zach_01]

Solved the pump question!
I was correct

View attachment 143813

Hey @Zach_01 or anyone else of course when I finally make the case swap and get things water cooled and then hopefully lower temps, a couple of questions.
Do you feel there is much more to be got out of this memory, or is it near enough at its limit?
I'm sure that you guys know way better than myself but with the hopefully better cpu temps that will result from water cooling, any idea how much of a performance gain there might be?

Man how do I miss that one? Sorry!
For the RAM question I personally cant really say what is going to happen. RAM it self cant be affected by water cooling but given the fact that the CPU mem controller will be cooled also anything is possible. You will have to try again with all the RAM settings to either increase speed further or tighten timings. I prefer speed...
For CPU, well you can see @PolRoger CPU... nothing to say more...

------------------------------------

@PolRoger, how do you see it by now? It has more to give as it is, if you further uncap PPT from 145W. Give it 150W try it, see results.
Then try this:

#4
PPT: 150W
TDC:100A
EDC: 125A
PBOscalar: X2

PS:
From screenshots I cant really see anything else usefull than max values of PPT/TDC/EDC because you didnt follow what I said to isolate capture screen strictly to bench time.

 

[Waldorf]

outside matching If speed, amds usually like timings (over speed).
or did that change with ryzen?

 

[mtcn77]

#1
PPT: 147W
TDC:100A
EDC: 145A
PBOscalar: Auto

#3
PPT: 145W
TDC:100A
EDC: 130A (feel free to reduce it more)
PBOscalar: X2 (Auto is X1, X2 mean longer and sustained clock and voltage. May increase temp a few degrees = 3~4C)

The thermal transfer has improved(146w>148w) when edc has been restricted, though performance has suffered... Any change you make more of these with wider edc-ppt gaps? 3900X has run much hotter in the latter run. It could be a thermal throttling issue. Also, you haven't included the coretemp monitor charts for the first run. What's obvious is the chips have not maintained a perfect boost.

What else, ryzen 3000 ihs is dished inwards and there are heat load issues.
I haven't seen much debate on this discussion in this forum, so laying it out there.

Dotting For Better Ryzen 3000 Thermals? | Level One Techs

As a follow-up to this – why might the spread method not be desirable? If your CPU has a concave ihs (or your cooler for that matter) it is possible to spread the Tim too thin. On stock wraith prism coolers it is pretty thick and the bottom of the cooler isn’t super even. On the h115i pro I...

forum.level1techs.com

 

[lorry]

What else, ryzen 3000 ihs is dished inwards and there are heat load issues.
I haven't seen much debate on this discussion in this forum, so laying it out there.

That very thing is talked about and displayed in GN video on lapping a 3000 with bearded hardware, admittedly a 3970 but they say that they are all the same

 

[Zach_01]

The thermal transfer has improved(146w>148w) when edc has been restricted, though performance has suffered... Any change you make more of these with wider edc-ppt gaps? 3900X has run much hotter in the latter run. It could be a thermal throttling issue. Also, you haven't included the coretemp monitor charts for the first run. What's obvious is the chips have not maintained a perfect boost.

What else, ryzen 3000 ihs is dished inwards and there are heat load issues.
I haven't seen much debate on this discussion in this forum, so laying it out there.

Dotting For Better Ryzen 3000 Thermals? | Level One Techs

As a follow-up to this – why might the spread method not be desirable? If your CPU has a concave ihs (or your cooler for that matter) it is possible to spread the Tim too thin. On stock wraith prism coolers it is pretty thick and the bottom of the cooler isn’t super even. On the h115i pro I...

forum.level1techs.com

We don’t care about thermal transfer increase out of PBO settings, and we are not doing any of this stuff to improve thermal transfer. This is a different matter.
What we like is more performance, and if this comes with a small increase in temp while the chip is within and under internal controller it’s ok. And I see that performance has increased from previous run.
The CPU run hotter in latter run because he set PBO scalar X2 from the default X1. This preserves voltage and clock in a more sustainable way.
Temp raise is necessary evil, one to control only with better cooling, and whatever that includes. Better cooling, surface lapping... etc...

Please read all information here before posting and don’t just state what is on your head. Don’t ruin this thread... And try to use normal language. Leave journalese aside...

 

[mtcn77]

Please read all information here before posting and don’t just state what is on your head.

??
Just pointing out the obvious and you may be missing out on my part to play here.
You won't get very far in case of higher duty, if it serves to hit temperature limits before the benchmark is offline. Which is the case, as you can see, looking at clocks. Restricting ranges has only served to lift up the thermal window despite higher temperatures have developed. This will continue until the new thermal equilibrium has been established at the peak temperature limit in a lower pbo setting.
GN made it clear the chip does not reliably boost up sequentially from one run to the next, so temperature loading is a no go. The necessary evil here is either to downgrade pbo ranges, or improving thermals.

One other thing I discovered today, that graphite works great with ryzen 3000; however it is not simple as that. The junction dish will cause uneven contact patch which is not something compression sensitive conductors can take care off by themselves. YMMV, essentially.

Thermal paste & pad comparison (MX4, Graphite thermal pads from IC and Panasonic)

I bought some Panasonic Graphite Thermal pad material from Arrow this week. I’ve had good experience with IC Graphite for LED testing, since they are reusable and require no cleaning unlike traditional thermal pastes. Panasonic datasheet...

budgetlightforum.com

 

[biffzinker]

One other thing I discovered today, that graphite works great with ryzen 3000; however it is not simple as that.

I'm using the IC graphite pad with a Ryzen 5 3600. How does tight in the graph relate to a third party's heatsink mounting mechanism that is screw down? Would over tightening the two screws help at all?

 

[Zach_01]

I’ve done extensive benchmarking with different PBO settings and I can confirm that the CPU, with EDC reduction and PPT release, sustainably preserves higher clocks in the exact same way with stock settings. Even with a slight increase in temperature. Of course you can’t keep decreasing EDC nor increasing PBOscalar. There is a point that this stops returning positive results. We haven’t find it yet for this CPU. Until PPT stops increasing by itself you keep going. After this only way to improve performance is with further cooling.

 

[mtcn77]

I'm using the IC graphite pad with a Ryzen 5 3600. How does tight in the graph relate to a third party's heatsink mounting mechanism that is screw down? Would over tightening the two screws help at all?

Great question. I think it goes hand in hand with diy ihs lap project. They are great products, just not for the faint of heart. One has to go fully custom. Just wanted to point it out. I think this is a good showcase out of the products on display. It goes hand in hand with the on going discussion, too. They have their detractors, but are we overclocking or not - ic diamond even leaves scuff marks on the die itself. How more determined could you be to try such a compound...

Panasonic here is the 'graphite'. I haven't noticed it in the legend and have missed mentioning it myself, so there it is!