Question Rocket Lake cooling/turbo question

[mikeymikec]

I've built a couple of PCs lately with i5-11600 (65W TDP) and i5-11600K (125W TDP) processors. I'm aware of the general word that paying attention to TDP figures with Rocket Lake in particular is a bit of a fool's game and I've experienced it myself when attempting to use a Be Quiet! Pure Rock Slim/2 cooler (rated 130W by BQ): Prime95'ing the 11600K results in ~80C temps in a matter of seconds, going from very low idle temps (IIRC 30C or therabouts) and near-silent CPU fan speeds.

Prime95'ing the 11600 wasn't as bad but still climbed fast (the sort of climb I'd traditionally associate with an insufficient HSF), but it was evident within a minute or so that the temps would continue to creep up beyond 80C, but what gets me is that the 11600 comes with its own HSF which is clearly lesser in capabilities than the Pure Rock Slim 2 (it's the size I would have expected for say a Skylake i3, nothing special at all, though with a copper plate at the base; some Intel HSFs don't have that).

From what I saw, both CPUs were turbo'ing basically the entire time (e.g. 4.5GHz and higher) during the short period of stress testing, way more than I'd expect from say my Haswell i5 and older Intel CPU gens generally which in my experience would turbo for a few seconds at a time. Do Rocket Lake CPUs basically turbo until they hit some pre-defined thermal limit then switch off turbo until they've cooled down again?

 

[Geegeeoh]

Isn't a MoBo setting to keep turbo indefinitely?

 

[arandomguy]

Only SKUs with Thermal Velocity Boost (I think only i9 Rocket Lake's have this) factor in temperature in their clock algorithm (as additional higher clock rate). There is a safety thermal throttle point (off hand 100C? 105C? default, not sure) but it's much higher than what desktop "enthusiasts" are typically comfortable with temperature wise so essentially a non factor/consideration.

Otherwise the turbo boost methodology is based upon power limits and time limits. They'll basically boost as high as possible given the SKU configuration as long as those trip points aren't met. There are "stock" settings but motherboards settings can override them. Motherboards in newer generations have been more typically configured to have more aggressive limits, preferring performance over efficiency as a default. This should be adjustable in bios however if you prefer a less aggressive setting.

Since you bought up the stock HSF and why it's considered "acceptable." The stock HSF is almost certainly sourced based on default boost trigger points and "typical" workloads (it's something like 2x the "power consumption difference between something like gaming and Prime95 for Intel CPUs with uncapped power limits).

 

[coercitiv]

Do Rocket Lake CPUs basically turbo until they hit some pre-defined thermal limit then switch off turbo until they've cooled down again?

It's not the CPU making these decisions, it's the motherboard stock UEFI settings. Motherboard makers are under no obligation to respect stock TDP settings, hence they set the highest power limits that their VRM can handle, and sometimes even disable those timers you were accustomed with during the Haswell era (called "Tau").

Therefore, if you want to have control over 11600(K) boosting & temps I suggest manual adjustments to PL1, PL2 and Tau, otherwise you'll have to rely on thermal throttling at very high temps. They should all be available in UEFI.

 

[mikeymikec]

I found this setting in the BIOS manual for the H570M-plus:

Dual Tau Boost
Allows you to enable or disable Dual Tau Boost feature. This is only applicable for
Desktop 35/65/125W sku. When DPTF is enabled, this feature will be ignored.
Configuration options: [Auto] [Disabled] [Enabled]

It looks like the one I'm potentially interested in, and perhaps I'd disable it if the typical usage involved extended periods of near-complete CPU saturation.

 

[dullard]

1) Like others have said, this should be adjustable in your motherboard settings.

2) Yes, if given sufficient work, it will turbo as much as possible up until it reaches a hot temperature.

3) Because of #2, temperatures don't matter: chasing low temperatures is the fool's errand. This is because it will keep going into turbo if temperatures are low, then when temperatures are high the turbo will turn off. The goal of the system is to give the best performance which has the effect of keeping temperatures higher than you are used to.

4) Because of #2, you will get high temperatures with a bad heat sink and a good heat sink. The difference between a bad heat sink and a good one is how much turbo you get, not temperatures. This is a shift from what you might see in much older CPUs.

5) Copper plates helps spread heat out more evenly, in case you had one core working hard and other cores idle. But the plate itself doesn't actually do anything for cooling capability: that is all fin design and air flow.

 

[DrMrLordX]

Isn't a MoBo setting to keep turbo indefinitely?

Yes, if you change tau to 0 it's effectively indefinite. Some motherboards come that way @ stock. Intel spec is 56s if I recall correctly.

@dullard

He can manipulate PL1 and PL2 values and/or tau if he wants lower temps.