Verify logic: low base clock speed - CPU boosting = CPU Longevity?

Can Someone verify my logic here. Planning on buying a new laptop in the coming weeks - and expensive one, hoping it will last a long time.

I'm planning on buying low base frequency CPU 1.8 - 2 GHz and turning off overclocking/boost to increase the longevity of the laptop. The PC will have daily medium load of the non-gaming variety

Or should I be thinking higher base clock speed without CPU boosting??? Help!

For the most part, you want your processor to be running at that slower speed. -

The 2 I'm currently looking at:

AMD Ryzen 7 5700U 8C / 16T 1.8GHz Up to 4.3 8
AMD Ryzen 7 4700U 8C / 8T 2.0GHz Up to 4.1GHz 7 Radeon Cores

To confuse things more a colleague mentioned that any laptop you buy today won't be limited in lifespan by the CPU. Is this accurate?

Modern processors have a lot of power saving features. Boost is one of them, others include power gating to shut down unused parts of the CPU when not in use.

The boost feature of modern CPUs allows systems that are mostly only lightly loaded to get rid of transient tasks as quickly as possible and thus get back down to their lowest power state as quickly as possible. Modern CPUs might have a 1.5GHz "base" clock speed, but can boost all the way up to 3.5GHz depending on thermal and power budgets, in theory over two times the theoretical power.

With boost disabled you will only even see the 1.5GHz speed, so whenever a task happens it mean that the CPU could be running it for twice as long. While it may be a similar amount of work and generate more heat to boost higher, the thermals means that a short high clock speed may be able to be soaked up in the short term and dissipated over a longer period than a more prolonged lower peak clock speed generaing heat over a longer time. That the CPU will be "fully on" for a longer time can also generate more heat.

The theory is then that you boost higher for a shorter length of time so you can fall back harder and shut down parts sooner to work down to a lower average temperature and power budget.

This principle was championed by Intel some years ago as "Race to Idle" or "Race to sleep" (as parts of the proceesor can sleep during idle).

That AMD and even ARM have versions of this kind of "low base clock, high peak clock, moderate thermally soaked clock speed" should tell you what the manufacturers think of this approach.

Lower average temperatures can extend lifetimes of more than just the CPU.

Clocking fast and letting the heatsink soak up the heat burst can mean you don't actually need to spin the fan as the average heat hasn't had time to increase that high before you shut back down. A longer period with more parts left on could mean more heat soaking into the heatsink and towards a thermal sensor that triggers the fan speed increase. Lower fan speeds (or even fans turned off) can mean they live longer too.

I've had CPUs that far outlived any mechanical part of a computer. Fans die a lot sooner than solid-state silicon in my experience, and I've had mechanical hard drives work themselves to death while the CPU happily chugged along doing it's job long after the hard disk suddenly vanished.

Sure, if your CPU is going to be running flat out 24 hours a day, 7 days a week, 365 days of the year then you might care about longevity and wear and tear, but it is far more likely to be struck by mechanical or other component failure before the actual CPU.

If all you are doing is browsing the internet and small bursty tasks then just leave your computer to do what the engineers set it up to do. Even if you had a task using the CPU at 100% then allowing boost could mean it simply finished in 4.5 hours instead of 5 as boost will let it use extra thermal headroom if it is there.

For normal use, I don't see a real reason to disable boost unless you know it causes problems for your workload, or it causes fan noise you don't want.

Limiting boost can have a benefit in limiting performance and overall heat output, so that might be the only time I'd personally consider disabling it. If I didn't care how long something took, but wanted "cool and quiet" then that might be a time to disable it.

There isn't really any evidence to suggest that disabling all this dynamic adjustment of boost and power saving actually extends processor life in any meaningful way. Silicon switching is not particularly harmful to the silicon itself (flash memory is a special case) and the most harmful thing will be heat, which is managed via thermal sensors and limits based on those devices.

Laptops have been known to limp along for years with clogged or broken CPU fans causing thermal throttling but remaining essentially functional. The CPU itself often long outlives mechanical parts like fans and thermal paste. So long as nothing sudden or catastrophic happens CPUs are quite hardy devices.

The simple fact is that the modern CPU is always turning itself on and off, hundreds of times a second. Blocking boost does not stop power saving modes unless you also disable C-states (more below) and then you will actually be generating more heat.

Intel processors have what are called C-states. Essentially C0 is "fully awake and running" and C7 is "taking a heavy nap". The C-states are what the CPU runs up and down in and while C7 is essentially a core shutdown, it is ready to start back up at a moments notice. It is softer than a sleep/shutdown as the core can be brought back up relatively quickly, but consumes far less power than an active core.

From Intel Energy Efficient Platforms Whitepaper there is this image on page 15 showing the rough idea of what happens in each state

Along with

Typical Core level C-states supported are:

• C0 – Active state executing code
• C1 – Halted, snoops serviced
• C3 - Core (L1/L2) caches flushed
• C6 – Core state saved and Core voltage reduced to ~0
• C7 – When last core enters C7, LLC is flushed progressively

Typical Package level states supported are:

• PC0 – Active state
• PC1 – Low latency state
• PC3/PC6 – LLC ways valid, retention voltage
• PC7 (Deep Power down) – LLC fully shrunk, No snoops, aggressive package power reduction
• C2 Popup – For Bus Master Traffic

And sure enough, in HWinfo you can see cores entering C7, and the package occasionally fully entering C7 and essentially shutting down for a moment. Some caches will be valid, but the cores will be effectively shut down and in minimal power mode. Ready to run, but waiting with only half an eye open.

When running a benchmark you do see cores fully enter C0, on my system if I run a single core benchmark I get Core0 at full speed with the other 3 cores mostly shut down. This is where the power saving is that allows single-core boosting to go higher than full system loading.