Windows reports two CPU speeds because one would be too simple
- Reference: 1747844893
- News link: https://www.theregister.co.uk/2025/05/21/chen_windows_cpu_speed/
- Source link:
The behavior can be found in the System > About section of the Settings app where Windows displays some basic hardware information such as the installed RAM, the name of the device, and the CPU.
However, there can often be two numbers for the CPU speed. In Chen's example, the first number is 3.60 GHz and the second is 3.71 GHz. Why?
[1]
"This field," [2]he explained , "is actually showing two pieces of information, one after the other." The first is the processor brand string, which, in Chen's example, reads "Intel(R) Core(TM) i7-4790 CPU @ 3.60GHz", and the second is the speed calculated by Windows, in this case "3.71 GHz".
[3]
[4]
The number in the brand string could be anything; it's lifted from the CPUID function codes 0x80000002 through 0x80000004 , which return arbitrary strings.
[5]The 12 KB that Windows just can't seem to quit
[6]The passive aggression of connecting USB to PS/2
[7]Why did the Windows 95 setup use Windows 3.1?
[8]Microsoft vet laments a world where even toothbrushes need reboots
"Some processor manufacturers include a CPU speed as part of that string," he said. Hence that "@ 3.60 GHz" suffix. However, "Windows itself does a rough calculation of the CPU speed and shows it after the processor brand string."
Chen put forward some suggestions as to why a manufacturer might include the number in the string. "Perhaps they do it to make it easier to detect overclocking," he said, "or systems being marketed as faster than they really are."
Or perhaps Windows engineers are as cynical as the rest of us when it comes to claims made by some hardware manufacturers.
[9]
We asked Intel why the number in the string did not always match the value that Windows came up with, and will update this piece should the chipmaker respond.
Microsoft would prefer users look to Task Manager to get an idea of CPU frequency. Clicking on CPU in the Performance tab will give the base speed alongside whatever the CPU is doing at the moment (which can fluctuate above and below that base speed depending on factors such as the current workload). ®
Get our [10]Tech Resources
[1] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aC5M_N2VQXiXubhiu0cKYgAAAlM&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[2] https://devblogs.microsoft.com/oldnewthing/20250520-00/?p=111201
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aC5M_N2VQXiXubhiu0cKYgAAAlM&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aC5M_N2VQXiXubhiu0cKYgAAAlM&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[5] https://www.theregister.com/2025/05/08/moricons_dll_raymond_chen/
[6] https://www.theregister.com/2025/03/27/raymond_chen_usb/
[7] https://www.theregister.com/2025/02/17/windows_95_windows_three_point_one/
[8] https://www.theregister.com/2025/02/02/raymond_chen_restarts_updates/
[9] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aC5M_N2VQXiXubhiu0cKYgAAAlM&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[10] https://whitepapers.theregister.com/
Re: The explanation is simpler, and far less sinister
Addendum, because I missed the edit window.
Just to be clear, what I wrote above is the explanation for the seemingly-random second speed that the Task Manager displays.
In other words, what your system has might be a "Intel(R) Core(TM) i5-11520K 3.8GHz" (numbers made up because I can't be bothered to pick a real example from ark.intel.com); that is the brand string returned by the CPUID instruction. The current speed of the processor, averaged across cores, is appended to that, giving something like "Intel(R) Core(TM) i5-11520K 3.8GHz 800MHz "; a moment later, as system load increases, it might be "Intel(R) Core(TM) i5-11520K 3.8GHz 1400MHz ", for example.
It's GiHz, which is 1024 MiHz, which in turn ... you know the drill
I dropped my CPU, it cut my leg and landed on my foot. Two megahurtz
Well at least the CPU packaging didn’t fall apart in the process, exposing the silicon; you wouldn’t want to die on the floor.
> Well at least the CPU packaging didn’t fall apart in the process, exposing the silicon
One of my first jobs was maintaining Apple ][ s. The motherboard RAM could often be fixed by pressing on every DIP chip, cut the tarnish on the pins. Magic fingers. These ][s had external floppy drives. One was sick. I opened the case to apply magic fingers...... hey! That's pretty! Like a tiny mirror inside the epoxy chip. This was the step-motor driver chip, hard work, it had overheated and blown its lid. Naked Silicon! So long ago that I got the chip (ULN2004) at Radio Shack and it worked forever after (or until I left that place).
FWIW, the ULN2004 is still in Full Production, 47 years later.
The stock frequency listed is one thing, but the frequency reported in your OS will vary with power management governors, usage conditions with "turboboost" and of course user settings in BIOS etc.
Say what you want about Windows (hatred!), but this is just people not understanding how things work. My 5.2 GHz processor cores are reporting "800 MHz" in my OS right now. If I start a compile job they'll ramp up to maximum because I have overridden turboboost ratios.
CPU's speed fluctuate
News at 11.
As much as I utterly loathe M$, the explanation is correct.
I thought..
Didn't we get over this clock frequency fixation at least 20 years ago?
In my youth I was most happy that my Z80-powered rig motored at 4MHz, whereas a BBC B slogged along at a solitary single megahertz. And yet the B could do some things a bit better than my Einstein. And thus I learned that clock speed is close to useless in judging performance.
-A.
Re: I thought..
Yep, hence the (very welcome) move about 20+ years ago, away from fixating on the clock speed as the be-all-and-end-all of processor performance. Which would you rather use, a 3.6GHz “Prescott” Pentium 4, or a 2.0GHz Core i7? On thread count, IPC, performance per watt, cache size, the latter trounces the former, and that’s even before you start looking at hardware video encoding/decoding and other specialist functions now commonly included in even base models. And Prescott, for the young’uns, was the one that pushed the P4P microarchitecture to its limits and as a result needed a heatsink the size of Yorkshire.
Clock speed is an irrelevance. In around ‘96, I built my first PC, with a Pentium 75MHz; with judicious and nerve-wracking adjustment of jumpers I was able to overclock it by a colossal and impressive-sounding 62%… or, more prosaically, a now-rather-piffling 45MHz. What’s 45MHz these days but a rounding error? If your shiny new multi-gigahertz CPU ran 45MHz less than its nominal speed (never mind the aforementioned built-in opportunistic underclocking) would you even notice? Or care?
CPUs don't have a "speed" anymore
They vary their speed based on load, heat and other factors. The best you can say is the "maximum" some of its core may run at, but you aren't guaranteed to get all the way there and depending on the CPU, board and cooling solution its possible you might get a little better than that.
The explanation is simpler, and far less sinister
FTA: Chen put forward some suggestions as to why a manufacturer might include the number in the string. "Perhaps they do it to make it easier to detect overclocking," he said, "or systems being marketed as faster than they really are."
Far be it from me to contradict Raymond Chen, whose metaphorical tea I am not worthy to fetch, but the explanation for the two speeds is that modern CPUs rarely if ever run at their "nominal" speed. At least in the x86 world, most have built-in, opportunistic overclocking and underclocking.
As an example of the former, suppose that a two-core CPU is running one very demanding thread, maxing out one core and leaving the other idle. The CPU will choose to boost the speed of that busy core temporarily above the "maximum" speed; as long as the total power envelope of the package isn't exceeded, no harm done.
On the flipside, why keep a 3.6GHz CPU ticking along at that speed if it's running nothing more demanding than Notepad? Dropping the frequency of underused core(s) down to a bare minimum to save power consumption makes sense, whether on battery or AC platforms.
Intel calls this combination of behaviours Turbo Boost & SpeedStep - probably with lots of (TM)s and (C)s and (R)s, this being Intel - and it's been around for years; certainly since latter generations of the Pentium 4. And I'm sure AMD has an equivalent.
A real-time CPU frequency monitor like CPU-Z will make this clearer and show the actual speed(s) as processor load changes.