Re: Raspberry Pi 3 & 3+ speed issues
John Klos <[email protected]> Fri, 31 Jul 2026 20:47:15 +0000 (UTC)
| Newsgroups | gmane.os.netbsd.ports.arm |
|---|---|
| Message-ID | <[email protected]> |
>> So they're basically both running at the same speed. At 600 MHz, they both
>> take about 24.5 seconds to count the first million prime numbers, meaning
>> that the Pi 3+, which says it's running at 1400 MHz, is really running at
>> 1200 MHz.
>
> Maybe another test shows a difference. If that 'primes' program is
> a common sieve, it will be limited by memory speed, and that is the
> same for pi3 and pi3+.
It counts primes in cache on most CPUs, so I use it to gauge clock speeds,
even "turbo burst" speeds, by comparing it with results from known clock
speeds. It's good to ask, and I should've mentioned the kind of counting,
but even if it were memory bound, the count running at 600 MHz wouldn't be
exactly half the speed of it running at 1200 MHz :) [1]
dry2 of 10000000 runs:
Microseconds for one run through Dhrystone: 0.3
Dhrystones per Second: 3436426.0
Microseconds for one run through Dhrystone: 0.3
Dhrystones per Second: 3424657.5
That's the same, too.
Does anyone out there with a Raspberry Pi 3+ want to check their results
at (supposedly) 1400 MHz?
John
[1]:
/* Find the nth prime.
John Klos, [email protected] */
#include <stdio.h>
long a=1000000, b=4, c=7, d, e, f=4, g=4, h=4, i=9;
int main() {
while (b<a) {
c+=f;
d=1;
f=6-f;
g=4;
if(e<c) {
h++;
e+=i;
i+=2;
}
if(e==c)
d=c;
while (d<h) {
d+=g;
g=6-g;
if (!(c%d))
d=c;
}
if (c!=d)
b+=1;
}
printf("%ld, %ld\n",b,c);
}