I think it's reasonable to think of the processor in the Pi 5 as 58 times faster than the original Pi. At the same time, total system performance depends not only on CPU but also the speed, latency and size characteristics of memory, cache, storage, network and graphics among others.Possibly. I took what I recall were "N times faster than the original Pi" and "N times faster than the latest model" announcements."rounding down error" - is it because the Pi2B (2014) is missing?
This is my full data set -I don't recall the 3A+/3B+ being said to be 12 times faster than originals, nor 4B being 12 times faster than 3A+/3B+.Code:
.-----------. .----.| A/B/A+/B+ |--.-> x 3 ->| 2B |`-----------' | `----' | .----. }-> x 4 ------>| 3B | | `----' x 18 x 45 | .---------. .----. .----. }-> x 4.5 --------->| 3A+/3B+ |-> x 4 ->| 4B |-> x 2.5 ->| 5B | | `---------' `----' `----' | .---------. `-> x 4.5 --------->| Zero 2W | `---------'
I suppose I'll have to trawl Google to see what multipliers l can find stated at the time of release.
For example the network speed of a Pi 5 is exactly the same as a Pi 4 and about 10 times faster than the original Pi.
The stream copy test shows 738.7 MB/sec for the original Pi memory bandwidth
viewtopic.php?p=1643919#p1643919
and 12305.7 MB/sec for a Pi 5.
viewtopic.php?p=2245109#p2245109
That is a 16.6 times improvement.
While it's now possible to use an m.2 SSD with the Pi 5, people using SD cards have likely seen a less than 10-fold improvement in storage speeds.
Bit banging the GPIO on a Pi 5 may be slower than doing the same on any other model of Pi.
On the other hand, since 2GB is less likely to swap than the 0.5GB of the original, the thrashing that can happen when a system runs out of RAM can sometimes be avoided.
As with all computers, one has to test the exact use case to know what the relevant speedup is.
Statistics: Posted by ejolson — Mon Aug 19, 2024 5:25 pm