32 or 64 bits with old CPU

Antoine8 wrote:

Hello

My computer is old.

The CPU is AMD Sempron 145 (64 bits, but only 1 core)

It is used with Windows 10 32 bits.

Is it useful to switch to Windows 10 64 bits with such an old CPU ?

Will i get an improvement in performance or the opposite ?

I finally found a comparison of 32-bit versus 64-bit
for a single program - Firefox. And the 32-bit version
has the edge, which is strange. This would be an
example of one tool you might use on a daily basis
(although a lot of people use Chrome of course).

https://www.raymond.cc/blog/mozilla-...ared-to-32bit/

That article was first written in 2012, and has been
updated for various releases of Firefox. The current
web page reflects Firefox 42 32-bit versus 64-bit.

Now, what that result reflects, is when code is not
specifically optimized for 64-bit, and the code has
a lot of small counters and branch constructs. The 64-bit
version does not have to run faster in all cases.

The Quantum browser version (57?) could be quite different,
because they put a lot of effort into trying to improve
the performance, and eventually, they'll go 64-bit only.

Paul

Paul wrote:
Antoine8 wrote:

Hello

My computer is old.

The CPU is AMD Sempron 145 (64 bits, but only 1 core)

It is used with Windows 10 32 bits.

Is it useful to switch to Windows 10 64 bits with such an old CPU ?

Will i get an improvement in performance or the opposite ?

I finally found a comparison of 32-bit versus 64-bit
for a single program - Firefox. And the 32-bit version
has the edge, which is strange. This would be an
example of one tool you might use on a daily basis
(although a lot of people use Chrome of course).

https://www.raymond.cc/blog/mozilla-...ared-to-32bit/

That article was first written in 2012, and has been
updated for various releases of Firefox. The current
web page reflects Firefox 42 32-bit versus 64-bit.

RaymondCC

Jetstream

Firefox 42 32-bit 104.45 (Phenom II 3.2GHz X4 955 CPU)
Firefox 42 64-bit 105.74 (Phenom II 3.2GHz X4 955 CPU)

*******

So I ran that benchmark on my single core AMD 2.2GHz laptop.
I used a 64-bit OS, to compare 32-bit and 64-bit program
runs. The 32-bit program run on the 32-bit OS would run at
the same speed as the 32-bit program on a 64-bit OS. So
running these on the 64-bit OS is sufficient to compare
them.

I ran a set of tests, rebooted, and reversed the run order.
The cache was not cleared between runs, which means the very
first run could have been slower if the files hadn't been
in the cache already. And it looks like the benchmark
has properly handled when to time the benchmark so
that doesn't happen. The same profile folder is used
for both copies of Firefox, and the browser has only
seen the benchmark site when it runs the benchmark, and
not any other sites.

Firefox 57.0.2 32-bit 39.876
39.763
64-bit 42.408
42.987
reboot

64-bit 41.284
41.691
32-bit 40.828
40.495

At least for that benchmark result, with Quantum, you
wouldn't be able to tell the difference. And they
were even closer in Firefox 42.

It's pretty hard to guarantee the OS CPU activity
won't screw up the benchmark results, and if I was
serious about doing a bench like this, I'd have to
switch to WinXP or something :-)

Win10 is no good for benchmark work. No good at all.
A number of times now, I've had maintenance activity
start up, right in the middle of a benchmark run.
A Windows Defender update came in, just before I
rebooted for the above test. I had to wait until
things settled down, before running the second
test set.

When the results are that close, there's no
reason to get excited.

Paul