Like most places, we have an “expert” in our village. This morning in a group discussion he said I have a really fast computer he had upgraded from 32 bit to 64 bit by changing the processor, bells started to ring for me as he went on about how he first updated the memory on his 32 bit from 8 to 16, but was not happy with performances so then changed the processor to a 64 bit, but now had some issues so was looking to get a 128 bit processor to change his 15 year old windows 7 to run 11. Was I interested in doing it for him as he was really busy.
I tried to explain
32 bit only addressed 4 gb memory no more (and not all of that)
64 bit is not twice as fast as 32 bit also uses bigger addressing structures so needed more.In fact, some 64-bit applications are slower because data structures tend to be larger in 64-bit applications (again, due to the larger pointers) and moving them around can slow things down. 64-bit uses approximately 200 - 500 MB more ram at idle than 32-bit
When we got to 128 bit processor decided he had no idea what he was talking about. I did check in case I had missed something There are no 128-bit CPUs on the market and there may never be because there is no practical reason for doubling the basic register size.
Dont think even windows 11 will run on it.
He did not change the motherboard just the intel chips !
Occasionally I listen to these conversations, add my little bit but get talked over as the English dont understand……
That does not add up. Motherboards support specific families of processors.
Finding out which component of a computer is the bottleneck in terms of speed is really difficult. The data-bus on the motherboard used to commonly be a limitation.
To cite an English nursery rhyme, you might have advised him to pull your other leg because that’s the one with the bells on.
This is as good an excuse as any to recall that some of the computers that come in for recycling to Linux date from 2007-2010, when SATA disks became common, though most need a 32 bit OS. The ones that still work are often business machines of a certain quality. There are also some more recent Atom-based ones that can’t take 64 bit Linux.
I’ve used Debian-based Emmabuntus, MxLinux and Q4OS, but can’t decide which to learn to the point where we would be able to support potential users. You can sometimes install extra RAM recovered from scrapped machines, and if you don’t break something while opening the case, a 25 euro SSD makes them go nice and fast.
Think he was totally incorrect but loud “experts” know more than me.
Occasionally I go down a track, but then question myself. Used to teach this stuff but its some time back and although i try hard to keep up not always possible.
Occasionally I have installed a 32 bit system of linux on av64 bit machine because it was over heating so showing it down not using the full data bus etc has worked out. But never run 64 or 32 bit will not install and changing a processor on an old motherboard except to see what happens is a recipe for disaster as well as expensive
I have never owned a 32 bit PC or Laptop. Always 64 Bit and my first ever one only had 2GB of Rammage. Ran slower than a snail, but back then spinning drives were the norm. I have a Atom Windows 8 Tablet Laptop, with a 64 bit processor, but the UEFI is 32 bit. I don’t think there is any Linux on the market that runs a 32 bit UEFI?
Microsoft made damn sure that nothing else was going to get installed on it. It is put away in a cupboard, inside it’s original protective case, along with it’s charger, which cost me a further £70.00 pounds, as the original was missing when I bought it at a fund raising event, outdoor fair thing, back in 2017. I tried Peppermint OS 10 on it, a version that Mark Greaves wrote, for extraordinary, weird setups. Unfortunately it never worked. It installed, but never booted.
I would not write Mint off.
There is also Peppermint. It is a nice clean minimal setup.
I think your biggest decision might be the DE rather than the distro.
I mentioned in another thread about using RFEfind on an old iMac with a coreduo (not core2duo) - so that (coreduo), and the firmware - restrict to running only 32 bit O/S.
But - you can update the firmware, and replace the coreduo CPU with a pin compatible Core2Duo and it will go 64 bit…
32 bit kernels can address more than 4 GB with PAE…
Also - pretty sure 32 bit Linux (and UNIX etc) could address the whole 4 GB (I know Solaris x86 could - and I’m sure when I ran 32 bit Linux on a 4 GB machine it saw everything) - but the 32bit Windows kernel restricted addressable use to a max of 3.5 GB (and I think by default it was only 3GB, so you had to hack the registry or something to get 3.5 GB)… Why? Did Redmond (i.e. Microsoft) get nostalgic for the 640 K memory limit?
Lack of vision. They could not see beyond the ram requirements of their own OS. They had no idea that someone might want to run a C program that addressed a large array.
Even Linux has a soft limit… you have to use ulimit to bypass it.
Windows version dependencies
Main article: Physical Address Extension § Operating system support
The “non-server”, or “client”, x86 SKUs of Windows XP and later operate x86 processors in PAE mode by default when the CPU present implements the NX bit. Nevertheless, these operating systems do not permit addressing of physical memory above the 4 GiB address boundary. This is not an architectural limit; it is a limit imposed by Microsoft as a workaround for device driver compatibility issues that were discovered during testing.[18]
Thus, the “3 GB barrier” under x86 Windows “client” operating systems can therefore arise in two slightly different scenarios. In both, RAM near the 4 GiB point conflicts with memory-mapped I/O space. Either the BIOS simply disables the conflicting RAM; or, the BIOS remaps the conflicting RAM to physical addresses above the 4 GiB point,[citation needed] but x86 Windows client editions refuse to use physical addresses higher than that, even though they are running with PAE enabled. The conflicting RAM is therefore unavailable to the operating system whether it is remapped or not.
I seem to remember - by default - 32 bit Windows server systems could address up to 3.5 Gb - but Windows XP was restricted to a hard limit of 3 Gb… I usually keep an XP VM around (for nostalgia) - it’s perfectly happy on 128 MB anyway…
That3_the stuff i used to teach with memory addressing limits
Never seen or heard of this being done.
If we step back further in time to msdos and the 760 limit, at that time I was teaching lotus 123 and needed extended or expanded memory (cannot remember now which) to get to the next limit 1024, and it was a real pain to configurer correctly. 123 had a add in to make it more WYSIWYG….. but that 40 years ago hence my memory faided
