As you would expect from something with a standard BIOS I had no problems running Tiny Core 7.2 from a pen drive.
Out of interest I then fitted an 8GB SSD and installed Tiny Core 7.2 to that. I then proceeded to compile the Tiny Core kernel on the D-class machine - something I'd done previously on the Z-class hardware.
Z90D D10DP real 43m 23.11s 49m 16.16s user 41m 21.27s 46m 56.87s sys 1m 46.47s 2m 03.65s
So, as one would expect from the different clock rates (1.65GHz vs 1.4GHz), the Z-class is a little faster than the D-class!
In April 2020 I tried out an 8GB RAM part, and then a 4GB RAM part. Both of these were correctly identified by the BIOS but Tiny Core 11.1 reported there was less than 3GB of RAM available which surprised me. That the figure was the same in both cases (4GB/8GB) was no surprise as my workhorse is the 32-bit version version of Tiny Core and hence cannot address more than 4GB of memory.
I did program a couple of other USB flash drives with BionicPup32 and BionicPup64. The 64-bit OS behaved as expected seeing almost all the available memory.
Comparing Tiny Core (non-PAE kernel) with BionicPup32 (PAE Kernel) I got the following figures:
OS Fitted
RAMVideo
Buffer sizeReported
MemoryBionicPup32 4GB 64MB 4,048,544 Tiny Core 64MB 2,692,856 BionicPup32 256MB 3,849,056 Tiny Core 256MB 2,497,784 BionicPup32 2GB 256MB 1,772,384 Tiny Core 256MB 1,717,496
From this you can see the advantage of a 32-bit PAE kernel when you are close to the 4GB addressing limit. In this instance BionicPup is able to map the memory-mapped IO space outside of the 4GB RAM space and so only loses the memory commandeered by the video buffer.
Tiny Core loses out as the memory-mapped IO space has to overlay some of the existing 4GB memory address range. Having said that I'm still surprised by the amount of memory address space it has lost.
Drop the fitted RAM to 2GB and then the difference is small as the non-PAE kernel has 2GB of unused address space that it can map the memory-mapped IO addresses into.