PDA

View Full Version : Xedex Baby Blue CPU Plus speed + compatibility testing



DeathAdderSF
August 18th, 2017, 11:33 PM
Hello, all.

I acquired a Xedex Baby Blue CPU Plus a while back, which is an 8-bit PC ISA card that allows CP/M-80 software to be used from MS-DOS. Essentially the card has a Z80 chip on it, and a small header is added to each CP/M executable to tell MS-DOS to turn control over to the Z80 chip. I personally find it fascinating that such a thing exists, especially given the existence of the NEC V20 and various software-only solutions to the problem.

Anyway! My card works fine with the limited amount of CP/M-80 software I've thrown at it, but I'm wondering ...

1). Are there any tools for CP/M that allow something like benchmarking of the Z80 CPU? I'd like to see how it performs in realtime.
2). Is there any CP/M software that uses any "dirty tricks," i.e. does some "nasty" (unexpected/undocumented) things that may throw off a piece of kit like this, that's not actually running CP/M as the host OS?

There's barely any info on this card online, at least from the searches I've done, so I'm interested to do what research/testing I can do on my own.

JonB
August 19th, 2017, 12:23 AM
1) See if you can find out what the clock speed of the Z80 is. If it is 4Mhz, performance should be on a par with pretty much any similar CP/M machine. What aspects are you hoping to test? CPU execution speed? Disk access timings?

2) CP/M has a standardised internal API (called the BDOS) and a second, lower level one (the BIOS) so most software should "just work", providing it interacts excusively with these - with perhaps configuration required to match the terminal emulation. Meaning by that, programs like Wordstar do things like setting the screen cursor position using escape sequences, and you may need to set it up with codes matching the PC's escape sequences. This was all part of the CP/M experience.

Some software is a bit machine dependant but I can't recall any specific examples other than some games written for machines like the Amstrad 8256 that had bitmapped graphics. CP/M was really built to interact with a character based I/O device like a terminal, so it doesn't natively support bitmapped displays.

Anyway, check this out:

https://groups.google.com/forum/#!searchin/comp.os.cpm/benchmark%7Csort:relevance/comp.os.cpm/GTbD0tmDQNA/9oGrS9CEpfEJ

Particularly this: https://groups.google.com/forum/#!searchin/comp.os.cpm/benchmark%7Csort:relevance/comp.os.cpm/TkWxSWRO00E/s1f8U4ugoSgJ

I found this program that might work : http://kildall.apana.org.au/~cjb/CPM/ - it appears to be an ASCII mandelbrot generator. Another common benchmark was Dhrystone and there's an interesting comment from one of the members: "I got either 180 or 200 Dhrystones per minute on a 4 MHz Z80 (Kaypro 2X)". The context is in this thread: https://groups.google.com/forum/#!searchin/comp.os.cpm/benchmark%7Csort:relevance/comp.os.cpm/Xym_4irqR6o/GB-QPf2-cikJ

Wikipedia has a good article on the Dhrystone benchmark here: https://en.wikipedia.org/wiki/Dhrystone

There are links to example implementations (one in C that you should be able to build, if you have a C compiler on your CP/M machine).

d5aa96
August 21st, 2017, 01:12 PM
Hello, all.

... I personally find it fascinating that such a thing exists, especially given the existence of the NEC V20 and various software-only solutions to the problem....



Having worked a large IBM PC dealer (ComputerLand) in 1983, I can tell you 1st hand that the Baby Blue was very popular at the time. The BB card was faster than the software-only solutions, and the BB card's RAM was usable by MSDOS. The folks who would spend several thousands of dollars on a IBM PC system had no problem with the cost of the BB card. If the V20 CP/M solution was available at the time, we were not aware of it. I did not start seeing V20's for sale until 1986 or 1987.

glitch
August 21st, 2017, 02:11 PM
Good to know there's actually copies of the DOS utilities out there, I picked one of these up at VCF East this year!