First post, by 386_junkie
The Systempro Project
1) Beginnings (The modified System): - 386_Junkie’s duo core 386; The modified Systempro!
2) Build (The Systempro): - <----------You are here
3) Bios: - The Systempro Project; The Bios
4) Bugs: -
5) Benching (386 class): - The Systempro Project; Benchmarking (386 class)
6) Benching (386 class) round 2: - The Systempro; Benchmarking (TI486SXL2-66)
7) Benching (4/586 class): -
Since the thread I did earlier in the year where I salvaged and bought a number of Systempro parts over the last 18 months and assembled them in a modified Zenith case, I have moved to a new city with a new job and found this: -
Quite incredible after putting in all the time and energy into making a modified Systempro... and then goes the universe and puts one right in front of me!
It was not in this condition when I received it of course, and with getting lucky with the weather over the weekend that I took her home... I went ahead and stripped out the 486 insides and gave her a good clean. She was loaded inside with all sorts of goodies... full complement of memory, original Connor HD's, EISA graphics card.
Still, with all the spares I have floating about... I of course made her into another 386 to play alongside her little sister. Will be bringing one of the girls back down to where i'm working... i'm in a new city, so it'll be good to have some computer company!
More to follow...
It's 18/09/16 and time for an update…
Since writing last I’ve managed to escape from work long enough to play with, and conduct some tests. I have also taken some memory from a Deskpro I have and upgraded the system to 80MB DRAM (10 x 8MB sticks) all 80ns.
I have a couple of faster sticks shown below which I have swapped over into the Deskpro, there is little point in having only 2 or 3 (60ns / 70ns) sticks being faster than the rest… it made sense to use them in the Deskpro which has lower memory demands.
As with the modification system I did... I again upgraded the FSB = de-soldering the original 66.66MHz crystal and solder in a new DIL socket so that crystals can be interchangeable depending on system reqs. In this case, I have inserted an 80MHz crystal to bring the FSB up to 40MHz.
So, the tests. Before I go on I should say that the normal benchmark software most folk here on Vogons use do not work unfortunately under Windows NT 3.1, only DOS. Multiprocessor performance is only possible under a supporting OS which DOS is not. It is because of this that the DOS benchmark tools will only see and perform against one of the 386 CPU’s. If anyone can recommend any Windows benchmarks that are likely to run under NT 3.1… I’m all ears.
The results from the short lived tests using a 100MHz oscillator to give a 50MHz FSB are below; -
Back in goes the 80MHz oscillator...