Archive for category Uncategorized
Added a small fix for video cards that are so slow it takes the vidram metric more than 10,000 usecs to finish (by comparison, the PCjr takes about 4,000). This was necessary to add a russian system to the database that has some serious speed issues 🙂
As the title says, this is just a yearly update with newly-submitted systems. One interesting system is the “FrankenTandy”, a Tandy SL that had its CPU and crystal swapped out. For a system originally built on an 8086, it’s now faster than a 286.
New version released with a bugfix and more systems added. 242 systems total in the database, including some very odd beasts including a PCjr running an FPGA instead of a CPU, and an XT upgraded all the way to a 486.
For TOPBENCH’s 5th anniversary, I thought I would out the easter egg I snuck in around version 0.33: The Star Benchmark. In the 1990s, I always thought it would be cute to see how many 60Hz 3-D stars would be possible on a system, but I didn’t implement it until TOPBENCH. To see how many “stars” your system can produce, use “starbench” as the only command-line argument:
…and your system will display increasingly more and more 3-D stars until doing so would cause flicker or slowdown. After viewing the display a bit, hit a key and the number of stars your system could produce will be displayed. If the number is followed by “(maximum!)”, then congratulations, your system “wins”. (There is no prize for winning!)
Thanks to many helpful people, there are now 200+ systems in the database. Head to the downloads page to snag the newest version.
I found a way to work around 8254 (and clones) bugs that sometimes return 65535 if an interrupt occurs while trying to read values, so with that, I’d say that TOPBENCH is now officially a beta. Woohoo! Head to the Downloads area to snag it.
After a long coding session, the full tool is now in a state that I feel comfortable to release for testing. Head over to the downloads section to grab it.
After resuming coding, I’ve gotten as far as datastructure handling, a method of comparing systems against each other that is reliable, and data/file handling routines. I had forgotten how it is simultaneously fun and irritating to have to code every single thing yourself. Need to manage an in-memory database? Brush up on your balanced binary tree knowledge from college. Need to manipulate .INI files or export to .CSV? Guess what, you’re writing all that drudgery yourself. Ugh.
In any case, I’ve got a console/text-only utility that can do all of the above, so I was able to import all 80+ contributed system entries into a single database, and it’s available on the Downloads page.
The next major chunks for me to write are the realtime benchmark/display portion and the ability to delete and edit systems in the database. After that’s done, all it needs is a halfway-decent interface, but that comes last.
I’ve begun development on the main utility, finally. I have eschewed any hopes of becoming a Turbo Vision master in less than a few months and will just throw a quick’n’dirty interface together to Get The Tool Out There. It will have full functionality, just not be as elegant behind the scenes and pretty in front of them. More in a few days.
While the main TOPBENCH full tool is not yet available, the helpful folks over at the Vintage Computer Forum have been helping me test the base metric code, and already some decades-old questions for me have been answered. Comparing (by hand, for now) over 50 submissions, I’ve been able to determine:
- Yes, a 386DX-40 was faster than a 486sx-25. (Who was faster was claimed by both sides at the same time, so it’s nice to finally know the truth.)
- A stock IBM PCjr is roughly half the speed of an original IBM PC. But you can double the PCjr’s speed by getting a memory expansion (the expanded memory does not have wait states), and you can double it further by replacing the 8088 with an NEC V20 CPU, and upgrading the crystal so that the system runs at 8MHz. So a fully-modded PCjr is twice as fast as a PC. (Wouldn’t it be incredible if we could do that with modern machines? $20 in parts and an afternoon, and your 3GHz Core i7 could now suddenly run at 6GHz?)
- Forgetting to turn the cache on in a 386DX machine could cut the system’s speed in half.
- Replacing CGA with VGA could, with fast VGA cards, increase graphics speed by 50%.
I hope to have an alpha version of the full tool ready in the next few weeks.