VOGONS


First post, by Jed118

User metadata
Rank Oldbie
Rank
Oldbie

I ran some benchmarks on these two machines, and I am a little intrigued, but not that surprised:

The 286 runs at 16MHz and has 1 Mb RAM, and a Seagate ST-157-A (which came out of below 386) and has a 512 KB Trident C-knockoff. I/O is handled by an integral 16 bit I/O ISA card that has built in COM LPT FDC and IDE. Overall this computer seems to have the more modern take on older hardware (Sound cards are identical in both systems though) aside from the older hard disk design.

The 386 also runs at 16MHz and has 4Mb RAM (via 36 individual DIP chips) and a Conner CP-3184 drive (which seems to have a weird delay - sometimes during POST it fires off a C: drive failure, but when you skip the RAM count it works. Every time. Also seems to have abnormally slow access times in certain tests, but performs *adequately* otherwise - No bad sectors and common DOS usage is not abnormally slow, but annoying. Still, 85MB > 43 MB) with an ATI VGA Wonder (originally with 256k, upgraded to 512k). I/O is handled by a 16 bit card, controlling just the hard disk and floppy disk. COM and LPT functions are handled by a Soviet-era 8 bit PCB. This computer is also equipped with an 80387.

These are the results: (TTX monitor = 286, Datatrain = 386)

xygtIwnh.jpg

GquSV3kh.jpg

Heading columns: CPU SPEED MHz Wait CRT BUS (Vid t.put) (BIOS t.put) Size MB (hdisk tput) (avg acc.) (T->T accs)

aq4xFCyh.jpg

Heading columns: CPU SPEED MHz Wait CRT BUS (Vid t.put) (BIOS t.put) Size MB (hdisk tput) (avg acc.) (T->T accs)

oBZCRboh.jpg

Just a side by side:

99WhPowh.jpg

I'm not really surprised, as the computer BIOS dates are one year apart, and the 286 board really seems like a tightly-integrated VLSI affair (still won't take SIPPS and DIP RAM together grrrrrrr could have had 2 MB) whereas the 386 board seems dated with a thousand discrete things. I'm not sure if that matters. What I can see is that the 286 seems to load era-correct games faster albeit slightly, than the 386. I've also noticed this difference between my P166 and PPRO200 - I think someone just optimized the 286 better.

The VGA All In One destroys whatever thing I slapped into the 286 though.

What do you guys think/can add from experience?

Youtube channel- The Kombinator
What's for sale? my eBay!

Reply 1 of 2, by xjas

User metadata
Rank l33t
Rank
l33t

Kinda hard to parse the results from your screenshots, but I'm not surprised that a late, mature, integrated chipset beats an older one with a bunch of discrete logic regardless of CPU revision. The 386SX was really just a gimped attempt to replace the 286 anyway (IIRC it was originally supposed to drop in to 286 boards.)

Wonder how a 486/16 with the cache disabled would stack up? 😜

twitch.tv/oldskooljay - playing the obscure, forgotten & weird - most Tuesdays & Thursdays @ 6:30 PM PDT. Bonus streams elsewhen!

Reply 2 of 2, by elianda

User metadata
Rank l33t
Rank
l33t
Jed118 wrote:

What do you guys think/can add from experience?

It is not much new information. You benched a 286 vs a 386SX with 8086 code. The 386SX is more or less the same speed, sometimes a bit slower than the 286. That is to be expected, even if you look at the cycles required for instructions considering the 386SX gets penalties for memory accesses due to the half width bus.
It would be more interesting if you compile a bench with 32 bit code where the 286 has to use multiple instructions to do the same operation. Also the 386SX can run 32 bit protected mode with paging as well as virtual 8086 mode what the 286 lacks completely.
Going back to DOS this means you can use a generic memory manager like QEMM386 or EMM386 to manage UMBs for loading TSRs to high memory as well as software using DOS Extenders such as PMODE/W, DOS4G, DOS4GW etc.

Retronn.de - Vintage Hardware Gallery, Drivers, Guides, Videos. Now with file search
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool