• Please review our updated Terms and Rules here

Compaq Deskpro 386/16 Restoration

Great job on restoring that DeskPro! IBM must have been pissed that Compaq beat them to the punch on a 386 machine! :D It's hard for me not to laugh a little imagining that.

Those old DeskPros look to be built really well. They look very nice too; styled just like the original Compaq Portable. As much of a Compaq fan as I am, I really can't believe I don't have one yet (such as an 8086, 286 or 386). If all works out though, I should be getting an 8086 soon. That's a machine I've wanted for years. I think it's so funny that in response to IBM making a portable (the 5155) to compete with Compaq's own portable, Compaq said "well, we can just make a desktop then!" and released the original DeskPro 8086 in 1984. And, as usual, Compaq took it a step further and made the DeskPro not only more powerful (using a full 8086 CPU versus the 8088 ), but also have more features as well.
 
Thank you sir. Message me with the memory you are looking for and I'll take a look through the loot I brought home. I grabbed a large quantity of random memory related cards and sticks and might have gotten lucky. I'd be glad to help out your cause if I have what you need, especially since it will likely be a year or 6 before I get to my Compaqs (about 8 other PCs and Zenith laptops higher on the priority list for me to work on).
 
Great job on restoring that DeskPro! IBM must have been pissed that Compaq beat them to the punch on a 386 machine! :D It's hard for me not to laugh a little imagining that.

Those old DeskPros look to be built really well. They look very nice too; styled just like the original Compaq Portable. As much of a Compaq fan as I am, I really can't believe I don't have one yet (such as an 8086, 286 or 386). If all works out though, I should be getting an 8086 soon. That's a machine I've wanted for years. I think it's so funny that in response to IBM making a portable (the 5155) to compete with Compaq's own portable, Compaq said "well, we can just make a desktop then!" and released the original DeskPro 8086 in 1984. And, as usual, Compaq took it a step further and made the DeskPro not only more powerful (using a full 8086 CPU versus the 8088 ), but also have more features as well.

That seemed to be Compaq's MO at the time. Build a completely compatible system and give you more features and performance for the money. Considering how well built they also were that strikes me as a potent combination.

Have you read the book Open by Rod Canion?
 
Very NICE!!! (to quote Borat)



That TB score can be pretty dependent on the quality of the VGA card. My 386/25 has quite the pedestrian VGA (the one in the TB database).

Time for some COVOIDS! - actually I wold like to know how it runs on your 386/16.

I gave COVOIDS a spin and it seems to run really well. I was unable to see any difference in frame rate from levels 1-13. When I get better I'll be able to give a more comprehensive assessment. ;-)
 
Yeah, I watched that documentary as well. I found it quite fascinating and a good companion piece to the book. You, of course, had a first hand account.

Do you know if that story about the executive getting a helicopter to go for lunch is true?
 
Yeah, I watched that documentary as well. I found it quite fascinating and a good companion piece to the book. You, of course, had a first hand account.

Do you know if that story about the executive getting a helicopter to go for lunch is true?

Couldn't say one way or the other, but probably Mike Swavely if it was true. One thing I can say is that everyone at Compaq was pretty low key. No ostentatious displays. A great place to work up until '92 when I left.
 
Great job on restoring that DeskPro! IBM must have been pissed that Compaq beat them to the punch on a 386 machine! :D It's hard for me not to laugh a little imagining that.

Actually wasn't so much that they were pissed, more like they didn't think it would be an advantage (thinking the 286 was all one needed). IBM was pretty much tone-deaf after the AT...and trying to jerk the market back with the introduction of MCA was their suicidal move. Intel themselves didn't even know for sure what the 386 would mean....

https://www.infoworld.com/article/2638538/intel-s-disruptive-move-with-the-386.html
 
Last edited:
Actually wasn't so much that they were pissed, more like they didn't think it would be an advantage (thinking the 286 was all one needed). IBM was pretty much tone-deaf after the AT...and trying to jerk the market back with the introduction of MCA was their suicidal move. Intel themselves didn't even know for sure what the 386 would mean....

https://www.infoworld.com/article/2638538/intel-s-disruptive-move-with-the-386.html

That article is a little weird. Why would Intel do a product launch of the 386 simultaneously with a licensee just because they had second sources for prior processors? There is no reason too! I'm sure that it never crossed their mind as a decision to make. And besides, there still continued to be licensing / cross-licensing agreements that came later. So all the article is speaking about is the launch. No they did not make a conscience decision to launch that way and that wasn't a "turning around". They were just bringing another product to market. Andy Grove said in the article it was never thought to be monumental.

Of course, IBM made mistakes... but I didn't own a 386 until the 90s and clones existed, so I dunno. Good for compaq for trying to make a splash though.
 
What would be an example of such a decision in your mind?

The delayed development of the 486 and Pentium in favor of the supposedly sexy RISC i860. There is an interesting behind-the-scenes section of Open about Rod Canion and Bill Gates pressuring Andy Grove to refocus on the x86 architecture.

Starting in the mid '80s, Compaq would send hardware and engineers to Microsoft for their engineers to use to ensure MS code would work flawlessly on Compaq hardware. During the CISC/RISC wars, I was sent to MS to make sure NT OS/2 (yes, before it was Windows NT) would run on the 386. NT OS/2 was originally a RISC only version of MS's operating system. However, one fine day I saw the fallout of a meeting between Cutler and Intel when the i860 was trashed for not being able to support NT. The x86 architecture had languished and fallen even further behind the RISC options. This led to Compaq working a deal with SGI and created the ACE initiative to make a common RISC based platform to compete with Intel. I was tasked to port Windows NT (rechristened after the MS/IBM fallout) to the prototype ACE machine which was really a warmed-over SGI IRIS Indigo with entry level graphics.

All these actions finally got Intel to get their head out of their ass and double-downed on the x86 development. Exciting times.

An old souvenir:
IMG_2392.jpg
 
That seemed to be Compaq's MO at the time. Build a completely compatible system and give you more features and performance for the money. Considering how well built they also were that strikes me as a potent combination.

Have you read the book Open by Rod Canion?

Absolutely!

I have not read that yet, but I did see it a few weeks ago. I plan on reading it eventually! I have watched Silicon Cowboys though, great documentary. I’ll probably rewatch it soon.
 
Last edited:
The delayed development of the 486 and Pentium in favor of the supposedly sexy RISC i860. There is an interesting behind-the-scenes section of Open about Rod Canion and Bill Gates pressuring Andy Grove to refocus on the x86 architecture.

Starting in the mid '80s, Compaq would send hardware and engineers to Microsoft for their engineers to use to ensure MS code would work flawlessly on Compaq hardware. During the CISC/RISC wars, I was sent to MS to make sure NT OS/2 (yes, before it was Windows NT) would run on the 386. NT OS/2 was originally a RISC only version of MS's operating system. However, one fine day I saw the fallout of a meeting between Cutler and Intel when the i860 was trashed for not being able to support NT. The x86 architecture had languished and fallen even further behind the RISC options. This led to Compaq working a deal with SGI and created the ACE initiative to make a common RISC based platform to compete with Intel. I was tasked to port Windows NT (rechristened after the MS/IBM fallout) to the prototype ACE machine which was really a warmed-over SGI IRIS Indigo with entry level graphics.

All these actions finally got Intel to get their head out of their ass and double-downed on the x86 development. Exciting times.

An old souvenir:
View attachment 65221

This is the kind of inside baseball information that makes this forum such a special place. The book only makes a basically cursory mention of the delay in the 486 in favour of the i860, but this provides so much more context. I was quite young during that time so the details of everything going on in the PC market at that time are not completely clear in my head, but I do distinctly remember the hype surrounding RISC in general. Based on my recollections it strikes me as a case of it being the new and sexy thing on the block which was basically christened as the next evolution of the market without a ton of data to back it up. In the long run elements of RISC would prove beneficial to the x86 architecture, but I'm not sure it was ready for prime time during that time frame; and like you said resources were misdirected.

Kind of reminds me of the endless recommendations to wait for OS/2 to break big, or VLIW being the next revolution in processor design with only a press release to get the hype train going. Sometimes it seems like the tech "press" can't get out of their own way.

It seems Compaq and Microsoft had a much closer relationship than most people are aware of. It would certainly explain why my Deskpro 386 has been so stable so far.

Some more improvements to show on that front in the near future BTW. ;-)
 
Because U.S. Government procurement contracts required it?

Government procurement contracts tell companies how to launch their future products? Or that the government only is requiring secondary sourcing for the products they are purchasing?
 
Hello everyone, I hope that you have had a nice holiday so far and that everyone is staying nice and safe as we wait for the new year to come around. :)

With some time off during the holiday, I've spent more time fiddling with the Deskpro to get it set up just as I'd like. The first thing that needed sorting was the CD-ROM drive. As shown a couple of posts ago I was very excited to have gotten a working external single-speed NEC drive. It seems that working was only a relative term since in short order the drive was no longer able to read almost any discs and was severely hunting on those that it did. I took the drive apart and cleaned the lens with some alcohol and a cotton swab to see if that easy fix would take care of it, however there was no improvement. I then spent many hours fiddling with the laser diode output, and while I could get some improvement I could never get it to reliably read discs. I spent many more hours swapping in and testing another internal SCSI single speed drive in the enclosure (NEC CDR-83). However it turns out that while the drive reads discs fine, that particular model had a problem with anything that streamed video or audio from the disc. The buffer setup was very strange in that it would quickly fill the internal buffer and then empty it before filling it again. This had the side effect of making video hitch and pause during playback. NEC offered a hardware fix for this back in the day, but obviously this does me no good. I'm going to hold on to the drive for some possible use in the future.

In any case, having sunk a lot of time and money into getting a working CD-ROM I was loath to try again when a local classified seller was offering the following drive for peanuts, so I jumped on it.
IMG_20201227_161059.jpg
What we have here is an NEC Multispin 3Xe. While it's a couple of years later (1993) than I was originally looking for I've been very impressed. As expected video runs smoothly, the seek times (195ms) are excellent and it has played every disc so far without issue. I also appreciate the audio controls on the front of the drive which allow me to easily play audio CDs while working in DOS programs. I even like the early 90s aesthetic of the case. This would have been incredible when it was first released.

For sound I've finally hooked up my Roland MT-100 that I picked up NIB about a year ago.
IMG_20201227_161124.jpg
The is a fully MT-32 compatible LA Synthesis sequencer and module and works perfectly in games that support the MT-32. Once again this was a local pickup which is incredibly lucky. Even more interesting was that it also came with a bunch of piano books and QuickDisks filled with music. When I opened the cover of the books I found a stamp for the music store that I used to go to for piano lessons as a kid. Based on the timeframe of when this device was available it is likely that this exact unit was one of the ones I would lust over when going to a lesson. A very small world indeed. :cool:

To solve the problem of getting files to and from my modern PC I hooked up a parallel Zip100 drive that I picked up thrifting a while ago.
IMG_20201227_161141.jpg
I have a USB Zip drive that I use on the Windows 10 machine to complete the chain. This setup represents something that could have been used on this computer back in the day and makes transferring files back and forth a snap. The only thing to keep in mind is that the Zip disks need to be formatted on the DOS machine for it to work on both. Much like floppies.

I was able to use a NIB 16550 serial card that I've been sitting on for years.
IMG_20201112_221245.jpg
I hooked up the mouse to the card on the theory that the 16550 UART can take care of much of the work polling the mouse and reduce the number of interrupts going to the CPU and thus lower utilization on a system that needs every cycle it can get.

To the existing serial port I hooked up this beauty.
IMG_20201227_161213.jpg
It's attached to a serial cable to allow me to optimize the placement for signal strength. I've already used this on a variety of retro machines from my 286 to my Amiga 500 and it works perfectly on every one. Browsing BBSes on the Deskpro is an absolute treat with a nice CD playing in the background and a cup of tea. 4800 BPS seems to be the sweet spot for the original serial port, and that's just fine. :)
IMG_20201227_193719.jpg
IMG_20201227_193852.jpg

Finally, the issue with the realtime clock seems to have somehow sorted itself out. I kept resetting it every time I used the machine and at some point around 3 weeks ago the time stopped going out of sync with the rest of the world and has been holding just fine, on or off, ever since. I'm not sure what was wrong with it, but perhaps simply using the computer regularly has shaken off some of the cobwebs as it were.

Thanks for continuing to show interest in this machine. Next up, some benchmarks!
 
Now that I've gotten all the show and tell that I can think of our of the way, it's time to run some benchmarks on this thing and see exactly how fast it is. Just a reminder that I have upgraded the video card in this to a CL-GD5428 based model, so that will affect the results as compared to its factory configuration.

I downloaded Phil's DOS benchmark pack and copied it to a ZIP disk. All of the tests were run straight off the ZIP with no problems. It turns out that many of the benchmarks will not run on a machine with only 2 MB of RAM, but I was able to complete a few of them.

Here they are, in no particular order.

Superscape Benchmark
IMG_20201227_205300.jpg

Landmark Speed Test
IMG_20201227_205326.jpg

Sysinfo (Look at the early date on the BIOS!!!!)
IMG_20201227_205815.jpg
IMG_20201227_205709.jpg

Topbench
IMG_20201227_205842.jpg

The landmark test is interesting because it is actually a bit faster than I expected. In theory a 286 can process 16-bit instructions as fast as the 386, so one would expect the machine to come up as a 16 MHz AT equivalent. Perhaps the faster video card is helping here as well and lifting the computer higher than it would have originally been.

Please let me know if there is anything else you would like me to run and I'll post the results; or even to try and record video so you can all see how things perform.

I see this as an opportunity for our entire community to explore this historically important computer together!! :)
 
Government procurement contracts tell companies how to launch their future products? Or that the government only is requiring secondary sourcing for the products they are purchasing?

Government contracts are lucrative and manufacturers will jump through many hoops to land an award. Case in point: Back in the 70's Monsanto, among others, was awarded a contract for mil a spec "frequency counter". Did they actually assemble them on the same line along with seeds and solvents? Who knows, most probably outsourced. Also, 3M comes to mind. Source: - cal lab experience.
 
Back
Top