• Please review our updated Terms and Rules here

Please help with real iron, I need the 486 and first Pentium systems

vol.litwr

Experienced Member
Joined
Mar 22, 2016
Messages
321
I am gathering data for my project. I need data from the 80486 based systems below 50 MHz. Above all, I really want to get data from the first 486 @16-25 MHz. I am also seeking data from the first Pentiums below 90 MHz. I attach a zipped COM-file for DOS (it and its sources are also available the project page). I gather data for runs with 100, 1000, and 3000 digits.
Please help. Thanks.
View attachment pi-pc386.zip
 
I am gathering data for my project. I need data from the 80486 based systems below 50 MHz. Above all, I really want to get data from the first 486 @16-25 MHz. I am also seeking data from the first Pentiums below 90 MHz. I attach a zipped COM-file for DOS (it and its sources are also available the project page). I gather data for runs with 100, 1000, and 3000 digits.
Please help. Thanks.
View attachment 52029

I've got a 486sx 33mhz and 25mhz, plus a PS/2e (I forget what 486 it has) I can try it on and my friend has a 60mhz Pentium if that would help
 
I am pretty sure the first 486s were the DX 25 and 33. Wikipedia reports a 20MHz version, but I think it might be an error. Anyone have proof?

As far as I know, the 16 and 20MHz parts were only available as SX chips, and those came along about a year and a half later. From what I remember, the 16MHz part was a special OEM part for Dell.
 
I am pretty sure the first 486s were the DX 25 and 33. Wikipedia reports a 20MHz version, but I think it might be an error. Anyone have proof?

As far as I know, the 16 and 20MHz parts were only available as SX chips, and those came along about a year and a half later. From what I remember, the 16MHz part was a special OEM part for Dell.

I've seen a 486DX/20 before, it was in the lab CPU drawer back when I was in high school. We had several oddball x86 chips like the i487DX, 486SX/2-66 and a Kingston Turbochip that ran at 200 MHz. We also had one 486SX/16, but I don't think anyone ever tried using it because it was so slow.

I can't tell you which speed grades came first, but the SX was released more than a year later than the original DX parts to compete with AMD's faster 386 chips.
 
I've got a 486sx 33mhz and 25mhz, plus a PS/2e (I forget what 486 it has) I can try it on and my friend has a 60mhz Pentium if that would help

All this hardware is very interesting. Thanks in advance.

Does this require an FPU?

No

Yeah, what Chuck said. I have a 486SX 20 that I can pop into a board.....

Quite a rarity. But you need also to get 20 MHz or less clock frequency for it.
 
Last edited:
I have some Intel 486 sx/2 chips and was wondering who the hell would buy those (must be an OEM thing).
 
So what exactly do you need? The numbers for 100, 1000, and 3000 digits and the type of computer? Am I missing anything?
 
So what exactly do you need? The numbers for 100, 1000, and 3000 digits and the type of computer? Am I missing anything?

The benchmark calculates given amount of digits of the number pi and prints the time. I am gathering these times for various types of computers. I don't need digits. So I need only three timer results for each type of computer.
 
I have some Intel 486 sx/2 chips and was wondering who the hell would buy those (must be an OEM thing).

Most people wouldn't have been able to tell the difference, other than the large price disparity. Nearly all applications at the time were still integer based and didn't benefit from an FPU.

Games were still majority 2D side or top scrollers, or at best Doom, which was still technically a 2D engine that used integer math. It was still a couple of years before Quake came along and the 3D FPS genre killed off the 486.

If I were in the market for a computer in 1993/1994, I'd probably go for a 486 SX/2.
 
I worked in engineering in the 90'd so my home machine would have to have the FPU (even my home built 386 did).
 
I suppose now would be a bad time to notice that this program could be made a little faster through some 8086 optimization?

Are you sure that it will be easy? IMHO the program is quite optimized. Indeed, I will be surprised if somebody can find a way to make it a bit faster. However don't try to optimize 6502, 6509, or Z80 code - it can take too much time. ;)
 
Code:
         xor ax,ax
         sub ax,ra
         mov bx,7
         xor dx,dx
         div bx
         and al,0fch
         mov [maxnum],ax
I don't know what this code does but it's all constants and so should be calculated at assembly time, not during runtime.
 
There are also some MOV AL,reg that can be replaced with XCHG reg,AL due to "reg" being explicitly trashed later. So yes, the code can be optimized to be measurably faster. The question is whether or not you're interested in that, and how it changes your project results.
 
Back
Top