| View previous topic :: View next topic |
| Author |
Message |
xsecret
Joined: 01 Feb 2004
Posts: 1846
Location: France
|
|
| Back to top |
|
 |
debs3759
Joined: 18 Jan 2006
Posts: 9477
Location: Northampton, Divided Kingdom
|
 Posted: Tue Aug 14, 2018 1:33 pm Post subject: |
 |
|
I have quite a few 386, 387 and 486 plus a few 5x86. I wouldn't say a lot when compared to John though 
I have plenty of free time for testing.
_________________
My graphics card database can be found at http://www.gpuzoo.com.
I can resist anything except temptation.
Debs |
|
| Back to top |
|
|
H3nrik V!
Joined: 15 Apr 2014
Posts: 1246
Location: Denmark
|
| Posted: Tue Aug 14, 2018 2:19 pm Post subject: |
|
|
| xsecret wrote: |
| H3nrik V! wrote: | | Will it run 50 MHz bus also? |
I don't know, but unlikely. One week ago, I was unable to go higher than a 10 MHz bus without write memory corruption. Three days ago, I find a trick to reach 25 MHz, but with insane frequencies inside the FPGA (400 MHz). Yesterday, I finally found what's wrong with the design and tested a new address decoder. Now it works just fine with a DX2-66 or a DX4-100 at 33 MHz, and the maximum clock inside the FPGA is limited to 66 MHz.
Unfortunately, I don't have a 486 with a 40 or 50 MHz bus to try right now  But I will try to find one ASAP. Signal integrity might be an issue, especially with a 50 MHz bus. |
Yeah, the signal integrity would be a concern, especially when using a daughter board as the unevitable shift in impedance going from board to board could mess up a lot .. |
|
| Back to top |
|
|
xhoba
Joined: 20 Apr 2009
Posts: 4596
Location: Picardie, France
|
| Posted: Tue Aug 14, 2018 2:28 pm Post subject: |
|
|
| xsecret wrote: | I just bought a 486 DX-50 NOS from john
But the same problem will pop up again: I don't really want to start a non-ES CPU collection. So I will probably need a betatester with a lot of different 80186-80486 in hand and willing to spend some time testing them.  |
I could help when I'll be back in France (around mid-october)...
_________________
http://www.x86-guide.net/Xhoba/en/collection.html |
|
| Back to top |
|
|
xsecret
Joined: 01 Feb 2004
Posts: 1846
Location: France
|
| Posted: Tue Aug 14, 2018 2:40 pm Post subject: |
|
|
| debs3759 wrote: | I have quite a few 386, 387 and 486 plus a few 5x86. I wouldn't say a lot when compared to John though
I have plenty of free time for testing. |
| xhoba wrote: | | [I could help when I'll be back in France (around mid-october)... |
Thanks! That will be useful.
| H3nrik V! wrote: | | Yeah, the signal integrity would be a concern, especially when using a daughter board as the unevitable shift in impedance going from board to board could mess up a lot .. |
Very true. I try to keep signal trace as short as possible, but it's not always possible. Much more decoupling caps can also help. I only have 4 tantalum & 8x 100 nF ceramic caps for the entire CPU right now. But this very early PCB has been only designed as a proof of concept with a 16 MHz max clock in mind
_________________
ES-Only Collector : http://www.engineering-sample.com
Universal Chip Analyzer (UCA) : https://x86.fr/uca / http://www.cpu-world.com/forum/viewtopic.php?t=34349 |
|
| Back to top |
|
|
xsecret
Joined: 01 Feb 2004
Posts: 1846
Location: France
|
| Posted: Tue Aug 14, 2018 2:48 pm Post subject: |
|
|
Just found a nice Am486DX2-80NV8T. Configured the PLL for 40 MHz bus, and after eliminating a crosstalk in a adjacent signal, it seems to works fine! 
Interesting stuff:
CPUID@BOOT: 0x432
eFLAGS (after CPUID enable attempt): 0x2
Seems it doesn't support the CPUID instruction.
PS : No way to run it at 100/50 MHz, but the debug connectors and logic analyzer plugged add a lot of noise.
_________________
ES-Only Collector : http://www.engineering-sample.com
Universal Chip Analyzer (UCA) : https://x86.fr/uca / http://www.cpu-world.com/forum/viewtopic.php?t=34349 |
|
| Back to top |
|
|
debs3759
Joined: 18 Jan 2006
Posts: 9477
Location: Northampton, Divided Kingdom
|
| Posted: Tue Aug 14, 2018 3:18 pm Post subject: |
|
|
How did you test the Xeon you posted in another thread if the max fsb is so low? Is that planned to be released in the distant future?
_________________
My graphics card database can be found at http://www.gpuzoo.com.
I can resist anything except temptation.
Debs |
|
| Back to top |
|
|
xsecret
Joined: 01 Feb 2004
Posts: 1846
Location: France
|
|
| Back to top |
|
|
xsecret
Joined: 01 Feb 2004
Posts: 1846
Location: France
|
| Posted: Tue Aug 14, 2018 4:22 pm Post subject: |
|
|
I could not resist any longer and I was brave enough (or stupid enough) to plug some of my DX2-66 ES on the alpha PGA UCAS.
Good news : all of them work properly! I grabbed some informations about four or them :
The Intel "Internal Samples Only" CPUID is 0x433. It doesn't support the CPUID instruction. So it's basically a B1-Step, similar to the commercial SX645. Would have expected an A-Step, but nope.
The Q0652 supports CPUID instruction with CPUID = 0x436. Late DX2, ES counterpart for SX955 (w/ Write-back).
The SYE36 ES is basically the same than the "Internal Sample Only" : no CPUID instruction support with CPUID = 0x433.
The AMD A80486DX-66NV8T don't support CPUID instruction and ID itself as 0x432.
I also found near them on the tray one blank 486 that I have never been able to identify. It don't boot on a 486 motherboard. Can't remember where does it come from, but probably a Journey kit.
Plugged on the UCA, it worked for 2-3 seconds @ 8 MHz before crashing horribly with multiple failures. But it was enough to grab ID information. This is indeed a 486 that supports CPUID instruction and ID itself as 0x480. So it's a defective early DX4-100 (without WB).
Interesting investigation... 
_________________
ES-Only Collector : http://www.engineering-sample.com
Universal Chip Analyzer (UCA) : https://x86.fr/uca / http://www.cpu-world.com/forum/viewtopic.php?t=34349 |
|
| Back to top |
|
|
Neology
Joined: 25 May 2011
Posts: 2412
Location: VIC, AU
|
|
| Back to top |
|
|
|
