• Please review our updated Terms and Rules here

AST Rampage 286 - unusual memory size detected

J

jh1523

Experienced Member
Joined
Jan 15, 2009
Messages
298
Location
coastal New England
I've been playing a bit with a Rampage 286 card - precisely this one: http://artofhacking.com/th99/i/A-B/52501.htm

According to the TH99 page, it has switch settings to set memory size all the way down to 128k. That would mean one bank (2x9 chips) populated with 2164 chips. The documentation doesn't mention precisely that the board supports 2164 chips - only 21256 chips are mentioned.

yet I wanted to see what happens... so I populated bank0 with 2164 chips (I have a bunch), set sw1/1-4 to "on" which as per the docs would mean 128k... set the sw2 switches to E000, and fired it up.

And something funny happens. Remm.sys (I tried both version 4.30 and 4.70) tells me I have 384k of expanded memory available on the board. Doesn't report any errors. So where did the extra 256k come from? I definitely put on 18 x 2164 chips only, i.e. 128k + parity.

Any ideas?

(note that I haven't tried to actually use the 384k RAM reported)
 
modem7 said:
That table (titled "DRAM SWITCH CONFIGURATION") will be for the starting address of the RAM.
Click to expand...

I don't think that's correct. Why would one want to map the memory after 2MB? Besides there is already a bank of switches used to set the mapping address (sw2, 1-7 as documented elsewhere on this forum and verified by me)
 
Obviously, you don't really have an extra 256K--and it should be obvious that 64Kb chips aren't supported. But it's good enough to confuse the test diagnostics. Here's why:

Small-DIP DRAM (as opposed to some early 22-pin stuff) uses the same set of address lines for both row and column addressing (RAS and CAS). So, on a 4164, you have 8 bits of row select and 8 bits of column select, giving you 64 bits. In a 41256, pin 1, which is unused on a 4164 gives you an extra bit (making 9) of both row and column addresses, giving you 18 bits of addressing, or 256K. Since the 4164 doesn't even see the extra bit of address on pin 1, it merely repeats its contents 4 times throughout the 256K addressing range.

If you could look at that area of memory, you'd see that data pattern repeated 3 times after the original.

Let me anticipate your next question. Why 384 and not 512K then?

Remember what I said about 128K DRAM chips. I suspect that if you could find the manual for this board, it would say that you could use 128Kx1 DRAM but only in the last row. So the diagnostics really think that you've got a row of 256K, followed by a row of 128K chips.

Now, quit fooling around...

So why 128K and not 256K? Probably because the original PC AT used 128Kx1 DRAM chip "stacked" and there exist 128Kx1 DRAM chips (Micron MT1128, for example) that use 9 bits for CAS and 8 for RAS. Strictly a historical oddity that seemed important to the engineers at AST.
 
Last edited:
Nice explanation, thanks.

Now to find some 256k chips... :)

(actually you've opened my appetite for something more exotic, so now I HAVE to get some 128kx1 chips... :D)
 
You must log in or register to reply here.
Back
Top