• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

IIfx | IINTX 64-pin SIMMs

olePigeon

Well-known member
@joethezombie  I've source NOS chips as well, but I didn't order them pending your SIMMs.  Do you need more DRAM packages?  I can forward you the link if you do.

 
Last edited by a moderator:

jjclay

Active member
Hi Joe. I'm keen to get some IIfx 64-pin SIMMs for my IIfx, if and when these may become available. Sign me up :)  

 

trag

Well-known member
Is Mac IIfx and LaserWriter RAM proprietary indeed? Not some sort of relatively uncommon but still standard RAM?

Does anybody have knowledge about this?
The memory chips are standard.  The way they are wired to the SIMMs is proprietary, in the sense that only Apple ever did it that way.   There's a thread where this is discussed at length.   If you follow the linked threads upstream from Joethezombie to Doug's, to the link he provides on page one and two of his six page thread, you'll find that the link doesn't work.  Sigh.  I hope it comes back.

Memory chips have separate D and Q pins for Data.  One pin is Data In.  The other pin is Data Out.   On the vast majority of SIMMs in the universe, the D and Q pins are just wired together and treated as a single I/O pin.

On the IIfx, Apple wired the D and Q pins of each memory chip to a unique pin on the SIMM.   This is why they can't use a 30 pin SIMM.   There aren't 8 unused pins available.  And I guess they were already using the 64 pin sockets for ROMs, or maybe there's nothing between 30 and 64.  

The reason Apple did this, is that on operations that Write to the RAM, the IIfx buffers the Write data in some register/buffer chips on the logic board.  The memory controller retains the address information.   And the CPU is told that the operation is done.    Then the CPU goes on about its business while the memory controller uses the stored data to complete the Write operation on the SIMMs.

If another RAM operation comes along before the Write is completed, then the memory controller might have to stall the CPU a bit, but generally, this frees up a lot of clock cycles for the CPU.   

The Read from SIMM pins are wired directly to the Data bus of the IIfx, but during a Write operation, the Read pins are all High-Z so they don't interfere.    It's having the Write pins off in their own little culd'e sac that makes things zoom.

Memory chips which are wider than 1 bit combine the Data-in (Write) and Data-Out (Read) pins on the chip.  So they cannot easily be used for the IIfx buffered write scheme.

However, by adding a bus switch chip between the SIMM pins and the memory chip pins and controlling it with  the Write Enable signal to the SIMM, one can allow the buffered Write to pass onto the SIMM while keeping the Read pins of the SIMM in high-Z.

There's a lot more detail in that thread that's not available.    I ran off a batch of 16MB IIfx SIMMs back around 2005 or so.  I built half the standard way with X 1 chips and built the other half with the above scheme using the much more common X4 chips.

IIfx_Front.jpg.ab4db618a6895b0acd5986546e692382.jpg


IIfx_Rev2_Front.jpg.ccf701047967935f7c551932b7168df2.jpg


 
Last edited by a moderator:

maceffects

Well-known member
@trag Those are interesting, it appears that the chips on the second are still available from one of my contacts in China, which is a big surprise.  Not a big fan of their price but you only need 2...  

 

trag

Well-known member
@trag Those are interesting, it appears that the chips on the second are still available from one of my contacts in China, which is a big surprise.  Not a big fan of their price but you only need 2...  
As mentioned in that thread that is currently missing, sigh, the SIMMs made with X4 chips work reliably four at a time, but not eight at a time.  They'll also work in a 4 & 4 configuration with the "conventional" SIMMS, and actually, there's a config of 6 & 2 that works.   I suspect too few chips is not sinking enough current in one of the address or control lines and causing ringing, but haven't experimented yet.  It's on the list, right after that PEX ROM DIMM...

Still trying to figure out where the last fifteen years went.  Of course the eighteen-year-old might have something to do with it.

 
Last edited by a moderator:

trag

Well-known member
That's it.  I even tried searching on the thread title but couldn't get the forum to admit it existed.   My forum foo seems to be weakening.

One of these days I will:

1)  Find the top of a flat work surface in my house, somewhere.

2)  Get that PEX ROM built based on the decoding work we did back in 2013....   Sheesh.

3)  And then build out the rest of my IIfx SIMMs.   

But my capacity for not doing the above, has so far, been infinite.

 

maceffects

Well-known member
@trag If you get the PEX ROM built, what do you think development wise can be done with such a ROM?  Do you think it could be used in a 8600/9600?  By the way, number 1 will never happen...  I've been trying for years.

Also, I should mention that I have no intention of creating anything myself since 2 people here already have similar projects.  My comments were mainly to offer to help source the needed memory chips.  I just wanted to clarify. 

 

trag

Well-known member
@trag If you get the PEX ROM built, what do you think development wise can be done with such a ROM?  Do you think it could be used in a 8600/9600? 


I offered to try to build one for a list member who owns a PEX back in 2008.   Our original plan was to build the module with sockets, because there was a lot of uncertainty about how to interleave the PEX dumped ROM file into ROM chips.  Sockets would allow us to experiment without desoldering/soldering the ROM chips repeatedly.

The ROM DIMM for the X500/X600 series is the same as for the PEX.  The code on the chips is different, but the edge connector, and wiring is all the same.

I have built ROM DIMMs for the X500/X600 and the Beige G3 and have the individual ROM chip files available for those machines.   But for the PEX all we had to work with was ROM dumps from the machine.  And from a previous run, I had two blank (unused/unsoldered) ROM DIMM boards left to work with.

I built a couple of ROM modules with sockets which the other fellow supplied, and tested then with PM9600 code on the chips in a 9600.    And I could never get the thing to work.   Note, that the circuit board I was using works fine when I solder the same chips directly to the board.   But there's 44 finicky little pins on each of those chips and there's four of those chips, so it's not really a wonder that the sockets didn't work -- in hindsight.

Anyway, in 2013 I had a little mental time available and did some comparing of the ROM chip files for the X500 to the ROM Dump file from the same machine.    Then compared the ROM Dump files for the X500 and PEX, which were very similar.   From that we (the list working on a thread in Hacks) was able to figure out the scheme to convert the ROM Dump file into four ROM chip files.    Doug B. wrote a little routine to convert the files and so PEX ROM chip files, which will almost certainly work, became available.   His routine was tested by applying it to X500 ROM Dump (or maybe the Beige) and comparing the output to the X500 ROM chip files we have available.

Since then, I started to prepare some ROM circuit boards for the project, but the flat surface I was using for that was overcome by events.

So, the TL/DR for all that, is that I owe someone a PEX ROM and that's my first project in the queue.   It's not as fraught, as it might sound.  At the time, before we embarked, I explained that I didn't have a lot of time available to work on it.  And I did build the DIMMs we originally discussed, which failed.  And the list member has been a saint about it.   But I want to get it done one of these days, darn it.

By the way, number 1 will never happen...  I've been trying for years.


I hear you.   I'm almost convinced.  But folks like you and Bolle are going to do all the fun projects while I'm procastinating/raising my Son, the last eighteen years.   I just need to clean this room at the back of the house and set up a table a friend gave me, and I'll have a place.   In reality, I could make the time, I think.  I just seem to have some block about starting.  I think I don't want to face figuring out what to do with a lot of the stuff in that room.   There's a bunch of fiddling aquarium equipment and supplies and I'm not quite sure what to do with them.  

 

dan.dem

Well-known member
Thank you trag for the in-depth explanation, even when I'm missing the knowledge necessary to follow it. At least it became clear to me, that IIfx-RAM is proprietary but recreatable.

However, I cannot think about many applications where you would need more than 32 MB RAM (using 8 pcs of 4 MB SIMMs) and at the same time like to use a Mac II-class machine. The most I had installed in 68k-times was 36 MB (on a Quadra-class Mac). What comes to my mind when I think about a Mac IIfx is running system 6-applications on the fastest machine possible, thus being limited to 8 MB anyway.

@trag

Still trying to figure out where the last fifteen years went.  Of course the eighteen-year-old might have something to do with it.
What better way to spend one's time?

 
Top