Hacking a RAM module

6 posts / 0 new
Last post
doug-doug the mighty's picture
Offline
Last seen: 3 weeks 1 day ago
Joined: Apr 14 2004 - 17:52
Posts: 1396
Hacking a RAM module

WHAT: I noticed that some of the 256MB SGRAM modules I have solder pads for more RAM (specifically two more 32MB chips). I have a few extra sticks that could become donors. The thought is that I could add two 32MB chips and a few micro-caps/resistors to my memory stick and upgrade my 256MB to a 320MB stick.

WHY: Because I have the parts to do it.

WHAT ELSE: I know this theory will be best put to practice if the chips are all of the same make and the risk of failure is high.

SO WHAT DO I WANT: Has anyone tried this and (whether you have or not) do you have any words of wisdom?

--DDTM

jt
Offline
Last seen: 17 years 1 month ago
Joined: Dec 20 2003 - 10:38
Posts: 447
Sounds like provision for parity . . .

. . . to me. AFAIK, if there were enough empty pads to double the memory size of the module you might have a shot . . . going from 256MB->320MB doesn't sound quite right.

Piccies/Scans of the modules would be helpful if you really want feedback, but a breakdown of the IC count and type is absolutely essential to get any worthwhile advice. I'm just winging it here, I really don't understand memory configs yet, that's why I'm exploring hack possibilities in that area, to learn why they're impossible . . . ignore the facts an . . .

jt Beee

Eudimorphodon's picture
Offline
Last seen: 3 months 3 weeks ago
Joined: Dec 21 2003 - 14:14
Posts: 1207
ECC

As JT noted, the extra solder pads are for "Parity", or in the modern case, ECC. (The ECC code for a 64 bit word requires another 8 bits. Which is why you'll find that virtually all SDRAM DIMMs have 8 8-bit wide chips per side, with room for a ninth. Whether that ninth is populated or not dictates whether it's an ECC DIMM.)

So in short, no, you're not going to end up with a "bigger" DIMM by populating those blank spots. You'll have a DIMM with more RAM on it, sure, but it won't be addressable.

--Peace

doug-doug the mighty's picture
Offline
Last seen: 3 weeks 1 day ago
Joined: Apr 14 2004 - 17:52
Posts: 1396
Clarification:

I did miscount when I added the RAM. I said 320 and meant 288. The configuration of the RAM stick has 8 chips (16MB each) on each side. Both sides of the stick have solder pads for an extra RAM chip. In other words, the stick has 8 chips on each side and room for a ninth on each side (total being 18 chips vice the existing 16). I figured if I add a chip to each side, I would be okay with the whole parity thing.

Thoughts?

Eudimorphodon's picture
Offline
Last seen: 3 months 3 weeks ago
Joined: Dec 21 2003 - 14:14
Posts: 1207
Re: Clarification:

I did miscount when I added the RAM. I said 320 and meant 288. The configuration of the RAM stick has 8 chips (16MB each) on each side. Both sides of the stick have solder pads for an extra RAM chip. In other words, the stick has 8 chips on each side and room for a ninth on each side (total being 18 chips vice the existing 16). I figured if I add a chip to each side, I would be okay with the whole parity thing.

Thoughts?

Um, read my post again. You're describing exactly what I said.

All adding those chips will do is turn a non-ECC DIMM into an ECC DIMM. In essence, you'll be turning a 32Mbit long by 64 bit wide array of RAM into a 32Mbit long by 72 bit *wide* array. Remember of course that the memory bus of "modern" (Pentium/PowerPC/etc) computers is 64 bits wide. That extra 8 bits is only accessible to chipset memory controllers which understand ECC, and use it to store the error correcting code. It will *not* make your DIMM "bigger".

Clear?

--Peace

doug-doug the mighty's picture
Offline
Last seen: 3 weeks 1 day ago
Joined: Apr 14 2004 - 17:52
Posts: 1396
Oh.

Oh.

Log in or register to post comments