identifying PC Simms

Joe Stevenson ikvsabre at comcast.net
Sun Oct 9 21:43:29 CDT 2005


Hey, thanks for the info.  I followed your advice, and determined that I have 16x 4MB-SIMM (parity) 70ns

Not much street value anymore, I know, but better 4MB than 1BM :-)

Joe


On 9/29/2005 at 9:19 PM Jeff Walther wrote:
*********** REPLY SEPARATOR  ***********

>>Date: Wed, 28 Sep 2005 23:53:00 -0400
>>From: "Joe Stevenson" <ikvsabre at comcast.net>
>
>>I have 16 30-pin simms left over from various past incarnations of 
>>my PCs, and I'm
>>  trying to figure out what I've got.
>
>>  I no longer have a motherboard to test them, so I have no idea what is
>what. 
>>
>>Is there any not-to-painless way to figure out what I've got?
>
>Not all that painless, but the only way I know that works...
>
>Take a SIMM.  Count the number of chips.  Find the model markings on 
>one of the chips.  There are usually two or three lines of writing on 
>a chip.  One of these will be a date or batch code and is irrelevant. 
>The line you want will start with a one, two or three (usually two) 
>character manufacturer code (e.g., K or KM for Samsung, TC for 
>Toshiba, M(numeral)M for Mitsubishi, HM or HN for Hitachi, etc.), 
>followed by some longish, about four to eight, alphanumeric code 
>which is mostly numerals, then a dash or space and a speed number in 
>nanoseconds, which may or may not have the trailing zero truncated.
>
>For example: HM5116400BS-8, MSM511000C-7,  KM44C16100B-5, TC514400AJ-6.
>
>Then go to a datasheet archive such as 
><http://www.datasheetarchive.com/> and enter the part number in the 
>search field.  It often helps to truncate the trailing characters 
>back to the first number in the body.  E.g. HM5116400, MSM511000, 
>KM44C16100, etc.
>
>The datasheet will tell you the capacity and organization of the 
>chip.  For example, a 1 MB 30 pin SIMM with eight chips on it will be 
>composed of 1M X 1 chips.  These have one million addresses with 1 
>bit at each address.  Eight of them working in parallel provide 1 
>million addresses with eight bits at each address or 1 megabyte.
>
>Multiply the total capacity of the chip by the number of chips on the 
>SIMM.  Remember that you're working with bits here, not bytes. 
>Divide by 8 and you've got the capacity in megabytes--except...
>
>Some SIMMs are parity SIMM and they are based on 9 bits of data 
>rather than 8 bits of data, so you'll need to divide that capacity by 
>9, not by eight for a parity SIMM.  A 30 pin parity SIMM will have 
>nine or three chips instead of eight or two, so they're fairly easy 
>to identify.
>
>However, a three chip 30 pin SIMM will have two chips with a certain 
>capacity and a third chip with 1/4 the capacity or either of the 
>other two.  In this case, calculate the total capacity of the two 
>larger chips and divide by eight.  Or find the capacity in bits of 
>one big chip and divide by four.
>
>In most cases, if the SIMM has eight or nine chips, then the capacity 
>in bytes is equal to the number of addresses any of the chips 
>supports (see the datasheet).  If the SIMM has three chips, then the 
>capacity in bytes is still equal to the number of address which any 
>of the three chips supports.
>
>For example, you find a three chip SIMM with two 4M X 4 chips and one 
>4M X 1 chip on board.  The capacity of this SIMM is 4MB or 4 
>Megabytes.  You find a SIMM with eight or nine 4M X 1 chips on board, 
>its capacity is also 4MB.
>
>The real trick is figureing out the capacity of the chips from the 
>markings on them.  Google searches sometimes help, but often (almost 
>always) just lead you to chip distributers spamming the search engine 
>space with part numbers to lead part searches to their sites.  They 
>often don't even have the chip in question, and rarely have any 
>useful information available on their website.
>
>SIMMs that can steer you wrong are composite SIMMs where groups of 
>smaller capacity chips are used to build a higher capacity SIMM.  For 
>example, building a 16 MB 30 pin SIMM out of eight 4M X 4 chips. 
>These are rare and should be easily identified because there should 
>be a non-memory chip on board to handle the address translations.
>
>Jeff Walther

*********** REPLY SEPARATOR  ***********







More information about the cctalk mailing list