home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!news.centerline.com!noc.near.net!news.Brown.EDU!qt.cs.utexas.edu!cs.utexas.edu!sun-barr!ames!haven.umd.edu!darwin.sura.net!spool.mu.edu!sdd.hp.com!usc!sol.ctr.columbia.edu!ucselx!crash!hale!russ
- From: russ@hale.cts.com (Russ Golish)
- Newsgroups: comp.sys.ibm.pc.hardware
- Subject: 3-Chip SIMM Modules
- Message-ID: <a3a4qB3w165w@hale.cts.com>
- Date: Mon, 14 Sep 92 04:57:21 PDT
- Organization: Hale Telecommunications San Diego CA
- Lines: 50
-
- There most definitely IS a difference between 9-chip SIMMs and the newer
- 3-chip variety. The 3-chip SIMMs will not work (correctly or at all)
- in many motherboards and memory boards. This incompatibility is most
- likely to occur with somewhat older systems and memory boards, where
- "older" is inherently a function of the hardware manufacturer. For AST,
- "older" is about two years and is typically isolated to low-end systems.
-
- The problem has been explained to me many times DD it involves DRAM
- refreshing. At first I was skeptical of this problem definition, but
- experience has changed my mind. The difference is that the new 3-chip
- SIMMs utilize a different, faster, and much more sophisticated refresh
- method. As "older" machines and memory boards did not incorporate this
- faster refresh timing into their designs, they are more likely to exhibit
- problems. Myself and several SIMM brokers have noticed that big brand
- vendors such as IBM, Compaq, AST, etc., exhibit problems more frequently
- than typical "OEM/Clone" boards. One possible explanation is that these
- vendors design to much tighter specs DD all system components are known.
- Another is that there are More of those systems out there, and/or More
- of those systems are having RAM upgraded, etc., among many others.
-
- Reports of software-related problems were fairly heavy 2-6 months ago.
- Many users reported that MS Windows experienced parity errors when under
- heavy load, and 3-chip SIMMs seemed to be a fairly common element. I
- heard detailed reports that some OS/2 users could not use the 3-chip
- SIMMs AT ALL. The system would boot DOS-5 fine, however OS/2 would not.
- These were truly ordinary systems, and apparently a quick swap to 9-chip
- SIMMs solved the problem completely. I know, I know: Solar Flares!
-
- Another problem with 3-chips is strictly hardware and well understood.
- The problem is caused by a mismatch between the (2) 4-bit wide DRAMs and
- the Parity DRAM. The mismatch is the speed, NOT that the ICs are from
- different vendors. The parity chip MUST BE the Same Speed or Slower than
- the other ICs (within a given tolerance.) The stamped or rated speed is
- not necessarily the true IC speed. One notable case is DRAM made by NMB,
- which routinely performs 20ns faster than rated speed. 3-chip's which
- contained (2) 4-bit wide DRAMs (non-NMB) and a NMB parity DRAM were prone
- to fatal parity anomalies. Many Brokers blacklisted this configuration.
- Definitely not NMB's fault DD strictly the fault of the SIMM assembler.
-
- Like all new "HW standards" in our multi-vendor world, it takes a while
- for everyone to get their timing signals in sync. This "standard" was
- better than most DD nearly everyone is in sync now. In general, there
- are very few compatibility problems using 3-chip modules in later Rev.
- motherboards. Good news indeed DD 3-chips are significantly cheaper...
-
- - Russ -
-
-
- HALE TELECOMMUNICATIONS - Public Access Node, San Diego CA 619/660-6734 HST
- russ@hale.cts.com (Russ Golish) 14-Sep-92 at 5:01a
-