SUMMARY --Memory for SS2--

From: Randy Born (randy@ai.iit.nrc.ca)
Date: Tue Jun 25 1991 - 11:01:09 CDT


        I guess the answer is " BUYER BEWARE". It seems that the simms
must be real 80 ns SIMMS. It is best to get the assurance from your
supplier that the simm WILL WORK IN THE SS2.

I am going with Clearpoint 4MB simms ($345) with the contract specifying
that the memory must work in the SunSPARCStation2.

Thanks to all 20 respondents ( their response are attached)
*******************************************************************************
* Randy V. Born EMAIL: born@ai.iit.nrc.ca *
* Technical Officer randy@ai.iit.nrc.ca *
* National Research Council of Canada PHONE: (613) 993-8549 *
* Montreal Road , Bldg M-50 FAX: (613) 952-7151 *
* Ottawa, Ontario, Canada, K1A 0R8 *
* *
* * ** ** ***** *** *** * *
* * *** ** ** ** ** * ** * * *
****************** ** * ** **** ** ** ********************
                * ** *** ** ** ** * ** * *
               * ** ** ** ** *** *** *

Subject: Re: MEMORY specs for SparcStation 2

Randy Born <randy@ai.iit.nrc.ca>:
>
> I know that SIMM should be 80 ns for my SS1's and IPC's but
> I need to know what speed is required for the SS2.
> A CAUTION in the mem install manual says SIMMs installed in the SS2 must
> be specifically designed to operate in it. DO NOT REMOVE SIMMs from a
> different SUN system and install in a SPARCSTATION 2.
>
> I am about to order 3rd party SIMMs to upgrade the workstations memory and
> I need this info. Is the caution above due to speed and if so can I use
> the faster SIMMs for SS2 in my IPC or SS1 workstation.
>
RESPONSES:
1.
The SS2 uses the same size and shape of SIMMs as the SS1, SS1+, and
IPC. The difference is that 80ns SIMMs for an IPC may or may not work
in an SS2. The reason is speed. 80ns memory is 80ns plus or minus
some percentage. If you get 80ns "minus" (faster chips) they work
fine. However, if you get 80ns "plus" (slower chips) your sparc-2 will
die.

We bought Sun memory with our SS2, but by now I would expect ClearPoint
and other vendors to have SS2 memory. Be sure you specify that this
memory is for an SS2, and that you must be able to return it for a full
refund if it doesn't work. Also be sure to say that "doesn't work"
will be decided exclusively by your company.

Hope this helps,
Ric (ric@cs.arizona.edu <Ric Anderson>)
2.
a company called Clearpoint sells 1MB and 4MB
SIMMS which they claim work in the SS1, SS1+,
IPC, and SS2. These are 80ns SIMMS. I've not
yet tried the chips in an SS2, but will within
the next couple of months.

george nassiopoulos
nassio@cfa.harvard.edu
3.
No idea.

Selina
4.
Scare tactics. Normal 80ns simms work just fine.

Use your wheels: that is what they are for.
datri@convex.com
5.
People within Sun have told me that 80 ns chips are required for the SS2
and they have to really be 80ns -- no fudging. These unnamed sources
recommended ClearPoint as being a reliable source.

Randy Garrett
6.
We have 4 SS2s here all filled with 80ns 4MB SIMMS. This memory does work
in an SS1, SS1+ and the IPC.. But take note; if you have a Mixture of 4MB
and 1MB SIMMS in the SS1 or SS1++ the 4MB SIMMS MUST be in the lower banks
or else you will get memory errors. I am not sure about the IPC or SS2 when
you have a mixture.

  One of the SS2s has been giving us memory parity errors on a regular
basis. Since I suspected 3rd party memory and the sindiag memory tester
could not generate the error I decided to write my own memory tester. I have
ran this memory tester on all 4 SS2s every evening for the last month
and only receive error reports from that one SS2. I am therefore convinced
that this SS2 has a flakey MMU on the motherboard (now to convince SUN).

  Whenever you order the SS2 memory specify in the contract that the memory
must work in an SS2. Since you are in Ottawa here are some supplieres:

Clearpoint Memeories T.O. Robert Stroud (416) 620-7242
Tracan 301 Moodie Dr. suite 201 Sun simm memory 596-9760

Drop me a line if you want my memory diags.

-- 
Phil Blanchfield
Internet: phil@dgbt.doc.ca	OR	phil@dgbt.crc.dnd.ca
7.
You're almost correct.  SIMMs for SS1's and IPC's must be at least 100ns and
SIMMs for SS2's 80ns.  We've found that most third party vendors now deal
exclusively with 80ns SIMMs and don't bother with the slower variety.

You can safely use 80ns SIMMs in any of the Sparc systems. If you've got any 100ns SIMMs however, you can only use them in the older systems.

John Kilheffer amp19263@garfield.amp.com 8. The questions is: What does 80 ns really mean?

80 ns industry standard SIMMS will fetch in nominally 80 ns. The SS2 requires that they absolutely make it in 80 ns every time. A lot of SIMMS can't do this reliably. So, the answer is that SOME 80 ns simms work fine, but some don't. The ones SUN sells have been qualified to work in an SS2 (if bought for an SS2). I'd either request that the 3rd party guarantee they work in an SS2, or I'd buy 70 ns SIMMS. There isn't any reason why you shouldn't be able to use SS2 certified SIMMS in the slower Sparcstations. 9.

Sparc 2's require 80 ns or better memory. AND IT HAS TO BE GRADED 80 ns.... ie +- 5% at worst. You can use 70 of course. I've seen it as low as ~180. There is nothing else unusual about the simms that go into the ss2.

We use 3'rd party simms without problem on the Sparc2. Initially we had some failures due to their not being graded 80 ns. When you order be very specific about what you want ( graded 80 ns simms to go in the sparc2 ) or specify 70 ns devices.

Hope this helps...

Jim Ray Harris Semiconductor Internet: jdr@semi.harris.com PO Box 883 MS 62B-022 10. Boy did you ever open a can-of-worms. The problem is that the SPARC2 need 80ns simms as well BUT and this is a big butt some simms manufactures and resellers sell simms as 80ns that really fall out side the tolerances of an 80ns simms. I got stuck with a bunch from Samsung that Andataco sold me. When I moved them from the SPARC 1's to SPARC2's they would fail with parity errors. To compound the problem this does not show up under diags. So the only way to tell you have a bad one is to put it in and let it run until it fails. Do NOT buy samsung simms for your sparc 2's. This is a warning.

flp 11.

70 ns for SS-2. SIMMs rated at 80 ns may or may not work - depends on how far they deviate from the nominal 80 ns. SS-2 supposedly takes SIMMs with speed close to or equal to 80 ns, but not slower; so SIMMs deviating too much from 80 ns will not work.

Sun puts 100 ns in their 4/3xx series (at least they put 100 ns in our systems) - that's much too slow for SS-2. I'm pretty sure that SS-1 and SS-1+ could take 100 ns, too. I was moving SIMMs from 4/3xx to SS-1 and SS-1+ and so far had no problems.

Margaret Mikulska UCSD, Dept. of ECE

12. >From what I've heard and understand the SPARCstation 2 requires 80ns *graded* SIMMS. Graded SIMMs are those that have been tested and sorted as being AT LEAST the specified cycle time. Usually components are allowed to fudge a few percent either way of their specifications.

In short you need SIMMs that are literally 80ns or faster.

Ian Reddy, UNIX Systems Consultant Internet: Ian_Reddy@ucs.sfu.ca 13. Ooooops. I stand corrected. It was a representative from ClearPoint (where we ordered memory from) that told us that some SS2's require 70ns simms... Sorry 'bout that...

Brent Bice bice@atlantis.ees.anl.gov "Visualize Whirled Peas!" 14. Our SUN rep said that some SS2's require 70 ns, and some only require 80 ns... Haven't verified this for factual tho. 15.

order 80A instead of 80. There were some problems with the 80nano in SS2's even. 16.

Net wisdom seems to be they must be NO WORSE THAN 80 nS - so if you buy 80 nS simms that are actually 83 nS, they won't work.

Having said that, I have tried 48 Mbytes of 3rd party, 80 nS simms (from Stratum Technology) in our SS2 and they all worked fine.

Hope this helps. = Tony Dale (tony@cosc.canterbury.ac.nz) = 17.

I have been told the same story by several different sources both at Sun and at memory vendors. The SS1, etc, uses 80ns +/-2ns SIMMs. By contrast, the SS2 uses 80ns +0ns -2ns. Since this is within +/-2, the SS2 SIMMs should work just fine in SS1s, IPCs, etc. The reverse is *not* true. 70ns SIMMs might work in a SS2. 18.

My memory supplier sez you need 70ns simms, low-cut style. Yah, you can use these in the slower IPC and SS1. Regards, John -- John R. F. McMacken mcmacken@watserv1.waterloo.edu 19.

Well, my local Sun engineer recommends Parity Systems for SPARCstation 2 third party memory. Evidently, the SS2 is VERY sensitive to the speed of its SIMMS. If the chips on the SIMM run at all slower than 80 ns, the SS2 is very unhappy. My engineer also stated that Clearpoint memory for the SS2 is causing the most trouble. Helios memory has caused a few complaints. I would stick with Parity Systems. Their number is (408) 378-1000. They will probably want to give you the number of a local group that can give you prices quotes. I personnally have not had the opportunity to experience what you are about to....

Robin Moore (moore@nlm.nih.gov) National Library of Medicine Bethesda, Maryland 20.

> My Sun "pre-sales" tech support guy told me that the SS2 was designed with > very close tolerances for memory access time. Most "80ns" DRAM is > actually 80 +/- 2 ns, but the SS2 will have problems if the access time is > even slightly greater than 80 ns; "80 +0/-2" is the way he characterized > the requirement. This is a matter of quality control in DRAM manufacture, > so hopefully there are some open-market sources of certified "80 +0" ns > SIMMs out there. If you have some, give them a try; if there are no > problems, then no problem. When shopping for them, you should insist that > the vendor warrant them for use in the SS2. Of course, you may pay a > premium for this degree of assurance; you may even find that Sun's prices > are competitive 8^). > > Karl A. Anderson | Internet: karl@forest.gsfc.nasa.gov > NASA/GSFC code 923 (STX) | voice: (301) 286-3815 > Greenbelt, MD 20771 | #include "std_disclaimer"

Hope it helps . . .

-- Phil Ritzenthaler The Advanced Computing Center for the Arts & Design (ACCAD) (614) 292-3416 ARPA: phil@grumpy.cgrg.ohio-state.edu



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:15 CDT