🚀 go-pugleaf

In <4kecjv$fsl@geraldo.cc.utexas.edu>, ignasi@ccwf.cc.utexas.edu (Ignacio Garcia) writes:
>Hi there!
>
>I'm planning to upgrade my computer's memory. Right now I have 16MB (70ns)
>installed. I wanted to put 16 or 32 more MB. I've got a good deal for 60ns
>SIMMS, but I would like to know if there are problems if you put 60ns SIMMS
>together with 70ns SIMMS. Any advice?
>
>TIA
>
>Ignacio

Just change your CMOS settings to reflect 70ns memory and you should be fine.
Warp should not care what is the speed of your memory, but CMOS settings
should reflect the slowest speed you are using.

-- Harri
---------------------------------------------------------------------
Harri Kaukovuo                          "Nuoruuden voi tuhlata muttei saastaa"
142 Elm Street #204                             -Pelle Miljoona
San Mateo, Ca 94401,USA             Email: tiijahar@dsp.com

In <4kecjv$fsl@geraldo.cc.utexas.edu>, ignasi@ccwf.cc.utexas.edu (Ignacio Garcia) writes:
>Hi there!
>
>I'm planning to upgrade my computer's memory. Right now I have 16MB (70ns)
>installed. I wanted to put 16 or 32 more MB. I've got a good deal for 60ns
>SIMMS, but I would like to know if there are problems if you put 60ns SIMMS
>together with 70ns SIMMS. Any advice?
>
>TIA
>
>Ignacio
>
>______________
>Ignacio Garcia    -Team OS/2-
>ignasi@ccwf.cc.utexas.edu
>http://ccwf.cc.utexas.edu
>"OS/2 is what DOS and Windows should have been. Windows NT
>aspires to become what OS/2 is." - Lee Reiswig, May, 1993.
>______________
>
>Computers are useless.  They can only give you answers.  - Pablo Picasso
>

I have 2 16M SIMM in my Pentium 100Mhz. One is 60ns and the other is 70ns. The 
thing that you have to care about is which one you will put in the first bank!

You must put the most rapid one in the first bank. If you put the 70ns in the first
bank, all your memory will works at 70ns. If you put the 60ns in the first bank, 
and if with luck, the 70ns can work at 60ns, your system will work at 60ns.
If not, your system will simply switch at 70ns!

Don't forget that the string written on the chip means that the manufacturer 
garantie that speed. In many case, the chip can work at better speed.

If you want to be sure, bring your 70ns SIMM at your store and ask the vendor to
test the speed of your SIMM.

In my case, my 70ns SIMM test very good at 60ns!!!!

[=======================================
[ Yves Boisjoly
[ OS/2 Warp technical support graduate at IAT.
[ OS/2 Team.
[ E-Mail: boisjol@ibm.net
[ Phone:  (514) 522-5079
[=======================================

In <4kecjv$fsl@geraldo.cc.utexas.edu>, ignasi@ccwf.cc.utexas.edu (Ignacio Garcia) writes:
>Hi there!
>
>I'm planning to upgrade my computer's memory. Right now I have 16MB (70ns)
>installed. I wanted to put 16 or 32 more MB. I've got a good deal for 60ns
>SIMMS, but I would like to know if there are problems if you put 60ns SIMMS
>together with 70ns SIMMS. Any advice?

Warp per se doesn't care, but the machine may; check your manuals to find
out whether it will tolerate such a mix. There _is_ such a thing as a chip that
responds too fast for the rest of the circuit. Odds are that in most designs
this won't be an issue, but only your manufacturer can say for sure.


 ----------------------------------------------------------------------- 
Joseph J. Kesselman             http://pages.prodigy.com/NY/keshlam/
"This note is a production of Novalabs Consulting, which is solely       
responsible for its content. Opinions not necessarily those of IBM."

In article <4khecg$3g8e@news-s01.ny.us.ibm.net>,
boisjol@ibm.net (Yves Boisjoly) says:

>I have 2 16M SIMM in my Pentium 100Mhz. One is 60ns and the other is 70ns. The 
>thing that you have to care about is which one you will put in the first bank!

>You must put the most rapid one in the first bank. If you put the 70ns in
>the first bank, all your memory will works at 70ns. If you put the 60ns in
>the first bank, and if with luck, the 70ns can work at 60ns, your system will
>work at 60ns. If not, your system will simply switch at 70ns!

This is untrue, even with EDO memory. The machine will work at whatever
speed is selected for the memory (and in the case of EDO, at the speed
per individual bank that you have selected).

No motherboard chipset can "know", without being told via CMOS, the speed of
a memory chip. Even if some manufacturer were to be stupid enough to make
one that would "guess", it probably wouldn't hold true over various chips
in different banks.

	-Kenny

-- 
Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting
kenny@panix.com		CI$: 75032,3044	  +1 617 524 5929/4949 Home/Office
16 Plainfield St, Boston, MA 02130-3633	  +1 617 983 9410      Fax
OS/2 box: pkenny.tiac.net (when I'm online) Get Warp-ed! OS/2 3.0 is here NOW!

Hi, guys!

First of all, thanks to all of you who replied to my post.

Second, I finally got the SIMMS. I decided to get the 70ns SIMMS (they're the
same price). I upgraded from 16 MB to 40MB (had to take off 2 4MB SIMMS :-(  )
Anyway, the system FLYES!!! It's a 486/66, and I assure you my system is
even faster than this P90 - 16MB from a friend of mine. Although Warp is
almost using all memory (just has 3-8MB free at normal use), swapping
has almost dissappeared!! Before, even when I maximized programs like
PMMail, swapping was active, reducing performance. Now everything comes
instantly. Multitasking is MUCH faster, and even redrawing the WPS is almost
instantly. I have set a heavy bitmap for the background, and before the upgrade
I usually had to wait for 1-2 seconds to get the bitmap redraw (of course I had
it disabled, it wasn't worth it). Now it comes instantly. Even WebExplorer is
faster (I can't belive it ;-)   ). And I have no words for Lotus Ami Pro. Before
it was faster to load Word (including the Windows sub-system) than Ami Pro.
Now Ami Pro works like an OS/2 program should.

Well, as you see, I'm very happy. For all the people who are thinking about
getting a faster processor, I have to say: Upgrade memory first. It's worth it
even if you have 16MB.

Again, my thanks to you all for all the replies I've got.

Cheers,

Ignacio

______________
Ignacio Garcia    -Team OS/2-
ignasi@ccwf.cc.utexas.edu
http://ccwf.cc.utexas.edu
"OS/2 is what DOS and Windows should have been. Windows NT
aspires to become what OS/2 is." - Lee Reiswig, May, 1993.
______________

New for Christmas: The John Bobbitt doll (Some assembly required)

In <4kgg09$71n@news.wco.com>, tiijahar@dsp.com  (Harri Kaukovuo) writes:
>Warp should not care what is the speed of your memory, but CMOS settings
>should reflect the slowest speed you are using.

	Assuming of course your PC will work with slower memory...we've sold
boards that required 60ns memory for Pentium 100s or faster...

Steve

In <4khecg$3g8e@news-s01.ny.us.ibm.net>, boisjol@ibm.net   (Yves Boisjoly) writes:
>I have 2 16M SIMM in my Pentium 100Mhz. One is 60ns and the other is 70ns. The 
>thing that you have to care about is which one you will put in the first bank!
>
>You must put the most rapid one in the first bank. If you put the 70ns in the first
>bank, all your memory will works at 70ns. If you put the 60ns in the first bank, 
>and if with luck, the 70ns can work at 60ns, your system will work at 60ns.
>If not, your system will simply switch at 70ns!
>

Forgive me, but  ..... huh???

The above makes no sense?? Is this something you have seen personally?
I suppose anything can happen, but I can't think of a reason why the speed
and the slot matter, especially since, as you say, most 70ns will run at 60ns.

Sounds like something else went wrong in your case.

What kind of local organization can check the speed of a RAM chip? That is
no small request.

Regards,

John

In <4kecjv$fsl@geraldo.cc.utexas.edu>, ignasi@ccwf.cc.utexas.edu (Ignacio Garcia) writes:
>Hi there!
>
>I'm planning to upgrade my computer's memory. Right now I have 16MB (70ns)
>installed. I wanted to put 16 or 32 more MB. I've got a good deal for 60ns
>SIMMS, but I would like to know if there are problems if you put 60ns SIMMS
>together with 70ns SIMMS. Any advice?
>

Go for it. With Ram prices falling so fast lately, keep an eye out and make 
sure you are getting a good deal. A 60ns chip can replace a 70ns with no
problem.

I think there are other potential caveats with OS/2 regarding mixing of 3 chip
and 9 chip simms, maybe someone who has tried it knows the scoop on that.

John

In <4kecjv$fsl@geraldo.cc.utexas.edu>, ignasi@ccwf.cc.utexas.edu (Ignacio Garcia) writes:
>Hi there!
>
>I'm planning to upgrade my computer's memory. Right now I have 16MB (70ns)
>installed. I wanted to put 16 or 32 more MB. I've got a good deal for 60ns
>SIMMS, but I would like to know if there are problems if you put 60ns SIMMS
>together with 70ns SIMMS. Any advice?
>
>TIA
>
>Ignacio
>
>______________
>Ignacio Garcia    -Team OS/2-
>ignasi@ccwf.cc.utexas.edu
>http://ccwf.cc.utexas.edu
>"OS/2 is what DOS and Windows should have been. Windows NT
>aspires to become what OS/2 is." - Lee Reiswig, May, 1993.
>______________
>
>Computers are useless.  They can only give you answers.  - Pablo Picasso
>

You will have no problems adding faster memory.  Just don't add parity
and non-parity memory

_+_+_+_+_+_+_+_+_+_+_+_+_+_+_+_+_+_+_+_+_

"The bitterness of poor quality lingers long after
the sweetness of low price has been forgotten.
                                danny rubis
                                Certified OS/2 Engineer
LooK Incorporated
76505.125@compuserve.com

In article <316d9bc7.6591325@news.demon.co.uk>,
john@jken.demon.co.uk (John Keniry) says:

>Only 3-8 Meg free out of 40 Meg total?!? Nurse I think I feel faint!

Don't be fooled. This ain't DOS/Windows.

	-Kenny

-- 
Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting
kenny@panix.com		CI$: 75032,3044	  +1 617 524 5929/4949 Home/Office
16 Plainfield St, Boston, MA 02130-3633	  +1 617 983 9410      Fax
OS/2 box: pkenny.tiac.net (when I'm online) Get Warp-ed! OS/2 3.0 is here NOW!

In article <316d9bc7.6591325@news.demon.co.uk>,
john@jken.demon.co.uk (John Keniry) wrote:
>ignasi@ccwf.cc.utexas.edu (Ignacio Garcia) wrote:
>
>>Second, I finally got the SIMMS. I decided to get the 70ns SIMMS (they're the
>>same price). I upgraded from 16 MB to 40MB (had to take off 2 4MB SIMMS :-(  )
>>Anyway, the system FLYES!!! It's a 486/66, and I assure you my system is
>>even faster than this P90 - 16MB from a friend of mine. Although Warp is
>>almost using all memory (just has 3-8MB free at normal use),
>
>Only 3-8 Meg free out of 40 Meg total?!? Nurse I think I feel faint!
>

 Amazing, isn't it? <s> OS/2 actually *manages* the memory people spend 
good money on, instead of letting it lie fallow like a different OS might.









-- 
     A. = aknight@wyoming.com
           aknight@yvv.com
/////////////////////////////////////\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
Something I cannot see puts upward libidinous prongs,                       
Seas of bright juice suffuse heaven.                              --Walt Whitman
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\///////////////////////////////////////

1)  You should be able to tell your motherboard that all of your memory 
is 70ns, and it will run it all at that speed.

2)  In your area, there should be computer swap meets.  I would take 
your 16 MB of memory there (presumably two 8 MB simms), take your credit 
card (CRUCIAL!) and trade your 16 MB + $$$ for two 16 MB 60 ns simms.  
Most memory vendors at computer swap meets will give you a reasonable 
price for your old memory, and you can get the upgraded memory for a 
good price.

This will save you simm slots on your mother board and get you 60 ns 
parts (so you can actually USE the 60 ns performance).

Use the credit card so that if the parts don't work, and you can't get 
ahold of the guy who sold you the memory in the first place, you can 
call your credit card company and have them return your money.  This 
protects you a bit more than using cash does.

Good luck. - k.

In <316d9bc7.6591325@news.demon.co.uk>, john@jken.demon.co.uk (John Keniry) writes:
>ignasi@ccwf.cc.utexas.edu (Ignacio Garcia) wrote:
>
>>Second, I finally got the SIMMS. I decided to get the 70ns SIMMS (they're the
>>same price). I upgraded from 16 MB to 40MB (had to take off 2 4MB SIMMS :-(  )
>>Anyway, the system FLYES!!! It's a 486/66, and I assure you my system is
>>even faster than this P90 - 16MB from a friend of mine. Although Warp is
>>almost using all memory (just has 3-8MB free at normal use),
>
>Only 3-8 Meg free out of 40 Meg total?!? Nurse I think I feel faint!
>
>Now where did I put that C64?
>
>johnk


=======================================================
I believe that OS/2 only uses the 1st 16Mb for processing and will use the 
additional RAM for swapping.  Once this is full, it will swap to disk.  This is 
why only 3-8Mb is free.  I have 32Mb in my Pentium 133 and I have only 512K 
available after booting.  I autostart about 15 programs, however!

TTYL, ROg!
rogerf@opi.mt.gov
Member TEAM OS/2 and proud of it!!
=======================================================
Running WARP Fullpack on a Packard Bell 486DX2-50 and loving it!
(My other PC is a Pentium!)
=======================================================
If it weren't for this darned computer, I'd be Mountain biking!!
=======================================================

Let me ask you a question. How can you say this:

In article <4kmeke$d1p@umt.umt.edu>, rogerf@opi.mt.gov (Roger Fuchs) says:

>Member TEAM OS/2 and proud of it!!

Then spout the following lie?

>I believe that OS/2 only uses the 1st 16Mb for processing and will use the 
>additional RAM for swapping.

Please, if you spread falsehoods like this, you're only *hurting* OS/2 and
the Team, and not helping.

The 16MB restriction has been discussed to death here, and a couple of posters
have even *gotten it right*. Look for my post on it in another thread in
.setup.misc .

	-Kenny

-- 
Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting
kenny@panix.com		CI$: 75032,3044	  +1 617 524 5929/4949 Home/Office
16 Plainfield St, Boston, MA 02130-3633	  +1 617 983 9410      Fax
OS/2 box: pkenny.tiac.net (when I'm online) Get Warp-ed! OS/2 3.0 is here NOW!

That is only true on ISA bus systems with a bus mastering device (such as
Adeptec SCSI).  Otherwise under VLB, and PCI the whole thing is useable under
OS/2

In <4kk4tf$ksq@news.aros.net>, John Brush@aros.net writes:
>In <4khecg$3g8e@news-s01.ny.us.ibm.net>, boisjol@ibm.net   (Yves Boisjoly) writes:
>>I have 2 16M SIMM in my Pentium 100Mhz. One is 60ns and the other is 70ns. The 
>>thing that you have to care about is which one you will put in the first bank!
>>
>>You must put the most rapid one in the first bank. If you put the 70ns in the first
>>bank, all your memory will works at 70ns. If you put the 60ns in the first bank, 
>>and if with luck, the 70ns can work at 60ns, your system will work at 60ns.
>>If not, your system will simply switch at 70ns!
>>
>
>Forgive me, but  ..... huh???
>
>The above makes no sense?? Is this something you have seen personally?
>I suppose anything can happen, but I can't think of a reason why the speed
>and the slot matter, especially since, as you say, most 70ns will run at 60ns.
>
>Sounds like something else went wrong in your case.
>
>What kind of local organization can check the speed of a RAM chip? That is
>no small request.
>
>Regards,
>
>John

Well every body, please accept my excuses for this stupid thing...

My salesman told me this when I was excited to bring this new 16M SIMM at home.
After your post and a little bit of thinking... I admit that this not sound so very
practicle.

I must admit that I don't have any trouble with those SIMM anyway, and I am now
testing it at 0 WAIT STATE without trouble...
Again, excuse my old advice and please forget it!!!!

[=======================================
[ Yves Boisjoly
[ OS/2 Warp technical support graduate at IAT.
[ OS/2 Team.
[ E-Mail: boisjol@ibm.net
[ Phone:  (514) 522-5079
[=======================================

In <316d9bc7.6591325@news.demon.co.uk>, john@jken.demon.co.uk (John Keniry) writes:
>ignasi@ccwf.cc.utexas.edu (Ignacio Garcia) wrote:
>
>>Second, I finally got the SIMMS. I decided to get the 70ns SIMMS (they're the
>>same price). I upgraded from 16 MB to 40MB (had to take off 2 4MB SIMMS :-(  )
>>Anyway, the system FLYES!!! It's a 486/66, and I assure you my system is
>>even faster than this P90 - 16MB from a friend of mine. Although Warp is
>>almost using all memory (just has 3-8MB free at normal use),
>
>Only 3-8 Meg free out of 40 Meg total?!? Nurse I think I feel faint!
>
>Now where did I put that C64?
>
>johnk

Just next to my DATASET!!  ;-)

[=======================================
[ Yves Boisjoly
[ OS/2 Warp technical support graduate at IAT.
[ OS/2 Team.
[ E-Mail: boisjol@ibm.net
[ Phone:  (514) 522-5079
[=======================================

It's never wise to mix memory types or speeds.  Although it sometimes
works, many times you can get very subtle or latent errors.  Or, it can
work for six months and then get flaky as components age and parameters
drift.  Or the memory may work fine within a limited temperature range,
but fail if it gets a little warmer or cooler than normal.

You have to decide if the risk is worth it.  Keep in mind that backups
may not help, as your backup may be what got corrupted.

Alan

boisjol@ibm.net   (Yves Boisjoly) writes:

>In <4kecjv$fsl@geraldo.cc.utexas.edu>, ignasi@ccwf.cc.utexas.edu (Ignacio Garcia) writes:
>>Hi there!
>>
>>I'm planning to upgrade my computer's memory. Right now I have 16MB (70ns)
>>installed. I wanted to put 16 or 32 more MB. I've got a good deal for 60ns
>>SIMMS, but I would like to know if there are problems if you put 60ns SIMMS
>>together with 70ns SIMMS. Any advice?
>>
>>TIA
>>
>>Ignacio
>>
>>______________
>>Ignacio Garcia    -Team OS/2-
>>ignasi@ccwf.cc.utexas.edu
>>http://ccwf.cc.utexas.edu
>>"OS/2 is what DOS and Windows should have been. Windows NT
>>aspires to become what OS/2 is." - Lee Reiswig, May, 1993.
>>______________
>>
>>Computers are useless.  They can only give you answers.  - Pablo Picasso
>>

>I have 2 16M SIMM in my Pentium 100Mhz. One is 60ns and the other is 70ns. The 
>thing that you have to care about is which one you will put in the first bank!

>You must put the most rapid one in the first bank. If you put the 70ns in the first
>bank, all your memory will works at 70ns. If you put the 60ns in the first bank, 
>and if with luck, the 70ns can work at 60ns, your system will work at 60ns.
>If not, your system will simply switch at 70ns!

>Don't forget that the string written on the chip means that the manufacturer 
>garantie that speed. In many case, the chip can work at better speed.

>If you want to be sure, bring your 70ns SIMM at your store and ask the vendor to
>test the speed of your SIMM.

>In my case, my 70ns SIMM test very good at 60ns!!!!

>[=======================================
>[ Yves Boisjoly
>[ OS/2 Warp technical support graduate at IAT.
>[ OS/2 Team.
>[ E-Mail: boisjol@ibm.net
>[ Phone:  (514) 522-5079
>[=======================================

-- 
Alan J. Brumbaugh    Neither Motorola nor I speak for each other.
Motorola General Systems Sector    email: brumba@cig.mot.com
1501 West Shure Drive              voice: +1 847 632 5040
Arlington Heights, IL  60004         fax: +1 847 632 4552

In message <4kgg09$71n@news.wco.com> - tiijahar@dsp.com  (Harri Kaukovuo) write
s:
->>I'm planning to upgrade my computer's memory. Right now I have 16MB (70ns)
->>installed. I wanted to put 16 or 32 more MB. I've got a good deal for 60ns
->>SIMMS, but I would like to know if there are problems if you put 60ns SIMMS
->>together with 70ns SIMMS. Any advice?
->>
->Just change your CMOS settings to reflect 70ns memory and you should be fine.
->Warp should not care what is the speed of your memory, but CMOS settings
->should reflect the slowest speed you are using.
->-- Harri
->---------------------------------------------------------------------
->Harri Kaukovuo                          "Nuoruuden voi tuhlata muttei saastaa"
->142 Elm Street #204                             -Pelle Miljoona
->San Mateo, Ca 94401,USA             Email: tiijahar@dsp.com

Isn't 70 ns a velocity measurement, therefore the larger the number the 
faster the access....

Damn....where's the off button..... :>

===============================================================
stonet@pathcom.com				Toronto, Canada

In <4klhs4$2fh@panix2.panix.com>, kenny@panix.com (Kenneth Crudup) writes:
>In article <316d9bc7.6591325@news.demon.co.uk>,
>john@jken.demon.co.uk (John Keniry) says:
>
>>Only 3-8 Meg free out of 40 Meg total?!? Nurse I think I feel faint!
>

Before you, or anyone else, jumps all over the OS, take a little time and learn
how OS/2 handles memory. It IS NOT AT ALL like dos or windows. The
fact that you SEEM to be using all that memory is like complaining that your
gas tank is full, and you have no room for more. Check it out first. It matters.

John

In message <4kmop1$2f7g@news-s01.ny.us.ibm.net> - boisjol@ibm.net   (Yves Boisj
oly) writes:
:>
:>In <4kk4tf$ksq@news.aros.net>, John Brush@aros.net writes:
:>>In <4khecg$3g8e@news-s01.ny.us.ibm.net>, boisjol@ibm.net   (Yves Boisjoly) writes:
:>>>I have 2 16M SIMM in my Pentium 100Mhz. One is 60ns and the other is 70ns. The 
:>>>thing that you have to care about is which one you will put in the first bank!
:>>>
:>>>You must put the most rapid one in the first bank. If you put the 70ns in the first
:>>>bank, all your memory will works at 70ns. If you put the 60ns in the first bank, 
:>>>and if with luck, the 70ns can work at 60ns, your system will work at 60ns.
:>>>If not, your system will simply switch at 70ns!
:>>>

That is a BIG NO-NO!!!

The recommended thing to do is put the SLOWEST memory in the first bank!  The 
reason is this: the fast chip can always slow down to the speed of the slow 
chip, but the slow chip can't keep up with the fast one...

OS/2 will access ALL memory at the speed of the chips in bank 1.  So if you 
have the 60ns chips in bank 1, OS/2 will access the 70ns chips at 60ns... and 
you'll probably bomb.

It's never recommended to mix n match speeds, but if you have to, follow the 
above rule.

BTW - when marching, the military always puts the smallest guys in front, so 
that the big guys won't run away from the small ones.  Same reasoning.  Slow 
in front, fast in back, and for much the same reason.

Bill
Team OS/2

I got this info from page 804, Inside OS/2 Warp

In message <4kmvo1$ol3@news.pathcom.com> - stonet@pathcom.com (stonet) writes:

:>
:>In message <4kgg09$71n@news.wco.com> - tiijahar@dsp.com  (Harri Kaukovuo) write
:>s:
:>->>I'm planning to upgrade my computer's memory. Right now I have 16MB (70ns)
:>->>installed. I wanted to put 16 or 32 more MB. I've got a good deal for 60ns
:>->>SIMMS, but I would like to know if there are problems if you put 60ns SIMMS
:>->>together with 70ns SIMMS. Any advice?
:>->>
:>->Just change your CMOS settings to reflect 70ns memory and you should be fine.
:>->Warp should not care what is the speed of your memory, but CMOS settings
:>->should reflect the slowest speed you are using.
:>->-- Harri
:>->---------------------------------------------------------------------
:>->Harri Kaukovuo                          "Nuoruuden voi tuhlata muttei saastaa"
:>->142 Elm Street #204                             -Pelle Miljoona
:>->San Mateo, Ca 94401,USA             Email: tiijahar@dsp.com
:>
:>Isn't 70 ns a velocity measurement, therefore the larger the number the 
:>faster the access....
:>
:>Damn....where's the off button..... :>
:>
:>===============================================================
:>stonet@pathcom.com				Toronto, Canada
:>

If I didn't have a sense of humor, I'd say that the speed of the chip (70ns, 
or whatever) refers to the time it takes for the memory to go from a 0 to a 1 
and back again.  Commonly refered to as the ramp time.  Ideally, a chip 
should have a ramp time of 0ns, but that's impossible to achieve.

We would like to see a ramp that looks like this on an O'scope:

   __
__|  |__

but they always look more like:

   __
__/  \__

Bill
Team OS/2

And up too late...  ;)

In <4kmvo1$ol3@news.pathcom.com>, stonet@pathcom.com (stonet) writes:
>>Isn't 70 ns a velocity measurement, therefore the larger the number the 
>faster the access....
>

To make it simple (like that's possible<g>) 70ns refers to the amount of time 
it takes the RAM to respond to requests for data reads or writes. 
From the point of view of the CPU, its the amount of time that the CPU must 
spend at that device to either read or write data. Therefore, 60ns is faster 
than 70ns, since the CPU get/put data from/to that device 10ns quicker, 
allowing the CPU to get on with the instruction faster.

If the RAM access speed is too slow, then the BIOS must be set to provide 
what are called wait-states. A wait-state is used to tell the CPU that the 
RAM (or any other device) is too slow and that the CPU will have to wait
for a pre-determined time in order for the accessed device to 'catch-up' to 
the CPU.

If your system uses slow RAM, it must insert wait-states, which by their
definition, tell the CPU to 'wait for me' Obviously, this will impact the 
performance of the system. 

How fast is fast, and how slow is slow? That can only be answered by 
the individual system clocks and RAM speeds. 

Usually, and I say that with a great deal of trepidation <g>, you will find
70ns devices in a 486 machine, while the higher speed pentiums will
be found with 60ns. There are always exceptions, but its something to keep
in mind if you intend to buy RAM for your 486, with the intent to one day
upgrade to a faster pentium. In that case, you would want to get 60ns so that
when you upgrade the CPU, you won't have to turn on any wait-states.

Did I say that clearly???

John

In article <4kq5f0$ehl@usenet6.interramp.com>,
ir003870@interramp.com (Billy Ko) says:

>The recommended thing to do is put the SLOWEST memory in the first bank!  The 
>reason is this: the fast chip can always slow down to the speed of the slow 
>chip, but the slow chip can't keep up with the fast one.

*sigh*

Where do people *get* this bullsh*t?

>BTW - when marching, the military always puts the smallest guys in front, so 
>that the big guys won't run away from the small ones.  Same reasoning.  Slow 
>in front, fast in back, and for much the same reason.

So?

>I got this info from page 804, Inside OS/2 Warp

I don't believe it.

I've worked with MBs where the *largest* bank of memory had to be bank 0, (due
to refresh issues) but that has nothing to do with speed. I've mixed speeds
between banks all the time.

A MOTHERBOARD CANNOT GUESS THE SPEED OF A BANK OF MEMORY! It will *not*
know until you program it to CMOS. Set the speed in non-EDO systems to that
of the slowest memory *in any bank*, and it will all behave fine!

	-Kenny

-- 
Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting
kenny@panix.com		CI$: 75032,3044	  +1 617 524 5929/4949 Home/Office
16 Plainfield St, Boston, MA 02130-3633	  +1 617 983 9410      Fax
OS/2 box: pkenny.tiac.net (when I'm online) Get Warp-ed! OS/2 3.0 is here NOW!

In message <4ksq7v$kob@panix2.panix.com> - kenny@panix.com (Kenneth
Crudup)15 Apr 1996 02:27:11 -0400 writes:
:>>And exactly how do you do this?  The CMOS on many of the boards I've seen
:>>have no option to do this.
:>
:>Yes, they do- it's just not called "bank speed" or anything like that.
:>
:>In most AMI BIOS', the "Advanced Chipset Setup" page has entries for
things
:>like DRAM Read Wait States, DRAM Writes WS, CAS time, RAS time, and the
like.
:>These control the timing for memory on the motherboard.
:>
:>If you don't know what you're doing (i.e., don't know exactly what each of
:>those terms mean), then use the factory settings (newer BIOS' have an
"Auto-
:>Configure" option.

And this is the exact point I was making.  You were raving about setting the
speed and no such option exists.  Had you stated what you did in your
response to my first message, I would then have only taken exception with
your last statement above.  You say it can't sense the speed and then you
turn right around and say to use the default.  If you can use the default
then you don't need to change the setting in the first place.

Your whole tirade makes absolutely no sense, you've totally contradicted
what you said first.

-------------------------------------------------------------------
Floyd Drennon <fdrennon@pobox.com>
OS/2 & Lan Server Certified Instructor
Comp-U-Comm - Computer & Communication Consultants
-------------------------------------------------------------------

In <4kmvo1$ol3@news.pathcom.com>, stonet@pathcom.com (stonet) writes:
>Isn't 70 ns a velocity measurement, therefore the larger the number the 
>faster the access....

	No, nano-SECONDS is a time measurement.  :)

Steve

In article <4ktfu9$1mu4@mule2.mindspring.com>,
fdrennon@pobox.com (Floyd Drennon) says:

>And this is the exact point I was making.  You were raving about setting the
>speed and no such option exists.

There is no simple "speed" option for the people who have no business in there
in the first place. Memory timing is done by a series of operations, several
of which can be set.

>You say it can't sense the speed and then you turn right around and say to
>use the default.  If you can use the default then you don't need to change
>the setting in the first place.

>Your whole tirade makes absolutely no sense, you've totally contradicted
>what you said first.

*sigh*

The key word is "If you don't know what you're doing". Memory timing is not
for people who can't understand simple electronics.

Some people, like yourself, evidently, don't mind driving a family station
wagon. Others like to soup up their engines. I don't leave *any* of my MBs
on the "default"- that's the *slowest*, "most safe" speed and guaranteed
to leave your machine running at far less than it's potential because it
uses the most pessimistic settings for memory timing. There is no "contra-
diction, because the machine is still only assuming the memory inserted
is at least fast enought to keep up with it's slowest speed (100 ns?). I put
that in there to avoid tripping up the adventurous, so those who don't
know WTF they're doing won't come to me looking for help.

	-Kenny

-- 
Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting
kenny@panix.com		CI$: 75032,3044	  +1 617 524 5929/4949 Home/Office
16 Plainfield St, Boston, MA 02130-3633	  +1 617 983 9410      Fax
OS/2 box: pkenny.tiac.net (when I'm online) Get Warp-ed! OS/2 3.0 is here NOW!

On 15 Apr 1996, Floyd Drennon wrote:

> In message <4krr9c$nbv@panix2.panix.com> - kenny@panix.com (Kenneth
> Crudup)14 Apr 1996 17:38:52 -0400 writes:
> :>A MOTHERBOARD CANNOT GUESS THE SPEED OF A BANK OF MEMORY! It will *not*
> :>know until you program it to CMOS. Set the speed in non-EDO systems to
> that
> :>of the slowest memory *in any bank*, and it will all behave fine!
> 
> And exactly how do you do this?  The CMOS on many of the boards I've seen
> have no option to do this?
> 
> -------------------------------------------------------------------
> Floyd Drennon <fdrennon@pobox.com>
> OS/2 & Lan Server Certified Instructor
> Comp-U-Comm - Computer & Communication Consultants
> -------------------------------------------------------------------
> 
In my bios, you have the option of setting the number of wait states the 
system applies on a bank-by-bank basis.  There are no specific speed 
callouts, but if you are installing memory that is slower than the design 
speed, and having memory problems, I would start kicking in wait states on 
the slow bank until the memory problems went away.

Regards -- Ed

In message <Pine.BSD/.3.91.960415114357.13045B-100000@matrix.eden.com> -
edwin bailen <n5kzw@eden.com>Mon, 15 Apr 1996 11:48:56 -0500 writes:

:>In my bios, you have the option of setting the number of wait states the 
:>system applies on a bank-by-bank basis.  There are no specific speed 
:>callouts, but if you are installing memory that is slower than the design 
:>speed, and having memory problems, I would start kicking in wait states on
:>the slow bank until the memory problems went away.

Guess I must be using a different type of board.  Of all the boards I've
seen, none have had this feature to set by bank.  That's interesting to
know.  What type BIOS do you have

-------------------------------------------------------------------
Floyd Drennon <fdrennon@pobox.com>
OS/2 & Lan Server Certified Instructor
Comp-U-Comm - Computer & Communication Consultants
-------------------------------------------------------------------

In message <4krr9c$nbv@panix2.panix.com> - kenny@panix.com (Kenneth
Crudup)14 Apr 1996 17:38:52 -0400 writes:
:>A MOTHERBOARD CANNOT GUESS THE SPEED OF A BANK OF MEMORY! It will *not*
:>know until you program it to CMOS. Set the speed in non-EDO systems to
that
:>of the slowest memory *in any bank*, and it will all behave fine!

And exactly how do you do this?  The CMOS on many of the boards I've seen
have no option to do this?

-------------------------------------------------------------------
Floyd Drennon <fdrennon@pobox.com>
OS/2 & Lan Server Certified Instructor
Comp-U-Comm - Computer & Communication Consultants
-------------------------------------------------------------------

>>Set the speed in non-EDO systems to that of the slowest memory *in any bank*,
>>and it will all behave fine!

In article <4ksenf$1sl0@mule2.mindspring.com>,
fdrennon@pobox.com (Floyd Drennon) says:

>And exactly how do you do this?  The CMOS on many of the boards I've seen
>have no option to do this.

Yes, they do- it's just not called "bank speed" or anything like that.

In most AMI BIOS', the "Advanced Chipset Setup" page has entries for things
like DRAM Read Wait States, DRAM Writes WS, CAS time, RAS time, and the like.
These control the timing for memory on the motherboard.

If you don't know what you're doing (i.e., don't know exactly what each of 
those terms mean), then use the factory settings (newer BIOS' have an "Auto-
Configure" option.

	-Kenny

-- 
Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting
kenny@panix.com		CI$: 75032,3044	  +1 617 524 5929/4949 Home/Office
16 Plainfield St, Boston, MA 02130-3633	  +1 617 983 9410      Fax
OS/2 box: pkenny.tiac.net (when I'm online) Get Warp-ed! OS/2 3.0 is here NOW!

In message <4krr9c$nbv@panix2.panix.com> - kenny@panix.com (Kenneth Crudup) wri
tes:
:>
:>In article <4kq5f0$ehl@usenet6.interramp.com>,
:>ir003870@interramp.com (Billy Ko) says:
:>
:>>The recommended thing to do is put the SLOWEST memory in the first bank!  The 
:>>reason is this: the fast chip can always slow down to the speed of the slow 
:>>chip, but the slow chip can't keep up with the fast one.
:>
:>*sigh*
:>
:>Where do people *get* this bullsh*t?
:>
:>>BTW - when marching, the military always puts the smallest guys in front, so 
:>>that the big guys won't run away from the small ones.  Same reasoning.  Slow 
:>>in front, fast in back, and for much the same reason.
:>
:>So?
:>
:>>I got this info from page 804, Inside OS/2 Warp
:>
:>I don't believe it.
:>
:>I've worked with MBs where the *largest* bank of memory had to be bank 0, (due
:>to refresh issues) but that has nothing to do with speed. I've mixed speeds
:>between banks all the time.
:>
:>A MOTHERBOARD CANNOT GUESS THE SPEED OF A BANK OF MEMORY! It will *not*
:>know until you program it to CMOS. Set the speed in non-EDO systems to that
:>of the slowest memory *in any bank*, and it will all behave fine!
:>
:>	-Kenny
:>
:>-- 
:>Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting
:>kenny@panix.com		CI$: 75032,3044	  +1 617 524 5929/4949 Home/Office
:>16 Plainfield St, Boston, MA 02130-3633	  +1 617 983 9410      Fax
:>OS/2 box: pkenny.tiac.net (when I'm online) Get Warp-ed! OS/2 3.0 is here NOW!

Excerpt from Inside OS/2 Warp, page 804:

Trap Errors and Memory Problems
Some basic things to check for when a customer is experiencing trap errors of 
any kind include:

1. All memory SIMMs must be the same size and speed...

2. All SIMMs should be from the same manufacturer.

3. If the customer has mixed memory SIMMs, it may be possible to get them to 
work, although this would be a configuration unsupported by OS/2

If mixed memory SIMMs are being used, the customer must place the SLOWER 
speed memory in the first or lowest address sockets.  This approach 
occasionally works because OS/2 tests memory in the lower address space to 
determine its speed, then runs ALL memory at that speed


And one more tip:

Please don't engage in personal type attacks, it is very unbecoming of a 
person.

Bill
Team OS/2

In article <4kuqs2$g76@usenet2.interramp.com>,
ir003870@interramp.com (Billy Ko) says:

>Excerpt from Inside OS/2 Warp, page 804:

>Some basic things to check for when a customer is experiencing trap errors of 
>any kind include:

>1. All memory SIMMs must be the same size and speed...
>2. All SIMMs should be from the same manufacturer.
>3. If the customer has mixed memory SIMMs, it may be possible to get them to 
>work, although this would be a configuration unsupported by OS/2

Unbelievable!

>If mixed memory SIMMs are being used, the customer must place the SLOWER 
>speed memory in the first or lowest address sockets.  This approach 
>occasionally works because OS/2 tests memory in the lower address space to 
>determine its speed, then runs ALL memory at that speed

This is utter garbage!

Who wrote this book?

>... because OS/2 tests memory in the lower address space to determine its
>speed, then runs ALL memory at that speed

Unbelieveable! OS/2 *cannot* "determine the speed of a bank of memory".
Period.

	-Kenny

-- 
Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting
kenny@panix.com		CI$: 75032,3044	  +1 617 524 5929/4949 Home/Office
16 Plainfield St, Boston, MA 02130-3633	  +1 617 983 9410      Fax
OS/2 box: pkenny.tiac.net (when I'm online) Get Warp-ed! OS/2 3.0 is here NOW!

On 16 Apr 1996, Kenneth Crudup wrote:

> In article <4kuqs2$g76@usenet2.interramp.com>,
> ir003870@interramp.com (Billy Ko) says:
> 
> >Excerpt from Inside OS/2 Warp, page 804:
> 
> >Some basic things to check for when a customer is experiencing trap errors of 
> >any kind include:
> 
> >1. All memory SIMMs must be the same size and speed...
> >2. All SIMMs should be from the same manufacturer.
> >3. If the customer has mixed memory SIMMs, it may be possible to get them to 
> >work, although this would be a configuration unsupported by OS/2
> 
> Unbelievable!
> 
> >If mixed memory SIMMs are being used, the customer must place the SLOWER 
> >speed memory in the first or lowest address sockets.  This approach 
> >occasionally works because OS/2 tests memory in the lower address space to 
> >determine its speed, then runs ALL memory at that speed
> 
> This is utter garbage!
> 
> Who wrote this book?
> 
> >... because OS/2 tests memory in the lower address space to determine its
> >speed, then runs ALL memory at that speed
> 
> Unbelieveable! OS/2 *cannot* "determine the speed of a bank of memory".
> Period.
> 
> 	-Kenny
> 
> -- 
> Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting
> kenny@panix.com		CI$: 75032,3044	  +1 617 524 5929/4949 Home/Office
> 16 Plainfield St, Boston, MA 02130-3633	  +1 617 983 9410      Fax
> OS/2 box: pkenny.tiac.net (when I'm online) Get Warp-ed! OS/2 3.0 is here NOW!

When I was running 2.1, I added some memory, and the system would not 
boot.  When I called the OS/2 tech support line, their statement was "Oh, 
all of the memory in the system must be of the same type and speed, and 
from the same vendor."  Right or wrong, they are at least consistent.

P.S. I haven't had an opportunity to check my Bios yet.

Regards,  Ed

In message <4l144f$75r@panix2.panix.com> - kenny@panix.com (Kenneth Crudup) wri
tes:
:>
:>In article <4kuqs2$g76@usenet2.interramp.com>,
:>ir003870@interramp.com (Billy Ko) says:
:>
:>>Excerpt from Inside OS/2 Warp, page 804:
:>
:>>Some basic things to check for when a customer is experiencing trap errors of 
:>>any kind include:
:>
:>>1. All memory SIMMs must be the same size and speed...
:>>2. All SIMMs should be from the same manufacturer.
:>>3. If the customer has mixed memory SIMMs, it may be possible to get them to 
:>>work, although this would be a configuration unsupported by OS/2
:>
:>Unbelievable!
:>
:>>If mixed memory SIMMs are being used, the customer must place the SLOWER 
:>>speed memory in the first or lowest address sockets.  This approach 
:>>occasionally works because OS/2 tests memory in the lower address space to 
:>>determine its speed, then runs ALL memory at that speed
:>
:>This is utter garbage!
:>
:>Who wrote this book?
:>
:>>... because OS/2 tests memory in the lower address space to determine its
:>>speed, then runs ALL memory at that speed
:>
:>Unbelieveable! OS/2 *cannot* "determine the speed of a bank of memory".
:>Period.
:>
:>	-Kenny
:>
:>-- 
:>Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting
:>kenny@panix.com		CI$: 75032,3044	  +1 617 524 5929/4949 Home/Office
:>16 Plainfield St, Boston, MA 02130-3633	  +1 617 983 9410      Fax
:>OS/2 box: pkenny.tiac.net (when I'm online) Get Warp-ed! OS/2 3.0 is here NOW!


How about David P. Both, lead OS/2 support person for IBM in Raleigh, NC (as 
of the writing of the book) since 1984, and has been with IBM since 1974, and 
is a Certified OS/2 Engineer FOUNDING MENBER.

Bill Stevens is an Accredited and IBM Certified OS/2 Engineer, authoring and 
conducting seminars on Supporting and T-shooting OS/2 internationally.

I would think that one of the original OS/2 Engineers and someone whose life 
revolves around troubleshooting OS/2 worldwide would know what they are 
talking about.

And, Page 175 of the User's Guide to OS/2 Warp (the one that came with Warp) 
states:

A TRAP0002 error is a hrdware error and is usually directly related to 
memory.  Mismatched memory can often cause these kinds of errors.  Your 
computer might have a variety of SIMMs that WERE PRODUCED BY DIFFERENT 
MANUFACTURERS OR THAT OPERATE AT DIFFERENT SPEEDS.

One thing that I haven't asked is... where did you learn YOUR facts?  I am 
showing ample evidence to support my claims, along with some creditable 
people who back up my claims.

All you do is argue and refute my claims, but do nothing to support yours.  
I am not stiff and unbendable.  If you can show me SOMEthing to support your 
claims, I will consider it, and if your arguement is sound, I am perfectly 
willing to accept it.

Time to put your money where your mouth is.

Bill
Team OS/2

>
>>The recommended thing to do is put the SLOWEST memory in the first bank!  The 
>>reason is this: the fast chip can always slow down to the speed of the slow 
>>chip, but the slow chip can't keep up with the fast one.
>

This must make it tuff on Pentiums with only two chips, since the CPU
loads from both chips at a time on with one big wide gulp!!

C'mon folks, lets get our facts right and just lose this bogus thread.

John

In message <4ksenf$1sl0@mule2.mindspring.com> - fdrennon@pobox.com (Floyd
Drennon) writes:
:>
:>In message <4krr9c$nbv@panix2.panix.com> - kenny@panix.com (Kenneth
:>Crudup)14 Apr 1996 17:38:52 -0400 writes:
:>:>A MOTHERBOARD CANNOT GUESS THE SPEED OF A BANK OF MEMORY! It will *not*
:>:>know until you program it to CMOS. Set the speed in non-EDO systems to
:>that
:>:>of the slowest memory *in any bank*, and it will all behave fine!
:>
:>And exactly how do you do this?  The CMOS on many of the boards I've seen
:>have no option to do this?
:>
I'd had many 486 boards over the last couple of years and I've never seen one 
that a setting for the memory speed.


Eric A. Erickson		- drowelf@vnet.net
Elvish Software Foundry		- Voice/Fax (704) 531-9022
Certified OS/2 Warp Engineer	- Team OS/2 Member   
************** Check out the OS/2 Betas Web Page **************
                http://www.teamos2.org/os2beta

In article <4l3to0$d44@usenet2.interramp.com>,
ir003870@interramp.com (Billy Ko) says:

>I would think that one of the original OS/2 Engineers and someone whose life 
>revolves around troubleshooting OS/2 worldwide would know what they are 
>talking about.

There's also a good chance that they are choosing their words carefully to
avoid support nightmares- there's a lot of hardware out there, and a lot
of people who don't know enough about their machines to give them enough
detail to get them into more trouble.

>"A TRAP0002 error is a hardware error and is usually directly related to
>memory."

In the contect of memory, it's an NMI (non-maskable interrupt) trap caused
by the parity detection circitry determining that what it stored in that 
location isn't what it's reading out.

>Mismatched memory can often cause these kinds of errors.  Your computer might
>have a variety of SIMMs that WERE PRODUCED BY DIFFERENT MANUFACTURERS OR
>THAT OPERATE AT DIFFERENT SPEEDS.

Notice use of the words "often" and "might".

>One thing that I haven't asked is... where did you learn YOUR facts?

I started off as a hardware designer. I've done memory system design.

I've also put together at least 100 systems for various clients, and in the
last 3 years since I've started using OS/2, fully 30% of those systems have
run OS/2, on various motherboards, with several different BIOS', and memory
of varying speeds, and different manufacturers. The main machine I now do my
OS/2 development work on is a 3-year old MB with 16MB of <somebody's> 70ns 30-
pin DRAM, and 16MB of <someone else's> 60ns 30-pin DRAM. I have my BIOS config
options hand-selected as far as the machine will tolerate, and it has never
trapped due to bad memory (normally TRAPs 2 or 3).

I will bet my bank account that all these alleged "mismatched memory" and
"different speeds" problems are due to *the BIOS being set improperly*, and
not due to who or what the manufacturer is, the speed of the memory, or
anything else like that. 

>I am showing ample evidence to support my claims, along with some creditable 
>people who back up my claims.

I have to be suspicious of *any* "evidence" by *anyone* who makes the
outrageous claim that "OS/2 measures the speed of memory in the first bank
and will run the rest of the machine at that speed" as I have seen reported
here as a "fact". Ludicrous.

OS/2 does not, in fact *can not*, "measure the speed of RAM and run the 
memory at that speed". All OS/2 or *any* other OS (be it DOS, NT, *NIX)
can do is expect that when it tells the CPU to "go out and get me memory
location XYZ", that the MB will *get* location "XYZ" without having to 
generate a parity trap (TRAP 0002), or return bad data (which leads to 
TRAP 3, D, and E mostly). That's all. It's up to your motherboard, and
provided you have your CMOS options set properly *to not expect to
say "The data is valid, send it to the CPU" before that is actually true*,
you will not get traps (on good memory) no matter *where* it came from.

>All you do is argue and refute my claims, but do nothing to support yours.  

See above.
I have hard-earned empirical evidence that tells me what I should know.

	-Kenny

-- 
Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting
kenny@panix.com		CI$: 75032,3044	  +1 617 524 5929/4949 Home/Office
16 Plainfield St, Boston, MA 02130-3633	  +1 617 983 9410      Fax
OS/2 box: pkenny.tiac.net (when I'm online) Get Warp-ed! OS/2 3.0 is here NOW!

Hello edwin!

Replying to a message of edwin bailen  to All:

 >>> Some basic things to check for when a customer is experiencing trap
 >>> errors of  any kind include:
 >> 
 >>> 1. All memory SIMMs must be the same size and speed...
 >>> 2. All SIMMs should be from the same manufacturer.
 >>> 3. If the customer has mixed memory SIMMs, it may be possible to get
 >>> them to  work, although this would be a configuration unsupported by
 >>> OS/2
 >> 
 >> Unbelievable!
 >> 
 >>> If mixed memory SIMMs are being used, the customer must place the
 >>> SLOWER  speed memory in the first or lowest address sockets.  This
 >>> approach  occasionally works because OS/2 tests memory in the lower
 >>> address space to  determine its speed, then runs ALL memory at that
 >>> speed
 >> 
 >> This is utter garbage!
 >> 
 >> Who wrote this book?
 >> 
 >>> ... because OS/2 tests memory in the lower address space to determine
 >>> its speed, then runs ALL memory at that speed
 >> 
 >> Unbelieveable! OS/2 *cannot* "determine the speed of a bank of memory".
 >> Period.   -Kenny  --  Kenneth R. Crudup, Unix & OS/2 Software
 >> Consultant, Scott County Consulting kenny@panix.com  CI$: 75032,3044  

 eb> When I was running 2.1, I added some memory, and the system would not 
 eb> boot.  When I called the OS/2 tech support line, their statement was
 eb> "Oh,  all of the memory in the system must be of the same type and
 eb> speed, and  from the same vendor."  Right or wrong, they are at least
 eb> consistent.
Well, that is correct, but that is because of timing problems, not with OS/2,
but with the mobo and the memory chips. These were problems of the earlier
mobo's (and some later mobos) but this problem is assumed to be over on 72-pin
simms. Obviously the 32-pin simms were not exactly enough specificated to
handle the timing problems.

As chips with same size/speed/manufacturer are to be considered compatible,
this is what was advised at the time.


Bye, Erik!
--
| Standard disclaimer: The views of this user are strictly his own.

In <4l144f$75r@panix2.panix.com>, kenny@panix.com (Kenneth Crudup) writes:
>In article <4kuqs2$g76@usenet2.interramp.com>,
>ir003870@interramp.com (Billy Ko) says:
>
>>Excerpt from Inside OS/2 Warp, page 804:
>
>>Some basic things to check for when a customer is experiencing trap errors of 
>>any kind include:
>
>>1. All memory SIMMs must be the same size and speed...
>>2. All SIMMs should be from the same manufacturer.
>>3. If the customer has mixed memory SIMMs, it may be possible to get them to 
>>work, although this would be a configuration unsupported by OS/2
>
>Unbelievable!
>
>>If mixed memory SIMMs are being used, the customer must place the SLOWER 
>>speed memory in the first or lowest address sockets.  This approach 
>>occasionally works because OS/2 tests memory in the lower address space to 
>>determine its speed, then runs ALL memory at that speed
>
>This is utter garbage!
>
>Who wrote this book?
>
>>... because OS/2 tests memory in the lower address space to determine its
>>speed, then runs ALL memory at that speed
>
>Unbelieveable! OS/2 *cannot* "determine the speed of a bank of memory".
>Period.
>

I won't be quite as emphatic, but this does sound a bit dubious.

John

ir003870@interramp.com (Billy Ko) writes:

>In message <4kmop1$2f7g@news-s01.ny.us.ibm.net> - boisjol@ibm.net   (Yves Boisj
>oly) writes:
>:>
>:>In <4kk4tf$ksq@news.aros.net>, John Brush@aros.net writes:
>:>>In <4khecg$3g8e@news-s01.ny.us.ibm.net>, boisjol@ibm.net   (Yves Boisjoly) writes:
>:>>>I have 2 16M SIMM in my Pentium 100Mhz. One is 60ns and the other is 70ns. The 
>:>>>thing that you have to care about is which one you will put in the first bank!
>:>>>
>:>>>You must put the most rapid one in the first bank. If you put the 70ns in the first
>:>>>bank, all your memory will works at 70ns. If you put the 60ns in the first bank, 
>:>>>and if with luck, the 70ns can work at 60ns, your system will work at 60ns.
>:>>>If not, your system will simply switch at 70ns!
>:>>>

>That is a BIG NO-NO!!!

>The recommended thing to do is put the SLOWEST memory in the first bank!  The 
>reason is this: the fast chip can always slow down to the speed of the slow 
>chip, but the slow chip can't keep up with the fast one...

>OS/2 will access ALL memory at the speed of the chips in bank 1.  So if you 
>have the 60ns chips in bank 1, OS/2 will access the 70ns chips at 60ns... and 
>you'll probably bomb.

>It's never recommended to mix n match speeds, but if you have to, follow the 
>above rule.

>BTW - when marching, the military always puts the smallest guys in front, so 
>that the big guys won't run away from the small ones.  Same reasoning.  Slow 
>in front, fast in back, and for much the same reason.

>Bill
>Team OS/2

>I got this info from page 804, Inside OS/2 Warp

FYI:  The Air Force does NOT put the shortest in front of a
marching column, but the tallest in front and to the left.
That way it gives the appearence of uniformity while marching.
The command sequence for this:  "If you are taller than the
person in front of you, tap his shoulder and take his place."
"RIGHT FACE!" "If you are taller than the person in front of
you, tap his shoulder and take his place."

Maybe there is a federal case for height discrimination?  (Just
kidding!)

In article <4l6rnv$753@netone.netonecom.net>, bbcat@netonecom.net   
(Michel Catudal) wrote:

> In <4l6kb3$p7n@news.aros.net>, John Brush@aros.net writes:
> >>
> >>>... because OS/2 tests memory in the lower address space to determine its
> >>>speed, then runs ALL memory at that speed
> >>
> >>Unbelieveable! OS/2 *cannot* "determine the speed of a bank of memory".
> >>Period.
> >>
> >
> >I won't be quite as emphatic, but this does sound a bit dubious.
> >
> >John
> 
> 
> I've heard that crap from several people. Actually you could find out how
> fast a simm can run but you have no guaranty that it could run that
> fast continuously (assuming you pick the high speed).
> Anyone done that is playing with fire, the 70nS part may work fantastic
> as a 60nS for 99.9% of the time then oooppss!
> As far as the manufacturers, too funny to comment.
> 
> Michel Catudal
> Proud user of OS/2 Warp

Considering the general audience for Warp, what benefit to the user would
there be for the OS to clock the memory subsystem?  And, at what cost?  I
could see putting monitor and control on the memory subsystem for a
real-time embedded application where timing is essential to the proper
function of the system, but I just don't see how that would provide
significant advantage on the desktop or on a server.  The demands just
aren't so critical that everything has to be in perfect sync and tune to
perform well.  Your much better off leaving the chores such as memory
timing up to those subsytems that are more concerned with it.

Now, the OS would be really concerned whether the memory gives it data
that is valid, so performing a couple checks to verify that the memory
basically checks out, and even to check out which CPU you're on, might
prove beneficial, but for desktop use, there is very little difference
between 50 and 80 ns once you get to end user perception.  At least, I
know I can't tell the difference between getting smashed in the face for
50ns v. 80ns.  It still happens pretty darned fast.

It would be really interesting to find out which OS task is so important
that it has to start at one edge of a memory cycles and end at the other
end.  It just doesn't seem that there would be anything so critical to the
OS that it would need to do this.

                 Kevin

-- 

In <4l6kb3$p7n@news.aros.net>, John Brush@aros.net writes:
>>
>>>... because OS/2 tests memory in the lower address space to determine its
>>>speed, then runs ALL memory at that speed
>>
>>Unbelieveable! OS/2 *cannot* "determine the speed of a bank of memory".
>>Period.
>>
>
>I won't be quite as emphatic, but this does sound a bit dubious.
>
>John


I've heard that crap from several people. Actually you could find out how
fast a simm can run but you have no guaranty that it could run that
fast continuously (assuming you pick the high speed).
Anyone done that is playing with fire, the 70nS part may work fantastic
as a 60nS for 99.9% of the time then oooppss!
As far as the manufacturers, too funny to comment.

Michel Catudal
Proud user of OS/2 Warp

In <4l6rnv$753@netone.netonecom.net>, bbcat@netonecom.net       (Michel Catudal) writes:
>
>I've heard that crap from several people. Actually you could find out how
>fast a simm can run but you have no guaranty that it could run that
>fast continuously (assuming you pick the high speed).
>Anyone done that is playing with fire, the 70nS part may work fantastic
>as a 60nS for 99.9% of the time then oooppss!
>As far as the manufacturers, too funny to comment.
>
>Michel Catudal
>Proud user of OS/2 Warp
>
>
Well, a 70ns chip is probably a 60ns one that had not pass the manufacturer 
60ns test.

Like a 75Mhz Pentium is probably a 166Mhz Pentium that has just pass the 75Mhz 
test...

[=======================================
[ Yves Boisjoly
[ OS/2 Warp technical support graduate at IAT.
[ OS/2 Team.
[ E-Mail: boisjol@ibm.net
[ Phone:  (514) 522-5079
[=======================================

In <4kuqs2$g76@usenet2.interramp.com>, ir003870@interramp.com (Billy Ko) writes:
>In message <4krr9c$nbv@panix2.panix.com> - kenny@panix.com (Kenneth Crudup) wri
>tes:
>:>
>:>In article <4kq5f0$ehl@usenet6.interramp.com>,
>:>ir003870@interramp.com (Billy Ko) says:
>:>
>:>>The recommended thing to do is put the SLOWEST memory in the first bank!  The 
>:>>reason is this: the fast chip can always slow down to the speed of the slow 
>:>>chip, but the slow chip can't keep up with the fast one.
>:>
>:>*sigh*
>:>
>:>Where do people *get* this bullsh*t?
>:>
>:>>BTW - when marching, the military always puts the smallest guys in front, so 
>:>>that the big guys won't run away from the small ones.  Same reasoning.  Slow 
>:>>in front, fast in back, and for much the same reason.
>:>
>:>So?
>:>
>:>>I got this info from page 804, Inside OS/2 Warp
>:>
>:>I don't believe it.
>:>
>:>I've worked with MBs where the *largest* bank of memory had to be bank 0, (due
>:>to refresh issues) but that has nothing to do with speed. I've mixed speeds
>:>between banks all the time.
>:>
>:>A MOTHERBOARD CANNOT GUESS THE SPEED OF A BANK OF MEMORY! It will *not*
>:>know until you program it to CMOS. Set the speed in non-EDO systems to that
>:>of the slowest memory *in any bank*, and it will all behave fine!
>:>
>:>	-Kenny
>:>
>:>-- 
>:>Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting
>:>kenny@panix.com		CI$: 75032,3044	  +1 617 524 5929/4949 Home/Office
>:>16 Plainfield St, Boston, MA 02130-3633	  +1 617 983 9410      Fax
>:>OS/2 box: pkenny.tiac.net (when I'm online) Get Warp-ed! OS/2 3.0 is here NOW!
>
>Excerpt from Inside OS/2 Warp, page 804:
>
>Trap Errors and Memory Problems
>Some basic things to check for when a customer is experiencing trap errors of 
>any kind include:
>
>1. All memory SIMMs must be the same size and speed...
>
>2. All SIMMs should be from the same manufacturer.
>
>3. If the customer has mixed memory SIMMs, it may be possible to get them to 
>work, although this would be a configuration unsupported by OS/2
>
>If mixed memory SIMMs are being used, the customer must place the SLOWER 
>speed memory in the first or lowest address sockets.  This approach 
>occasionally works because OS/2 tests memory in the lower address space to 
>determine its speed, then runs ALL memory at that speed
>
>
>And one more tip:
>
>Please don't engage in personal type attacks, it is very unbecoming of a 
>person.
>
>Bill
>Team OS/2
>
>
>
>

I got 60ns in the first bank, and 70ns in the second. works just fine. No
memory trap errors what so ever.. 

	Dennis

>>:>>>
>>:>>>You must put the most rapid one in the first bank. If you put the 70ns in the first
>>:>>>bank, all your memory will works at 70ns. If you put the 60ns in the first bank, 
>>:>>>and if with luck, the 70ns can work at 60ns, your system will work at 60ns.
>>:>>>If not, your system will simply switch at 70ns!
>>:>>>
>
>>That is a BIG NO-NO!!!
>
>>The recommended thing to do is put the SLOWEST memory in the first bank!  The 
>>reason is this: the fast chip can always slow down to the speed of the slow 
>>chip, but the slow chip can't keep up with the fast one...
>
>>OS/2 will access ALL memory at the speed of the chips in bank 1.  So if you 
>>have the 60ns chips in bank 1, OS/2 will access the 70ns chips at 60ns... and 
>>you'll probably bomb.
>
>>It's never recommended to mix n match speeds, but if you have to, follow the 
>>above rule.

I don't know who wrote this statement, and it really doesn't matter. However,
I am getting way sick of people who do not know have a clue 
making assumptions and giving the impression that they know what
they are talking about.

Now hear this. OS/2, W95, and even the mighty UNIX, have absolutely no
idea what the speed of the memory is. They don't care, it matters not at all,
and anyone who says that the OS knows, or cares, is wrong. Plain and 
simple!

And, if you have a Pentium, listen up. The pentium CPU will gobble up 2 simms
at a time when it accesses the memory. Its bus requires 2 simms to even work.
Put only one 32M device in your pentium, and it won't even run. Everyone
quit saying that it reads the first one, then adjusts everything to that one 
device. It reads 2 devices at a time. 2 at a time. Repeat after me, the 
P5 bus requires 2 simm devices, not one.

Anyone who tells you that you can't mix speeds (meaning a 60ns in place of
a 70ns) is a bigger liar than the person who fits the above scenario. If your
motherboard and bios supports 70ns RAM, you can put 10ns Ram in those 
slots and the system won't give a rat's ass! If you put a SLOWER device in,
then you have a potential problem.

have we got it now? Mix the speeds. faster won't hurt! No OS on the planet
can determine ram access speed! No OS cares!

Now for the corny line. I design embedded controllers and computer systems
and I know what I am talking about. Challenge me, flame me, I don't care.
I do know what I a talking about in this area. Can't we all just get along?

Thank you.

Regards,

John

In article <4l9eei$gil@news.aros.net>,  <jbrush@aros.net> wrote:
>have we got it now? Mix the speeds. faster won't hurt! No OS on the planet
>can determine ram access speed! No OS cares!

I'll give you the last two.

I suspect, though, that there probably are some crappy motherboards out
there that decide what the RAM speed is by what they find in the first bank
(whether that's 16, 32, or 64 bits wide).  *IF* such beasts exist in the
wild in any number, that alone is enough reason for IBM tech support to
tell you that mixing memory is a bad idea.

Lord knows there have been enough shit-stupid motherboard design decisions
made to shave a couple pennies off the price.

Adam
-- 
adam@phoenix.princeton.edu | Viva HEGGA! | Save the choad! | 64,928 | Fnord
"Double integral is also the shape of lovers curled asleep":Pynchon | Linux
Thanks for letting me rearrange the chemicals in your head. | Team OS/2
You can have my PGP passphrase when you pry it from my cold, dead brain.

In <Pine.BSD/.3.91.960417123353.8471C-100000@matrix.eden.com>, edwin bailen <n5kzw@eden.com> writes:
>>> 
>> Unbelieveable! OS/2 *cannot* "determine the speed of a bank of memory".
>> Period.
>> 
>> 	-Kenny>When I was running 2.1, I added some memory, and the system would not 
>boot.  When I called the OS/2 tech support line, their statement was "Oh, 
>all of the memory in the system must be of the same type and speed, and 
>from the same vendor."  Right or wrong, they are at least consistent.
>

I used to run OS/2 2.1 with mixed 8 meg of memory: 4 with 3 chip, 4 with 9-chip,
different manufacturers, no problem.

Since Warp I've upgraded to 20 meg: I've keep the 9-chip 4 meg and added
16 meg from yet another manufacturer. On top of that, the little plastic latch
for one of the simms broke on me (oh, those big fingers!), and I improvised a
plastic holder and a little tape.

It workers like a charm, every time.

                |       Ricardo Paoletti
                |       em Buffalo, NY