FWB's "Hard Disk Toolkit" will not run on OS X, so I have no way of
checking RAM thoroughly on G5 model hardware.
HDT is the only util' I know of that will do thorough RAM testing, the
kind of testing that takes hours to do.
(Apegio and Minor-March testing patterns, for example)
I have tried all the below with no luck:
1) Googling
2) MacFixit
3) VersionTracker
4) Apple Tech' Support (amusing story behind this one)
Thanks for any suggestions. There are repair outfits that will
thoroughly test RAM for a rather high price, however I prefer to do it
myself, because it takes a lot less time to track down occassional
random freezes of Macs due to borderline RAM and faulty memory-manager
chips.
Mark-
The best one that I have found is 'memtest' that runs from the command
line and it is free from versiontracker.com.
http://www.versiontracker.com/dyn/moreinfo/macosx/17156
--
-Ernie-
"There are only two kinds of computer users -- those who have
suffered a catastrophic hard drive failure, and those who will."
Have you done your backup today?
> In article <290820040645163442%NoSpam...@invalid.com>,
> Mark Conrad <NoSpam...@invalid.com> wrote:
>
> > Is there any utility capable of doing a thorough testing of RAM on
> > modern Mac models that can't run OS X?
> >
> > FWB's "Hard Disk Toolkit" will not run on OS X, so I have no way of
> > checking RAM thoroughly on G5 model hardware.
> >
> > HDT is the only util' I know of that will do thorough RAM testing, the
> > kind of testing that takes hours to do.
> > (Apegio and Minor-March testing patterns, for example)
> >
> > I have tried all the below with no luck:
> > 1) Googling
> > 2) MacFixit
> > 3) VersionTracker
> > 4) Apple Tech' Support (amusing story behind this one)
> >
> > Thanks for any suggestions. There are repair outfits that will
> > thoroughly test RAM for a rather high price, however I prefer to do it
> > myself, because it takes a lot less time to track down occassional
> > random freezes of Macs due to borderline RAM and faulty memory-manager
> > chips.
> >
> > Mark-
>
> The best one that I have found is 'memtest' that runs from the command
> line and it is free from versiontracker.com.
>
> http://www.versiontracker.com/dyn/moreinfo/macosx/17156
Opps -- sorry, I didn't read correctly. I saw OS X, but missed the
_cant run on_. Too early - haven't had my coffee yet.
The old TechTool Pro (version 2.5.5) will run any combination of:
Rotational
Arpeggio
Web
Leap
Minor March
Major March
plus a few simple byte patterns.
--
Tom Stiller
PGP fingerprint = 5108 DDB2 9761 EDE5 E7E3
7BDA 71ED 6496 99C0 C7CF
In article <290820040645163442%NoSpam...@invalid.com>,
Mark Conrad <NoSpam...@invalid.com> wrote:
> Is there any utility capable of doing a thorough testing of RAM on
> modern Mac models that can't run OS X?
Gauge Pro:
http://www.versiontracker.com/dyn/moreinfo/macos/15583
--
Mike Rosenberg
<http://www.macconsult.com> Macintosh consulting services for NE Florida
<http://bogart-tribute.net> Tribute to Humphrey Bogart
Toyota Prius fans: Check out alt.autos.toyota.prius
> Is there any utility capable of doing a thorough testing of RAM on
> modern Mac models that can't run OS X?
Damn! - Senility has struck here again.
I meant to write:
"Is there any utility capable of doing a thorough testing of RAM on
modern Mac models that can ONLY run OS X?"
Sorry everyone.
Anyhow, when I called Apple tech' support yesterday, they promptly
routed my call to India.<g> The gent' there did not have a clue, so
he routed my call back to the states. After much consultation among
themselves and 20 more minutes on hold, they informed me that they
could not reveal what utility they used to check RAM on the new G5s,
because that was proprietory information.
Heh heh, I got a kick out of that reply.
Mark-
> Opps -- sorry, I didn't read correctly. I saw OS X, but missed the
> _cant run on_. Too early - haven't had my coffee yet.
Don't be sorry - I made a big fat mistake on my original post, so your
info' is right on target.
It was 6:43AM, and I had been up all night trying to breathe life into
my iListen application. (no luck)
I MEANT my original post to read this way:
"Is there any utility capable of doing a thorough testing of RAM on
modern Mac models that can ONLY run OS X?"
Sorry about the mistake. The original post was entirely wrong.
Thanks, I will chase down your suggested "memtest" command-line app'
and check it out.
Hope it will run the hours-long "Minor-March" and the much longer
"Major-March" RAM test patterns.
Those tests are very good at detecting both borderline RAM failures and
they also are good at detecting problems with the "Memory Manager"
chip, which is located on the CPU board on some Mac models.
Solves some problems with intermittant freezes, in my experience.
Mark-
> Is there any utility capable of doing a thorough testing of RAM on
> modern Mac models that can't run OS X?
Tech Tool Pro does extensive RAM tests, amongst other things.
> > Is there any utility capable of doing a thorough testing of RAM on
> > modern Mac models that can't run OS X?
>
> Tech Tool Pro does extensive RAM tests, amongst other things.
As I mentioned in a previous post, I made a terrible mistake in my
initial post in this thread. I was very tired, having stayed up all
night battling another program.
My wrong statement:
"Is there any utility capable of doing a thorough testing of RAM on
modern Mac models that can't run OS X?"
I can't believe I wrote that. In the first place, there are probably
no "modern" Mac models that cannot run OSX, so that statement of mine
does not make sense at all.
I "intended" the statement to read correctly this way:
"Is there any utility capable of doing a thorough testing of RAM on
modern Mac models that can ONLY run OS X?"
Anyhow, TechTool Pro does some RAM testing, an it will indeed reveal
defective RAM that is real bad, however it is not all that great at
detecting borderline RAM problems.
The reason is, individual RAM cells are influenced by "nearby" adjacent
RAM cells, because they are packed together so tightly in the RAM chip.
Kinda like the old telephone lines of many years ago, where one could
hear the conversations of other people because of poor shielding of the
individual lines in a packed cable of many lines.
A really good RAM testing utility will therefore toggle every possible
combination of nearby RAM cells, in an effort to "corrupt" the
particular cell being tested at the time.
As you can imagine, this extensive testing of RAM takes a horrendous
amount of time, however it is capable of detecting RAM faults that are
missed by more simplified testing.
The specific test that toggles all possible combinations of nearby RAM
cells is called the "Major March" test, however that test can only be
ran on extremely fast computers, otherwise it would take forever to
run.
Other lesser testing patterns are called by various names like
"Minor-March" and "Apegio".
Years ago I had a problem of random freezes of my Pismo powerbook,
which was under warranty at that time. The usual RAM tests all passed
with flying colors, every time.
I had a program called "Hard Disk Toolkit", which did more extensive
RAM testing, among which the so-called "Arpegio" test pattern was used.
(HDT will still not work on an OSX-only type of computer)
Arpegio revealed a defect about 30% of the time. The text readout of
the results mentioned that the memory-manager chip might possibly be
bad, instead of the RAM itself.
I sent the computer to the Apple repair depot in Texas, and they sent
it back saying they could find nothing wrong with it.
That kept going on for three seperate turn-arounds over a three month
period, me insisting that freezes still occured, Apple doing things
like replacing the entire motherboard.<g>
Eventually, in desperation and to get me off their back, they finally
listened to the testing results I was trying to get across to them.
They replaced the CPU board, which also had the "memory manager" chip
on it. The CPU board is a seperate board, not part of the motherboard
in that particular Mac powerbook.
That fixed the freeze problem, and the Apegio test passed 100% of the
time.
I was not out one cent, Apple paid for everything, even shipping costs.
Mark-
Thanks everyone for the help, I would be lost if it were not for these
NGs.
The following routine worked for me. I am in no way advocating the use
of memtest for others, because like lots of Terminal programs, it could
be very dangerous in the hands of a novice.
That said, I am very impressed with this free utility, so it will have
a regular place in my periodic maintenance of my OSX Macs.
To do fairly extensive RAM testing on an OSX-only computer:
1) Download memtest from versiontracker.com
2) Draged unstuffed memtest folder from
the desktop into my hard disk icon.
3) Restarted computer while holding down shift key
to enter "SafeMode".
4) Logged in as root.
5) Double-clicked my hard drive icon to open it,
dragged the memtest folder onto the desktop.
The name of that folder in my case is
"memtest-4.03M"
(very important to know exactly what that
name is, because that name
will be used shortly)
6) Opened Terminal, then I typed the commands
listed a little further down.
The first command relies on Terminal being
opened in my home directory. If for any reason
that first command threw up an error message,
I would then have typed instead:
cd /Users/mark/Desktop
The second command uses the exact name of
the memtest folder, no spaces in that name.
The last command actually starts the
memtest program running. The first item in
the command is a period, the last item in the
command is the number one.
I typed the commands, each command followed
by pressing the return key:
cd Desktop
cd memtest-4.03M
./memtest all 1
In my case, the program took 3 hours to completely check "all" the RAM.
(actually, a portion of RAM is not checked, because that portion is
needed to actually run the memtest program)
If the number one of the last command was replaced by the number four,
the program will run the entire test four times, and take 12 hours to
finish in my case.
It is easy to monitor the progress of the program, because the Terminal
readout provides me with a blow-by-blow report of what is being checked
at any moment.
Below is a portion of the Terminal readout at the end of the 3 hour run.
It is very important that the lines below about the success of
"locking" memory be there, otherwise I might as well force-quit
Terminal, because the results will be invalid.
Last part of Terminal readout at end of testing.
*********************************************
Attempting to lock allocated physical
memory....memory locked successfully!
Test Pass 1/1:
Stuck Address : ok
Random Value : ok
Compare XOR : ok
Compare SUB : ok
Compare MUL : ok
Compare DIV : ok
Compare OR : ok
Compare AND : ok
Sequential Increment: ok
Solid Bits : ok
Block Sequential : ok
Checkerboard : ok
Bit Spread : ok
Bit Flip : ok
Walking Ones : ok
Walking Zeroes : ok
All tests passed.
********************************************
The first batch of tests through "Compare AND" only take about ten
minutes to run, and correspond to the testing that might be done by
utilities like TechTool Pro.
At the end of testing, I cleaned up everything, threw out the memtest
folder, restarted computer to get back into "regular" OSX.
Mark-
> Eventually, in desperation and to get me off their back, they finally
> listened to the testing results I was trying to get across to them.
I wish I couldn't relate to that kind of response from Apple...
--
Steven Fisher; sdfi...@spamcop.net
"Morituri Nolumus Mori."
Hi Mark,
Not sure if I did it correctly but d/I'd memtest and ran it according to
the readme file (but it's running in multiuser mode?) and hung up on
"block sequential: setting 59" and just didn't go any further. Wonder
if this has anything to do with the "multiuser mode"? I have AppleCare
and ran the memory test with that which showed nothing wrong with the
memory. Have been having a lot of problems with the machine (a PowerMac
G4 with a gb of RAM) locking up when watching a DVD (any DVD) and also
when just running very small audio files in ProTools (two MIDI tracks!).
The only way to stop the process is to do a cold shutdown as the
keyboard also locks up. My wireless mouse is off also.
Karen
> > The first batch of tests through "Compare AND" only take about ten
> > minutes to run, and correspond to the testing that might be done by
> > utilities like TechTool Pro.
> >
> > Mark-
>
> Hi Mark,
>
> Not sure if I did it correctly but d/I'd memtest and ran it according to
> the readme file (but it's running in multiuser mode?) and hung up on
> "block sequential: setting 59" and just didn't go any further. Wonder
> if this has anything to do with the "multiuser mode"?
You bring up several points, I will address them one at a time.
Multiuser mode is normal for memtest, it is very unlikely that it has
anything to do with your particular problem.
These are a few of the things that _could_ cause your problem with
memtest:
1) The "Block Sequential" test is the first of several tests that
"stress" the individual memory cell by changing nearby cells.
Not all of the RAM in your G4 gets tested by memtest. A portion of the
RAM about 100MBs in size is not tested at all, because it is needed
just to run memtest itself.
Now let's assume that somewhere in that 100MBs of RAM there is a "weak"
memory cell that only fails occassionally, causing the "lockups" that
you are experiencing.
When the 'Block Sequential" test is ran, it could very well change
nearby memory cells, inducing a failure in that "weak" memory cell that
is being used to run memtest, _even_ _though_ that weak memory cell
never gets formally tested.
(never gets tested because the "weak" cell is part of the 100MBs that
is needed to run memtest itself)
Once the weak memory cell is induced to fail by memtest "flipping"
nearby cells, the lockup could occur.
The only way I know of to eliminate this as the source of your problem
is to open up your G4 in order to get at the memory modules themselves.
Let's assume that your particular G4 has a total of two memory modules
installed in it.
With the computer unplugged from its power source, someone who is
familar with the damage that can be caused by static electricity can
_carefully_ switch the two memory modules.
That will move the "weak" cell away from the critical 100GBs of memory
needed to run memtest itself.
Subsequent running of memtest should then not cause any lockup when
"Block-Sequential" setting 59 is ran. With any luck, the weak memory
cell will be found, also.
Okay, with that possibility out of the way, let's continue with one
other thing that could be causing the problem.
2) All your RAM could be good, but the so-called "Memory-Manager" chip
could be bad. This actually happened to me, BTW.
In this case, switching memory modules around will do no good, the
lockup problem will still persist.
This memory-manager problem makes the RAM _appear_ to be bad.
If you run memtest several different times, and every time the lockup
occurs at a different place in the testing, then the memory-manager
chip is very suspect as being the culprit.
I do not know where the memory-manager chip is located on a G4.
On my Pismo powerbook, it is located on the CPU board, _not_ on the
motherboard.
The fact that your lockup _only_ occurs when you are running the
applications you mentioned puzzles me. Perhaps some other folks here
might have an idea about what is wrong.
Mark-