#MMU or No MMU?
9 messages in this thread
I have a VECTOR accelerator with what is suppose to be a 68EO30
processor (25 Mhz). I understand that this chip does not have an
MMU. Yet when I run SHOWCONFIG or SYSINFO, both programs show a
68030 processor and a 68030 MMU. Is this a short coming of these
programs? My main confusion is whether I should run ENFORCER or turn
off the CACHE when running the BridgeBoard.
Ronnie,
Why don't you check the chip on the board and see if it's a 68E030. I
think, however, that the advertising says it has a 68030 with MMU, and not an
EC030. Perhaps you had another product in mind.
No the VECTOR does come in various configurations including a 68EC030.
In fact the photo on the box shows the EC version. The label on my
box indicates that I have the EC version and I'm not certain but I
think my chip is labeled EC (I'd have to pull the board to double
check). The VECTOR does have a feature that allows it to copy
Kickstart ROM into 32bit RAM without an MMU. The User's Guide
recommends using this even for those with the 68030 because it's
faster. I've tried turning this feature off on reboot but still get
indication of the presents of a MMU. I guess there are three
possibilities.
1. SYSINFO and SHOWCONFIG aren't capable of differentiating between a
68030
and a 68EC030 on a Vector board.
2. The FASTROM feature of Vector is tricking the programs into thinking an MMU
is present.
3. Some cluts on night-shift mixed a 68030 in a batch of 68EC030's before they
were labeled.:)
I guess I'll start running ENFORCER and leave the CACHE's on unless I develop
a problem (Even ENFORCER reports that the MMU is ON when I run it!!!!????)
– via Whap!
Ronnie,
Suggest the following. Run CPU nofastrom/fastrom and see if
it reports the relocation of ROMs into RAM. CPU tries to take over
the MMU to do this, so it may report failure if no MMU is available.
If it's possible to boot without mapping ROM into the special 512k of
ram on the Vector, then boot that way and see what Sysinfo reports.
Try running AIBB, particularly a later version. When AIBB runs it
doesn't know how to tell which processor is present and asks you to
designate one. Choose one, then see if an MMU is still reported.
Lastly, check the chip or docs, or whatever. I suspect the cover on
the box means nothing. It's probably the same regardless of
configuration.
Regardless, you can easily determine if Enforcer QUIET is doing its job when
you run the BB. Just try doing some AREADs of files. If AREAD works without
problems, then I wouldn't worry about it, Enforcer must be doing what it needs
to do. If, on the other hand, AREAD locks up or fails, then Enforcer isn't
really in control of the MMU, so then you can turn off the data cache. You
don't need to turn both caches off, only the data cache is the problem.
68EC030 chips have the MMU logic, it's just that it's been
"decertified" by the factory. In other words, it's a busted 030.
Depending on how badly busted the MMU logic is, you may get hints
from software that the CPU has a working MMU. DON'T BELIEVE IT! If
you run Enforcer (or CPU FASTROM, etc.) on a EC030 with a sortof
working MMU, you'll be tearing your hair out in short order.
It is possible that they did an over speced run.
They do this with components in terms of speed.
Put out 5000 33MHz parts, mark some as 25, some as 33.
It is possible they left the MMU in a few chips because of an over stock
of MMU chips, or an understock of non MMU chips.
I think the new version of enforcer has a program to check for the MMU.
I believe it is called lawbreaker
Dave
Elite MicroComputers – Flying high on AutoPilot
The LAWBREAKER program is used to test to see if Enforcer properly traps
illegal memory accesses. It's and example program and does not test for an MMU.
Enforcer does that. But like I said, if the chip says "EC" than it doesn't have
a fully functional MMU. It may work most of the time, but there are no
promises.
Unless you're doing development work, there's little point in having or using
an MMU. No user programs use it, and neither does the OS. Er, well, "CPU
FASTROM" does, but that's it.
David,
That was exactly the problem that the other chap had reported. He
claimed that the board was supposed to have an EC type chip, but that Sysinfo
was reporting the MMU active. Perhaps you should pass this along to him.
Now it all comes together. Just this week I read something about an
EC just being a 030 with a defective MMU but I found it too
incredible to believe. I've been trying to run ENFORCER but have
been getting frequent total system lock ups. I was attributing them
to some public domain software I've been running but I would bet it's
as you say. I'll stop using ENFORCER and just turn the DATA CACHE
off when running the Bridgeboard. Thanks to all for the help!!