>> I've tried patching the idle loop to a halt. It never seems to get
>> executed. I thought when all else was said and done the idle loop
>> still was in use.
> But your patch means that CHARON-VAX is emulating the HALT
> instruction. It doesn't halt the emulation.
> Just before you execute the HALT, you have to send a message to the
> emulator itself, not the emulated VAX.
> As CHARON-VAX is right now, it's emulating a VAX 100%, so there's no
> way to talk with the emulator. SRI would have to provide a "hole"
> through which you could control the emulator. That doesn't exist
> right now.
> --Stan Quayle
> President, Quayle Consulting Inc.
> Stanley F. Quayle, P.E. N8SQ +1 614-868-1363 Fax: +1 614 868-1671
> 8572 North Spring Ct. NW, Pickerington, OH 43147
This same problem exists with other virtual machine emulators, too.
Using VMware on either a Linux or WinNT/2K/XP host, guest operating
systems such as Linux or any Win32 O.S. will "play nice" and idle the
CPU when they are not busy. However, guest operating systems like DOS
or NetWare don't do this so they cause VMware to consume 100% of the
host system's CPU. There is a TSR available [DOSIDLE.COM] and an NLM
[NW4-IDLE.NLM or NW5-IDLE.NLM as appropriate] that effectively solve
this problem by executing a halt instruction at times when the guest
O.S. has nothing else to do.
Interestingly enough, power consumption & heat output measurements have
shown that a physical IA-32 based system running NetWare will consume
less power and output less heat if it runs this CPU idling program.
This tends to lead to a longer life time for the hardware components as
well as having a more efficient & cooler-running system. There are now
a number of people running this CPU idler on dedicated servers to make
them more power efficient during their idle times.
Now, one thing that VMware is not doing is actually emulating a CPU;
instead, it virtualizes the CPU of the host but the guest O.S. will
detect the same CPU as the host system is using. With CHARON-VAX, it
might be tougher to force this type of CPU idling to happen because the
emulator has to emulate the CPU to even detect an invalid opcode that
might be used to signify that the emulator should throttle itself back
w/respect to consumption of the host system's CPU.
ICQ # 22321532
RTFM Consulting Services Inc. 864 801 2795 voice & voicemail
103 Autumn Hill Road 864 801 2774 fax
Greer, SC 29651 800 774 0718 pager