[Info-vax] Baremetal emulators, was: Re: Alpha emulator for OSX

Johnny Billquist bqt at softjar.se
Mon Feb 8 09:24:34 EST 2016


On 2016-02-08 14:27, Simon Clubley wrote:
> On 2016-02-08, lists at openmailbox.org <lists at openmailbox.org> wrote:
>> On Mon, 8 Feb 2016 04:10:38 -0800 (PST)
>> johnwallace4--- via Info-vax <info-vax at rbnsn.com> wrote:
>>>
>>> You mention bare metal. Arie says "vtAlpha runs on X86 without the
>>> need for an extra OS like Windows or Linux." He doesn't say "bare
>>> metal" here, but the term is used on their website.
>>>
>>> Unless something has changed since the last significant discussions
>>> I remember on this topic (several years ago), vtAlpha runs on top of
>>> a cut down Linux (it'd be unrealistic to expect it to run on random
>>> x86 systems *without* an underlying OS, or at least a HYPErvisor).
>>>
>>> So it may not fit everyone's definition of "bare metal". Others may be
>>> perfectly happy with a setup like that. Take your pick.
>>
>> Thanks, that's a good point. I probably misunderstood what was said in the
>> post although the fact you say "bare metal" is mentioned on the website
>> does muddy the waters. From the end-user point of view there isn't much
>> practical difference but from a technology point of view there is certainly
>> quite a distinction and it would be nice to clarify how it works.
>>
>
> There is potentially a major difference from the end-user viewpoint if you
> are trying to run real-time applications.
>
> Let's keep this simple for the benefit of those (_not_ you!) who try to
> confuse the issues here.
>
> Does the part of the Alpha emulator which is concerned with doing the
> emulation directly talk to the device registers in the hardware on the
> host system ?
>
> If yes, then it's bare metal. If no, and it talks to some intermediate
> OS's syscalls layer instead of the hardware then it's not bare metal.
> BTW, it doesn't matter if that OS layer is bundled with the emulator or
> not; it's still an intermediate OS layer either way.
>
> However, if the underlying OS has real-time guarantees (say it's a RTOS
> for example) then you _might_ still be able to meet the real-time
> requirements of the application.
>
> Don't forget that real-time doesn't mean fast; it means a _guaranteed_
> bounded response time. That's something which a surprising number of
> people don't seem to understand.

Very true.

One additional point where it might make a difference is that if you 
have an OS underneath, that you run the emulator on, then you are also 
potentially exposed to any problems/bugs/exploits that exist in that OS, 
which means you might need to do "upgrades" just because the underlying 
OS needs patching, while nothing in your environment, nor the hardware, 
have any issues. Not to mention if you actually do get compromised at 
the host OS level...

So, there sometimes is a huge difference, even to end users, whether you 
actually run on bare metal, or have an host OS for you emulator.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt at softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol



More information about the Info-vax mailing list