Can you turn off Pipeline in ARM Cortex M3

rickman <gnuarm@gmail.com> writes:

> On 9/9/2015 9:24 AM, John Devereux wrote:
>> Dimiter_Popoff <dp@tgi-sci.com> writes:
>>
>>> On 09.9.2015 &#1075;. 12:18, Tom Gardner wrote:
>>>> On 08/09/15 21:51, Klaus Kragelund wrote:
>>>>> Disabling the pipeline would make the code more determistic
>>>>
>>>> Not significantly.
>>>>
>>>> Even with i486s, with their tiny caches, the ratio between
>>>> mean and worst case (IIRC) interrupt latencies could be 10:1
>>>> (from memory 70us vs 700us).
>>>
>>> Hmmm, these are huge figures, even the 70uS is too huge I suppose.
>>> A 1 MHz 6800 had IRQ latency in the range of 30uS or so.
>>
>> We are talking about a CM3, that has sub-microsecond latency (12 cycles
>> from memory).
>
> Only if the interrupt is enabled.  Depending on what the code is doing
> it can be a lot longer.  That is the real concern in an ISR.

Yes if he is relying on a guaranteed interrupt latency he should not
turn them off. Also should be top priority irq!

He should not turn the clock down to 1 khz either. Or set his timer
output back to a GPIO. Etc.

I don't think it is a good idea to have the deadtime enforced by
software, regardless. OP is going to get mighty tired of changing blown
FETS during development. I was just responding to the talk above about
caches and pipelines and 70us vs 700us latencies. None of that matters
on a CM3 if a guaranteed <1us jitter is acceptable.


-- 

John Devereux

Tom Gardner <spamjunk@blueyonder.co.uk> writes:

> On 09/09/15 17:41, Dimiter_Popoff wrote:
>> On 09.9.2015 &#1075;. 19:25, Tom Gardner wrote:
>>> On 09/09/15 16:51, Dimiter_Popoff wrote:
>>>> On 09.9.2015 &#1075;. 18:07, Tom Gardner wrote:
>>>>> On 09/09/15 13:09, Dimiter_Popoff wrote:
>>>>>> On 09.9.2015 &#1075;. 12:18, Tom Gardner wrote:
>>>>>>> On 08/09/15 21:51, Klaus Kragelund wrote:
>>>>>>>> Disabling the pipeline would make the code more determistic
>>>>>>>
>>>>>>> Not significantly.
>>>>>>>
>>>>>>> Even with i486s, with their tiny caches, the ratio between
>>>>>>> mean and worst case (IIRC) interrupt latencies could be 10:1
>>>>>>> (from memory 70us vs 700us).
>>>>>>
>>>>>> Hmmm, these are huge figures, even the 70uS is too huge I suppose.
>>>>>> A 1 MHz 6800 had IRQ latency in the range of 30uS or so.
>>>>>
>>>>> The figures were from memory, but were definitely in units of
>>>>> time (s), not conductance (S).
>>>>
>>>> Hah! I have thought for decades second was supposed to be abbreviated
>>>> to a capital S.... may be it has been the case in the past, may be
>>>> I remembered wrongly.
>>>
>>> It never was.
>>>
>>> However, Tektronix used to get it wrong on their oscilloscopes - and
>>> as a metrology company they really ought to have know better.  I
>>> suspect they finally realised the error of their ways when specifying
>>> their digitising scope's sampling rate, e.g. 100MS/s.
>>> Not that S=samples is correct!
>>
>> May be I have taken it from them. I used to look at some of their
>> service manuals when I was making my first steps in analog design
>> (back then I wanted to build an oscilloscope... never built it,
>> not many projects I have left unfinished but this is one of them.
>> Learned a lot while trying though. I may still build one, now at
>> least I know how to :D ).
>
> One of my projects, when I get A Round Tuit, is to make a 2GS/s
> 4GHz scope. The trick is to find a way to do it with only trivial
> analogue front end components. Hence no amplifiers, no ADCs, just
> a 50ohm input with very simple analogue components. Yes, there
> will be limitations, but that's half the fun :)

Hmm, there was the "slideback sampler" that John Larkin and Mike Monet
used to fight about, a decade or two ago in sci.electronics.design.


> Yes, I know you can buy remarkably fast remarkably cheap ADCs
> nowadays. But I want to do it for tens of dollars, not thousands :)
>

-- 

John Devereux

On 10/09/15 07:42, Dimiter_Popoff wrote:
> On 10.9.2015 г. 00:32, Tom Gardner wrote:
>> On 09/09/15 22:30, Tom Gardner wrote:
>>> On 09/09/15 21:56, Dimiter_Popoff wrote:
>>>> [uS is what I would
>>>> have written, how am I supposed to do that now? "us" is more ambiguous?
>>>> Tom?]
>>>
>>> :(
>>>
>>> Prepend a number :)
>>>
>>> More seriously I have seen a "micro" character in some usenet posts.
>>> I wonder how it got there, and whether other people's clients rendered
>>> it correctly.
>>>
>>> Let's try "μ" which is simply a "special character" inserted into
>>> an OpenOffice document then copied/pasted into this message, which
>>> I'm composing in SeaMonkey. Since it can also be copied/pasted
>>> into a linux/bash command line in the xfce terminal emulator, there's
>>> cause for hope.
>>>
>>> If it works I don't know whether I'll use it every time and become
>>> a major nitpicker with those that don't!
>>
>> Well, it has apparently worked from my PoV.
>> Has it worked for you?
>> Do you see it as U+03BC or something?
>
> It worked for me, too. And I have the character available in my
> editors (text and graphics) under dps but not via a single key;
> I have used it earlier but I think lately I just use "u".
>
> I suppose I can get used to "us" instead of "uS", but may be an
> option would be se-6 or sE-6... I still can't quite swallow the
> thought that S is invalid for seconds. I looked at my very
> old tek 465B, well, its front panel is also designated with
> lower case around the timebase knob.

Don't throw that scope out; there's a thriving market
for them on ebay, /especially/ if they are working :)
I have the HP equivalent, the HP1740A, and the local
hackspace has a near relative, the Tek2213.

On 10/09/15 08:05, David Brown wrote:
> For Linux, you have other options.  Again, depending on the keyboard
> layout, AltGr-m is likely to work - certainly on my Norwegian layout I
> can get a wide range of characters with combinations of AltGr, shift,
> and different keys.  Then there is also the "compose" key - typically,
> this is not assigned by default, but you can set it up if you want it (I
> use the "scroll lock" key as compose).  Then µ is compose + m + u.

Oh, excellent.

Trying the other combinations is going to be a great way of
pointing out to idiot companies that valid email
addresses aren't limited to roman alphanumeric characters.
I think I'll use something like "þomas gardner"- they
won't notice the inaccuracy.

On 10/09/15 09:16, rickman wrote:
> On 9/10/2015 3:05 AM, David Brown wrote:
>> On 09/09/15 23:30, Tom Gardner wrote:
>>> On 09/09/15 21:56, Dimiter_Popoff wrote:
>>>> [uS is what I would
>>>> have written, how am I supposed to do that now? "us" is more ambiguous?
>>>> Tom?]
>>>
>>> :(
>>>
>>> Prepend a number :)
>>>
>>> More seriously I have seen a "micro" character in some usenet posts.
>>> I wonder how it got there, and whether other people's clients rendered
>>> it correctly.
>>>
>>> Let's try "μ" which is simply a "special character" inserted into
>>> an OpenOffice document then copied/pasted into this message, which
>>> I'm composing in SeaMonkey. Since it can also be copied/pasted
>>> into a linux/bash command line in the xfce terminal emulator, there's
>>> cause for hope.
>>>
>>> If it works I don't know whether I'll use it every time and become
>>> a major nitpicker with those that don't!
>>
>> µs works fine for me.  As long as readers have a reasonably modern
>> newsreader, there should be no problem with the letter "mu" for micro -
>> it is a standard character in the standard fonts on most systems.  If
>> someone has customised their desktop or newsreader to use a weird font
>> with a more limited character range, then it might show up oddly.  If
>> there are any fans of command-line newsreaders, they should also see it
>> if they have a utf-8 locale on their systems.
>>
>> As for typing it, that depends on your system and your keyboard.
>> Windows is (as always) more limited.  It should be easy for most
>> non-English keyboard layouts, which separate the "Alt" and the "Alt Gr"
>> keys - on my Norwegian keyboard layout, it's just AltGr-m.  But if that
>> doesn't work, character map is probably the least inconvenient method
>> unless you can remember the Alt + numeric keyboard sequence.
>>
>> For Linux, you have other options.  Again, depending on the keyboard
>> layout, AltGr-m is likely to work - certainly on my Norwegian layout I
>> can get a wide range of characters with combinations of AltGr, shift,
>> and different keys.  Then there is also the "compose" key - typically,
>> this is not assigned by default, but you can set it up if you want it (I
>> use the "scroll lock" key as compose).  Then µ is compose + m + u.
> 
> Along with the often used
> 
> ° - Alt-0176

Shift-AltGr-0 or Compose+o+o gives me °

> ± - Alt-0177

AltGr-+ or Compose+++- gives me ±

Both of these work in Linux, but not Windows.

> 
> mu is
> µ - Alt-0181
> 
> The zero is very important.  Without it you get other chars.
> 

On 10/09/15 10:46, Tom Gardner wrote:
> On 10/09/15 08:05, David Brown wrote:
>> For Linux, you have other options.  Again, depending on the keyboard
>> layout, AltGr-m is likely to work - certainly on my Norwegian layout I
>> can get a wide range of characters with combinations of AltGr, shift,
>> and different keys.  Then there is also the "compose" key - typically,
>> this is not assigned by default, but you can set it up if you want it (I
>> use the "scroll lock" key as compose).  Then µ is compose + m + u.
> 
> Oh, excellent.
> 
> Trying the other combinations is going to be a great way of
> pointing out to idiot companies that valid email
> addresses aren't limited to roman alphanumeric characters.
> I think I'll use something like "þomas gardner"- they
> won't notice the inaccuracy.
> 

UTF-8 characters should, in theory, work in email addresses - but only
in theory.  In practice, you can expect trouble somewhere along the line.

But UTF-8 characters should be fine in email bodies, as long as they are
reasonably common.  So letters like þðñçıß should be no problem, as well
as symbols like ½²“‘”’«»ħ .

How do these ones turn out?

Real numbers: ℝ
Mho (inverted Ohmega): ℧
Aleph: ℵ
Set membership: ∊
fi ligature: fi
alpha: α
beta: β
infinity: ∞

On 10/09/15 10:06, David Brown wrote:
> On 10/09/15 10:46, Tom Gardner wrote:
>> On 10/09/15 08:05, David Brown wrote:
>>> For Linux, you have other options.  Again, depending on the keyboard
>>> layout, AltGr-m is likely to work - certainly on my Norwegian layout I
>>> can get a wide range of characters with combinations of AltGr, shift,
>>> and different keys.  Then there is also the "compose" key - typically,
>>> this is not assigned by default, but you can set it up if you want it (I
>>> use the "scroll lock" key as compose).  Then µ is compose + m + u.
>>
>> Oh, excellent.
>>
>> Trying the other combinations is going to be a great way of
>> pointing out to idiot companies that valid email
>> addresses aren't limited to roman alphanumeric characters.
>> I think I'll use something like "þomas gardner"- they
>> won't notice the inaccuracy.
>>
>
> UTF-8 characters should, in theory, work in email addresses - but only
> in theory.  In practice, you can expect trouble somewhere along the line.

My objection is to web pages where I fill in my email address
and their stupid validation procedure rejects it. Even worse is
the case where one of the company's systems accept it and others
reject it - which can make it impossible for their internal staff
to change it!

I like using foo+srcotty@gmail.com when communicating with
Scrotty Co, so I can track leaks and easily filter messages.

On 10/09/15 07:42, Dimiter_Popoff wrote:
> I suppose I can get used to "us" instead of "uS", but may be an
> option would be se-6 or sE-6... I still can't quite swallow the
> thought that S is invalid for seconds. I looked at my very
> old tek 465B, well, its front panel is also designated with
> lower case around the timebase knob.

It more often occurred on their on-screen displays, e.g.
the first one I came across: http://www.jvgavila.com/t7d20_w1.jpg

On 10/09/15 08:51, John Devereux wrote:
> Tom Gardner <spamjunk@blueyonder.co.uk> writes:
>> One of my projects, when I get A Round Tuit, is to make a 2GS/s
>> 4GHz scope. The trick is to find a way to do it with only trivial
>> analogue front end components. Hence no amplifiers, no ADCs, just
>> a 50ohm input with very simple analogue components. Yes, there
>> will be limitations, but that's half the fun :)
>
> Hmm, there was the "slideback sampler" that John Larkin and Mike Monet
> used to fight about, a decade or two ago in sci.electronics.design.

As far as I can tell from mostly dead links, similar
but not quite the same.

Given that it is solely a hobby project, I would expect
my performance to be worse. For example, if someone mentions
jitter then I will shut my eyes, put my hands over my ears
and sing "la la la".

On 10.9.2015 г. 12:23, Tom Gardner wrote:
> On 10/09/15 07:42, Dimiter_Popoff wrote:
>> I suppose I can get used to "us" instead of "uS", but may be an
>> option would be se-6 or sE-6... I still can't quite swallow the
>> thought that S is invalid for seconds. I looked at my very
>> old tek 465B, well, its front panel is also designated with
>> lower case around the timebase knob.
>
> It more often occurred on their on-screen displays, e.g.
> the first one I came across: http://www.jvgavila.com/t7d20_w1.jpg

I now remember seeing that as well on newer scopes than my 465B
(which works and I am not planning to part with, it was THE scope
back in the day when I wanted to build one I was looking up to :-).
I may well have taken it from there.

I suppose the capital S is justified though. While invalid in "normal"
text, put into context - as on a scope display - it deals with the
"us" ambiguity which might be annoying at a subconscious level
not only to me. Well I may only be looking for an excuse to not
start changing the S in the spectrometry software on our devices :-).

Dimiter