[ntp:questions] NTP Bug 2328 - Vista/Win7 time keeping inaccurate and erratic

Brian Inglis Brian.Inglis at SystematicSw.ab.ca
Fri Nov 29 13:42:10 UTC 2013


On 2013-11-29 05:14, Martin Burnicki wrote:
> Brian Inglis schrieb:
>> On 2013-11-28 02:31, Martin Burnicki wrote:
>>
>>> On the other hand, as far as I know, there is no new API to apply
>>> adjustments
>>> to the Windows system time more smoothly.
>>
>> Mention of smooth reminded me of using Bresenham's algorithm, generically
>> called the digital differential analyzer, to manage and compensate for
>> quantized data during allocations (unfortunately not in finance systems,
>> so unable to perform direct personal compensation!) ;^>
>>
>> These techniques are best known for drawing smooth digital lines and arcs,
>> but also provide anti-aliasing input, and are useful for properly handling
>> residuals of quantized or rounded data, so the output data is as smooth (in
>> either a visible or virtual sense) as possible.
>
> Are you referring to my graphs, or to the way time adjustments are applied to the Windows system time? ;-)

Time adjustments - time graphs should be spikey or they are unbelievable, so for
your bottom figure I would average readings to be able to see any variations!

DDA is a well proven approach for handling residuals of data that has to be doled
out in lumps, whether that be single x,y pixels, 16 ticks, or 100 m3 ;^>
May be able to use it to consistently handle all input and output time periods.

Used it last night looking at phong shading plots to produce highlights (shape
size reduces along cosine curve while each rgb component shades to white),
figuring out how many shades are required and will reasonably fit in the space,
while producing correct sizes and shades that end with some pure white area.

-- 
Take care. Thanks, Brian Inglis


More information about the questions mailing list