Posts about hardwarehttps://mithrandi.net/categories/hardware.atom2018-05-19T08:14:04ZmithrandiNikolaClock drifthttps://mithrandi.net/blog/2007/01/clock-drift/2007-01-27T13:58:00Z2007-01-27T13:58:00Zmithrandi<div><p><a name="d27t1158"></a>
</p><div>
<h5 class="tags">tags:</h5>
<ul class="tags">
<li><a rel="tag" href="http://technorati.com/tag/hardware"><img src="https://mithrandi.net/images/techtag" alt="[tag]"> hardware</a></li>
<li><a rel="tag" href="http://technorati.com/tag/clock"><img src="https://mithrandi.net/images/techtag" alt="[tag]"> clock</a></li>
<li><a rel="tag" href="http://technorati.com/tag/drift"><img src="https://mithrandi.net/images/techtag" alt="[tag]"> drift</a></li>
</ul>
<p>A post on python-dev by Neil McLaren caught my eye today:</p>
<blockquote>
<blockquote>
<blockquote>
<p>A signal sent while the operating system is doing certain things to</p>
<p> the application (including, sometimes, when it is swapped out or</p>
<p> deep in I/O.)</p>
</blockquote>
<p>That sounds like an outright bug. I can’t think</p>
<p> of any earthly reason why the handler shouldn’t</p>
<p> be called eventually, if it remains installed and</p>
<p> the process lives long enough.</p>
</blockquote>
<p>See above. It gets lost at a low level. That is why you can cause</p>
<p> serious time drift on an “IBM PC” (most modern ones) by hammering</p>
<p> the video card or generating streams of floating-point fixups. Most</p>
<p> people don’t notice, because xntp or equivalent fixes it up.</p>
<p>Neil McLaren</p>
</blockquote>
<p>If you’ve been wondering why the clock drift on your PC seems so</p>
<p> bad, perhaps this is part of the problem?</p>
<p>In my next blog post, I intend to detail setting up dual-path</p>
<p> routing to take advantage of the peculiarities of the South African ISP</p>
<p> market to save yourself some money if you are using an ADSL</p>
<p> connection.</p>
</div></div><div><p><a name="d27t1158"></a>
</p><div>
<h5 class="tags">tags:</h5>
<ul class="tags">
<li><a rel="tag" href="http://technorati.com/tag/hardware"><img src="https://mithrandi.net/images/techtag" alt="[tag]"> hardware</a></li>
<li><a rel="tag" href="http://technorati.com/tag/clock"><img src="https://mithrandi.net/images/techtag" alt="[tag]"> clock</a></li>
<li><a rel="tag" href="http://technorati.com/tag/drift"><img src="https://mithrandi.net/images/techtag" alt="[tag]"> drift</a></li>
</ul>
<p>A post on python-dev by Neil McLaren caught my eye today:</p>
<blockquote>
<blockquote>
<blockquote>
<p>A signal sent while the operating system is doing certain things to</p>
<p> the application (including, sometimes, when it is swapped out or</p>
<p> deep in I/O.)</p>
</blockquote>
<p>That sounds like an outright bug. I can’t think</p>
<p> of any earthly reason why the handler shouldn’t</p>
<p> be called eventually, if it remains installed and</p>
<p> the process lives long enough.</p>
</blockquote>
<p>See above. It gets lost at a low level. That is why you can cause</p>
<p> serious time drift on an “IBM PC” (most modern ones) by hammering</p>
<p> the video card or generating streams of floating-point fixups. Most</p>
<p> people don’t notice, because xntp or equivalent fixes it up.</p>
<p>Neil McLaren</p>
</blockquote>
<p>If you’ve been wondering why the clock drift on your PC seems so</p>
<p> bad, perhaps this is part of the problem?</p>
<p>In my next blog post, I intend to detail setting up dual-path</p>
<p> routing to take advantage of the peculiarities of the South African ISP</p>
<p> market to save yourself some money if you are using an ADSL</p>
<p> connection.</p>
</div></div>