<div dir="ltr"><br><br><div class="gmail_quote">On Wed, Oct 1, 2008 at 7:18 AM, Robert G. Brown <span dir="ltr"><<a href="mailto:rgb@phy.duke.edu">rgb@phy.duke.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div class="Ih2E3d">On Tue, 30 Sep 2008, Nifty niftyompi Mitch wrote:</div></blockquote><div class="Ih2E3d"> .....</div><div class="Ih2E3d"> <br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
One site service of interest is AC power. A modern processor sitting<br>
in an idle state that then starts a well optimized loop will jump from<br>
a couple of watts to 100 watts in as many clocks as the set of pipelines<br>
is deep behind the instruction decode and instruction cache fill. A 1000<br>
processor (4000 cores) might jump from 4000 watts to 100000 watts in the<br>
blink of an eye (err did the lights blink). Buffer that dI/dT through<br>
the PS and it is less but still interesting on the mains which are synchronized.<br>
</blockquote>
<br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Interesting. I never have seen the lights blink although I don't run<br>
synchronous computations. One wonders if the power supply capacitors<br>
(which should be quite large, I would think) don't soak up the<br>
transient, though, even on very large clusters. Also, I think that the<br>
power differential is smaller than you are allowing for -- I don't think<br>
most idle processors draw "no" power...<div><div></div></div></blockquote><div><br>The dI/dT for processors can be quite high.<br>
<div class="compareResultFilterRowValueDescriptor"><span id="ctl00_cphBody_lblProcessor">AMD Phenomâ„¢ X4 Quad-Core is listed as a 140 watt part (thermal)<br>it is unlikely that all </span>450 million transistors are active in an idle<br>
loop. Tom's Hardware list the idle power at 21 watts. The speed at<br>which a modern processor can go from idle to full power is astonishing.<br>The local on board power supply regulation must respond very quickly. The delta<br>
from 21 to 140 fit inside one half cycle of a 50/60 Hz AC mains service. So inside<br>of one AC cycle the part can move from 21 to 140.... which is large when <br>multiplied by a 1000 node dual socket cluster. I do know of clusters and<br>
labs of workstations that power on hosts and disks in sequence to limit the <br>startup power surge. Lots of us have been at it long enough to know that <br>induction motors like elevators, refrigeration compressors and even vacuums can hit the mains<br>
hard enough to trigger errors. My home vacuum does dim the lights a little bit.<br><br>In normal practice I doubt that this is an issue but synchronization in the extreme <br>is interesting in its details and side effects. <br>
</div></div></div><br>-- <br> NiftyOMPI<br> T o m M i t c h e l l<br>
</div>