hachyderm.io is one of the many independent Mastodon servers you can use to participate in the fediverse.
Hachyderm is a safe space, LGBTQIA+ and BLM, primarily comprised of tech industry professionals world wide. Note that many non-user account types have restrictions - please see our About page.

Administered by:

Server stats:

8.9K
active users

#turbostat

0 posts0 participants0 posts today
Benjamin Carr, Ph.D. 👨🏻‍💻🧬<p><a href="https://hachyderm.io/tags/Turbostat" class="mention hashtag" rel="tag">#<span>Turbostat</span></a> Utility Bumps 1024 <a href="https://hachyderm.io/tags/CPU" class="mention hashtag" rel="tag">#<span>CPU</span></a> Core Limit To 8192 Cores After <a href="https://hachyderm.io/tags/HPE" class="mention hashtag" rel="tag">#<span>HPE</span></a> Breaches It With a 1152 Core unnamed system.<br />We currently aren&#39;t aware of any server CPU configurations that can exceed this limit (in terms of physical cores), so this may be a custom or next-generation solution from <a href="https://hachyderm.io/tags/Intel" class="mention hashtag" rel="tag">#<span>Intel</span></a> or <a href="https://hachyderm.io/tags/AMD" class="mention hashtag" rel="tag">#<span>AMD</span></a>. The utility currently only supports <a href="https://hachyderm.io/tags/x86" class="mention hashtag" rel="tag">#<span>x86</span></a> processors, which seemingly rules out an Arm system from causing the issue. <br /><a href="https://www.phoronix.com/news/Linux-6.15-Turbostat" target="_blank" rel="nofollow noopener noreferrer" translate="no"><span class="invisible">https://www.</span><span class="ellipsis">phoronix.com/news/Linux-6.15-T</span><span class="invisible">urbostat</span></a></p>
claude<p>another example (Newton-Raphson zooming for Mandelbrot set):</p><p>with 7 threads : 105.26 Watts * 12.2 seconds = 1284.6 Joules<br>with 1 thread : 54.17 Watts * 52.1 seconds = 2822.3 Joules<br>sleeping / idle overhead : 26.01 Watts<br>7 threads minus overhead (105 - 26.01)W * 12.2s = 964 Joules<br>1 thread minus overhead (54.17 - 26.01)W * 52.1s = 1467.2 Joules<br>if the machine would be on/idle anyway: 105.26W * 12.2s + 26.01 W * (52.1 - 12.2)s = 2321.9 Joules</p><p>thus using more threads saves energy even when parallelism efficiency is far from perfect: best to get in/out as quickly as possible so you can turn the machine off (ideal case) / leave it fully idle (second best)</p><p>power consumption doesn't scale linearly with load (a little load increases a lot vs baseline: 1 thread doubles idle power consumption, but high load doesn't increase much more: 7 threads quadruples idle power consumption)</p><p>measured with turbostat on Debian, AMD 2700X CPU with default CPU scaling governor, usual browser/email/etc running too</p><p><a href="https://post.lurk.org/tags/ParallelComputing" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ParallelComputing</span></a> <a href="https://post.lurk.org/tags/Energy" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Energy</span></a> <a href="https://post.lurk.org/tags/permacomputing" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>permacomputing</span></a> <a href="https://post.lurk.org/tags/turbostat" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>turbostat</span></a></p>