[英]What is better from an energy-saving perspective to run task minimum energy consumption
The CPU frequency and CPU usage are the main factors that impact energy consumption (as far as I know). CPU频率和CPU使用率是影响能耗的主要因素(据我所知)。 however what is better from an energy-saving perspective to run task minimum energy consumption:然而,从节能的角度来看,运行任务的最低能耗更好:
Option 1: Maximum CPU frequency with minimum usage选项 1:使用最少的最大 CPU 频率
Option 2: Maximum CPU usage with min frequency.选项 2:最小频率下的最大 CPU 使用率。
Work per time scales approximately linearly with CPU frequency.每个时间的工作量与 CPU 频率大致呈线性关系。 (A bit less than linear because higher CPU frequency means DRAM latency is more clock cycles). (比线性低一点,因为更高的 CPU 频率意味着 DRAM 延迟是更多的时钟周期)。
CPU power has two components: switching (dynamic) power which scales with f 3 (because voltage has to increase for higher frequency, and transistors switch are pumping that V^2 capacitor energy more often); CPU 功率有两个组成部分:开关(动态)功率,与 f 3成比例(因为电压必须增加以获得更高的频率,晶体管开关更频繁地泵送 V^2 电容器能量); and leakage power which doesn't vary as dramatically.和泄漏功率变化不大。 At high frequency dynamic power dominates, but as you lower the frequency, eventually it becomes significant.在高频时,动态功率占主导地位,但当您降低频率时,它最终会变得很重要。 The smaller your transistors, the more significant leakage is.晶体管越小,泄漏越严重。
System-wide, there's also other power for things like DRAM that doesn't change much or at all with CPU frequency.在系统范围内,对于 DRAM 之类的东西,还有其他功能不会随 CPU 频率变化太多或根本不会变化。
Min frequency is more efficient , unless the minimum is far below the best frequency for work per energy.最小频率更有效,除非最小频率远低于单位能量工作的最佳频率。 (Some parts of power decrease with frequency, others like leakage current and DRAM refresh don't). (功率的某些部分随频率降低,其他部分如漏电流和 DRAM 刷新则不会)。
Frequencies lower than max have lower work per energy (better task efficiency) up to a certain point.低于最大值的频率在一定程度上具有较低的单位能量工作(更好的任务效率)。 Like 800 MHz on a Skylake CPU on Intel's 14 nm process.就像英特尔 14 纳米工艺的 Skylake CPU 上的 800 MHz。 If there's work to be done, there's no gain from dropping below that;如果还有工作要做,那么低于该水平就没有任何好处; just race-to-sleep at that most efficient frequency.只是以最有效的频率争分夺秒。 (Power would decrease, but work rate would decrease more below that point.) (功率会降低,但工作率会在该点以下降低更多。)
https://en.wikichip.org/wiki/File:Intel_Architecture,_Code_Name_Skylake_Deep_Dive-_A_New_Architecture_to_Manage_Power_Performance_and_Energy_Efficiency.pdf is slides from IDF2015 about Skylake power management covered a lot of that general-case stuff well. https://en.wikichip.org/wiki/File:Intel_Architecture,_Code_Name_Skylake_Deep_Dive-_A_New_Architecture_to_Manage_Power_Performance_and_Energy_Efficiency.pdf是 IDF2015 中关于 Skylake 电源管理的幻灯片,很好地涵盖了很多一般情况。 Unfortunately I don't know where to find a copy of the audio from Efraim Rotem's talk;不幸的是,我不知道在哪里可以找到 Efraim Rotem 演讲的音频副本; it was up for a year or so after, but the original link is dead now.大约一年后它就出现了,但原来的链接现在已经死了。 :/ :/
Also in general about dynamic power (from switching, not leakage) scaling with frequency cubed if you adjust voltage as well as frequency, see Modern Microprocessors A 90-Minute Guide!如果您调整电压和频率,一般情况下动态功率(来自开关,而不是泄漏)随频率的立方缩放,请参阅现代微处理器 90 分钟指南! and和
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.