Jump to content

Talk:Dynamic voltage scaling

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

"Voltage decreases quadratically with power consumption - halving the frequency reduces the power consumption by a factor of four."

[edit]

...but this does not follow from the equation (and the equation is correct). GregorB 23:00, 2 November 2007 (UTC)[reply]

The equation is correct as is the conclusion. If P ~ V^2, then reducing V by a factor of 2 reduces P by a factor of 4 - one half squared is one fourth. Raul654 (talk) 06:55, 28 March 2008 (UTC)[reply]

Merge in Undervolt/Overvolt

[edit]

As to the proposal to merge in undervolting and overvolting - I was also thinking that there are several overlapping articles here. I was wondering about merging Dynamic voltage scaling into the CPU core voltage article. Don't see that need separate articles for core voltage and changing core voltage.

As to the proposal of merging undervolting/overvolting. Two details that ought to be dealt with. 1) As I understand it, undervolting and overvolting are often applied in a static, rather than dynamic sense. So perhaps moving the article to Voltage scaling (or merging into CPU core voltage) as part of the merge would make sense. (With appropriate coverage of dynamic scaling, static, etc.)

2) Overvolting is applied to things other than processors. (e.g. RAM, busses, northbridges.) It may be that the same applies to undervolting, although it is not covered in the article. As it stands, the Dynamic voltage scaling article does not appear to cover these usages. Could voltage scaling be extended to cover voltage adjustments to other such systems?

If those two small matters can be dealt with, it seems reasonable to merge undervolting and overvolting into voltage scaling. And possibly to merge voltage scaling into CPU core voltage as well. Zodon (talk) 04:58, 18 March 2008 (UTC)[reply]

I just found another page on a closely related subject - DVFS covers Dynamic voltage and frequency scaling. Seems like this would also be a good candidate for merging. Perhaps Frequency Scaling and Voltage Scaling should both be merged into it? The nice thing about covering the two together is that it gives clearer place to cover the interaction between voltage and frequency. (Saving the problem of where to cover the interaction between them.) Zodon (talk) 06:02, 18 March 2008 (UTC)[reply]

On the proposal to merge in Voltage and frequency scaling, see my comments in Talk:Dynamic frequency scaling#Re: Proposed merger from voltage and frequency scaling. Zodon (talk) 02:32, 28 March 2008 (UTC)[reply]

[edit]

1. There is no need to have the same concept reapplied millions of times. An example is in literary writings, there is no word bank as it is assumed that the reader(s) understand the vocabulary.

2. With Links, there is no need to have the audience copying and pasting the keyword into the search box. If a person wishes to further understand the concept of Dynamic voltage scaling, they would just go into the section where the technology is listed as a device to, in my case, save power. 1oooop (talk) —Preceding unsigned comment added by 98.154.33.140 (talk) 06:06, 29 September 2009 (UTC)[reply]

Overvolting is done in order to increase computer performance, or in rare cases, to increase reliability.

[edit]

No, as far as I understand, it is not rarely done to increase reliability, it is always done to increase reliability.

Of course the need to improve reliability normally arises after reliability has been adversely affected by overclocking(which in turn was done to increase performance). Still the overvolting itself is done to increase reliability, not performance. The operating voltage of a CPU does not affect its computational power at all as long as the microchip is stable.

Programs that modify voltage

[edit]

I think it would be rather useful to maybe list some different programs for different OS/platforms that are used to scale computer voltages.

For example there is: RMclock [Windows], CPUGenie [Windows], CoolBook [Mac OSX]

I'm sure there are more and it would be helpful if they could be compiled on this wiki.--Psychedelia (talk) 09:57, 20 March 2013 (UTC)[reply]

"By applying a higher voltage to the devices in a circuit, the capacitances are charged and discharged more quickly, resulting in faster operation of the circuit and allowing for higher frequency operation"

[edit]

Is that true? My understanding is that the time it takes to charge and discharge to final voltage level is not dependent on the applied voltage level. It depends on the time constants (R and C). They wont change with applied voltage level. 14.139.128.14 (talk) 13:42, 12 November 2014 (UTC)[reply]

Yes, you are correct, and that part of the article is written poorly. The rate of change in voltage per unit of time is higher for larger voltage differences, however the amount of time to fully reach a certain percentage of filling the capacitor with charge (like 63.2% full) is dependent on RC only. Different supply voltages will charge the capacitor to the same percent full the same amount of time.
I'm going to correct to give a little better explanation: the higher supply voltages result in a faster slew rate (change in unit of voltage per unit of time in volts per second, not percent per second). And that faster slew rate allows for quicker transitioning through the mosfet's threshold voltage.
(I think whoever wrote the article was trying to dumb it down maybe to make it easier for people unfamiliar with capactiance to understand, but they dumbed it down too much to become incorrect: for instance the explanation of capacitance was wrong...it is not "a measure of how long it takes for a given current to produce a given voltage change"). Em3rgent0rdr (talk) 04:05, 27 January 2023 (UTC)[reply]
I've rewritten to be like:
Toggling a MOSFET's state requires changing its gate voltage from below the transistor's threshold voltage to above it (or from above it to below it). However, changing the gate's voltage requires charging or discharging the capacitance at its node. And each node in a circuit has capacitance, arising from various sources, primarily transistor gate capacitance, diffusion capacitance, and wires (coupling capacitance).
Higher supply voltages result in faster slew rate (rate of change of voltage per unit of time) when charging and discharging, which allows for quicker transitioning through the MOSFET's threshold voltage. Additionally, the more the gate voltage exceeds the threshold voltage, the lower the resistance of the transistor's conducting channel. This results in a lower RC time constant for quicker charging and discharging of the capacitance of the subsequent logic stage.
This explanation could still be tweaked a bit, but it is much more correct than what the article had apparently said since 2014. Em3rgent0rdr (talk) 04:34, 27 January 2023 (UTC)[reply]