Ok, I have just installed eCS Silver and ACPI because I have SMP processor. Works fine. But I don't know much about ACPI. I understand it can help reducing power consumption, but how?
When I power on my PC the cooling fans start with full speed and lot of noise. If I select to boot to eCS the speed and noise will continue at the same level and never be reduced. But if I boot to Win Vista the fans can't be heard - what a lovely silence. I have tried to change some settings, but to no help.
So, is there anything I can do about it?
/einar
First, tell us your full PC setup. If it is a retail model, that will be much easier. If not, Motherboard/CPU, etc.
I'm betting that in Windows the ACPI driver (or maybe even a utility) is throttling the fans down based upon reported temperature. This is sometimes called QFan, SmartFAN, etc. Also, many BIOS have this functionality built right in (ASUS boards from my experience) and should be independent of the OS used.
I'm not sure our ACPI driver does much other than provide SMP support, high IRQ support. I don't know as if it allows for CPU frequency/voltage scaling or Fan control.
I guess that Eugene Gorbunoff will best be able to answer these questions.
It depends on your CPU and motherboard.
The current implementation of ACPI for eCS does not support AMD's PowerNow and Cool 'n Quiet features. Your best chance then would be to have a look in the BIOS setup of your motherboard. I have an ASUS board that supports something like fan profiles called QFan. If I select the lowest, the system is about as quiet as when booted to Windows. Still, the processor frequency is not reduced when idle.
If you have an Intel board, you can try the SetGetState and SetGetThrtl utilities that come with ACPI and the patched kernel from the betazone. Also have a look at your BIOS settings.
Most likely, ACPI is perhaps installed but it is not working. A few tests, that can tell you more:
(1) Power consumption. ACPI is responsible for "sleep cycles" when your CPU has nothing reasonable to do, Does not work for you because you are hearing fans in "heavy duty" all the time.
(2) Shut down. Are you able to power off your machine by software? You command "Shut down", the machine shuts down and powers off. If you are able then ACPI is running somehow.
(3) Lid, standby, hibernate (notebooks). If you are closing NB lid, the display should get turned off near the "closed" position. This is an ACPI function. Also, ACPI is responsible for "standby" and "hibernate" states. If something of this works, ACPI is functional somehow.
(4) High IRQs (above 16 or 10h) used by your HW. If your HW manager (more reliable is sysinfo) shows high IRQs then your ACPI is running somehow and APIC is in use.
If nothing of (1) - (4) works then ACPI is not working. On a UNI machine, you can comment out the PSD= and the BASEDEV=APM.ADD statements. On SMP, you must have ACPI installed even if it is not working so that you must not comment out the statements.
QuoteOn SMP, you must have ACPI installed even if it is not working so that you must not comment out the statements.
This is not always true (perhaps never true). I find that SMP works better with the old OS2APIC.PSD that it does with ACPI.PSD (AFIK you can't use both). With ACPI.PSD, I am lucky if I can get 4 hours, before the system hangs solid (sometimes, it takes only a few minutes). With OS2APIC, I get about a week, before odd things begin to happen, and then it is almost never a solid hang. I doubt if the problem is related to the actual SMP support in ACPI, but there is something in there that is deadly, and it has never left any tracks, that I can find. Of course, power control needs either ACPI, or APM, and a lot of new systems no longer support APM, or it is broken, and won't work properly. Again, you can't use both. All I can say, is that you need to experiment with the options, to find the best setup for your system. If ACPI doesn't work properly, be sure to report it (see the docs). The problems won't get fixed if you don't report them (they may not get fixed anyway, in my experience, but you need to try).
Well, the AMD's Power Now IS supported, as well as Intel's SpeedStep.
I can under-power my CPU on my (single core) Think Pad manually - from 1,7GHz (max.) to 600MHz (min.).
The real question is whether ACPI supports this on YOUR system.
The utility used for this is called: SetGetState.exe
You might also try setting up the ACPI throttling feature in X:\MPTN\ETC\acpid.cfg, this will de-crease the load (and temperature) of your CPU as well.
Quote from: DougB on 2009.09.16, 23:21:06
QuoteOn SMP, you must have ACPI installed even if it is not working so that you must not comment out the statements.
This is not always true (perhaps never true). I find that SMP works better with the old OS2APIC.PSD that it does with ACPI.PSD (AFIK you can't use both)...
Doug, I think it depends on the hardware. Most of my servers, being as old as they are, will doubtfully work with the new methods/drivers.
QuoteWell, the AMD's Power Now IS supported, as well as Intel's SpeedStep.
I know that some functions are supported, with some Intel processors. I have been completely unable to get much of anything (Power) to work with AMD, other than simple power off (Asus M3A78-EM motherboard, with quad core Phenom processor). I would love to be able to throttle down my desktop. It has two available states, according to one of the utilities, but nothing changes anything. From what I understand, there is a program that might work, but I have been unable to locate such a thing.
I have been able to use setgetState.exe to change the clock on my Thinkpad T43 (1871-W8M - Intel, with 4 states available), which makes a tremendous difference in temperature (according to the Reactor temperature widget), and in how long the battery will last (according to the Beer Battery widget). Performance is quite acceptable, even at the slowest speed. Throttling, on the other hand, is pretty flaky. It claims that it can be throttled to 13% in steps of 13%, but very little actually works, if I throttle it below 50%. Doing that doesn't make a lot of difference anyway, and it seems to work well at 50%. None of that works with ACPI 3.14, as comes with eCS 2.0 RC6a. ACPI 3.16 works , but seems to have a few other problems. ACPI 3.17 is very broken on all of the systems that i have tried it on. ACPI 3.18 works about the same as 3.16, on my desktop (which isn't good), but I haven't found the time to try it on the laptop, yet.
I have also been unable to get my T43 to resume. It appears to suspend properly, but I can't get it to wake up again. The screen comes on, but that is all. There is no display, just the background light. The only thing that I can do, is hold the power button, until it goes off. I haven't spent the time to try every possible combination with it, so I might have some settings that need to change.
I can see that ACPI is going to be necessary, someday, but it is barely usable, even on Intel machines (AMD is supposedly not supported, which is not acceptable). The main problem seems to be that the ACPI specs are not properly documented, and every manufacturer takes a different approach, with every machine they make, trying to get it right. Obviously, Microsoft totally ignores that stuff, and does it themselves. The ACPI team is trying to do it according to the specs, and that just doesn't work, because nobody seems to know how what the specs actually mean.
The original question was:
QuoteI understand it can help reducing power consumption, but how?
The answer seems to be that SetGetState.EXE, and SetGetThrottle.EXE, are the available tools. IF ACPI actually supports them, on your specific machine, they should help. If not, you are left with the option of reporting the problems to the ACPI developer, and hope that they will figure it out, someday.
FWIW, I get the impression that Mensys realizes that ACPI is a major problem. Whether they can do anything about it, or not, depends on a lot of things. One of those things is finding people who are willing to do a little work to report successes, and failures, to the ACPI developer. Unfortunately, that also means the users must have a valid Software Subscription (so you can get the software to test), and they need to do some serious reading, of poorly written, and sometimes misleading, and confusing, documentation. It also takes a LOT of time to try all of the possible combinations, to see what does, or doesn't, work. If you are not sure what you can do to help the project along, start by reading the docs (more than once), and start trying things, to see what works, and doesn't work. Then report your findings (after seeing if somebody else has reported the problems, so you don't duplicate things).
Have fun...
Quote
I have also been unable to get my T43 to resume. It appears to suspend properly, but I can't get it to wake up again. The screen comes on, but that is all. There is no display, just the background light. The only thing that I can do, is hold the power button, until it goes off. I haven't spent the time to try every possible combination with it, so I might have some settings that need to change.
I was able to suspend/resume my R51 pretty nicely, with ACPI 3.17, /VBE parameter and UniAudio and USB drivers removed.
Unfortunately the network stopped working (both Intel 100/1000 and Intel 2200BG wireless using GenMac). There was either the issue with the GenMac drivers or the networking subsystem needed "reinitializing". I did not find a way through though...