The Notebook Review forums were hosted by TechTarget, who shut down them down on January 31, 2022. This static read-only archive was pulled by NBR forum users between January 20 and January 31, 2022, in an effort to make sure that the valuable technical information that had been posted on the forums is preserved. For current discussions, many NBR forum users moved over to NotebookTalk.net after the shutdown.
Problems? See this thread at archive.org.

    Does EIST extend battery runtime significantly?

    Discussion in 'Hardware Components and Aftermarket Upgrades' started by tomdav20, Mar 21, 2009.

  1. tomdav20

    tomdav20 Notebook Enthusiast

    Reputations:
    0
    Messages:
    29
    Likes Received:
    0
    Trophy Points:
    5
    Just want to know how much power consumption can EIST reduce if the notebook is idle, i.e. 0-5% CPU load.
    For example, if your battery can last 3 hours with EIST on, then how the runtime will be shorten with EIST off? Is it signifcant?

    Any comments are appreciated!
     
  2. Greg

    Greg Notebook Nobel Laureate

    Reputations:
    7,857
    Messages:
    16,212
    Likes Received:
    58
    Trophy Points:
    466
    The amount of power the processor consumes is proportional to the square of the operating voltage and linear with respect to its run-time frequency. It is also somewhat dependent upon how much work the processor is doing at any given time, but not nearly as much.

    So while having a computer run at 0-5% load at all time will help with battery life, being able to lower the clock of the computer (e.g. EIST) will help out even more.

    The amount of power saved is different between any two lines of computers, mostly because all the hardware is so different. On an ultraportable EIST can make a huge difference, while on the latest 17" gaming machine it may not matter much at all.

    Anyway, I will say that EIST is definitely a good thing and that there is no reason to turn it off. If buying, I'd pay attention to the battery life more...even those EIST does affect battery life it shouldn't be the deciding factor.

    What computer doesn't come with EIST now though???
     
  3. tomdav20

    tomdav20 Notebook Enthusiast

    Reputations:
    0
    Messages:
    29
    Likes Received:
    0
    Trophy Points:
    5
    Thanks for you quick reply. My CPU is Celeron Dual-core T1600. According to Intel's site it should support EIST. I don't know if it is a BIOS/chipsets/OS problem, I can't get EIST work on both Windows or Linux.

    The BIOS doesn't provide any settings about EIST. (This crappy BIOS only has Time/Date and Boot order setting) I requested for a BIOS update but never get reply.

    I have also done a lot of searching about this CPU -- Celeron Dual-Core T1600. Still cannot find a single user with this CPU can enable EIST on his notebook.

    The battery life is so poor! It works no more than 1hr20min even though everything is minimal -- dimmest screen, no wifi, speakers muted, 0% cpu load, etc. If it is a CPU problem then I may consider buying a pentium dual-core or C2D, but it would void the warranty.. err..

    Hope that more people will share their experience.

    EDIT:
    spec
    Celeron Dual-Core T1600 (EIST supported according to Intel's stie)
    SiS 671MX Chipsets
     
  4. John Ratsey

    John Ratsey Moderately inquisitive Super Moderator

    Reputations:
    7,197
    Messages:
    28,841
    Likes Received:
    2,166
    Trophy Points:
    581
    I suspect it is a BIOS problem. What notebook do you have?

    You can download and run RMClock which should show if your CPU is changing speed to match the workload. Select Mobile CPU on the Advanced page in order to show the voltages correctly (if the option is available).

    If you are lucky, you may be able to adjust the voltages or manually throttle the CPU to a lower speed. See the undervolting guide for details of how to use RMClock.

    John
     
  5. tomdav20

    tomdav20 Notebook Enthusiast

    Reputations:
    0
    Messages:
    29
    Likes Received:
    0
    Trophy Points:
    5
    Thanks for reply. I have tried RMClock. I can only lower the CPU frequency by throttling the FSB. The voltage and the multipliers are still fixed. The power consumption and battery life are still the same. I have looked at the undervolting guide and nothing works on this computer.

    This brand of this notebook is "HASEE". Their customer supporting site is too bad. Cannot even find the spec of this computer.
     
  6. John Ratsey

    John Ratsey Moderately inquisitive Super Moderator

    Reputations:
    7,197
    Messages:
    28,841
    Likes Received:
    2,166
    Trophy Points:
    581
    I just saw the Hasee thread. I had never heard the name previously.

    Possibly better than nothing but the bigger netbooks such as the Samsung NC20 will start to compete for this market segment.

    John
     
  7. tomdav20

    tomdav20 Notebook Enthusiast

    Reputations:
    0
    Messages:
    29
    Likes Received:
    0
    Trophy Points:
    5
    No one ever did an experiment of EIST on/off on the impact of battery runtime?
     
  8. John Ratsey

    John Ratsey Moderately inquisitive Super Moderator

    Reputations:
    7,197
    Messages:
    28,841
    Likes Received:
    2,166
    Trophy Points:
    581
    A lot. It's the difference between running the CPU at slowest speed and a lower voltage and full speed at a higher voltage. Greg gave you the theory but, to keep you happy I am running a quick test on my Dell E6400 with its P8600 CPU.

    1. Light usage = power drain of 10.5W

    2. CPU fully loaded at 800MHz and 0.925V = 13.5W

    3. CPU fully loaded at 2400MHz and 1.05V (undervolted) = 22.5W

    4. CPU fully loaded at 2400MHz and 1.1375V (default) = 26W

    If we assume that around 7W is used by the rest of the system, then the CPU power on light usage is 3.5W increasing to 6.5W under full load 800MHz, increasing to 14.5W at 2400MHz and 1.05W increasing to 19W at 2400MHz and 1.1375V.

    The T series CPUs have a maximum design power of 35W compared to 25W for the P series (they are selected because of the lower power consumption and cost more to buy).

    It is likely that the limited BIOS in your HASEE notebook doesn't allow the CPU to go into a low power state when idle, so it is running at full speed even when it has no work to do. In this case the 80 minutes on battery looks very plausible.

    You can use the battery page of RMClock to see the power drain when running off battery so you check what happens under different load conditions.

    John
     
  9. tomdav20

    tomdav20 Notebook Enthusiast

    Reputations:
    0
    Messages:
    29
    Likes Received:
    0
    Trophy Points:
    5
    Thanks a lot for your valuable information!
    EIST really makes HUGE difference!

    I googled around and find out a HASEE users site (not official) and many of them claimed that T1600 doesn't support EIST. In their opinion Intel gives wrong data on its site. They said although the BIOS hides the setting of EIST from the user, the default setting of EIST is 'on'. Once they replace their T1600s with a Pentium Dual-Core / C2D (T2xxx, T5xxx, T7xxx, etc), EIST works perfectly.

    I don't know whether I should trust them or not. It is difficult to believe Intel is a liar. But 80 minutes of battery life is really too short for me. I need at least 2 hours but unfortunately I cannot find a AC-main socket while I am not at home.

    Now I have two choice:
    1. Buy an extra battery pack (~40USD)
    2. Buy an cheapest Socket P CPU with EIST and install it by myself

    The cheapest Pentium Dual-Core is T2310 which cost ($36 -40 USD). Replacing it actually is a downgrade as T2310 = 133 x 11 = 1.47Ghz but T1600 = 166 x 10 = 1.67Ghz. All benchmarks also tell T1600 is much faster than T2310. Replacing the CPU very possibly void the warranty, and, most importantly, I don't know if it could really get EIST work.

    Buying an extra battery pack is easy but it will be so troublesome for me to hibernating > change battery > reboot everytime. It adds extra weight too.

    I am in a dilemma. Look like a fool. This is a cheap machine so I don't want to spend so much on it.

    I am running Linux now so I cannot try the RMCLOCK.
    But this command also shows the power usage:
    Code:
    cat /proc/acpi/battery/BAT0/state
    present:                 yes
    capacity state:          ok
    charging state:          discharging
    present rate:            2446 mA
    remaining capacity:      1549 mAh
    present voltage:         10509 mV
    
    At this moment everything is off or minimal except wifi is on. CPU load 0%-2%. It still draws 26W!
    Thanks for your information again!
     
  10. tomdav20

    tomdav20 Notebook Enthusiast

    Reputations:
    0
    Messages:
    29
    Likes Received:
    0
    Trophy Points:
    5
    I just did a test.
    With 0% CPU load, everything is off, power usage is around 24W
    With 100% CPU load, everything is off, power usage is around 40W

    What I really want to know is with EIST OFF (your CPU should be at stock frequency and voltage) and the CPU is idle, what is the power consumption?

    I am really interested in.
     
  11. John Ratsey

    John Ratsey Moderately inquisitive Super Moderator

    Reputations:
    7,197
    Messages:
    28,841
    Likes Received:
    2,166
    Trophy Points:
    581
    OK. I rebooted my Dell E6400, went into the BIOS and disabled SpeedStep.

    The result is that the CPU runs at a constant 1.600MHz instead of the normal 800MHz to 2400MHz range.

    Idle power consumption showed as 9.5W and full load power consumption (at 1600MHz, 1.0V) as 17.5W. The latter is where I would expect but former is a little surprising. First I should note that I didn't have a lot of things running in the background as I did yesterday so that may account for the 1W difference. However, the CPU still isn't using much power when on idle.

    Therefore SpeedStep along doesn't do much for the power management when idle. The real control for that is ability for the CPU to go into low power states known as C1, C2, etc., when different parts of the CPU get shut down. This behaviour is managed by the BIOS. This exercise reminds me that when I got my Zepto 6224W (actually made by Incentec in China) in mid-2007 I was wondering why that had poor battery life.

    A little research revealed that the CPU wasn't going into the C3 low power state. A BIOS update which enabled C3 gave a useful increase in battery time. You can check what is happening using Perfmon. Start > Run > Perfmon to start it. Then right click on the graph window and select Add Counters. %C2 and %C3 are among the options. You may want to select both and see what shows. If both sit on the 0 axis then neither are enabled. Or you may see that %C2 runs at 80 to 90% but %C3 stays at zero. There's no option to see the even deeper sleep states (C4 etc., - these show as C3). When a computer is on idle the CPU can spend around 90% of its time asleep.

    Some notebooks only enable these low power states when running on battery and some enable them all the time. I prefer the latter since it keeps everything cooler although there is a minuscule loss of performance because of the possible delay while the CPU has to wake up when there is work to be done.

    So, getting back to the original topic, there are two issues: (i) Changing of CPU speed to suit the workload and (ii) Enabling the various CPU low power states. The latter is more critical for good battery time.

    John
     
  12. tomdav20

    tomdav20 Notebook Enthusiast

    Reputations:
    0
    Messages:
    29
    Likes Received:
    0
    Trophy Points:
    5
    Thanks so much for your help!
    While it is idle Perfmon shows C2 ~35% and C3 ~60%.
    I think C3 has been enabled as default by the BIOS, although it hides this item from the user. As I mentioned on my previous post, I am unable to reduce the voltage and the multiplier by RMCLOCK or in Linux. I can only throttle the FSB but it doesn't reduce power consumption at all.

    Would you give me some advice to reduce the power consumption? I am willing to halved the performance for getting more battery life.
     
  13. John Ratsey

    John Ratsey Moderately inquisitive Super Moderator

    Reputations:
    7,197
    Messages:
    28,841
    Likes Received:
    2,166
    Trophy Points:
    581
    There may be some useful tips in this thread.

    However, I am inclined to think that the T1600 is a power-hungry CPU. The specs here show a voltage range of 1.075V to 1.175V. The minimum operating voltage (and therefore power consumption) is quite high. These CPUs are what is left in the sorting bin because they fail some of the specification requirements to be sold as more expensive CPUs.

    That web page also shows:
    but this page does say it is supported. :confused:

    Another CPU which has a lower specified voltage range is probably the only way to significantly reduce the power consumption.

    John
     
  14. tomdav20

    tomdav20 Notebook Enthusiast

    Reputations:
    0
    Messages:
    29
    Likes Received:
    0
    Trophy Points:
    5
    Thanks for your suggestion.

    I am still considering whether I should buy an extra battery pack or upgrade the cpu.