Page 31 of 36 FirstFirst ... 212627282930313233343536 LastLast
Results 601 to 620 of 703
  1. navinag's Avatar
    Posts
    455 Posts
    Global Posts
    489 Global Posts
    #601  
    I completely agree. Those people using the patches that can't boot are probably hitting a frequency ceiling or vmin issue. Palm probably never intended to use these features for the first Pre...so they got a lower bin part.

    Quote Originally Posted by kigmatzomat View Post
    It's pretty standard for most chip manufacturers to build all the chips on the same line and then "bin" them based on the QA/QC testing. The difference between a 1.8 and 2.0Ghz Intel processor often is that the 1.8 didn't test out at 2.0Ghz. The single core CPUs like Core Solo are often Core Duo chips that have one defective core that's been sealed off. The overclock-friendly CPUs are when the production lines reliably produce CPUs that pass QA at a higher clock-speed than their price "bin."

    If palm didn't plan to use the smartreflex and scaling features (possibly because they feared the way it might interfere with video playback and multitasking), their order to TI didn't require smartreflex/scaling certified processors. It cuts down the price and doesn't affect their needs.
  2. #602  
    Quote Originally Posted by kigmatzomat View Post
    It's pretty standard for most chip manufacturers to build all the chips on the same line and then "bin" them based on the QA/QC testing. The difference between a 1.8 and 2.0Ghz Intel processor often is that the 1.8 didn't test out at 2.0Ghz. The single core CPUs like Core Solo are often Core Duo chips that have one defective core that's been sealed off. The overclock-friendly CPUs are when the production lines reliably produce CPUs that pass QA at a higher clock-speed than their price "bin."

    If palm didn't plan to use the smartreflex and scaling features (possibly because they feared the way it might interfere with video playback and multitasking), their order to TI didn't require smartreflex/scaling certified processors. It cuts down the price and doesn't affect their needs.
    This bears repeating. This is standard practice in the microchip industry. Anyone saying "but it's supposed to run at 600mhz!" needs to keep this in mind. It might work fine, it might not--it's a risk.
  3. stubbs's Avatar
    Posts
    425 Posts
    Global Posts
    442 Global Posts
    #603  
    Quote Originally Posted by clipcarl View Post
    I haven't been on this thread for a while... Sorry about that. To summarize my experiences and the information collected from users in this thread: none of this is worth it.

    First, there is no method that is completely stable on every Pre. Weirdly, Pres vary wildly when it comes to battery life and their ability to to run the various methods described here. It almost seems as if Palm buys its chips from TI's reject bin and they're only guaranteed to work when underclocked to the stock 500MHz. My method (the original post) is probably the most stable on the most Pres, but even it doesn't work on all devices. The other methods (ondemand governor and smartreflex) are a lot less stable but some Pres will run them just fine. Frankly, none of these techniques is stable enough for it to be generally available and whoever put it in Preware didn't do anyone any favors.

    Second, even when you can find a method that is stable on your Pre, all of these methods only give marginal improvements in battery life at best. I have tried all of the methods on all of my own Pres (and I've been through a lot of them) and at best I might get an extra 15% increase in battery life. In my opinion that's not worth the stability issues and headache.

    So my first recommendation is not to use any of this. But if you must and you want the best battery life with the most stability and fewest glitches I recommend using the method in the original post and have it lock the speed at 500MHz with the screen on and at 250MHz with the screen off.
    I agree. The three Pres I've had have reacted very differently to different methods of powersaving and overclocking. That said, the one I've had for months loves SmartReflex and has no issues.

    However, this really isn't stuff for people who haven't used linux and don't know how it all works to try to run, it's just too risky. I agree that having the patches in Preware isn't a good idea, it just invites all sorts of problems. (Especially considering that these are possibly the first patches in Preware that could actually damage a phone). I say if you want to do it, go to webos-internals.org and do your homework first. Especially since it's a simple thing to echo the changes to file and test them before writing a startup script that won't let you boot if all doesn't go well.

    In summary, DON'T USE THESE IF YOU DON'T KNOW WHAT YOU'RE DOING.
  4. drizek's Avatar
    Posts
    333 Posts
    Global Posts
    769 Global Posts
    #604  
    So is there a javascript version of Prime95 that we can use to test for stability?
  5. #605  
    why would smartreflex and the other one even be in there if they couldnt use them at some point ?? i dont get it lol maybe im just tired
    ĦṔ-Ḷṫ-Ŧḯη
    Here is a direct link to webOS Doc for all carriers
    http://www.webos-internals.org/wiki/...octor_Versions
    P.S. if i have helped you and you are thankful please hit the thanks button to the right---->
  6. #606  
    My phones varied. One I could run at anything above 500mhz.
    One sped up quite a bit but stuttered on Pandora. None gave better battery life and all hated smartreflex.
    My new phone has been running on 600mhz smartreflex since it landed and it's been totally stable and responsive.

    But for most it shouldn't be that easy too install (:
  7. #607  
    Quote Originally Posted by Shadavis08 View Post
    why would smartreflex and the other one even be in there if they couldnt use them at some point ?? i dont get it lol maybe im just tired
    Just because Linux brings a lot of universal components does not mean that Palm decided to use all of those components in webOS.
  8. #608  
    Quote Originally Posted by grndslm View Post
    Just because Linux brings a lot of universal components does not mean that Palm decided to use all of those components in webOS.
    And to add to that, just because it's there doesn't mean it will work.
    Quote Originally Posted by Brain_ReCall
    I'm an Embedded Software Engineer. My idea of a Good User Interface is printf().
  9. #609  
    dont do it
  10. #610  
    too crazy
  11. #611  
    Quote Originally Posted by denisonbeau01 View Post
    too crazy
    The SmartReflex patches are actually pretty damn awesome. You can pick 500, 600, or 550 if you can edit one file. I'll prolly end up going with 550 myself, tho I haven't exactly tested it. I just know that the 600 was running great earlier. It gets GREAT battery life either way. It's getting like 0.35 % drain/hour with radios off, which is amazing compared to what it does stock. The thing is toooo slow right now with the battery drain test, so I can't even turn it on to play some more.

    It was definitely stable, tho. So, I'd highly consider running it at 550 if 600 is stable for you. If 600's not stable, then just stick with the 500 SmartReflex patch.

    Stock is whack, B. Voltage scaling FTW!
  12. #612  
    Quote Originally Posted by grndslm View Post
    Just because Linux brings a lot of universal components does not mean that Palm decided to use all of those components in webOS.
    It's not Linux. It's built into the TI chip.

    Quote Originally Posted by Brain_ReCall View Post
    And to add to that, just because it's there doesn't mean it will work.
    As proven in the past it may not work still (:
  13. xsavior's Avatar
    Posts
    80 Posts
    Global Posts
    89 Global Posts
    #613  
    Did anyone have an issue with the screen brightness using the Smartreflex 600? It seemed like my base brightness level was about 15% higher, even with the lower brightness patch installed. I went to the Smartreflex 500, and it's back to normal. Over all, both have been working well.
  14. #614  
    For what it's worth, I've had occasional issues with screen brightness being too bright when I first turn it on, even before I installed Smart Reflex patch. I don't think it's related.
  15. Tikerz's Avatar
    Posts
    447 Posts
    Global Posts
    519 Global Posts
    #615  
    Quote Originally Posted by Xsavior View Post
    Did anyone have an issue with the screen brightness using the Smartreflex 600? It seemed like my base brightness level was about 15% higher, even with the lower brightness patch installed. I went to the Smartreflex 500, and it's back to normal. Over all, both have been working well.
    Yes, I noticed that too. Sometimes it would jump out of it back to normal, but it was annoying. I stopped using the patches because I started to notice weird things happening in different programs.
  16. #616  
    Here's a suggestion for the patch developers: put together a non-permanent patch for testing. Instead of adding a shell script to the /init directory that runs at every boot, just update/create the ondemand, max_cpu_freq and min_cpu_freq files on the current session so that a full power cycle puts it back to factory.
  17. #617  
    I think I am missing something here. I turned on Powersave Smartreflex 500 MHz. /sys/power/sr_vdd1_autocomp and /sys/power/sr_vdd1_autocomp are set to 1. scaling_min_freq is 125000, but my cat time_in_state is:

    600000 0
    550000 590
    500000 473713
    250000 0
    125000 0

    250000 and 125000 are always 0. What am I missing?
  18. #618  
    SmartReflex is voltage scaling, not frequency scaling. They are two different modes of power saving. And at the moment the possibility of things going horribly wrong when both are used is too high to mess with since both modes are still pretty alpha.

    SmartReflex attempts to reduce the voltage (and thus power consumed) when the chip is idle without decreasing the frequency.

    Frequency scaling instead changes the speed of the chip, which winds up decreasing the amperage consumed even though the voltage is constant.

    The power consumed is voltage x amperage x time. Voltage is analagous to water pressure or velocity while amperage is the cross section or area of the current. If you have water flowing at 0.5 m/s in a 2 m^2 channel you have (0.5 x 2=) 1 cubic meter/s water flow. If you have 5v in a 0.2 amp circuit it produces (5 x 0.2) = 1 W of power.

    If you think of the chip as a water wheel spun by the battery's current, by allowing the wheel to spin slower (the chip frequency) you can keep the same water pressure (voltage) but use shallower water flow (amperage). This is scaling.

    Alternately you can keep the same depth of water (amperage) but try to decrease the water pressure (voltage) to the lowest it can go while keeping the wheel spinning normally (standard frequency).

    IMO, smartreflex has more chances of crashing because circuits can do weird things when not fed enough power. That's why I'm here in the frequency scaling forum.
  19. #619  
    I uninstalled the smartflex 600mhz patch but now I want it back...where can I get the patch?
  20. #620  
    Quote Originally Posted by kigmatzomat View Post
    SmartReflex is voltage scaling, not frequency scaling. They are two different modes of power saving. And at the moment the possibility of things going horribly wrong when both are used is too high to mess with since both modes are still pretty alpha.

    SmartReflex attempts to reduce the voltage (and thus power consumed) when the chip is idle without decreasing the frequency.

    Frequency scaling instead changes the speed of the chip, which winds up decreasing the amperage consumed even though the voltage is constant.

    The power consumed is voltage x amperage x time. Voltage is analagous to water pressure or velocity while amperage is the cross section or area of the current. If you have water flowing at 0.5 m/s in a 2 m^2 channel you have (0.5 x 2=) 1 cubic meter/s water flow. If you have 5v in a 0.2 amp circuit it produces (5 x 0.2) = 1 W of power.

    If you think of the chip as a water wheel spun by the battery's current, by allowing the wheel to spin slower (the chip frequency) you can keep the same water pressure (voltage) but use shallower water flow (amperage). This is scaling.

    Alternately you can keep the same depth of water (amperage) but try to decrease the water pressure (voltage) to the lowest it can go while keeping the wheel spinning normally (standard frequency).

    IMO, smartreflex has more chances of crashing because circuits can do weird things when not fed enough power. That's why I'm here in the frequency scaling forum.
    kigmatzomat, thank you for your excellent response. No wonder I am confused, I am in the wrong forum . I was not seeing that much improvement in battery life with Smartreflex ,and was looking for a reason. I used the scaling data (incorrectly) to try and justify. Maybe, I'll try the scaling patch.

Posting Permissions