*** Official Clevo W110ER / Sager NP6110 Owner's Lounge ***

Prema's BIOS gives one the option to undervolt.. thus more battery power

 

Most of us are aware of this, but Prema's BIOS offers more than just disabling Turbo. You can turn hyperthreading on/off, speedstep on/off, undervolt the dedicated GPU, overclock your system RAM, even overclock/underclock the IGP (HD 3000/4000).

Plus Turbo is disabled anyhow with dedicated GPU enabled unless you use ThrottleStop.

 

Turbo still on for me with dGPU enabled, but I'm using sandy bridge ?
Won't it just disable on battery ?

Also, I wonder what conditions would make interesting to disable hyperthreading and/or speedstep ?

 

How to underclock IGP from bios ?

 

Nope, with dGPU enabled (as in 3D app using it with Optimus), Turbo does not work. No software conditions will disable hyperthreading or speedstep. Well you can pseudo turn it off by restricting number of cores in msconfig, but then the CPU speed is fixed.

 

Hmm had a blue screen yesterday - minimized ArmA2-DayZ and opened Skype to make a call, and while connecting, it "Blued out". I didn't read carefully what the blue screen said, since I was in a hurry to reset and make that call.

I'm not really worried about it, considering how glitchy ARMA 2 is and how DayZ makes it even more glitchy, and remembering how Skype used to crash two of my previous PCs, but just wondering what could have been another cause for the blue screen. Had Turbo On through Throttle Stop and no OC'ing done to CPU or GPU. Just a second before HWiNFO64 was showing CPU temps at 75 and Memory usage at ~60.

 

True. To check frequency I used throttlestop, without it beeing turned on, but I just realised it still forces turbo on if checked...

About hyperthreading/speedstep, allow me to reformulate:
In which cases would someone want to disable them ?

 

What is the latest on replacement screens for our favorite tiny terror? Do we have a definitive favorite as a drop in replacement? Any rumors about higher resolution or something other than TN?

 

I had the same with DayZ. I played tons of other games for hours, but DayZ was the only one I ever had trouble with: My machine just shut down, even though temps were very decent.

 

Not that I have heard about. It would be great if someone had but so far the Stock Glossy and the Matte version that Mythlogic originally found seem to be it.

I don't think anyone has torn apart one of the new Asus 11inch Zen books to tell us what kind of display connector it uses. That is going to be the main limitor as most of the higher rez displays are using eDP now and our laptop uses the older LVDS connector type.

 

Ok, good to know my assumption that it had been DayZ's fault was correct and I'm not alone in this. DayZ is a pretty fidgety piece of software after all, with ARMA's awkwardness and unreliability dating back to Operation Flashpoint days...

 

I don't know much about this but I recall reading something about the LVDS needing different numbers of lanes for different resolutions (like dual link DVI vs. regular) has anyone confirmed that our W110ERs could even run a higher res screen if one was made?

 

I dont see how everyone with the beta 304.48 driver doesnt seem to report any optimus bugs or delays when opening simple things like a text file. When ever I try to open something like 'Calculator' it takes nearly 10-15 seconds before it shows up (Clean install of win7 x64 and 304.48 driver)

This is for an SSD not a HDD (You shouldn't ever have to wait 10 seconds to open a 300kb executable file, especially on a SSD which should be instant)

From what ever HTWing said, when opening a program, the gpu is activated, then deactivated causing some delay. Then the program starts.

I would surely like to have the latest driver. But geforce.com doesn't seem to have an official driver for the 650m. This happened to me before when I had a 540m, there was no driver to be seen for months even when the chip has been out for a few months as well.

 

there's a new beta driver, try it.

 

Seriously? I was blaming Win8 for that annoyance. I'll try the new driver ASAP!

 

64bit link is corrupted, here's direct download 306.02 beta:
http://us.download.nvidia.com/Windo...n8-win7-winvista-64bit-international-beta.exe

 

I didn't have a problem as serve as described above but there was a 1-2 second delay. With the 306.02 drivers installed everything is snappy again! Thanks!

 

Is that new driver the best possible one? I am asking because it is still in beta stage. In fact the only three drivers they have on nvidea's page are in beta stage. Why is this, anyone know? Also, what was wrong with the stock driver that people thought it was best to update?

 

Prema Mod for this model updated to version v1.4 (UV)

http://forum.notebookreview.com/sag...ies-prema-bios-mod-review-31.html#post8811968

 

I just wanted to use one I could download from nvidia instead of from Clevo. Its crazy that there still isn't a non-beta driver for us to use though. FWIW I haven't had any problems with the 306.02 driver crashing like the last one.

Also, Anandtech just posted their review of the Asus UX31A and of course had to throw the insane battery life of their Eurocom W110ER in my face again as I read it. I couldn't help but comment and Jarred Walton replied with some insight:

So is there still some hope for a magical increase in battery life for our favorite little powerhouse? Has anyone contacted Eurocom to ask if they used some secret ingredient when cooking up their version?

 

Ok, my problem is that games refuse to play at their proper speed with the nVidia is engaged. Plays very slow and drags. Often takes 5 minutes just to get to the main menu. But it plays relatively fast on the intergrated graphics. Of course, not as fast or as detailed as when the nvidia was working properly but it is working. I am just confused on what would cause the nVidia to suddenly stop working as well and I am hoping the drivers will fix it.

 

Would be nice if a user here had that version !

 

Yeah! I'd like to have a few more data points confirming this is real and not some kind of glitch...

 

None of the other reviews for the eurcom monster were able to get that much run time. The new 1.0.4 bios helps with power draw. You should able to get 4.5 hours with light internet usage instead of the 3.5. I don't know why intel didn't put in the ability explicitly power gate or turn off cores when not being used.

 

You could also drop the max. TDP down to 24W with the mod BIOS. So the system will keep the CPU below even on max load.

 

Yeah they probably measured battery life when the time was set from Daylight Savings back to normal. And now they are proud, that they have a full hour more than their competitors. ;P

 

Maybe not the most convenient way, it would mean reboot when autonomy is needed, and another time after that, a software solution like throttlestop is better, unfortunately the GUI/options are pretty poor and it isn't open source

 

Throttlestop won't work for the TDP on this model though...limiting the multi won't necessarily drop the power consumption...too bad Clevo removed the parts required to drop it down to 14W and 9W...

 

What's so bad about throttlestop? It works great for me, you can use it to underclock or overclock, if allowed by your BIOS.

 

That's not what he is saying. He said, setting a TDP limit in BIOS is not as convenient as using a Software Solution like ThrottleStop, as it requires a reboot.

 

No, he said the GUI/options are pretty poor and Throttlestop isn't open source. Correct, it is not open source, but you could probably talk to the developer if you really wanted to see the code. I don't see what's wrong with the GUI.

Also, throttlestop can edit the TDP but I am not sure if that is locked down on the BIOS on these machines or not, so you may be stuck to just limiting the multiplier, which will help, but potentially not as much as a real TDP change.

 

lol Alright, so you were focusing on the second half of his post, whereas I focused on the first. ;P

 

How's that possible, I don't get it... plus it does indeed drop power consumption on mine

Told ya, GUI & options.
It works fine (well, almost) but you can't set it to do what you want automatically (not for everything), so it has to be operated manually.
In addition, it doesn't minimize correctly, won't show its main window if launched twice (shows "already running" instead, so usefull !) and other minor glitches..

Those make it quite a pain to use daily. It's also a bit cluttered, you can easily tell that uncle_webb only cares about core functionality...
It's so nice that this utility exists, too bad it's being leftover on closed sources though.

I tried, got an answer at first (I won't work on it anymore except bugfixes & cpu support), then asked if he could open sources / share code.... no answer ^^
The guy wanted to make money out of it, it didn't work well so he's a bit angry at the community now. Problem is, I couldn't even find a donation button

 

I don't see 1.00.04 or any BIOS listed on Clevo's site. I'd prefer to use the stock version before flashing to the modded one...

 

Yep it's quite a strange behavior from clevo, must have special clearance to access them, seems like they don't want their work to be installed by users, they only work on BIOS for their sellers :-/

Maybe it would be worth to write to their support, to better understand the situation..

 

You can also download the stock at Premas BIOS Mod Page

 

Forcing core parking in Windows Processor Power Settings may be a viable option to limit power consumption. See this post first that explains how to reaveal core parking settings in Windows 7 power plans Advanced power options:
http://forum.notebookreview.com/sager-clevo/416916-clevo-bios-thread-52.html#post8773184

E.g.: Change your power plan to "Power Saver" and try to limit the number of active cores with the following settings (Note: each setting can be specified for when the laptop is either "On battery" or "Plugged-in"): (For Clevos, it's probably a good idea to change Hotkey Power Conservation Mode to "Balanced" or "Energy Star" prior to changing windows power plan settings)

"Processor performance core parking min cores" = 0% <-- Allow all cores to be parked
"Processor performance core parking max cores" = 12% <-- maximum number of cores that are allowed to be "unparked", i.e. force 88% (7/8 cores) to be parked.
"Processor performance core parking core override" = Disabled

Also, the maximum processor frequency could be reduced with:
"Processor performance boost policy" < 100%
"Maximum processor state" < 100%
"Minimum processor state" < "Maximum processor state"

"System cooling policy" = Passive <-- Not sure this does anything on Clevos

 

Thanks, I didn't scroll down quite far enough before.

 

Is the power consumption exponential compared to the freq. increase ?
If there's a graph somewhere that would help locate the sweet spot.

I mean, if it's not exponential it should be better to get a job done by 2 cores @ 2Ghz in 1 minute 20, than by 1 core @ 2ghz in 2 minutes, and even more than 1 core @ 1ghz in 4 minutes, right ? Since the rest of the PC will also consume power during that time.

 

the equation for power consumption is CV^2*f. So power should be linearly related to frequency and have a square root relation to Voltage.

 

It is linear with respect to frequency but exponential with respect to voltage. Also note that changing the frequency will also change the voltage as well.

EDIT: To clarify, as the frequency lowers down the voltage does as well. For each frequency(multiplier) there is a different VID (voltage ID) used by the CPU. Since the heat/power with respect to voltage is squared, it is more efficient to take longer at a slower clockspeed. The relationship is such P = CFV^2, where P = power, C = scaling constant, F = frequency, and V = voltage.

 

wow, that's too much for a natural scientist. I am off.

 

so we'd need a graph or table to check what's the best freq/voltage combination to lower consumption in the long term..

 

Kevin (the TS developer) explained to us, that reducing the multis won't drop the power consumption because the CPU needs to remain longer at a lower multi to finish the same work. He suggested that a higher multi allows the CPU to drop faster back to a lower power state, which in return would yield in a lower overall power consumption.

 

Well, not really. Essentially the lowest voltage possible is the most efficient. Since frequency alone is linear, it doesn't affect the efficiency. Half the frequency = twice the time as half the power, IF the voltage is static, sp half the frequency and half the voltage = less than half the power for twice the time, that's a net gain.

 

Ok that's what I tought, but it's only true if it's linear, as debated above

I fail to see how that differs from what I said !
Since voltage only increases by levels, a graph/table would help =)

 

No, I am saying you don't need a graph. The answer is the lowest voltage (which occurs at the lowest speed) is the most efficient. That's all there is to it.

If you must, the graph for power used vs frequency would look like y=x, where power is the y axis and frequency is the x axis, and then the graph for power would be y=x^2 where y axis is again power and x axis is the voltage.

 

Ok, I initially believed that voltage will be the same for 8x or 9x multipliers, as an example.
Looks like it's not the case and voltage changes every step, as you edited afterwards.

Still, that theory only stands for the cpu, I don't want to make this more complex than it seems ^^, but let me demonstrate a case:

At time T0
your computer consumes 50W, your cpu 20W @20x
my computer consumes 40W, my cpu 10W @12x

At time T1
job done, your computer now consumes 25W, your cpu idles at 5W
my computer still consumes 40W, job is unfinished

At time T2
both have done the job, and idles @25W

If 1.5x(T0-T1) > (T1-T2), cpu theory becomes wrong and global power consumption is higher @12w, with a lower voltage.

Not sure if that verifies itself but you see where I'm aiming at ?

 

Well, you did your math wrong, if the base system without cpu is 30w then the 50w system only drops to 35w not 25w, at t1, while the second system remains at full 40w. If you do the math then, the second system comes out ahead, with 80 vs 85.

 

Here's the interesting paper I read about power consumption vs CPU speed.

Power Optimization - a Reality Check
http://www.cs.berkeley.edu/~krioukov/realityCheck.pdf

As for ThrottleStop, you're right, it has a few quirks and it won't win any beauty contests with its crowded GUI from 1998. I think most W110ER owners are too busy happily gaming at full speed to be too concerned about how ThrottleStop looks.

My goal was to make enough money so I could buy some new hardware so I could continue development. That never happened so I had to go out and get a real job like everyone else. Working on RealTemp and ThrottleStop for over 3 years has cost me a quarter of a million dollars in lost income and spent savings. I think I have sacrificed enough for the user community. If I was angry or bitter, I would not have released ThrottleStop 5.00 for the 3rd Generation Core i CPUs.