UMPCPortal.com Forum Index
   Other UMPC Hardware
  Benchmark results & Battery life tests on eo.

Browsing this Thread:   4 Anonymous Users

 

 Bottom   Previous Topic   Next Topic
1234...8>
Poster Thread
Benchmark results & Battery life tests on eo.
#1

Joined: 2006/3/8
From
Posts: 680
Offline


Guys, these are the results that I got. It seems to me that these two benchmarks are not doing a good job because this device is not performing like at half a level of the Celeron at 900 MHz. In another hand, 3DMark score is really bad and I guess again is because may be this program is not prepared for using the capabilities of the VN800 and fails to detect properly the hardware.

Opinions?

Update about the Battery Life Issue:

In these few days that I have been playing around with the TabletKiosk eo, I have learned a few things that I want share.

First of all, I want to quote here a comment of Martin Smekal, CEO of TabletKiosk, talking to GottaBeMobile regarding battery life: "The pre-production unit you have does not have it’s power management features implemented, and there are on-going bios modifications that will effect the battery performance in the production unit. However, this being said take into consideration this is a full X86 architecture system running full Windows XP operating system, while utilizing a three (3) cell battery. A standard Tablet PC or Laptop will have at least a six (6) cell battery to achieve 3.5 hours of battery life."

My guess is that they released these units without those BIOS modifications because we still see the same or worse battery life that Rob saw in the preproduction unit. In another hand, I have found that either there is not any Power Management features implemented in these units or they have been very poorly implemented. And the reason why I said that is because during the tests we have found that when switching to some of the Power Management Profiles that are supposed to save Battery the CPU stepping feature goes off and the CPU stays at 1 GHz all the time when it's not supposed to do that. In another hand, the display brightness is preset at a very high level when the unit is running on battery and I have not found a way to set it at 25% and keep it in that way every time that I'm running on battery. From the point of view of Windows Applications, I have not found anything in the eo that can come close to the concept of "Power Management". All I have seen is what you find in Microsoft Operating System.

Add to all these that this unit is not entering into C3 stage (one of the power saving stages that could report up to 15% of more battery life) due to a bug in Microsoft Operating System that has not been properly handled by AMTek/TabletKiosk/VIA in this case. And I said "properly handled" because the problem still there, and it's a known problem by OEMs because Microsoft have been talking about it for a while.

So taking in consideration all the above, and waiting for a solution... here is some tips that will help you to add at least 10 minutes more of battery life to your eo:
  • I know that many of you prefer DIVX format over other formats but... Our tests show that while playing movies in DVIX format the processor worked between 600 MHz to 1 GHz almost all the time. And the CPU load was higher. Playing movies in MPEG-2 format either copying the whole DVD into your HDD or converting the DVD to a smaller MPG is the best way to go because the video chip can decode MPEG-2 and that takes out part of the CPU Load. When using MPEG-2 files the CPU speed stayed most of the time at 400 MHz and the CPU Load was lower.
  • To get more of battery life while playing an MPEG-2 movie you need a software designed to use the VN800 video chip. I tested 3 programs and the best one so far is WinDVD 7.0 (earlier version do not support the VIA chip).
  • After installing Notebook Hardware Control I started to see a better battery life. There are two reasons for that. This program overwrites the Power Saving Profiles in your UMPC and forces a real stepping in the processor if the option for "Dynamic Switching" is selected. The second reason is that this program shows in the System Tray the CPU Speed, CPU Load and Battery Life. Having these meters all the time visible let you know if anything is going wrong and that gives you the chance to correct any settings before is too late.
  • When you unplug the unit from power the display brightness is automatically reduced but not enough. The best results without compromising too much the visibility were obtained when the display brightness was set around 25%.


Just by doing these 4 tips I was able to add around 10 to 15 mins more to my initial 1:25 hours. I know that 1:35 is not what you were expecting but.... Hey... Something is better than nothing.

It's a clear fact for us right now that this unit is suffering from a well known and documented bug in the OS that makes the battery to drain faster. That bug can only be fixed by updating the BIOS and doing some changes in the USB drivers. That means that we will have to wait for TabletKiosk final fix.. Meanwhile following these 4 tips will help you a little bit.

One more last thing: We contacted early this week TabletKiosk showing them our support giving them all the information that we had at that moment and letting them know about this link where all the test results have been posted. They confirmed to me that they got all this info and that it was forwarded to their engineers. I have not been contacted since then so I assume that they still working in this case but so far they do not have anything to add to what we have found.

--
Posted by CTitanic to ULTRA MOBILE PCs TIPS at 5/11/2006 09:34:24 AM
_________________
MY UMPC BLOG
Posted on: 2006/5/8 15:27
Create PDF from Post Print
Top
Re: Benchmark results
#2
Site Administrator
Joined: 2006/3/2
From Bonn, Germany
Posts: 3727
Offline
I expect the 3Dmark to be much less than the Intel GPU.

The PCMark is an unknown. VIA processors have often required specific drivers. I wonder if you have all the right drivers installed under Windows?

http://www.viaarena.com/default.aspx?PageID=420&OSID=1&CatID=1070

Steve
_________________
Dont forget to check out the LIVE page for live chat.
Posted on: 2006/5/8 16:03
Create PDF from Post Print
Top
Re: Benchmark results
#3

Joined: 2006/3/8
From
Posts: 680
Offline
Yes, I went and downloaded these drivers and installed them and they are the same that came with the unit.
_________________
MY UMPC BLOG
Posted on: 2006/5/8 16:07
Create PDF from Post Print
Top
Re: Benchmark results
#4
Site Administrator
Joined: 2006/3/2
From Bonn, Germany
Posts: 3727
Offline
OK.

I've also contacted Van of Vans hardware who has some experience with testing C7-M for Viaarena. Lets see if he can give us pointers.

Regards
Steve.
_________________
Dont forget to check out the LIVE page for live chat.
Posted on: 2006/5/8 16:18
Create PDF from Post Print
Top
Re: Benchmark results
#5

Joined: 2006/3/8
From
Posts: 680
Offline
I oponed a thread at via

http://forums.viaarena.com/messageview.aspx?catid=31&threadid=72092&enterthread=y

To see if somebody there knows anything that could help us to gain some idea of what is going on.
_________________
MY UMPC BLOG
Posted on: 2006/5/8 16:23
Create PDF from Post Print
Top
Re: Benchmark results
#6

Joined: 2006/3/8
From
Posts: 680
Offline
I tried yesterday to run other tests but without any luck.

Quake demo does not run under XP. At least I could not make it to work in my Laptop so I did not tried in the eo. The idea was to run it in a batch file to run the demo and get how many FPS it does. I did not try Quake II because in that one you need a keyboard to launch the Console and them run the demo showing how many FPS. I do not have a USB keyboard at this moment so I thought that installing it is going to be a waste of time. And I really short of time today and tomorrow.
_________________
MY UMPC BLOG
Posted on: 2006/5/8 16:27
Create PDF from Post Print
Top
Re: Benchmark results
#7
Site Administrator
Joined: 2006/3/2
From Bonn, Germany
Posts: 3727
Offline
Hopefully we'll get some info soon. I can't believe 1:25hrs is the most you're going to get.

Steve.
_________________
Dont forget to check out the LIVE page for live chat.
Posted on: 2006/5/8 17:04
Create PDF from Post Print
Top
  •  CrashDavis
      CrashDavis
Re: Benchmark results
#8
UMPC Curious
Joined: 2006/4/25
From
Posts: 4
Offline
Below is a link from the handtops forum talking about the via chipset and it's performance versus a celeron. It doesn't paint a bright picture. I had assumed the Via was closer to the Celeron than these benchmark results demonstrate. I still don't think I'll go with the q1 though. It is only running 500mb which is slow on an XP system and the power consumption of the celeron will be much greater than the Via chipset. If I could somehow get the EO I ordered with a celeron chip inside, it would be exactly the system that I am looking for.

http://www.handtops.com/forum/2073/0//Via_C7M___Slower_per_clock_than_Crusoe.html#
Posted on: 2006/5/8 21:02
Create PDF from Post Print
Top
Re: Benchmark results
#9
Site Administrator
Joined: 2006/3/2
From Bonn, Germany
Posts: 3727
Offline
Hi Crash Davis.

Good link there. I hadnt read that until now.

Yup, its true. The VIA is going to underperform the Celeron in processing power tests. VIA dont really try to hide the fact either. They just concentrate on power utilisation in their promotional material.

I've had some information from Van Smith now. I hope Ctitanic is reading:

Quote:
For both performance and battery life
issues, CTitanic's system is best evaluated with OpenSourceMark that can be
downloaded here:

http://prdownloads.sourceforge.net/opensourcemark/OSMarkV1b8b.zip?download

There are bugs in that release.  I will try to push out a new version before
the end of the month.

Abnormally short battery life could be due to the following issues:

1. P-States are not enabled.
2. C3/C4 is not being entered.

"P-States" (performance states) dynamically change the CPU's clock speed and
voltage to suit the current workload.  There is a utility in OpenSourceMark
that will show if P-States are working by charting the current CPU frequency
that the C7 itself reports.  

If P-States" are not working, it might still be possible to turn them on by
changing the power profile to "Portable/Laptop":

* right-click on desktop
* choose "Properties"
* click "Screen Saver" tab
* click "Power..." button
* change "Power schemes" to "Portable/Laptop"

If the CPU clock speed still remains pegged at max then the BIOS might not be
set up properly.

For C3/C4, open PerfMon to see if C3 is being entered:

* type <Windows key>+<r> to open the run dialog.
* open "perfmon"
* remove all counters by clicking on them in the table and click <delete>.
* right-click on chart, choose "Processor" for performance object, and then
add "% C3 Time".

If the graph stays at 0, then a vital power saving measure is not working.  
This could be due to a USB device.  Disconnect all USB devices.  If C3 is
still at 0, then the BIOS might not be set up properly.

Although I was involved in the initial development of PCMark and even though a
few of its tests are a result of my input, OpenSourceMark is a much more
balanced benchmark.

I strongly suggest that CTitanic download OpenSourceMark and obtain "Official
Run" results on both the Celeron-M and the C7-M.  To check for P-States,
he'll need to click on the "CpuSpeed" button.


Regards
Steve.
_________________
Dont forget to check out the LIVE page for live chat.
Posted on: 2006/5/8 21:51
Create PDF from Post Print
Top
Re: Benchmark results
#10

Joined: 2006/3/8
From
Posts: 680
Offline
I have that benchmark and I'll check what he is saying. I have not run it fully because it takes a long time to pass the 3 passes. So, basically for what he says, it seems that it's a bad idea to play the movie from the USB memory stick?
_________________
MY UMPC BLOG
Posted on: 2006/5/8 22:03
Create PDF from Post Print
Top
Re: Benchmark results
#11

Joined: 2006/3/8
From
Posts: 680
Offline
The stepping has been working from day one because that was one of the thing that I checked. To be sure t cheeked again and the processor goes from 400 to 1000 when it's needed.
_________________
MY UMPC BLOG
Posted on: 2006/5/8 22:08
Create PDF from Post Print
Top
Re: Benchmark results
#12

Joined: 2006/3/8
From
Posts: 680
Offline
_________________
MY UMPC BLOG
Posted on: 2006/5/8 22:14
Create PDF from Post Print
Top
Re: Benchmark results
#13

Joined: 2006/3/8
From
Posts: 680
Offline
Steve, I can't find the second part of his message. I cant find the PerfMon.
_________________
MY UMPC BLOG
Posted on: 2006/5/8 22:25
Create PDF from Post Print
Top
Re: Benchmark results
#14
Site Administrator
Joined: 2006/3/2
From Bonn, Germany
Posts: 3727
Offline
Perfon is a standard windows XP (possibly pro) tool.

Maybe it's not available in XP home.
I only have pro here and perfom is available.

Steve.
_________________
Dont forget to check out the LIVE page for live chat.
Posted on: 2006/5/8 22:32
Create PDF from Post Print
Top
Re: Benchmark results
#15

Joined: 2006/3/8
From
Posts: 680
Offline
Got it. I thought that was something else. I'll check that later, I'm running the whole long benchmark now.
_________________
MY UMPC BLOG
Posted on: 2006/5/8 22:51
Create PDF from Post Print
Top
 Top   Previous Topic   Next Topic
1234...8>

 


Welcome New Members
lailai031 2010/9/3
hanegaton 2010/9/3
oruytqeqw 2010/9/3
robertes91 2010/9/3
Ouisri 2010/9/3
tywonchek 2010/9/3
drayianh 2010/9/2
eflame 2010/9/2
runakotan 2010/9/2
ericmoses2 2010/9/2


Privacy Policy | Faq | Contact


A Carrypad website  |  Original Design by SevenDays Design additions by UMPCPortal.com