Dell Vostro 1310 questions. & RMClock advice?

Hello, I am new to forum. I bought Dell Vostro T8100,N8400M, 4GB, 250GB. And after month I have few questions.

- Is here a difference between T8100 and T8300 in the matter of VID change <0.950V? I can't set lower.

- The production of fan noise and heat became higher ( after upgrading to A13 BIOS). Are there specific fan RPM settings and tresholds in BIOS?

- Can Dell 1505 mini n card be used as soft Accesspoint?

- Can be LCD brightness even lowered?

- Has anybody tested power consumption with no HDD?

- Has anybody know approx power consumption of the components (LCD, backlight, HDD, system, CPU, fan...)

-Isn't the default APM value dangerous(too many unload load cycles)?

-My BIOS update failed, so technician replaced motherboard. I think he forgot to apply thermal pasta(e) on CPU and GPU. the temperature are higher. Can Copper mod be applied onto Vostro 1310? I looked briefly on the gpu and i can' see any obstacle.

-Are there differences in Win XP, Win7 and Linux( i would install kubuntu) in the matter of power saving?


-NOW THE MAIN QUESTION. ABOUT OPTIMAL SETTINGS IN RMCLOCK @ C2D T8100.

Is here a complete optimal parametrs set ? Can anybody give me hint (i read flipfire's posts and I think it is the right man with deep knowledge)


So: I will write my settings, you are welcome to correct me:
Settings tab- nothing to set
management:multi step, os load mgmt (because it is "accurate according to taskmgr), HLT off

Profiles: I use 10x100 (superflm) and 10x200 only ( according to http://forum.notebookreview.com/showthread.php?t=276661 , http://mjg59.livejournal.com/88608.html ) ,all 0.950V. What is the right Perf on demend cpu usgage level treshold?


ADV info:- Help me with this section,.
TM1,TM2,Sync,ExtThrott ,C1E,C2E,C3E,C4E, Hard C4E, IDA, DFFS, Mobile, starup.
Chipset - default,
throttle- off


THANKS FOR ANSWERS, Martin

 

Wrong forum . . .

 

The lowest VID you can set is determined by intel. Most 45nm Penryn can only be set to 0.950V, but not lower. The 65nm Meroms can be set to 0.900V in most cases.

Overriding with
[HKEY_CURRENT_USER\Software\RightMark\RMClock]
"UnlockVid"=dword:00000001
has no effect.

Fan speed, thresholds and control is nothing Dell gives to the end user. For good reasons. I doubt the Vostro 1310 will offer that in the BIOS settings.

Don't know, but replacing mini PCIe cards is easy. I have read somewhere in these forums that certain vendor and device ids are blacklisted in Dell BIOS and therefore these cards don't work. However this is no reliable information. It was just written in this forum.

M1330 has a pretty broad brightness range. And reducing the "maximum brightness level" in the BIOS I can even reach lower brightness levels with the "Fn" keys. LED displays consume less power than CCFL displays.

No but you can detach the HDD and boot a linux live distro.

difficult to measure. But as a general rule of thumb:


http://software.intel.com/en-us/art...with-windows-vista-on-intel-laptop-platforms/

Don't know what the default is, but you can set yourself. E.g. but there exist simpler tools for windows:
http://www.thinkwiki.org/wiki/How_to_make_use_of_Harddisk_Power_Management_features

Why not simply apply some thermal paste and leave it that way. Since you are looking for ways to reduce the power consumption to a very low level anyway, you shouldn't face any heat issues.

Yeah, many, many differences. Out of the box ubuntu most likely consumes more power than vista/xp. However lots of tweaking can be done powertop being a useful tool:
http://www.lesswatts.org/projects/powertop/
and some guides (sorry two are german):
http://www.thinkwiki.org/wiki/How_to_reduce_power_consumption
http://thinkpad-wiki.org/Vista_Stromverbrauch_senken
http://thinkpad-wiki.org/Linux_Stromsparen

A properly configured Vista system is supposed to reduce less power since vista has a more fine grained power management regarding the different system components. However Vista also has a big footprint of services running, which often result in a higher power consumption (see intel slides above).

You propably don't want to hear but optimal is a relativ term. If I want to get away with the least amount of heat I have to limit VID and FID at the expense of performance. If I want the least amount of heat with the maximum performance I end up with different settings.

The T8100 is a particular difficult processor in that respect: It uses half multipliers. It's maximum multiplier is 10.5x to reach 2.1GHz. Unfortunately RMClock 2.35 can't set half multipliers. The walkaround is to unable the IDA mode. The mode itself is pretty much useless (no longer true for Core i7/Nehalem architectures) e.g. see anand's observations:
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3195&p=3

If you enable the IDA mode the CPU under load will be be in 10.5x = 2.1GHz mode most of the time. The VID that is applied is the IDA mode one. And since this VID also has to accomodate for the "true" IDA overclock mode of I think 11.5x = 2.3 GHz, this VID needs to be higher than it needs to be for 2.1 GHz operation. What makes this even more difficult: I'm not aware of a way of running the CPU isolated in IDA mode, which makes it difficult to find the lowest possible IDA mode VID. You probably add some headroom for safety reasons.

The choice you have to make: Use 10.5x = 2.1 GHz, but with a rather high VID. Or use just 10x = 2 GHz with a lower VID.

Sorry, link will be german. But soon a final version of NHC will be available. Beta is already available:
http://www.notebook-treff.de/board/...control/18-aktuelle-nhc-beta-2009-oder-neuer/
This tool is supposed to have "true" half multiplier support and operate better with the T8100.

Since you link to a posting from me which hasn't been updated for quite some time:

Find the highest FID that runs with the lowest VID. Let's assume that is 9x normal mode 0.95V = 1.8 GHz. This should be your lowest CPU clock. CPU will sleep longer after finishing tasks. In that example you probably end up with two or three modes:

. 9x, 10x
. 9x, 10x, 10.5x (IDA)

I don't recommend fiddling with the advanced settings in RMClock. Some of the reasons I have explained in the bugs section of my posting you mentioned.

 

Thanks for reply, I will react.

1) T8100 vs T8300 - "if you choose multiplier 8 or 9" you can use lower voltage" (0.9 or 0.85) i forgot the post it was written in

2) My Vostro BIOS is very minimalistic - no fan nor LCD brightness control
I have only time; peripherals (usb,finger, express, bt), bootlogo, fast boot, enable speedstep; boot order ;and finally load default/save/exit

3) soft Acess point- I think it is problem of Dell Wifi control software. My second Asus has that option(STAtion mode (adhoc,infrastruc.) OR Soft AP)

4) ^2- BIOS - i googled something about linux "delllcdbrightness". It uses something from SMBIOS - i do not understand that much.

5) HDD - I tried "hdparm -Y hda" (windows). But due to windows' background processes & services the hdd spinned up every minute so I had not chance to "approximate"

6) there are "complaints" and missatisfacion about Dell's heat management system. I wasn't sure about dissasembly of fan system, I had looked to service manual and surprised there is no need do detach lower part of chasis (the part that includes HDD and memory access screws)

7) I use on all states 0.950V (except IDA- It is not clear). IDA is disabled, i run at 2Ghz.

**NEW***

8) Do you know any program to manage fan RPM( better numerically than I8kfangui's no/slow/fast). I tried I8KFanGui, it is particulary 'incompatibile'. It led to no automatical fan control after close and in "auto control" to 0.5s freezes every 2seconds.

9) RMCLock-I use only Superflm 10x and normal 10x. Performance on demand- which percent treshold is ideal to save energy?

if i summarize the options: can you say which is best to save most energy in idle and have allowed full speed potential of the cpu
Remeber flipfire's thesis " Use SuperFLM only for idling"
- use all p-states with superflm x6
- use all p-states with superflm x10
- use superflm x6 and normal x10 only
- use superflm x10 and normal x10 only
- use normal x10 only -- I think it is not so effective(but i could be partialy(!) according to flipfire), because the half fsb energy saving is higher than C1,c2,c3,C4 saving.

 

Each CPU is individually marked for a minimum voltage. You will find some T8100 and T8300 that go lower as 0.95 Volt. I do know one user which can go down to 0.925V with a T8300. But most 45nm Penryns really go only as low as 0.95V regardless of the multiplier you choose in software.

It's not a common feature to use WiFi cards as access points. Maintaining such a feature costs money - especially regression testing. Just as an example. Linux W-Lan drivers for intel WiFi boards have that feature disables as well:
http://git.kernel.org/?p=linux/kern...it;h=33fd8195529d74c0fe23cddd1c76fe7e03bbd324
It think it's broken on those cards. In general I think: If you need to use your WiFi card as an access point, then linux will provide you with much more stable operation and documetation which card has support for it.

I don't think it's a good idea to spin down hard drives in windows.

Tell me one dell notebook, where people don't complain about the heat management. On the M1330 we saw like 5-6 changes to the thermal management during BIOS updates: If the fan kicked in earlier, then people complained the notebook is too loud. If it kicked in later, others or the same complained temps being too high.

There are some things you can do: Undervolting, notebook cooler, lift notebook from desk ...

Controlling the fans with tools such as i8kfanui is not advised. Besides as you mentioned: it doesn't really work anymore. The code of i8kfanui is reverse engineering from dell diagnostics tools. Some SMBIOS functions similar to those of "delllcdbrightness" you found. But those changed over time and those functions derived from inspiron 8000 lines don't work reliable anymore.

I don't think it's a good idea for the even advanced user to manually override the thermal management - you can easily break your system which raises costs for dell. Especially with a GPU known to break fast (8400M in Dell Vostro 1310):
http://en.community.dell.com/blogs/...ment-to-all-affected-customers-worldwide.aspx
But if you want to do that, somebody would have to go ahead and disassemble more recent Dell diagnosics code (again).

I don't know the best threshold. There are different advises out there depending whether you are on AC or DC. Default is fine..

Then just use one (!) p-state namely the highest: 10x = 2.0GHz @0.95V. This will save you the most energy.

The reason why I suggest using just one p-state in your case (only applies to your CPU):
. switching does cost time and consumes energy without doing anything
. look at intels formula again: power is proportional to clock if VCore is constant. If you work at only half the clock speed you need to work twice as long. Formula says: Same energy. However formula doesn't reflect sleep states. And the faster CPU wins, since it can stay much longer in low power sleep states.

Neither does this make sense nor do I find a source for this. I can only see that he said that:

http://forum.notebookreview.com/showthread.php?t=235824

This is something completely different. I'll get to that soon:

Okay, here's the catch for SuperLFM: Minimum multiplier of intel CPUs is 6x. At normal clock this is 1.2 GHz. Now for 1.2 GHz operation intel doesn't want to guarantee that all their CPUs run at 0.95V or less. So they program their Core2Duo with voltages such as 1.000V at 1.2GHz.

In order to go down with the voltages even more, they require a lower multiplier. Instead of using a lower multiplier, they have chosen to cut down the FSB speed in half. Which gives 600 MHz. This can safely be performed at 0.95V. Windows by default uses 8x = 800 MHz as the lowest p-state.

Operating at low clock speeds is othorgonal to C1,C2, C3, C4 sleep states. That means: Either your CPU is working at low clock speed OR it is sleeping. If your CPU is idling, windows immediately puts the CPU in one of the sleep states. That's the reason why flipfires statement "This ingenious feature lets your processor idle at an even lower clock speed" doesn't make sense. The CPU isn't idling when it's performing work at low clock speeds.

It's definitely better to use higher clock speeds if you don't raise the VID to allow the CPU to stay longer in low power sleep states c3, c4.

Windows performance monitoring allows you to display the battery drain current when on DC. Doesn't work on AC. Display that information and play e.g. a mp3 or DVD simulating constant light work (it's otherwise very difficult to play back a scenario with light constant workload such that your battery drain observations are reliable).

Now switch with RMClock between two modes:
. SuperLFM 12x = 1.2GHz
. Normal 6x = 1.2 GHz
Unfortunately you don't have a 12x multiplier and can't do that. The result is: You won't see a different in power consumption. SuperLFM doesn't reduce power to a measurable amount just by cutting down FSB speed to half.

Now a second test, which you can peform:
. SuperLFM 10x = 1 GHz
. Normal 10x = 2 GHz
Again playback DVD and watch. You will recognize that at 2GHz the power consumption is lower.

Therefore it doesn't make sense to use any lower FID than 10x on your machine.

Since you didn't do a lot of reading, it's best you try these yourself. This will give you more accurate answers than trusting what some people wrote that you don't even know and didn't provide backup data.

 

5) hdaparm - It was just experiment- not for real use.

8) fan RPM - my case: there is 120 seconds repeating interval (in idle or slight work). : 86 sec - Fan OFF, 34 sec fan running.
My idea was to make the fan rotating at lower RPM (quit), but for longer time.

7) P-State config: I will experiment. After disabling intermediate states and using only 10x superflm and highest i noticed lower consumption. I will continue more.

Quick test: (playing music in FlStudio)(values not exact!)(C3,C2 , C1= Cx Time% in perfmon)
SuperFLM10 Load 30%, C3: 50%,C2: 15%,C1: 0%
SuperFLM6 Load 45%, C3: 30% ,C2: 30%,C1: 3%
Norm10 Load 16%, C3 Time: 85%,C2: 0-2%, C1:0%
it means that in max perf(2Ghz)= More time in powersaving mode= more energy saved. Am I right?

NEW:
What about throttling? One notebook 'tester'(or fan) says it is the past for today processors and was useful for Pentium4M.