Page 3 of 4

Posted: Sat Jan 02, 2010 11:12 am
by electrodacus
Hi MtnHermit welcome to SPCR.
The best computational power vs power consumption (for the CPU) is at about 1.8Ghz but 2Ghz is much better number :)
The problem is not the CPU it only uses less than 4W at idle the problem is the motherboard and chipsets. The chipsets are always two or more generation behind the CPU they use old manufacturing technology for the chipset.
The most energy efficient CPUs if I exclude the manufacturing process is the one with the higher number of cores without HT. Hyper Threading is not efficient.
I'm also interested in low power topics. I will probably soon move offgrid and is easier to reduce the power consumption than to produce. Now I'm working on an electric mini tractor with front end loader and a wind turbine.

Posted: Sun Jan 03, 2010 8:04 am
by MtnHermit
electrodacus wrote:Hi MtnHermit welcome to SPCR.
Thanks.
The problem is not the CPU it only uses less than 4W at idle the problem is the motherboard and chipsets. The chipsets are always two or more generation behind the CPU they use old manufacturing technology for the chipset.
Certainly true with Atom, the chipset consumes more than Atom. That's about to change with Clarkdale, only one generation behind. Early number are very encouraging. You mention MB power, have you measured and/or seen test. MB power tends to be lumped into some big power envelope and not broken out. How does one know which MB to select for lowest power?
The most energy efficient CPUs if I exclude the manufacturing process is the one with the higher number of cores without HT. Hyper Threading is not efficient.
Unless you have other test results, the HT on Atom has no power impact, see here. Perhaps you're referring to HT as it once was on the P4? BTW, you can get HT or not on various versions of Clarkdale.
I'm also interested in low power topics. I will probably soon move offgrid and is easier to reduce the power consumption than to produce.
Interesting observation and the foundation of practical off the grid living. One poster on another forum remarked that since CFL's now had dimmers, they were much more practical. I use mostly incandescents and find that CFL's and incandescents use the same power when they're both turned off. :)

Posted: Sun Jan 03, 2010 11:09 am
by mark19891989
@ MtnHermit: would you be willing to post any information about you off grid setup?

thanks mark.

Posted: Sun Jan 03, 2010 11:18 am
by MtnHermit
mark19891989 wrote:@ MtnHermit: would you be willing to post any information about you off grid setup?
What do you want to know?

Pretty standard setup, solar panels, big lead acid batteries, 4KW Trace inverter, 6KW generator that has 192 hours in 11-years, everything downstream of the inverter runs 120VAC.

Posted: Sun Jan 03, 2010 12:42 pm
by mark19891989
MtnHermit wrote:
mark19891989 wrote:@ MtnHermit: would you be willing to post any information about you off grid setup?
What do you want to know?

Pretty standard setup, solar panels, big lead acid batteries, 4KW Trace inverter, 6KW generator that has 192 hours in 11-years, everything downstream of the inverter runs 120VAC.
I wanted to know general setup costs, and how much you are able to run of off it. in real world rather then Kwh

thanks for the reply.

do you have your pc running from 12v dc or is that also powered by the inverter?

i would love to go off the grid

Posted: Sun Jan 03, 2010 6:22 pm
by MtnHermit
mark19891989 wrote:I wanted to know general setup costs, and how much you are able to run of off it. in real world rather then Kwh
Not sure I understand, but . . .
Roughly $20K installed, verses $135K for a wire, I'm roughly 4-miles from the nearest utility wires @ $35K/mi. The solar > battery > inverter system runs everything in my house: well, furnace, refrigerator, microwave, TV, DVR, this PC, wireless internet, furnace, washer, dryer, dishwasher, lights . . . everything, but limited to 3KWH/day assuming it's sunny.
do you have your pc running from 12v dc or is that also powered by the inverter?
EVERYTHING is 120V/60C AC, no DC. If I need DC I'll use a AC/DC converter such as a battery charger for a car. DC is only available at the battery box whereas AC outlets are everywhere, to code.

In 11+ years I've had one failure, last year, 2 of my 12 - 6V batteries died. I shuffled them around such that I'm now running only 8 batteries. They're very heavy ~150# and very expensive ~$300 each and you can't replace two, I'd have to replace all twelve.

Posted: Mon Jan 04, 2010 11:15 am
by mark19891989
wow, that really cool, cant believe how much it is to get the power company to run a cable 4 miles!

i would love the idea of living of the grid, i havnt seen/read about any people in the uk doing it

it must be a good feeling having the your power free after the setup costs

i dont have a clue how much power we use in this house a day as i dont pay the bills, do you find the 3kwh daily limit restricts your appliance usage?

Posted: Wed Jan 06, 2010 10:15 pm
by electrodacus
Certainly true with Atom, the chipset consumes more than Atom. That's about to change with Clarkdale, only one generation behind. Early number are very encouraging. You mention MB power, have you measured and/or seen test. MB power tends to be lumped into some big power envelope and not broken out. How does one know which MB to select for lowest power?
Not really true Clarkdale is integrates the GPU and memory controller on the same chip but there will still be more power used at idle from Memory controller and GPU than CPU for example I have a Q8400s undervolt at 0.925V and the idle power is 4W for the CPU and about 8W for the Northbridge G31 (memory controller and GPU) other 2W+ for the south bridge and the rest up to about 30W idle for my system is Network chip probably 1W+ audio memory another 4W the SSD is only less than 0.48W and a lot of loss on the DC-DC converters for CPU ,RAM and Northbridge.
Unless you have other test results, the HT on Atom has no power impact, see here. Perhaps you're referring to HT as it once was on the P4? BTW, you can get HT or not on various versions of Clarkdale.
You can not relay disable HT even if not used it still take power my argument is that a Qxxxx is more efficient than a core i7 core i5 for the same amount of work.
Interesting observation and the foundation of practical off the grid living. One poster on another forum remarked that since CFL's now had dimmers, they were much more practical. I use mostly incandescents and find that CFL's and incandescents use the same power when they're both turned off. :)
:) I will use LED and they also use the same power when turned off.
I have some land in the end probably 10 acres (4 hectares) and I will use mainly wind and self made savonius wind turbines for a total of 400kwh/month there will be some solar but wind is still cheaper. I will need a lot of electricity since all will be electric including heating and is Canada :) right now is -32C. In summer I will use the excess power for an electric tractor and irrigation. I have may plans I do not know when I will be able to finalize everything.

Posted: Tue Mar 09, 2010 1:34 pm
by Telstar
electrodacus wrote:I hope I will find a solution to eliminate the only fan in my system (it is not noisy since is only at 5V under 800rpm) but it get a lot of dust in my system I will probably be able to use a filter but I will prefer to remove this completely even if I need to change the motherboard.
If you know a motherboard with LGA775 and integrated video that will be able to run completely passive please let me know.
I dunno if you solved this, but i AM running my system completely fanless now. The undervolted nexus is gone :)
The gigabyte GA-EG45M-UD2H m-atx board is what i'm using. Please note that I'm not usually going with high cpu usage as i only play music, with some filtering and equalization. CPU is around 30-40% max and temps are in the low 50s(°C)

Posted: Sat Mar 13, 2010 10:30 am
by electrodacus
Unfortunately I was not able to remove the Fan the Northbridge G31 in my case is to hoot for my liking.
Interesting that it works for you with G45 that is 24W TDP and it will not work for me on G31 with 15.5W TDP.
Can you put your hand on the Northbridge heatsink after one hour or more in use?.

Posted: Sun Mar 14, 2010 4:35 am
by Telstar
electrodacus wrote:Can you put your hand on the Northbridge heatsink after one hour or more in use?.
I dont know but the system sensor says usually 45°, while the cpu averages 51°C with light usage (Music playback only -with equalization)

Posted: Mon Mar 15, 2010 11:30 am
by electrodacus
Telstar wrote: I dont know but the system sensor says usually 45°, while the cpu averages 51°C with light usage (Music playback only -with equalization)
If this is the case then is relay great 45°C is excellent but maybe the temperature sensor is faulty. If you have access to heatsink ca you measure with the finger :) usually if is under 60C it will be tolerable.
I may buy the same board or similar if this is the case I will love to get rid of the fan.

Rear-mount Thermal Blocks

Posted: Sat Mar 20, 2010 7:33 pm
by PlanetOfTheApes
Image
Zalman page

"Heat Transfer via Rear-mount Thermal Blocks
When the Rear-mount Thermal Blocks are installed on the back-side of the motherboard in line with the position of the FETs (Field Effect Transistor) and the Northbridge chipset, each can lower the FET temperature by 10 to 30°C and the Northbridge chipset by 5 to 10°C."


These thermal blocks are designed to be used with Zalman TNN cases. I haven't tried them in a regular case, but I thought I'd mention it here in case anyone's interested in experimenting.

Sold as a pack of 8 blocks I think, don't know if they're still available. I suspect they work best with an aluminium uni-body frame type case.

Re: Rear-mount Thermal Blocks

Posted: Sat Mar 20, 2010 10:12 pm
by ces
PlanetOfTheApes wrote: These thermal blocks are designed to be used with Zalman TNN cases. I haven't tried them in a regular case, but I thought I'd mention it here in case anyone's interested in experimenting.

Sold as a pack of 8 blocks I think, don't know if they're still available. I suspect they work best with an aluminium uni-body frame type case.
You mean those little blue tubes? Are the white ends made of something that will not short the board?

What are they made of? Where can you buy them?

Re: Rear-mount Thermal Blocks

Posted: Sun Mar 21, 2010 4:29 am
by PlanetOfTheApes
You mean those little blue tubes? Yes

Are the white ends made of something that will not short the board? sticky thermal tape/pad I think

What are they made of? Aluminium

Where can you buy them? google: Zalman ZM-RTB1

I heard installation can be tricky because the ends aren't sticky enough.

Posted: Tue Mar 23, 2010 12:42 pm
by Ksanderash
Why pay to Zalman for such low quantities of goods ;)

Image

Official.
At Amazon.

Posted: Tue Mar 23, 2010 1:15 pm
by ces
Ksanderash wrote:Why pay to Zalman for such low quantities of goods ;)
]
Surely you jest

Re: Rear-mount Thermal Blocks

Posted: Tue Mar 23, 2010 1:26 pm
by ces
PlanetOfTheApes wrote:I heard installation can be tricky because the ends aren't sticky enough.
Are the ends electrically insulated?

Re: Rear-mount Thermal Blocks

Posted: Tue Mar 23, 2010 4:00 pm
by PlanetOfTheApes
ces wrote: Are the ends electrically insulated?
I don't have any reason to suspect they're not insulated.

Note - they may not fit under your north bridge if it has a backplate.

Posted: Wed Mar 24, 2010 12:19 pm
by Ksanderash

Posted: Wed Mar 24, 2010 1:16 pm
by ces
Ksanderash wrote:
ces wrote:Surely you jest
I'm not :lol:
What exactly is under the hot-cold pack?

What is the impact of using the hot-cold pack: on temps and on sound?

Posted: Wed Mar 24, 2010 2:32 pm
by Ksanderash
ces wrote:What exactly is under the hot-cold pack?

What is the impact of using the hot-cold pack: on temps and on sound?
The gel inside the pack absorbs vibrations, thus doesn't allow moving parts to generate noise. It also has low temperature resistance coefficient, so the heat is transmitted to massive surface outside very quick and fully. Just as thermal grease does.

There is a HDD under the gel pack, if you go through link to other pictures.

Posted: Fri Mar 26, 2010 11:43 am
by electrodacus
NEW You can see photos with the 260W PSU and 220W brick LINK

Image

Posted: Sat Mar 27, 2010 7:01 am
by ces
[quote="electrodacus"]NEW You can see photos with the 260W PSU and 220W brick LINK[/quote

It would help me, and others, to better understand how this works if you could add a few close up shots showing the wiring and what is connected to what.

Would you please do that?

Posted: Sat Mar 27, 2010 9:26 am
by electrodacus
ces wrote:
electrodacus wrote:NEW You can see photos with the 260W PSU and 220W brick LINK[/quote

It would help me, and others, to better understand how this works if you could add a few close up shots showing the wiring and what is connected to what.

Would you please do that?
Yes I will do that no problem. I will add all the info to that Link.
At this time I do some tests and is not working perfect there is a chance that it will not start after a power down if I try to restart from 10 to 25 seconds after the power down if sooner or later then it will 95% of the time work if is in the 10 to 25 sec interval is a chance of 50% that it will not start when I push the power ON button.
The connexion are simple now I use a Y ATX cable to connect the PSU in parallel and a Y4 cable to connect the input 20V together this is all.
But in the end if I do not find a better solution I will conect to the second PSU only the GND and PowerON pin on the ATX cable an 20V input together then use the first PSU to supply the motherboard and videocard and second PSU to supply the CPU and HDD . I already test this and is working perfectly.

Posted: Sat Mar 27, 2010 10:44 am
by ces
electrodacus wrote: The connexion are simple now I use a Y ATX cable to connect the PSU in parallel and a Y4 cable to connect the input 20V together this is all.

But in the end if I do not find a better solution I will conect to the second PSU only the GND and PowerON pin on the ATX cable an 20V input together then use the first PSU to supply the motherboard and videocard and second PSU to supply the CPU and HDD . I already test this and is working perfectly.
I am not certain I understand. When you do the pictures. Can you show it the current way and also this second way, even if the first way doesn't end up working?

Posted: Sat Mar 27, 2010 9:21 pm
by electrodacus
ces wrote: I am not certain I understand. When you do the pictures. Can you show it the current way and also this second way, even if the first way doesn't end up working?
Just go to the same LINK and you see that I added another photo that shows how everything is connected there is also an explanation read PPPS Now everything works just fine.

Posted: Sun Mar 28, 2010 12:14 am
by ces
electrodacus wrote:Just go to the same LINK and you see that I added another photo that shows how everything is connected there is also an explanation read PPPS Now everything works just fine.
It's making a little more sense. But some more close ups at different angles would help.

One thing that clearly didn't make sense is the 24 wire y-splitter. It looks like one of the plugs has a number of wires pulled out of it. Why did you do that? Why would that help anything? It seems counter-intuitive to me.

Posted: Sun Mar 28, 2010 5:08 pm
by electrodacus
ces wrote: It's making a little more sense. But some more close ups at different angles would help.

One thing that clearly didn't make sense is the 24 wire y-splitter. It looks like one of the plugs has a number of wires pulled out of it. Why did you do that? Why would that help anything? It seems counter-intuitive to me.
Is a 20 wire Y splitter and the only wires that are in the second PSU are GND wires and Power ON the green wire the rest are removed so that the PSU will start and stop at the same time but the different voltage as 12V,5V,3.3V are not connected together. I explained also that the PSUs are starting not exactly at the same time even if they use the same wire for power on the delay is not big 10 to 20ms but this is enough to have one of the PSU that starts first to be unable to provide enough inrush current without triggering overcurrent protection. So the solution is to use one PSU for CPU and HDD and the other one for MB and video card in this way is working great.

Posted: Sun Mar 28, 2010 5:40 pm
by ces
electrodacus wrote:Is a 20 wire Y splitter and the only wires that are in the second PSU are GND wires and Power ON the green wire the rest are removed so that the PSU will start and stop at the same time but the different voltage as 12V,5V,3.3V are not connected together. I explained also that the PSUs are starting not exactly at the same time even if they use the same wire for power on the delay is not big 10 to 20ms but this is enough to have one of the PSU that starts first to be unable to provide enough inrush current without triggering overcurrent protection. So the solution is to use one PSU for CPU and HDD and the other one for MB and video card in this way is working great.
Oh that makes sense.

If that is so, then could you get by with just connecting the two boards through that 4 wire plug? The extra connections on the 20 wire must add something, or you wouldn't be doing it. What do they add?

I didn't know that winmate only has a 20 connector ATX plug. What do you do if you have a 24 connector motherboard?