How efficient is your cruncher?

[Norton]

Hi guys Joined in order to add info in this thread:

OldChap| Seasonic 1000p | 2*xeon e5 2650 @2.4(turbo)| 1 SSD | 240W @ 100% | 19600 on VINA | 81.89 | Linux Mint 17 | Watercooled (d5) 4 fans 24/7 crunching
OldChap| Seasonic 1000p | 2*xeon e5 2650 @2.4(turbo)| 1 SSD | 240W @ 100% | 13813 on non-VINA | 57.55 | Linux Mint 17 | Watercooled (d5) 4 fans 24/7 crunching
OldChap| Seasonic ss330 bronze| celeron j1900 @2.4(turbo)| 1 SSD | 17.5W @ 100% | 1809 on VINA | 103.37| Linux Mint 17 | passive cooled 24/7 crunching
OldChap| Seasonic ss330 bronze| celeron j1900 @2.4(turbo)| 1 SSD | 17.4W @ 100% | 1330 on non-VINA | 76.43| Linux Mint 17 | passive cooled 24/7 crunching

Points of note: the bigger rig is about 95% efficient in that the power from the socket is only 5% higher at 252.7w
The same cannot be said of the bronze psu trying to run at 5% load. This is closer to 65% efficient.
Close attention must therefore be paid to psu selection or multiple boards could be run from one psu or maybe one of these j1900 rigs could be piggybacked onto an existing rig.

Welcome and thanks for your contribution to the thread!

 

[OldChap]

Thanks for the welcome

Yeah well it was brought to my attention on WCG forums https://secure.worldcommunitygrid.org/forums/wcg/viewthread_thread,38054_lastpage,yes#494689 and I do like a good/useful resource such as this.

I have a mind to try to replicate this on my home forum. If successful, I propose we consider merging the info for the greater good.

Just a thought

 

[Heaven7]

Welcome @OldChap !

 

[OldChap]

OK, after an edit or two here and a bit more thought about the aims of this I am going with sharing, that being easier than tables on XS. Just now waiting to see if folks will add their info there....

https://docs.google.com/spreadsheets/d/1ll78DE12LBlZkP-sy-9h3kdO_XQD_9Xazh6Bvo2xlII/edit?usp=sharing

You may have noted that I consider the difference between output of normal and VINA projects to be of note, now, after some thought, I think it might be some useful if users look at the difference between Watts and the Volts * Amps readings shown on their meters. This might give an insight into PSU efficiency.

Anyway, that is the direction I would like to take our chart so feel free to watch and see how it goes (if at all)

Happy crunching

OC

 

[CAPSLOCKSTUCK]

VINA

?

 

[OldChap]

Some Vina details can be found at http://www.worldcommunitygrid.org/help/viewSearch.do?searchString=Vina+Projects but the bottom line is that there are currently only two and the FAHV wu's from the FA@h project are all now just about done.

This just leaves work from the Outsmart Ebola Together project.

Why is this significant? Well, if you run Linux on your rig then the Vina wu's run much more efficiently and therefore pay more points, so much so that I feel there should be a separate column to record this. The two different figures I have shown for my rigs should demonstrate this clearly. The mixed column is for anyone who just takes what they are given and doesn't chase one project.

Hope this helps

 

[OldChap]

Gentlemen, please forgive what may seem to be (and possibly is) both an idea and thread hijack but I wanted to bring my post at WCG to your attention.

https://secure.worldcommunitygrid.org/forums/wcg/viewthread_thread,38126

Hopefully this will add even more info on the subject. I am thinking that it should be possible in the fullness of time to copy/paste info from this into your own spreadsheets in order to sort according to the criteria of your choice

OC

 

[OneMoar]

mine is as follows
idle/ 119w
crunching cpu only 223w
folding /w gpu 340W
standy-by/off 15W
as for points ... non existent because I suck lol
temps are 50/60c when crunching ambient increase in temps from machine is about +15 degrees f with both window fans running so thats >90F with 80 degree outside temps O_O

 

[xvi]

Bump! This is a reminder for all the "Oh, I'll get some numbers soon" posts!

Ninja Edit: I love statistical data like this. The reason why can be explained with a Calvin and Hobbes comic.

 

[twilyth]

I still can't add the 2x14=28c rig because the points have been all over the map. One day it was as high as 58k but the next 29k and a couple of days later 20k. You can see the rig's history here

http://boincstats.com/en/stats/15/host/detail/3357630/lastDays

If it ever stabilizes I'll come back and post. I can say that power consumption seems to be steady at around 380watts under full load.

 

[Norton]

I still can't add the 2x14=28c rig because the points have been all over the map. One day it was as high as 58k but the next 29k and a couple of days later 20k. You can see the rig's history here

http://boincstats.com/en/stats/15/host/detail/3357630/lastDays

If it ever stabilizes I'll come back and post. I can say that power consumption seems to be steady at around 380watts under full load.

Probably best if you go by this:
http://stats.free-dc.org/stats.php?page=host&proj=bwcg&hostid=3357630

Look at your 7 day avg vs RAC, or 28 day rolling average- when these are close to equal then your ppd is reasonably stable. Check these after you've been running for at least another week or two.

 

[xvi]

Look at your 7 day avg vs RAC, or 28 day rolling average- when these are close to equal then your ppd is reasonably stable. Check these after you've been running for at least another week or two.

The ARM chip has been running untouched, non-stop for a long time, but I haven't been able to get RAC and 7-day matched. RAC is actually higher, iirc.
Just mentioning an oddity of mine.

 

[twilyth]

Probably best if you go by this:
http://stats.free-dc.org/stats.php?page=host&proj=bwcg&hostid=3357630

Look at your 7 day avg vs RAC, or 28 day rolling average- when these are close to equal then your ppd is reasonably stable. Check these after you've been running for at least another week or two.

I just switched to all UGM so we'll see what happens. If my points spike, you'll know why.

 

[t_ski]

I still can't add the 2x14=28c rig because the points have been all over the map. One day it was as high as 58k but the next 29k and a couple of days later 20k. You can see the rig's history here

http://boincstats.com/en/stats/15/host/detail/3357630/lastDays

If it ever stabilizes I'll come back and post. I can say that power consumption seems to be steady at around 380watts under full load.

Are you running CEP2? I found on my 2P that I would get a bunch of CEP2, then they would take 12+ hours to complete. My PPD would go down during this, and spike back up when I was running other projects. I lowered the amount of WUs I could take (down to 6) and the swings seem like they are gone.

 

[twilyth]

No, I was only getting one per but I changed that to 3 to help get the diamond badge sooner.

I opted out of OET but I still seem to have 128 pages of them so I'm going to have to clean out some machines

 

[t_ski]

When I was trying to get only OET, the work seemed to run out, clearing my buffer and forcing me to get other work. Then the numbers would swing down and I'd start getting OET later. My numbers would spike then even out, but I'd run out of OET again and the cycle would repeat.

 

[twilyth]

Did you clear the default option that says you'll take work from other projects if work for your chosen one runs out? I had that checked for the longest time and kept getting work from unchecked projects.

 

[t_ski]

Yes, I did have it on because I didn't want it to run out of work completely. The downside is the swings in PPD.

 

[xvi]

@krusha03 Would you mind correcting 4x4n's entry? Looks like CPU and GPU were separated in to separate columns.

 

[[Ion]]

[Ion] | Supermicro 1400w Gold | 4x Opteron 6376| 2x SSD | 460w @ 100% | 38500 | 83.7 | Linux Mint 17.2 | Dedicated WCG

 

[krusha03]

@krusha03 Would you mind correcting 4x4n's entry? Looks like CPU and GPU were separated in to separate columns.

Thanks for reminding me about this thread. I have been busy lately and I have completely forgot to monitor it. Updated with all entries and fixed 4x4n's entry

 

[Nordic]

james888 | Cougar S850 Silver rated | 2500k @ 4GHz 1.2v| 1 HDD | 121W @ 100% (4 threads) | 5,934 | 49 | Linux Kernel 3.19.3-k17 | Dedicated Cruncher

james888| Cougar S750 silver rated| Celeron g1620 @ 2.7GHz 0.8v | 1 HDD | 38W @ 100% (2 threads) | 1,978| 52.1 | Linux Kernel 3.19.3-k17 | Dedicated Cruncher

Been meaning to test this. I think I have some really efficient cruncher, but have not checked yet. I am hoping I am near the top. I need to trade those big PSU's for something smaller and more efficient. I would like to find some 350w-600w gold or better psu's.

The g1620 celeron could gain the most from a smaller psu. The motherboard is a feature rich p67 board which adds an additional watts compared to a more basic motherboard.

 

[krusha03]

james888 | Cougar S850 Silver rated | 2500k @ 4GHz 1.2v| 1 HDD | 121W @ 100% (4 threads) | 5,934 | 49 | Linux Kernel 3.19.3-k17 | Dedicated Cruncher

james888| Cougar S750 silver rated| Celeron g1620 @ 2.7GHz 0.8v | 1 HDD | 38W @ 100% (2 threads) | 1,978| 52.1 | Linux Kernel 3.19.3-k17 | Dedicated Cruncher

Been meaning to test this. I think I have some really efficient cruncher, but have not checked yet. I am hoping I am near the top. I need to trade those big PSU's for something smaller and more efficient. I would like to find some 350w-600w gold or better psu's.

The g1620 celeron could gain the most from a smaller psu. The motherboard is a feature rich p67 board which adds an additional watts compared to a more basic motherboard.

Added

 

[[Ion]]

[Ion] | Dell 1000w 80+ | 2x Xeon E5540| 1x 3.5" HDD | 345W @ 100% | 5900 | 17.1 | 2008 R2 Enterprise | Nearly dedicated WCG

 

[OneMoar]

ARM chips are the way of the future for crunching/folding
they offer lower power consumption,they scale better than x86,and the licensing its more liberal than that hot mess with AMD/Intel
and they are cheaper a 4 or 8 core ARM chip on your board of choice is less than 1/4 the cost of a desktop