How efficient is your cruncher?

[xvi]

I think my FX rig is in the 280w range (I leave it plugged in to the kill-a-watt). Does about 4k average when I don't leave games of Civilization open for days at a time (seems low, runs 100%, but is my main rig). Very rough guess, but 14.29 ppd/watt puts it very close to the FX-6300's efficiency. I'll try to verify actual numbers soon and get wattage down as low as possible to give it more of a "true cruncher" test.

Edit: I was very, very wrong. See below.

 

[manofthem]

Sorry, I've been very swamped lately and have barely touched my pcs. However, I do intend to get you some numbers on my 3 rigs to add in. I doubt it'll be by tonight but I shall try. By the weekend sounds more like it. Yall know how family life is, crazy

 

[ThE_MaD_ShOt]

Well this month I used 2290kw of power per electric bill for the whole house. Just under $250. 13 rigs I'm happy with that.

 

[Caring1]

Well this month I used 2290kw of power per electric bill for the whole house. Just under $250. 13 rigs I'm happy with that.

That's roughly how much I use per month for the whole house, but mine adds up to nearly $600

 

[agent00skid]

agent00skid | Silverstone Strider Gold S series 850W | Athlon 860K @ 3.7GHz + R7 250| 1 SSD | 100W (4 threads) | 2813| 28.13 | Fedora 22 Alpha x86_64 |
agent00skid | 19V external and MB onboard | Celeron J1900 | 1 SSD | 15W (4 threads) | 1409 | 93.93 | Fedora 22 Alpha x86_64 |
agent00skid | 19V external and MB onboard | Athlon 5350 @ 2.1GHz | 1 HDD + 1 SSD | 31W (4 threads) | 1740 | 56.13 | Fedora 21 x86_64 |

The power measurements for the 2 19V are a bit old. And the one for the Athlon 5350 is when the HDD is spun down.

 

[xvi]

My memory was wrong. Like, not even close.

xvi | Corsair AX750 | FX 8350 @ 4.6GHz 1.5v + GTX680 | 1 SSD + 2 HDD | 335W @ 100% (8 threads) | 4101 | 12.24 | Win 8.1 Pro x64 | Not dedicated, but close

*cringe*
More results to come.

Edit: 3930k rig blows sig rig out of the water.
xvi | BFG LS-550 | Core i7 3930k @ 3.2/3.8GHz + HD 7870 | 1 HDD | 190W @ 100% (12 threads) | 9774 | 51.44 | Ubuntu x64 | Dedicated
xvi | HP 750w | 2x Xeon E5450 @ 3GHz + Matrox MGA G200e | 8 HDD | 271W @ 100% (8 threads) | 7777 | 28.70 | Debian x64 | Dedicated, HP DL180 G5

 

[krusha03]

Added the results from @xvi and @agent00skid

 

[[Ion]]

My memory was wrong. Like, not even close.

xvi | Corsair AX750 | FX 8350 @ 4.6GHz 1.5v + GTX680 | 1 SSD + 2 HDD | 335W @ 100% (8 threads) | 4101 | 12.24 | Win 8.1 Pro x64 | Not dedicated, but close

*cringe*
More results to come.

Edit: 3930k rig blows sig rig out of the water.
xvi | BFG LS-550 | Core i7 3930k @ 3.2/3.8GHz + HD 7870 | 1 HDD | 190W @ 100% (12 threads) | 9774 | 51.44 | Ubuntu x64 | Dedicated
xvi | HP 750w | 2x Xeon E5450 @ 3GHz + Matrox MGA G200e | 8 HDD | 271W @ 100% (8 threads) | 7777 | 28.70 | Debian x64 | Dedicated, HP DL180 G5

That your 3930k is so efficient is really incredible. I had suspected more. The old C2Q ones, not so much

 

[R-T-B]

This isn't quite "crunching" but I figure I'll post this here for shits and giggles:

I have a 3TH bitcoin farm that consumes a little under 2.5KW. Pretty nice as that's under a watt per Megahash, and is considered very efficient in bitcoin terms. It costs me around $200 a month to run and generates about $250.

It's time is running short though. I used to have over 10TH in hardware but the inefficient ones have already been shut down and I barely make that $50, and that's with some of the most affordable electricity in the country.

If I end up shutting it all down, I have a 240V 20A line ready to run some heavy duty crunching gear and will likely get into that to give back a little.

 

[krusha03]

This isn't quite "crunching" but I figure I'll post this here for shits and giggles:

I have a 3TH bitcoin farm that consumes a little under 2.5KW. Pretty nice as that's under a watt per Megahash, and is considered very efficient in bitcoin terms. It costs me around $200 a month to run and generates about $250.

It's time is running short though. I used to have over 10TH in hardware but the inefficient ones have already been shut down and I barely make that $50, and that's with some of the most affordable electricity in the country.

If I end up shutting it all down, I have a 240V 20A line ready to run some heavy duty crunching gear and will likely get into that to give back a little.

FYI there is a bioinc project that gathers bitcoins to donate actual money to research groups. Haven't really looked into it how it works but just letting you know that you can still put those ASICs to good use

 

[R-T-B]

FYI there is a bioinc project that gathers bitcoins to donate actual money to research groups. Haven't really looked into it how it works but just letting you know that you can still put those ASICs to good use

A lot of the inefficient ones are gone but I might consider it for the remaining ones.

 

[xvi]

That your 3930k is so efficient is really incredible. I had suspected more. The old C2Q ones, not so much

I was really surprised. The server in that post is using the lower power (aka: better binned) 3GHz quad core E5450 with an 80w TDP instead of the otherwise identical X5450 with a 120w TDP too. I figured it would be really efficient even though the clock speeds were relatively high. I think the Core 2 series is just too behind-the-times these days.

 

[[Ion]]

I was really surprised. The server in that post is using the lower power (aka: better binned) 3GHz quad core E5450 with an 80w TDP instead of the otherwise identical X5450 with a 120w TDP too. I figured it would be really efficient even though the clock speeds were relatively high. I think the Core 2 series is just too behind-the-times these days.

Yeah, I had a dual-E5420 system and it always heated up the room terrifyingly fast. Particularly given that performance wasn't anything special. The dual X5472 also puts out lot of heat too--more than I would expect for dual 95W CPUs and basically nothing else.

 

[manofthem]

I hope to be posting some numbers tonight. I'm waiting for Free-DC's final update on my main rig tonight, and I'm going to try to get the power reading on my other pcs, just have to re arrange a few things to do that

 

[manofthem]

Ok here we go... (sorry for the double post)

manofthem | Corsair AX1200 | i7 4770k @4.3 @1.28v | 1 SSD + 1 HDD | 179w @100% (8 threads) | 6024 | 33.65 | Win 8.1 Pro x64 | Crunches 24/7, except random gaming

manofthem | Seasonic 660XPS2| i7 4790 @3.8 @1.095v | 1 SSD | 93w @100% (8 threads) | 5346 | 57.48 | Win 7 Pro x64 | Crunches 24/7

manofthem | XFX 550Pro | i3 2100 @3.1 @1.11v | 3 HDDs | 94w @100% (4 threads) | 2118 | 22.53 | Win 7 Ult x64 | Crunches 24/7


I have to say that I'm impressed with the 4790 with its low power consumption. I might add a few of those eventually: replace the i3 with one and then add one more.

Had a little scare on the 4790 rig. I thought I plugged the pc in, but apparently I had plugged the Plasma HDTV in. I was getting a reading of 340w and it was freaking me out! But then I realized it was not right when I shut off the pc and it was still reading 180w lol, figured it out all by my lonesome.

 

[krusha03]

Ok here we go... (sorry for the double post)

manofthem | Corsair AX1200 | i7 4770k @4.3 @1.28v | 1 SSD + 1 HDD | 179w @100% (8 threads) | 6024 | 33.65 | Win 8.1 Pro x64 | Crunches 24/7, except random gaming

manofthem | Seasonic 660XPS2| i7 4790 @3.8 @1.095v | 1 SSD | 93w @100% (8 threads) | 5346 | 57.48 | Win 7 Pro x64 | Crunches 24/7

manofthem | XFX 550Pro | i3 2100 @3.1 @1.11v | 3 HDDs | 94w @100% (4 threads) | 2118 | 22.53 | Win 7 Ult x64 | Crunches 24/7


I have to say that I'm impressed with the 4790 with its low power consumption. I might add a few of those eventually: replace the i3 with one and then add one more.

Had a little scare on the 4790 rig. I thought I plugged the pc in, but apparently I had plugged the Plasma HDTV in. I was getting a reading of 340w and it was freaking me out! But then I realized it was not right when I shut off the pc and it was still reading 180w lol, figured it out all by my lonesome.

Added to the table. The i7s seem really efficient at crunching. I guess if you could undervolt down to @Knoxx29 xeon your PPD/watt would be about the same. Also i wonder what is the efficiency of that AX1200 being used at 15% of it's rated value

 

[R-T-B]

Added to the table. The i7s seem really efficient at crunching. I guess if you could undervolt down to @Knoxx29 xeon your PPD/watt would be about the same. Also i wonder what is the efficiency of that AX1200 being used at 15% of it's rated value

If he's using it at 15%, I believe I read reviews indicate something odd: The non-digital HXi750W (the one i use) beats it at low loads by quite a bit in efficiency figures (well, 2-3% anyways, but at low wattages most people don't care).

 

[manofthem]

Added to the table. The i7s seem really efficient at crunching. I guess if you could undervolt down to @Knoxx29 xeon your PPD/watt would be about the same. Also i wonder what is the efficiency of that AX1200 being used at 15% of it's rated value

The i7 4790 is only running at 3.8, won't hit the turbo of 4.0 and I'm guessing that's a limitation of the H97 board it's in; only a 4 pin cpu plug. All in all, it's running pretty efficiently in a rig with not a lot in it so it makes sense that it's lower power usage. The 4770k however is pulling more since it's overclocked a little with more in the pc and case. Still, not horrible.

Additionally, the AX1200, though overkill for most of this pc's usage, is more efficient than the HX1000 that it replaced. I took a glance at the power thread that Mussels made, and I posted results there with both PSUs: after I swapped in the AX, power draw dropped by ~8% I haven't really loaded this PSU up much, max maybe hit 600w, but it's all good

 

[xvi]

power draw dropped by ~8%

o.0
That's a lot more than I would have expected from just a PSU swap. Nothing else changed?

Edit: Nevermind. Looked up spec sheets. The HX1000 isn't even Bronze? -__-
HX1000 "80 PLUS Certification Level: Certified"
AX1200 "80 PLUS Certification Level: Gold"

 

[HammerON]

Extremely efficient. Now stop bothering me

 

[krusha03]

Well i decided to take my power meter and test one of the i5 desktops i am using at uni. And man was i surprised power consumption while crunching is 59.3W and when run 24/7 it makes ~4000PPD on average. That's 67.75 PPD/W And it's as (actually more) efficient as @Knoxx29 60W xeon

Extremely efficient. Now stop bothering me

No one forced you to post here but if you are claiming that show the numbers to support it

 

[Heaven7]

Heaven7 | Corsair HX1000 | 2x Xeon X5690 @ 3.6GHz 1.224v + 7970 | 3 HDD + 4 SSD | 620W @ 90% (24 threads) | 9672 | 15.60 | Win Vista x64 | Test cruncher
Heaven7 | Corsair RM750 | 2x Xeon E5620 @ 2,5GHz 1.2v + Matrox MGA-G200e | 1 SSD | 212W @ 99% (16 threads) | 3787 | 17.86 | Win7 x64 | Dedicated 24/7
Guess this sucks...

 

[twilyth]

I'd love to get some efficiency numbers on the v3 Xeons. Here are the top ppd producers from Boincstats.

Using the TDP numbers for the 2699v3 as an example, it looks like 36k ppd/190 watts or about 190ppd/watt.

If that's accurate, I should be able to retire most of my servers and get almost 4 times the output with the same wattage - assuming I can find some chips at a reasonable price.

Does this sound right or did I make some stupid math error?

edit: yup stupid math error - 36/290 not 190 gives 124ppd per watt or about 2.5 times what I get now.

 

[krusha03]

Heaven7 | Corsair HX1000 | 2x Xeon X5690 @ 3.6GHz 1.224v + 7970 | 3 HDD + 4 SSD | 620W @ 90% (24 threads) | 9672 | 15.60 | Win Vista x64 | Test cruncher
Heaven7 | Corsair RM750 | 2x Xeon E5620 @ 2,5GHz 1.2v + Matrox MGA-G200e | 1 SSD | 212W @ 99% (16 threads) | 3787 | 17.86 | Win7 x64 | Dedicated 24/7
Guess this sucks...

Added

I'd love to get some efficiency numbers on the v3 Xeons. Here are the top ppd producers from Boincstats.

Using the TDP numbers for the 2699v3 as an example, it looks like 36k ppd/190 watts or about 190ppd/watt.

If that's accurate, I should be able to retire most of my servers and get almost 4 times the output with the same wattage - assuming I can find some chips at a reasonable price.

Does this sound right or did I make some stupid math error?

edit: yup stupid math error - 36/290 not 190 gives 124ppd per watt or about 2.5 times what I get now.

Well that's the CPUs only, the motherboard, gpu, hdd and ram sticks will also consume some power. Especially dual socket server boards usually have bunch of extra's on them like raid controlers and SAS controllers that consume extra power so you need to add another 50-60W on top of that. @Knoxx29 has a dual socket 60W xeon that ends up using ~190W with 6 or 8 sticks of ram (not sure), hdd and the onboard GPU. With that being said you will probably end up at 100-110 PPD/W. But look at that E5-2666 v3. It's almost as good with half the cores/threads but can't find any info about the CPU

 

[twilyth]

The couple of servers that I've measured tend to be pretty close to the TDP number but then I normally use the minimum amount of ram, only one hard drive, minimal graphics, etc.

Maybe that 2666 is part of the 2660 line - http://ark.intel.com/products/81706/Intel-Xeon-Processor-E5-2660-v3-25M-Cache-2_60-GHz