ANTMINER S7 is available at bitmaintech.com with 4.86TH/s, 0.25J/GH

[VirosaGITS]

Has anyone received B8 and measured the power consumption at the wall yet? 1300w + 10% seem a bit on the high side, completely capping a 15a120v breaker and leaving no headroom for OC or anything.

My B8 pulled 1362w from the wall with a platinum PSU running on 240v at 700MHz. Would be closer to 1400w if running on 120v instead due to PSU efficiency loss on 120v.

Yeah, the 240v do get 2%~ more efficiency. Thanks for the number, but i'm not sure i want to buy a 400$ ATX PSU to get 94% on 120v.

Maybe i could find some quiet 1600w sever PSU that is quiet?

The two 1100w PSU i have would not work very well here.

you can get the 2000 watt plat

links
coming up

https://bitcointalk.org/index.php?topic=1284563.0

https://bitcointalk.org/index.php?topic=1257327.msg13383917#msg13383917

https://bitcointalk.org/index.php?topic=738527.0

http://holybitcoin.com/product/dps-2000bb-breakout-adapter-board/

Right, i would like to go with this. However i only have 120v, which is what i meant, when i vaguely mentioned that the S7 would max a 120v15a breaker. Sorry for being unclear.

So i'd need a 1600w~ 120v PSU to have enough efficiency to run 1300w DC under 1440w, 1400w would be prefered, not sure why but i went through half a dozen surge protector (Belkin, APC) and all the one i ran at 1440 or higher melted or such.

My only alternative is having a big cord extention, about 15 meters, bringing the 240v to my living room window, plugging a rated 40A PDU or whatever and then plugging some 2000 as recommended or 2880 PSU that Finksy. Dunno how doable that all is, i never used 240v.

yeah if you go atx you need the expensive evga 1600 models  or a leadex 1600 model all cost good money. 300 usd plus

could you try 2 1000 watt evga's

http://www.newegg.com/Product/Product.aspx?Item=N82E16817438044   this is about 225 for 2

I found a good setup  use this rebate :


http://images10.newegg.com/uploadfilesfornewegg/rebate/SH/EVGA3MIRsDec16Dec3115cd12us1.pdf

buy this 1000 watts http://www.newegg.com/Product/Product.aspx?Item=N82E16817438062

buy this 750 watts  http://www.newegg.com/Product/Product.aspx?Item=N82E16817438060

both get a 30 dollar rebate so the

 750 watts = 95- 30 or 65 usd
1000 watts - 135-30 or 105 usd   total = 170 usd

use the 750 to run the controller and 1 board
use the 1000 to run the other 2 boards.

use 2 power strips to the same 20 amp line 120 volt line  or 2 power strips to 2 15 amp 120 volt lines.

Have all power off plug in everything    then turn the 750 watt on first and the  1000 watt on second.

do a software configuration like freq ? not sure what batch so freq ?        and do the fan at 61%

check hashing    good to go.

gold psu's giving a max of 1750 watts for 170 usd after rebates  that is really fucking good.

I ordered a GS 1050 yesterday for the same price as the 1000w. I'm in Canada so price might be a bit different.

Anyhow, those rebate normally take 2 months to get received. Is there a way to request them online now or you still need to print the form and mail it? I don't have a printer.

[philipma1957]

Has anyone received B8 and measured the power consumption at the wall yet? 1300w + 10% seem a bit on the high side, completely capping a 15a120v breaker and leaving no headroom for OC or anything.

My B8 pulled 1362w from the wall with a platinum PSU running on 240v at 700MHz. Would be closer to 1400w if running on 120v instead due to PSU efficiency loss on 120v.

Yeah, the 240v do get 2%~ more efficiency. Thanks for the number, but i'm not sure i want to buy a 400$ ATX PSU to get 94% on 120v.

Maybe i could find some quiet 1600w sever PSU that is quiet?

The two 1100w PSU i have would not work very well here.

you can get the 2000 watt plat

links
coming up

https://bitcointalk.org/index.php?topic=1284563.0

https://bitcointalk.org/index.php?topic=1257327.msg13383917#msg13383917

https://bitcointalk.org/index.php?topic=738527.0

http://holybitcoin.com/product/dps-2000bb-breakout-adapter-board/

Right, i would like to go with this. However i only have 120v, which is what i meant, when i vaguely mentioned that the S7 would max a 120v15a breaker. Sorry for being unclear.

So i'd need a 1600w~ 120v PSU to have enough efficiency to run 1300w DC under 1440w, 1400w would be prefered, not sure why but i went through half a dozen surge protector (Belkin, APC) and all the one i ran at 1440 or higher melted or such.

My only alternative is having a big cord extention, about 15 meters, bringing the 240v to my living room window, plugging a rated 40A PDU or whatever and then plugging some 2000 as recommended or 2880 PSU that Finksy. Dunno how doable that all is, i never used 240v.

yeah if you go atx you need the expensive evga 1600 models  or a leadex 1600 model all cost good money. 300 usd plus

could you try 2 1000 watt evga's

http://www.newegg.com/Product/Product.aspx?Item=N82E16817438044   this is about 225 for 2

I found a good setup  use this rebate :


http://images10.newegg.com/uploadfilesfornewegg/rebate/SH/EVGA3MIRsDec16Dec3115cd12us1.pdf

buy this 1000 watts http://www.newegg.com/Product/Product.aspx?Item=N82E16817438062

buy this 750 watts  http://www.newegg.com/Product/Product.aspx?Item=N82E16817438060

both get a 30 dollar rebate so the

 750 watts = 95- 30 or 65 usd
1000 watts - 135-30 or 105 usd   total = 170 usd

use the 750 to run the controller and 1 board
use the 1000 to run the other 2 boards.

use 2 power strips to the same 20 amp line 120 volt line  or 2 power strips to 2 15 amp 120 volt lines.

Have all power off plug in everything    then turn the 750 watt on first and the  1000 watt on second.

do a software configuration like freq ? not sure what batch so freq ?        and do the fan at 61%

check hashing    good to go.

gold psu's giving a max of 1750 watts for 170 usd after rebates  that is really fucking good.

I ordered a GS 1050 yesterday for the same price as the 1000w. I'm in Canada so price might be a bit different.

Anyhow, those rebate normally take 2 months to get received. Is there a way to request them online now or you still need to print the form and mail it? I don't have a printer.

evga make you print it. also I dont know if they work for Canada since you have a Canadian  newegg.

Still if you could get them they should be good the 750 has 4 separate pcie cables and the 1000 has 6 separate cables.

that matches up nice for the s-7

[KeezAZA]

Sorry. RSA (South Africa).
I was considering the Batch 9. but with all these issues of boards not running makes me concerned that i might get a dud and I am then stuck with more delays.

Have you considered hosting? I host all of my BTC miners and only run Scrypt miners at home.

https://bitcointalk.org/index.php?topic=1283287.0

I have not considered hosting before. It may be a better solution considering the shipping delays and electricity costs.
 

[VirosaGITS]

Has anyone received B8 and measured the power consumption at the wall yet? 1300w + 10% seem a bit on the high side, completely capping a 15a120v breaker and leaving no headroom for OC or anything.

My B8 pulled 1362w from the wall with a platinum PSU running on 240v at 700MHz. Would be closer to 1400w if running on 120v instead due to PSU efficiency loss on 120v.

Yeah, the 240v do get 2%~ more efficiency. Thanks for the number, but i'm not sure i want to buy a 400$ ATX PSU to get 94% on 120v.

Maybe i could find some quiet 1600w sever PSU that is quiet?

The two 1100w PSU i have would not work very well here.

you can get the 2000 watt plat

links
coming up

https://bitcointalk.org/index.php?topic=1284563.0

https://bitcointalk.org/index.php?topic=1257327.msg13383917#msg13383917

https://bitcointalk.org/index.php?topic=738527.0

http://holybitcoin.com/product/dps-2000bb-breakout-adapter-board/

Right, i would like to go with this. However i only have 120v, which is what i meant, when i vaguely mentioned that the S7 would max a 120v15a breaker. Sorry for being unclear.

So i'd need a 1600w~ 120v PSU to have enough efficiency to run 1300w DC under 1440w, 1400w would be prefered, not sure why but i went through half a dozen surge protector (Belkin, APC) and all the one i ran at 1440 or higher melted or such.

My only alternative is having a big cord extention, about 15 meters, bringing the 240v to my living room window, plugging a rated 40A PDU or whatever and then plugging some 2000 as recommended or 2880 PSU that Finksy. Dunno how doable that all is, i never used 240v.

yeah if you go atx you need the expensive evga 1600 models  or a leadex 1600 model all cost good money. 300 usd plus

could you try 2 1000 watt evga's

http://www.newegg.com/Product/Product.aspx?Item=N82E16817438044   this is about 225 for 2

I found a good setup  use this rebate :


http://images10.newegg.com/uploadfilesfornewegg/rebate/SH/EVGA3MIRsDec16Dec3115cd12us1.pdf

buy this 1000 watts http://www.newegg.com/Product/Product.aspx?Item=N82E16817438062

buy this 750 watts  http://www.newegg.com/Product/Product.aspx?Item=N82E16817438060

both get a 30 dollar rebate so the

 750 watts = 95- 30 or 65 usd
1000 watts - 135-30 or 105 usd   total = 170 usd

use the 750 to run the controller and 1 board
use the 1000 to run the other 2 boards.

use 2 power strips to the same 20 amp line 120 volt line  or 2 power strips to 2 15 amp 120 volt lines.

Have all power off plug in everything    then turn the 750 watt on first and the  1000 watt on second.

do a software configuration like freq ? not sure what batch so freq ?        and do the fan at 61%

check hashing    good to go.

gold psu's giving a max of 1750 watts for 170 usd after rebates  that is really fucking good.

I ordered a GS 1050 yesterday for the same price as the 1000w. I'm in Canada so price might be a bit different.

Anyhow, those rebate normally take 2 months to get received. Is there a way to request them online now or you still need to print the form and mail it? I don't have a printer.

evga make you print it. also I dont know if they work for Canada since you have a Canadian  newegg.

Still if you could get them they should be good the 750 has 4 separate pcie cables and the 1000 has 6 separate cables.

that matches up nice for the s-7

Yeah thats the plan right no, while i wait to figure out the 240v stuff. I already have like 8 of those PSU's and i really like them. But i was hoping to do not have to do multiple PSU per miner.

[Prelude]

The 2000w PSU is not platinum, in fact it's silver but pretty close to gold. It's younger and stronger 2500w brother is gold rated, though. The 2880w beast is not actually platinum either, it is solidly in silver territory. I measured about 87% efficiency. Everyone confuses it with it's younger, slightly more buff 2980w brother the 39Y7414 is platinum, and much more expensive on ebay. We're all buying the 39Y7349 which has no official efficiency numbers anywhere that I can find as it's too old to have been put through the 80+ testing program. My own testing though, puts it at about 87% efficiency at 50% load.

The 2980W 39Y4714 shows a little over 94% efficiency at 50% load verses the 1880W 39Y7349 at your stated 87% at 50% load.

So, let me make sure I understand this correctly;

If we had an S7 that we knew consumed 1210 watts [And used a 2980W 39Y7414 with 93% efficiency with approximately 40.6% load] we should see about 1,301 Watts at the wall on a watt meter.

AND

If we had an S7 that we knew consumed 1210 watts [And used a 2880W 39Y7349 with 86% efficiency with approximately 42.0% load] we should see about 1,407 Watts at the wall on a watt meter.

105 watts difference.  It would be about the same with 2 x S7's on each PSU I believe.  Let me see:

If we had 2 x S7's we knew consumed 2420 watts combined [And used a 2980W 39Y7414 with 93% efficiency with approximately 81% load] we should see about 2,602 watts at the wall on a watt meter.

AND

If we had 2 x S7's we knew consumed 2420 watts combined [And used a 2880W 39Y7349 with 86% efficiency with approximately 84% load] we should see about 2,813 watts at the wall on a watt meter.

So, my math is showing about 105 watts extra has to be burnt to power an S7 with the 2880W verses the 2980W.  If we had 20 x S7's and burnt 105 watts more per S7 with the 2880W PSU, that would be 2100 more watts burnt.  The price difference between each PSU is approximately $75 to $85.  Let's use the $75 difference...

Converting that 2100 watts (2.1 kW) more power into dollars when using 10 x 2980's to power 20 x S7's:

730 hours in a month x 2.1 kW per hour = 1,533 kWh's in a month

1,533 kWh's x $0.10 per kWh = $153.30 saved in a month [All 10 x 2980W's paid for in 9.785 months; the price difference between 2980W & 2880W made up in 5 months]
1,533 kWh's x $0.09 per kWh = $137.97 saved in a month [All 10 x 2980W's paid for in 10.87 months; the price difference between 2980W & 2880W made up in 5.435 months]
1,533 kWh's x $0.08 per kWh = $122.64 saved in a month [All 10 x 2980W's paid for in 12.23 months; the price difference between 2980W & 2880W made up in 6.115 months]
1,533 kWh's x $0.07 per kWh = $107.31 saved in a month [All 10 x 2980W's paid for in 13.98 months; the price difference between 2980W & 2880W made up in 7 months]
1,533 kWh's x $0.06 per kWh = $  91.98 saved in a month [All 10 x 2980W's paid for in 16.31 months; the price difference between 2980W & 2880W made up in 8.15 months]
1,533 kWh's x $0.05 per kWh = $  76.65 saved in a month [All 10 x 2980W's paid for in 19.57 months; the price difference between 2980W & 2880W made up in 9.785 months]
1,533 kWh's x $0.04 per kWh = $  61.32 saved in a month [All 10 x 2980W's paid for in 24.46 months; the price difference between 2980W & 2880W made up in 12.23 months]
1,533 kWh's x $0.03 per kWh = $  45.99 saved in a month [All 10 x 2980W's paid for in 32.62 months; the price difference between 2980W & 2880W made up in 16.31 months]

That's about the gist of it. When I was looking into using those 2880w-ers for my farm, the 2980w-ers were quite a bit more expensive. I picked up 2880w-ers for ~$50 on average, and the 2980w-ers were not available under $300. They seem to have come down a bit now, though. I of course went with the 2880w-ers as I assumed they were platinum also since that's what I've seen everybody here call them. Measuring DC output and comparing it to AC input tells a different story, though. I've since converted pretty much every PSU in my farm to server PSUs with 93.5-94% efficiency (Which is pretty much Titanium rated) at 100% load as I tend to run them close to or pretty much maxed out. My power only costs about 6c USD, but the savings still add up with the gain in efficiency. The lower heat output is nice too, along with opening up capacity for a few more rigs to run on my pretty much maxed out 200A service.

I may look into the 2980w-ers now that they've come down in price along with j4abber's awesome breakout boards. The ease of use is unparalleled in the server PSU world thanks to his breakout board. My current PSUs require a lot of labour... Mostly soldering, but also rigging up the wires so they're "modular". An interesting fact to note is that the 2880W-ers are made by Astec, but IBM decided to go with Delta as the OEM for the 2980W-ers. Delta are also the OEM on the popular 2000W PSUs. I'm a huge Delta fan boy, every server PSU I run is made by them. They're pretty much top dog as far as I know.

[Prelude]

Has anyone received B8 and measured the power consumption at the wall yet? 1300w + 10% seem a bit on the high side, completely capping a 15a120v breaker and leaving no headroom for OC or anything.

My B8 pulled 1362w from the wall with a platinum PSU running on 240v at 700MHz. Would be closer to 1400w if running on 120v instead due to PSU efficiency loss on 120v.

Yeah, the 240v do get 2%~ more efficiency. Thanks for the number, but i'm not sure i want to buy a 400$ ATX PSU to get 94% on 120v.

Maybe i could find some quiet 1600w sever PSU that is quiet?

The two 1100w PSU i have would not work very well here.

you can get the 2000 watt plat

links
coming up

https://bitcointalk.org/index.php?topic=1284563.0

https://bitcointalk.org/index.php?topic=1257327.msg13383917#msg13383917

https://bitcointalk.org/index.php?topic=738527.0

http://holybitcoin.com/product/dps-2000bb-breakout-adapter-board/

Right, i would like to go with this. However i only have 120v, which is what i meant, when i vaguely mentioned that the S7 would max a 120v15a breaker. Sorry for being unclear.

So i'd need a 1600w~ 120v PSU to have enough efficiency to run 1300w DC under 1440w, 1400w would be prefered, not sure why but i went through half a dozen surge protector (Belkin, APC) and all the one i ran at 1440 or higher melted or such.

My only alternative is having a big cord extention, about 15 meters, bringing the 240v to my living room window, plugging a rated 40A PDU or whatever and then plugging some 2000 as recommended or 2880 PSU that Finksy. Dunno how doable that all is, i never used 240v.

yeah if you go atx you need the expensive evga 1600 models  or a leadex 1600 model all cost good money. 300 usd plus

could you try 2 1000 watt evga's

http://www.newegg.com/Product/Product.aspx?Item=N82E16817438044   this is about 225 for 2

I found a good setup  use this rebate :


http://images10.newegg.com/uploadfilesfornewegg/rebate/SH/EVGA3MIRsDec16Dec3115cd12us1.pdf

buy this 1000 watts http://www.newegg.com/Product/Product.aspx?Item=N82E16817438062

buy this 750 watts  http://www.newegg.com/Product/Product.aspx?Item=N82E16817438060

both get a 30 dollar rebate so the

 750 watts = 95- 30 or 65 usd
1000 watts - 135-30 or 105 usd   total = 170 usd

use the 750 to run the controller and 1 board
use the 1000 to run the other 2 boards.

use 2 power strips to the same 20 amp line 120 volt line  or 2 power strips to 2 15 amp 120 volt lines.

Have all power off plug in everything    then turn the 750 watt on first and the  1000 watt on second.

do a software configuration like freq ? not sure what batch so freq ?        and do the fan at 61%

check hashing    good to go.

gold psu's giving a max of 1750 watts for 170 usd after rebates  that is really fucking good.

Slight correction to the above advice: Controller needs to be the last thing powered on, so you'd turn the 1000w PSU on first and then the 750w PSU on after.

[philipma1957]

The 2000w PSU is not platinum, in fact it's silver but pretty close to gold. It's younger and stronger 2500w brother is gold rated, though. The 2880w beast is not actually platinum either, it is solidly in silver territory. I measured about 87% efficiency. Everyone confuses it with it's younger, slightly more buff 2980w brother the 39Y7414 is platinum, and much more expensive on ebay. We're all buying the 39Y7349 which has no official efficiency numbers anywhere that I can find as it's too old to have been put through the 80+ testing program. My own testing though, puts it at about 87% efficiency at 50% load.

The 2980W 39Y4714 shows a little over 94% efficiency at 50% load verses the 1880W 39Y7349 at your stated 87% at 50% load.

So, let me make sure I understand this correctly;

If we had an S7 that we knew consumed 1210 watts [And used a 2980W 39Y7414 with 93% efficiency with approximately 40.6% load] we should see about 1,301 Watts at the wall on a watt meter.

AND

If we had an S7 that we knew consumed 1210 watts [And used a 2880W 39Y7349 with 86% efficiency with approximately 42.0% load] we should see about 1,407 Watts at the wall on a watt meter.

105 watts difference.  It would be about the same with 2 x S7's on each PSU I believe.  Let me see:

If we had 2 x S7's we knew consumed 2420 watts combined [And used a 2980W 39Y7414 with 93% efficiency with approximately 81% load] we should see about 2,602 watts at the wall on a watt meter.

AND

If we had 2 x S7's we knew consumed 2420 watts combined [And used a 2880W 39Y7349 with 86% efficiency with approximately 84% load] we should see about 2,813 watts at the wall on a watt meter.

So, my math is showing about 105 watts extra has to be burnt to power an S7 with the 2880W verses the 2980W.  If we had 20 x S7's and burnt 105 watts more per S7 with the 2880W PSU, that would be 2100 more watts burnt.  The price difference between each PSU is approximately $75 to $85.  Let's use the $75 difference...

Converting that 2100 watts (2.1 kW) more power into dollars when using 10 x 2980's to power 20 x S7's:

730 hours in a month x 2.1 kW per hour = 1,533 kWh's in a month

1,533 kWh's x $0.10 per kWh = $153.30 saved in a month [All 10 x 2980W's paid for in 9.785 months; the price difference between 2980W & 2880W made up in 5 months]
1,533 kWh's x $0.09 per kWh = $137.97 saved in a month [All 10 x 2980W's paid for in 10.87 months; the price difference between 2980W & 2880W made up in 5.435 months]
1,533 kWh's x $0.08 per kWh = $122.64 saved in a month [All 10 x 2980W's paid for in 12.23 months; the price difference between 2980W & 2880W made up in 6.115 months]
1,533 kWh's x $0.07 per kWh = $107.31 saved in a month [All 10 x 2980W's paid for in 13.98 months; the price difference between 2980W & 2880W made up in 7 months]
1,533 kWh's x $0.06 per kWh = $  91.98 saved in a month [All 10 x 2980W's paid for in 16.31 months; the price difference between 2980W & 2880W made up in 8.15 months]
1,533 kWh's x $0.05 per kWh = $  76.65 saved in a month [All 10 x 2980W's paid for in 19.57 months; the price difference between 2980W & 2880W made up in 9.785 months]
1,533 kWh's x $0.04 per kWh = $  61.32 saved in a month [All 10 x 2980W's paid for in 24.46 months; the price difference between 2980W & 2880W made up in 12.23 months]
1,533 kWh's x $0.03 per kWh = $  45.99 saved in a month [All 10 x 2980W's paid for in 32.62 months; the price difference between 2980W & 2880W made up in 16.31 months]

That's about the gist of it. When I was looking into using those 2880w-ers for my farm, the 2980w-ers were quite a bit more expensive. I picked up 2880w-ers for ~$50 on average, and the 2980w-ers were not available under $300. They seem to have come down a bit now, though. I of course went with the 2880w-ers as I assumed they were platinum also since that's what I've seen everybody here call them. Measuring DC output and comparing it to AC input tells a different story, though. I've since converted pretty much every PSU in my farm to server PSUs with 93.5-94% efficiency (Which is pretty much Titanium rated) at 100% load as I tend to run them close to or pretty much maxed out. My power only costs about 6c USD, but the savings still add up with the gain in efficiency. The lower heat output is nice too, along with opening up capacity for a few more rigs to run on my pretty much maxed out 200A service.

I may look into the 2980w-ers now that they've come down in price along with j4abber's awesome breakout boards. The ease of use is unparalleled in the server PSU world thanks to his breakout board. My current PSUs require a lot of labour... Mostly soldering, but also rigging up the wires so they're "modular". An interesting fact to note is that the 2880W-ers are made by Astec, but IBM decided to go with Delta as the OEM for the 2980W-ers. Delta are also the OEM on the popular 2000W PSUs. I'm a huge Delta fan boy, every server PSU I run is made by them. They're pretty much top dog as far as I know.

While I can not confirm

86 or 87% for the 2880 watter I can confirm it is not as efficient as the evga 1600 t2
I can also confirm the 2880 watter running 1 s-7 will allow higher hashing since it provide 12.2 volts to the s-7 and the titanium evga 1600 t2 provide 11.97 volts under full load.

basically the 2880 watter will give more gh then the 1600 watt evga t2. 

the delta 2000 with the 2x break out board is also less efficent then the titanimum evga 1600 t2 but   since the deltas allow easy volts of 11.83 12.17 and 12.51  they are really good plus really quiet.

if you want to slightly down clock the s-7 for noise issues and go with the 11.83 volt setting you can do 3 s-7's in 4500 gh range  you will do very well with that setup.

[Sweminer777]

How about on a B6 ?.


What could i overclock it whit all 10 pins and 2000bb IBM ?.

[dmwardjr]

So it's better to buy the 2980W 39Y4714 if you are into mining long term. It should be compatible with the Finksy 2880W board, right? If so I may buy 39Y4714 model from now on.

I think J4bberwock's board should still work fine with it.  I'm going to find out.  If it works fine, that's all I'm getting from now on.  I wish I knew about it sooner before I got the 18 x 2880's I have now.

[Prelude]

So it's better to buy the 2980W 39Y4714 if you are into mining long term. It should be compatible with the Finksy 2880W board, right? If so I may buy 39Y4714 model from now on.

I think J4bberwock's board should still work fine with it.  I'm going to find out.  If it works fine, that's all I'm getting from now on.  I wish I knew about it sooner before I got the 18 x 2880's I have now.

They will work fine, the pinout looks to be the same.

I've known for a long time about the efficiency, but I hesitated to put it out there since no one else had ever mentioned it; I figured my units might be old or tired. After further testing, I'm just about sure I'm right. Now I wait for someone to either confirm my findings, or prove me wrong.

[dmwardjr]

That's about the gist of it. When I was looking into using those 2880w-ers for my farm, the 2980w-ers were quite a bit more expensive. I picked up 2880w-ers for ~$50 on average, and the 2980w-ers were not available under $300. They seem to have come down a bit now, though. I of course went with the 2880w-ers as I assumed they were platinum also since that's what I've seen everybody here call them. Measuring DC output and comparing it to AC input tells a different story, though. I've since converted pretty much every PSU in my farm to server PSUs with 93.5-94% efficiency (Which is pretty much Titanium rated) at 100% load as I tend to run them close to or pretty much maxed out. My power only costs about 6c USD, but the savings still add up with the gain in efficiency. The lower heat output is nice too, along with opening up capacity for a few more rigs to run on my pretty much maxed out 200A service.

I may look into the 2980w-ers now that they've come down in price along with j4abber's awesome breakout boards. The ease of use is unparalleled in the server PSU world thanks to his breakout board. My current PSUs require a lot of labour... Mostly soldering, but also rigging up the wires so they're "modular". An interesting fact to note is that the 2880W-ers are made by Astec, but IBM decided to go with Delta as the OEM for the 2980W-ers. Delta are also the OEM on the popular 2000W PSUs. I'm a huge Delta fan boy, every server PSU I run is made by them. They're pretty much top dog as far as I know.

Thanks again for sharing your opinion.  I believe I've pretty much convinced myself to get the 2980's from this day forward after listening to you and doing some research myself.

[dmwardjr]

86 or 87% for the 2880 watter I can confirm it is not as efficient as the evga 1600 t2

The main reason I chose the 86% number for the 2880 is after looking at the numbers on the data sheet for the 2980.  The wattage on a single S7 falls below the sweet spot for the 87% efficiency and 2 x S7's are too high for the sweet spot for best 87% efficiency.  Hence, the 86% number instead of the 87% number.

[dmwardjr]

So it's better to buy the 2980W 39Y4714 if you are into mining long term. It should be compatible with the Finksy 2880W board, right? If so I may buy 39Y4714 model from now on.

I think J4bberwock's board should still work fine with it.  I'm going to find out.  If it works fine, that's all I'm getting from now on.  I wish I knew about it sooner before I got the 18 x 2880's I have now.

They will work fine, the pinout looks to be the same.

I've known for a long time about the efficiency, but I hesitated to put it out there since no one else had ever mentioned it; I figured my units might be old or tired. After further testing, I'm just about sure I'm right. Now I wait for someone to either confirm my findings, or prove me wrong.

I believe you're right.  You've had the 2880 long enough to know.  You also provided the data sheet on the 2980.  I'm quite certain you are correct.  I'm planning on purchasing at least 16 of them over the next 3 to 4 months.  It depends on how soon I get my power upgraded (increased) here at the house.  

I'm hoping to get a $0.062 per kWh rate myself for the following facility:  I need to talk with my power company to see what the requirements will be for that rate.  Otherwise, I'm not going to lease the 5,000 to 6,000 feet I would like to lease.  If I can get that rate, I won't have enough rigs to justify leasing the place until 3 or 4 months from now; unless I was to partner up with someone.  I've covered up the address to it's location in the screenshots.

I'm interested with anyone's thoughts on this data center being able to keep it cool enough for 200+ S7's?

[rockyforever]

So it's better to buy the 2980W 39Y4714 if you are into mining long term. It should be compatible with the Finksy 2880W board, right? If so I may buy 39Y4714 model from now on.

I think J4bberwock's board should still work fine with it.  I'm going to find out.  If it works fine, that's all I'm getting from now on.  I wish I knew about it sooner before I got the 18 x 2880's I have now.

They will work fine, the pinout looks to be the same.

I've known for a long time about the efficiency, but I hesitated to put it out there since no one else had ever mentioned it; I figured my units might be old or tired. After further testing, I'm just about sure I'm right. Now I wait for someone to either confirm my findings, or prove me wrong.

I believe you're right.  You've had the 2880 long enough to know.  You also provided the data sheet on the 2980.  I'm quite certain you are correct.  I'm planning on purchasing at least 16 of them over the next 3 to 4 months.  It depends on how soon I get my power upgraded (increased) here at the house.  

I'm hoping to get a $0.062 per kWh rate myself for the following facility:  I need to talk with my power company to see what the requirements will be for that rate.  Otherwise, I'm not going to lease the 5,000 to 6,000 feet I would like to lease.  If I can get that rate, I won't have enough rigs to justify leasing the place until 3 or 4 months from now; unless I was to partner up with someone.  I've covered up the address to it's location in the screenshots.

I'm interested with anyone's thoughts on this data center being able to keep it cool enough for 200+ S7's?

Simple google search brought up the add btw.

Looks interesting. let me know if your looking for additional investors!

[dmwardjr]

Simple google search brought up the add btw.

Looks interesting. let me know if your looking for additional investors!

I suppose I should have covered up the sign in the photo and AT&T former data center.

[rockyforever]

Simple google search brought up the add btw.

Looks interesting. let me know if your looking for additional investors!

I suppose I should have covered up the sign in the photo and AT&T former data center.

Eh, no biggie

[nhando]

How is the noise level on the IBM 2990W PSU? Did they improve on the FAN used or same as the IBM 2880W? What else about it that makes them better than the IBM 2000 combo other than more power and connections?  I need to prepare for another batch of PSU order soon.

[philipma1957]

How about on a B6 ?.


What could i overclock it whit all 10 pins and 2000bb IBM ?.

Well my issue is my garage was too hot due to really warm weather.   but cooled off so look below

here is a batch 2 freq 575 boards  using the 2x 2000 watt break out board  the 2 psu's are doing 2 avalon sixes and 1 s-7  17 amps 240 volts = 4080 watts 24/7

the s-7 is doing 5150 freq 637  volts are 12.5 or so




the avalon sixes are oveclocked volts are 12.5 but there is a bit of sag

[notlist3d]

How about on a B6 ?.


What could i overclock it whit all 10 pins and 2000bb IBM ?.

Well my issue is my garage was too hot due to really warm weather.   but cooled off so look below

I have the opposite around here it's winter weather and cold.  Can I ask what did you do to cool it off on your garage?

I'm always interested in seeing cooling, and how it varies so much.

[klondike_bar]

it possible to get the PWM to go lower? If i enter 1%, 5%, 10% there's no noticed difference in volume.

noise isnt bad, about equal to an S3 in a warm room, maybe a tad more 'penetrating'. With the cold air from outside, I can achieve good hashrates with a low fanspeed, but if i could get it 20% quieter at the cost of ~200GH I'd be happy to do so.

Without the cold air from outside, low-rpm is far from sufficient if i am tageting a 55-60C operational temperature
20C (ambient) = 3.6TH
10C = 4TH
5C = 4.3TH
0C = 4.6TH
-5C = 4.8TH (and 0.011% errors, havent pushed further yet)

It is possible to go lower than 20% PWM with my modified bitmain_spi file :
The only modification is that it can go to 10% min instead of 20% min. (You still need to set the fan percentage to what you want  : example 18%)
This allows to effectively  decrease the noise from the s7 (totally silent if set at 10%) if you are in a very cold ambient temperature or if you downclock the frequency seriously
It should not be used if the temperature is > 70°C. Check that the stop miner when temp is over 80°C box is checked in the configuration page.

Download here :
https://mega.nz/#!8BUVVa7K!pSyaXVVjxfkthPdgJUSabKQQaRiVi1s6-zct5Ob32iM
The original is here if you need it :
https://mega.nz/#!MctVRR7D!lEEb0NvQQNtAemzYNawEAtp8qMVi75dxOLqwNFqv7To

To use it :
scp bitmain_spi_s7_fan_10.ko root@192.168.0.16:/config/           #Copy the file to the miner
then ssh to the miner :
ssh root@192.168.0.16
then :

Code:

cp /lib/modules/3.8.13/kernel/drivers/bitmain/bitmain_spi.ko /config/bitmain_spi_original.ko        #Makes a backup
cp /config/bitmain_spi_s7_fan_10.ko /lib/modules/3.8.13/kernel/drivers/bitmain/bitmain_spi.ko         # Copy the new file to the good place
/etc/init.d/cgminer.sh stop ; /etc/init.d/cgminer.sh start                                    #restart cgminer

The modification will be gone if you restart your miner.
If you want to do it again after a restart you only need to do the two last command :

Code:

cp /config/bitmain_spi_s7_fan_10.ko /lib/modules/3.8.13/kernel/drivers/bitmain/bitmain_spi.ko         # Copy the new file to the good place
/etc/init.d/cgminer.sh stop ; /etc/init.d/cgminer.sh start                                     #restart cgminer

If my new binaries help you and you are satisfied, consider helping me by sending some bitcoins to my address:
1NfUptoShEtSD2SQV2f45Xr1bTysFnsHiU

Warning: Do nothing you don't understand ; No warranties whatsoever are given by me ; this could not work or be useless ; this could burn your S7 or make it useless ; this could void your warranty with bitmain.

It works! looking at the two files, seems the difference is 2-3 hex values that are modified. The method provided above works, but heres the easy way for those on windows:

CAUTION: comparing the modified file against the original THAT WAS ON MY BATCH 8 shows a large number of differences, as does comparing my original with the "original" file uploaded above.  The vast majority of changes seem to be incremental hex values: 4F becomes 4E, 83 becomes 82, 9A becomes 99, etc. not sure why though

SOLUTION: using a site like www.diffnow.com, compare the existing bitmain_spi.ko file with the modified one provided above. If there are only 2-3 changes its safe to change. If Its a batch8 system, Ive made a modified file here:  https://mega.nz/#!wAcA3SrZ!2cVcpIR4qjJY2xmFvRREOMlVYFLk6YtzIxQNteGdWe0  (works to 10% fan, and with this the system runs fans at max when changing settings, as it did when stock) (based on this original (S7, B8): https://mega.nz/#!kYdC2bTD!kcBid6lMeN9ygYxJ2sprMhRBmDJ-bwaBO3ldZPCHWiU)


1) download the modified file above.
2) download and run winscp.
3) connect to miner (SCP mode, port 22, root/admin). navigate to folder:   /lib/modules/3.8.13/kernel/drivers/bitmain/
4) rename bitmain_spi.ko to bitmain_spi_original.ko
5) transfer the bitmain_spi_s7-fan_10.ko file to that directory, and then rename as bitmain_spi.ko
6) login to the miner's webUI. change fan setting there and "save&apply"
Not sure if changing it this way will survive a reboot, but the process is simple to repeat.

OPTIONAL:
7) copy the bitmain_spi_s7_fan_10.ko file to /config/bitmain_spi_s7_fan_10.ko
if modifications are lost on system restart, SSH in and type:    

Code:

cp /config/bitmain_spi_s7_fan_10.ko /lib/modules/3.8.13/kernel/drivers/bitmain/bitmain_spi.ko         # Copy the new file to the good place
/etc/init.d/cgminer.sh stop ; /etc/init.d/cgminer.sh start                                     #restart cgminer

interesting sidenote, when changing settings in the lanfeusst version, the fan goes quiet, then spins up (before it went full blast, then reduced speed)