[GUIDE] GridSeed GC3355 5 Chip Setup/power/windows/linux/rpi by UnicornHasher

[usao]

Im not trying to take away from the OP, but I just wonder why there is such a wide variation in reported power consumption...
Even if you take the PF into account, it wouldn't cause a variation of a 2x unless im missing somthing here...

[IchibahnSLC]

Lol Thanks Eric

Wrong calculator:

Assume 10W per unit as a safe load so 36 units would hit 80-100% on your power supply at 900 clock.

I'll edit my above post.

And think a little next time.

Guys from HashMaster are telling me the individual 300KH/s units (with fan) draw 20W per unit.
Other folks are telling me 7-8W per unit, you are telling me 10W...
Bit confused here. Has anyone actually measured the power or current when doing LTC only and when doing BTC and LTC together?
I would really like to get a more solid number. I have several HP 12V @100A server PSU's, which I thought I could purpose for these gridseed miners, since the included PSU is reported to be crappy.

Also, do some folks remove the fan and use a box-fan to cool a bunch at once?

Look in the power and temps section of the OP.

I clocked from 600-900 in single and dual mode and took a reading.

Also, to be fair to everyone that gave you information, the wattage depends on the efficiency of the power supply.

As stated in the OP and with usable measurements.  My readings are at about 80% efficiency.  I was running at 99 watts in single mode at 850 clocks.

Part 6: Power Usage & Temps

Power Usage and Speeds: Power is based on killowatt readings for the controller, USB hub and 10 devices using a Corsair AX1200 PSU.  Divide out the power factor for actual usage.  The speed reports are only based on a 5-10 minute sample time so add +/-10% variance.  You can see in the screen cap that the per device speed is completely crazy this is mostly the controller and device quality.  

Power Supply Efficiency: Corsair AX1200 80 Plus GOLD

I was just relaying little known info, I agree with your findings and they look accurate for your setup. I'm not doubting your findings. 

[daemonfox]

Im not trying to take away from the OP, but I just wonder why there is such a wide variation in reported power consumption...
Even if you take the PF into account, it wouldn't cause a variation of a 2x unless im missing somthing here...

There are a handful of factors dictating other people's power readings...

1. If they are using a single unit on a USB right off their PC/laptop and measuring off the included power supply... they arent getting the full read as the board inside is powered by the USB port itself.
2. If they are measuring multiple units on a powered USB hub attached to the controller, then you are getting a slightly inflated rate based on all the extra hardware that is not the device itself.
3. If they are measuring this using the included power brick, that is going to be different than the power usage from a PSU using a molex to barrel jack connection.
4. If they are using a single unit, then that can still only give you an idea because some of these units have already shown they are faster/slower than others which means they pull more/less power than others.
5. Depending on your power source, you have the efficiency factor
6. Depending on what kind of USB hub you are using when measuring multiple devices can cause a variation in power rate based on what kind of power supply you have for the hub
7. Out of the box the miners have a script for running at 600 frequency which draws very little power... some people are increasing this to 850/900/more so it draws more power... these increases can vary in power draw based on how well the miner is put together and the 5 chips quality.

Just a handful or things that make everyone's readouts different for these things...

[worldlybedouin]

@miaviator:  Would you please update the OP which has a link (under #5) for the easy 5 step process to running gridseeds off of a RasPi?

The new link should be this one:  https://litecointalk.org/index.php?topic=9908.msg140876#msg140876

User "misc" has created a new Scripta+CGMiner RasPi image and it works quite well.

[wolfey2014]

I've been having a problem with my miners disconnecting and just showing "dispatching new work toGC3355 LTC core" and nothing else. WTF???

It started happening after I pointed them to ScryptGuild.com
Before that, they had pretty much "yay" hashed full time without any interruptions.

Does anyone know if a pool server can cause this to happen? Make them quit running by perhaps, quit sending work to them for a few minutes which causes them to just - hang?

I don't think it's the USB hub causing it.

This happened 6 hours after I signed up with ScryptGuild, then another 2 hours after that, then another 18 hours after that. It's not the USB hub or lack of power causing it.

To get them running again just now, I have re-set all the ports using disable-enable which saves me from having to physically unplug and re-plug each one. I'm going to figure out a way to make this problem 'auto detect this failure and run disable-enable for each port automatically so they keep on hashing.

I have also disabled the power saving feature on each 'root hub' in device manager just to make sure WinBlow$ doesn't cause it to happen too.

I'm using Win7. I don't know if XP allows one to make these types of config changes or not.

I"m not interested in putting up with this disconnect shyt! I am sure no one else is either

Wolfey2014

Hi I'm running windows 7
With 10 gridseeds running on good usb 3 10 port hubs
I have pointed them at scryptguild for the last 24 hours and I have to say I'm very impressed with the returns I'm getting.

I have never seen the problem you are having are you running cgminer or cpu miner?

what brand / model name and number are you using for the 10 port hub?
Interesting take on SG,,,sounds almost corporate PR like....

[usao]

Im not trying to take away from the OP, but I just wonder why there is such a wide variation in reported power consumption...
Even if you take the PF into account, it wouldn't cause a variation of a 2x unless im missing somthing here...

There are a handful of factors dictating other people's power readings...

1. If they are using a single unit on a USB right off their PC/laptop and measuring off the included power supply... they arent getting the full read as the board inside is powered by the USB port itself.
2. If they are measuring multiple units on a powered USB hub attached to the controller, then you are getting a slightly inflated rate based on all the extra hardware that is not the device itself.
3. If they are measuring this using the included power brick, that is going to be different than the power usage from a PSU using a molex to barrel jack connection.
4. If they are using a single unit, then that can still only give you an idea because some of these units have already shown they are faster/slower than others which means they pull more/less power than others.
5. Depending on your power source, you have the efficiency factor
6. Depending on what kind of USB hub you are using when measuring multiple devices can cause a variation in power rate based on what kind of power supply you have for the hub
7. Out of the box the miners have a script for running at 600 frequency which draws very little power... some people are increasing this to 850/900/more so it draws more power... these increases can vary in power draw based on how well the miner is put together and the 5 chips quality.

Just a handful or things that make everyone's readouts different for these things...

Thanks. Im looking for the power draw at the barrel plug on an individual miner device. I have several HP 1200W 12V@100A server PSU's and I was planning to use these to power the actual miners. My goal was to figure out roughly how many I could plan to wire to each PSU. If im off by a factor of 2, it could result in a melt-down, so I was hoping to pin down a more precise value. I didn't want to include the rpi or hub or any of that stuff in the calculation.
On another note, has anyone had failures attributed to overclocking? How "safe" are they to overclock?

[vabchgent]

Im not trying to take away from the OP, but I just wonder why there is such a wide variation in reported power consumption...
Even if you take the PF into account, it wouldn't cause a variation of a 2x unless im missing somthing here...

There are a handful of factors dictating other people's power readings...

1. If they are using a single unit on a USB right off their PC/laptop and measuring off the included power supply... they arent getting the full read as the board inside is powered by the USB port itself.
2. If they are measuring multiple units on a powered USB hub attached to the controller, then you are getting a slightly inflated rate based on all the extra hardware that is not the device itself.
3. If they are measuring this using the included power brick, that is going to be different than the power usage from a PSU using a molex to barrel jack connection.
4. If they are using a single unit, then that can still only give you an idea because some of these units have already shown they are faster/slower than others which means they pull more/less power than others.
5. Depending on your power source, you have the efficiency factor
6. Depending on what kind of USB hub you are using when measuring multiple devices can cause a variation in power rate based on what kind of power supply you have for the hub
7. Out of the box the miners have a script for running at 600 frequency which draws very little power... some people are increasing this to 850/900/more so it draws more power... these increases can vary in power draw based on how well the miner is put together and the 5 chips quality.

Just a handful or things that make everyone's readouts different for these things...

Thanks. Im looking for the power draw at the barrel plug on an individual miner device. I have several HP 1200W 12V@100A server PSU's and I was planning to use these to power the actual miners. My goal was to figure out roughly how many I could plan to wire to each PSU. If im off by a factor of 2, it could result in a melt-down, so I was hoping to pin down a more precise value. I didn't want to include the rpi or hub or any of that stuff in the calculation.
On another note, has anyone had failures attributed to overclocking? How "safe" are they to overclock?

To error on the side of caution take the max amps and x by 2. They overclock to 850 rather well. I have mine dual mining at 850 with no problems.  I personally use 12v 6a power supplies per unit saves the time and the trouble of the math and the cost about 11.00 a piece shipped.

[scryptographer]

Im not trying to take away from the OP, but I just wonder why there is such a wide variation in reported power consumption...
Even if you take the PF into account, it wouldn't cause a variation of a 2x unless im missing somthing here...

There are a handful of factors dictating other people's power readings...

1. If they are using a single unit on a USB right off their PC/laptop and measuring off the included power supply... they arent getting the full read as the board inside is powered by the USB port itself.
2. If they are measuring multiple units on a powered USB hub attached to the controller, then you are getting a slightly inflated rate based on all the extra hardware that is not the device itself.
3. If they are measuring this using the included power brick, that is going to be different than the power usage from a PSU using a molex to barrel jack connection.
4. If they are using a single unit, then that can still only give you an idea because some of these units have already shown they are faster/slower than others which means they pull more/less power than others.
5. Depending on your power source, you have the efficiency factor
6. Depending on what kind of USB hub you are using when measuring multiple devices can cause a variation in power rate based on what kind of power supply you have for the hub
7. Out of the box the miners have a script for running at 600 frequency which draws very little power... some people are increasing this to 850/900/more so it draws more power... these increases can vary in power draw based on how well the miner is put together and the 5 chips quality.

Just a handful or things that make everyone's readouts different for these things...

Thanks. Im looking for the power draw at the barrel plug on an individual miner device. I have several HP 1200W 12V@100A server PSU's and I was planning to use these to power the actual miners. My goal was to figure out roughly how many I could plan to wire to each PSU. If im off by a factor of 2, it could result in a melt-down, so I was hoping to pin down a more precise value. I didn't want to include the rpi or hub or any of that stuff in the calculation.
On another note, has anyone had failures attributed to overclocking? How "safe" are they to overclock?

For me a single unit (w/ fan running, clocked at 800MHz) running off of the PSU (80+ Bronze) and one of the PC's USB ports draws ~10W at the wall. Without the fan it draws ~7W.

My units run best at either 800 or 850 depending in the unit. When clocked to high they start producing hardware errors.

[daemonfox]

Im not trying to take away from the OP, but I just wonder why there is such a wide variation in reported power consumption...
Even if you take the PF into account, it wouldn't cause a variation of a 2x unless im missing somthing here...

There are a handful of factors dictating other people's power readings...

1. If they are using a single unit on a USB right off their PC/laptop and measuring off the included power supply... they arent getting the full read as the board inside is powered by the USB port itself.
2. If they are measuring multiple units on a powered USB hub attached to the controller, then you are getting a slightly inflated rate based on all the extra hardware that is not the device itself.
3. If they are measuring this using the included power brick, that is going to be different than the power usage from a PSU using a molex to barrel jack connection.
4. If they are using a single unit, then that can still only give you an idea because some of these units have already shown they are faster/slower than others which means they pull more/less power than others.
5. Depending on your power source, you have the efficiency factor
6. Depending on what kind of USB hub you are using when measuring multiple devices can cause a variation in power rate based on what kind of power supply you have for the hub
7. Out of the box the miners have a script for running at 600 frequency which draws very little power... some people are increasing this to 850/900/more so it draws more power... these increases can vary in power draw based on how well the miner is put together and the 5 chips quality.

Just a handful or things that make everyone's readouts different for these things...

Thanks. Im looking for the power draw at the barrel plug on an individual miner device. I have several HP 1200W 12V@100A server PSU's and I was planning to use these to power the actual miners. My goal was to figure out roughly how many I could plan to wire to each PSU. If im off by a factor of 2, it could result in a melt-down, so I was hoping to pin down a more precise value. I didn't want to include the rpi or hub or any of that stuff in the calculation.
On another note, has anyone had failures attributed to overclocking? How "safe" are they to overclock?

To error on the side of caution take the max amps and x by 2. They overclock to 850 rather well. I have mine dual mining at 850 with no problems.  I personally use 12v 6a power supplies per unit saves the time and the trouble of the math and the cost about 11.00 a piece shipped.

Take the above with a grain of salt and reference #7.

Prime example... he may be getting great results clocked at SHA256d850/scrypt850 dual mining... but if I clock mine above SHA256d750/scrypt750, the cpuminer reports nothing but red nonce errors. SHA256d mines like a beast as you clock it up but it interferes with my scrypt core the higher it goes. My best results thus far are 6 GH/s SHA256d and 330 KH/s scrypt clocked evenly at 750/750. I can push to 383 KH/s solo at 850 but further it starts to yield less actual shares submitted. I have not yet attempted to OC the SHA256d engine to see its max... though it is listed some places it is capable of 11+ GH/s.

[suchmoon]

Thanks. Im looking for the power draw at the barrel plug on an individual miner device. I have several HP 1200W 12V@100A server PSU's and I was planning to use these to power the actual miners. My goal was to figure out roughly how many I could plan to wire to each PSU. If im off by a factor of 2, it could result in a melt-down, so I was hoping to pin down a more precise value. I didn't want to include the rpi or hub or any of that stuff in the calculation.
On another note, has anyone had failures attributed to overclocking? How "safe" are they to overclock?

10 miners, connected to the "stock" power supply at 850MHz draw 80W at the wall, measured with a Kill-a-Watt. This is Scrypt-only mode. I don't know the efficiency rating of the power supply, probably not more than 80%, but let's be generous and assume 90%. Let's be equally generous on my Kill-a-Watt's lack of accuracy and assume a 10% margin of error there. The end result is ~8W per miner. I think measuring 10 miners that way is fairly accurate, it averages out any individual variations. Overclocking Scrypt to 850MHz is safe, most people around here seem to be running at that speed without issues. 900MHz results in reduced accepted share rate.

[miaviator]

Im not trying to take away from the OP, but I just wonder why there is such a wide variation in reported power consumption...
Even if you take the PF into account, it wouldn't cause a variation of a 2x unless im missing somthing here...

Why not?

I took power readings based on USB HUB, 10 Devices, 1200w Power supply at ~80% efficiency (low load)

I figured that would give safe numbers to use with 20% headroom for anyone reading. 

I get a decent drop down (10-20%) when I switch PSU's.

Through people using single devices, externally powered hubs, different efficiency power supplies, different ambient temps, etc, etc, 2x is believable specifically in scrypt mode where 2x is 6watts and who knows how many people are using conditioned power or decent meters.

[vabchgent]

Im not trying to take away from the OP, but I just wonder why there is such a wide variation in reported power consumption...
Even if you take the PF into account, it wouldn't cause a variation of a 2x unless im missing somthing here...

There are a handful of factors dictating other people's power readings...

1. If they are using a single unit on a USB right off their PC/laptop and measuring off the included power supply... they arent getting the full read as the board inside is powered by the USB port itself.
2. If they are measuring multiple units on a powered USB hub attached to the controller, then you are getting a slightly inflated rate based on all the extra hardware that is not the device itself.
3. If they are measuring this using the included power brick, that is going to be different than the power usage from a PSU using a molex to barrel jack connection.
4. If they are using a single unit, then that can still only give you an idea because some of these units have already shown they are faster/slower than others which means they pull more/less power than others.
5. Depending on your power source, you have the efficiency factor
6. Depending on what kind of USB hub you are using when measuring multiple devices can cause a variation in power rate based on what kind of power supply you have for the hub
7. Out of the box the miners have a script for running at 600 frequency which draws very little power... some people are increasing this to 850/900/more so it draws more power... these increases can vary in power draw based on how well the miner is put together and the 5 chips quality.

Just a handful or things that make everyone's readouts different for these things...

Thanks. Im looking for the power draw at the barrel plug on an individual miner device. I have several HP 1200W 12V@100A server PSU's and I was planning to use these to power the actual miners. My goal was to figure out roughly how many I could plan to wire to each PSU. If im off by a factor of 2, it could result in a melt-down, so I was hoping to pin down a more precise value. I didn't want to include the rpi or hub or any of that stuff in the calculation.
On another note, has anyone had failures attributed to overclocking? How "safe" are they to overclock?

To error on the side of caution take the max amps and x by 2. They overclock to 850 rather well. I have mine dual mining at 850 with no problems.  I personally use 12v 6a power supplies per unit saves the time and the trouble of the math and the cost about 11.00 a piece shipped.

Take the above with a grain of salt and reference #7.

Prime example... he may be getting great results clocked at SHA256d850/scrypt850 dual mining... but if I clock mine above SHA256d750/scrypt750, the cpuminer reports nothing but red nonce errors. SHA256d mines like a beast as you clock it up but it interferes with my scrypt core the higher it goes. My best results thus far are 6 GH/s SHA256d and 330 KH/s scrypt clocked evenly at 750/750. I can push to 383 KH/s solo at 850 but further it starts to yield less actual shares submitted. I have not yet attempted to OC the SHA256d engine to see its max... though it is listed some places it is capable of 11+ GH/s.

I am running in dual mode 850 with no red none errors ?  everyone is different

[miaviator]

@miaviator:  Would you please update the OP which has a link (under #5) for the easy 5 step process to running gridseeds off of a RasPi?

The new link should be this one:  https://litecointalk.org/index.php?topic=9908.msg140876#msg140876

User "misc" has created a new Scripta+CGMiner RasPi image and it works quite well.

That post doesn't have steps, just an image posted by a newbie?  I'll build a scripta build and post instructions.  I need some more gridseeds to do testing on.

[miaviator]

any tips on how too hook up 26 units on one power supply?
that' a lot of wires ....

From someone else's post:

http://amzn.com/B004EAHJAG

https://i.imgur.com/HDheiZn.jpg

The link points to connectors which are 2.1mm in size, I thought that they were suposto be 2.5mm? Anyone tried this specific connector to see if it fits?

2.1mm plugs do fit and I would say they are better really as a slightly tighter fit is a good thing. I would strongly advise against using those octopus cables though as the wires in them are so thin its a joke. Definitely useless for BTC mode and not even good for LTC only mode as even if they dont get hot you dont need any extra impedance to increase power line noise.

These are designed to be power splicers for low current equipment..  You might be right though.  Looks like most of them are designed for 500ma at 12v power.  So they will likely melt.

[miaviator]

Bwa HaHa:

http://www.newegg.com/Product/Product.aspx?Item=N82E16811165213

15.0" x 19.0" x 3.5" 2U rackmount 15" Depth.



I think there is room for a Pi as well if I cut the power cords a bit shorter.

I could fit at least 30 in a 3U and 20 in a 2U long case based on this catastrophe.

[vabchgent]

okay 500watt power supply roughly 25a rail 10awg wire to keep it safe. A power supply delivering at 41amps 12 awg.  That's a lot of gridseeds  

[suchmoon]

... 31 pieces deliver a steady 9,5-10 Mh/s. I'm using two raspberry Pi's.

How many can you run on one Raspberry Pi?

Bought a set of 10, I hope they arrive soon

I'm going for a raspberry setup, does that thing need it's own powersupply?
How many greedseeds can I hook up on 1 Raspi, I've read 20, but it doesn't seem confirmed.

20 with the "stock" 10-port hubs (I'm currently running that way), 26 with 13-port hubs like this: http://www.amazon.com/gp/product/B00HL7Z46K

It gets more complicated after that, since you can only have 2 levels of hubs on the Pi, and the 10- and 13-port hubs already have 2 levels inside them. Something like this might work, depending on how it's set up internally: http://www.monoprice.com/Product?c_id=103&cp_id=10307&cs_id=1030702&p_id=6436&format=4

Monoprice 24-port hub has 7 ports at the first level, 3 of those connect to 3 internal 7-port hubs. The total comes to 25 ports (4 remaining at the first level, + 3*7 at the second level), not sure where the 25th port is though. Anyway, good news is that Raspberry Pi can access all 24 ports. I have not connected miners to it yet, but tested with a USB flash drive. Will follow up once I do a real mining test.

http://www.monoprice.com/Product?c_id=103&cp_id=10307&cs_id=1030702&p_id=6436&seq=1&format=2
http://www.amazon.com/dp/B008F28V7E

[jjj0923]

Bwa HaHa:

http://www.newegg.com/Product/Product.aspx?Item=N82E16811165213

15.0" x 19.0" x 3.5" 2U rackmount 15" Depth.



I think there is room for a Pi as well if I cut the power cords a bit shorter.

I could fit at least 30 in a 3U and 20 in a 2U long case based on this catastrophe.

I would be concerned about air flow. how about an open front and  when rack mounting put one of the 3U fan kits in the front of the rack.

[daemonfox]

Bwa HaHa:

http://www.newegg.com/Product/Product.aspx?Item=N82E16811165213

15.0" x 19.0" x 3.5" 2U rackmount 15" Depth.



I think there is room for a Pi as well if I cut the power cords a bit shorter.

I could fit at least 30 in a 3U and 20 in a 2U long case based on this catastrophe.

I would be concerned about air flow. how about an open front and  when rack mounting put one of the 3U fan kits in the front of the rack.

There is ZERO reason to be concerned about airflow in scrypt only mode unless your ambient is well above 90 F. The ~8W you burn up in scrypt only mode doesn't even bring these things to a point where you can't touch them while hashing... not even close.

[CartmanSPC]

...but they were born to be free..not hidden in a cage!