Topic: [GUIDE] GridSeed GC3355 5 Chip Setup/power/windows/linux/rpi by UnicornHasher - page 58. (Read 365630 times)

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.

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

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.

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.

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.

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.

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.

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?

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

@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.

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...

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. 

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...

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.

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

Look in the power and temps section of the OP.

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

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?

I don't have any GPUs but I can tell you I have successfully run two instances of cgminer on a RPi.  One cgminer for gridseed and one cgminer for dualminer each in their own screen session.  Both instances are hashing away with no issues.  I want to a cgminer that has support for gridseed and dualminer so I can run one but it's working so I can't complain.

Anybody have a mechanical drawing of the (currently shipping) controller board and its mounting holes?

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?