Topic: [0.1 Bounty Claimed] Please write a complete guide with screenshots on Gridseed - page 3. (Read 11325 times)

Has anyone been able to get these to dual mine under windows

Thanks for the BTC! I'm sorry about the language, but my windows is in Dutch, so I can't really do anything about it. Is there anything I can do to make it more clear? I tried to be clear in my description, and use the terms from the english windows there.

Evidently, all controllers are not the same. Is there only one source/person/programmer/program/standards for the firmware, software, configs, language, hardware..

Who makes them? I have heard - read, wii box and only recently realized it's not the same as the game 'wii' or some game, or is it? I know,idiot, right. I should know this stuff, ;[ not necessarily.

Anyway, they don't sound as stable as some have claimed, hence the question above.
Anyone know the answer/s to this?

Which one do 'I' buy? Who makes the best and most stable controller and software?

I think even the latest cpuminer for LTC is still a bit buggy, perhaps fifo drivers and settings?
I also think that someone needs to come out with a FULL DUPLEX UART that doesn't cause bottlenecks with these high speed networked UARTS all flowing out to one receive buffer in a 10 port hub...what were they thinking? Well, most peripherals today, the usb ones, don't all work and run at the same time...Our miners do!

Port speed isn't really the issue here, it's the ABILITY of a given UART chip (one in this case) to decipher and decide where to send ALL that traffic (aaaahhhhhh) at once....blockages are occurring at this point aka bottleneck in the pipe. Gets jammed up, does a buffer overflow dump, there goes all that data (POOF) , the system spits dummy - DOH! - and hangs.

The readout in the DOS window is only showing new work requests 'Dispatching new work to GC3355 LTC Core. That's as far as the downloaded data gets though, nothing is going out to the miner. Dispatches are not reaching the dispatch-ee! So it has nothing to do, so it doesn't report back. The receive buffer works fine because there are multiple receive buffers in a USB hub, but only ONE transmit buffer!....>DUMB!

One final point I almost forgot to make, speed is good but not at the expense of stability! We all want stable performance. Even though it isn't the fastest we can hash, it's still a superior solution because it gives long term results with few, if any, interruptions. Set it and forget it. It's stable and where needed, various redundancies can be built in on the support end of the game. That's the stability and reliability we're shooting for! Profits! Not chronic problems!

Well,that's my 55c!
Peace!
Wolfey2014

Well, maybe you figured it out 'in your case' as it is possible to under power the USB devices, especially with tiny 500mA wall warts.

By now, I'm pretty sure there is nothing wrong with my GC3355 5 chip miners.
It's definitely a communications problem.

I'm using a 2.5A wall wart to power my hub and 5 miners and I know it's way overkill which is good! You want a bit if not a lot of overhead (more amps available than ever needed) to keep things stable. I mean the current draw from these miners is very peaky.

But, my system is having what I believe is a FIFO buffer problem.
The transmit buffer settings are too high for most 2.0 USB hub UART's because the hubs were not designed for constant 'network' traffic type data throughput. They are designed for and are stable managing 'intermittent' traffic. Not 'constant' traffic. So when running several to many miners via USB, the transmit buffers crap out. Something called Transmit Buffer Overflow/overload occurs, data packets are lost and you end up with your miners having new work dispatched to them but nothing comes back. They hang. DOH!

Most USB hubs come with several receive pipes/channels but only 1 - that is ONE transmit pipe / channel.
So when your computer transmits (sends) many 'new work' requests at nearly if not the same exact time, well - within nano seconds of each other - , they plug up (bottleneck) and crap out...give up and just stop working.

Unplugging each port separately - waiting for the PC to ack - then plugging them back in and waiting for an ack' (acknowledgement) usually cures the immediate problem. But after running a few hours, the problem happens again and you're back to square one. What to do? Is it the hub's fault or is it the transmit buffer setting is too high?
To change this setting to a slightly lower one, - COM port PROPERTIES / PORT SETTINGS / ADVANCED.
Make the change, click OK out of there, unplug and re-plug each port on your hub - one at a time to get it freshly recognized.

Evidently the UARTs on these GC3355 miners are set to auto-detect port speeds etc. so nothing to do there. They are defaulted to run at 115200 but can run as high as 6.25Mbps. Most 2.0 hubs run at up to 480Mbps. 3.0 can run much higher speeds than that. But you don't need even 115200, evidently. I'm running mine at 38400 and they seem to flow data nicely.

If you want to picture what a FIFO buffer overflow looks like, its like backed up rush hour traffic. The road gets plugged up so no one can move, more and more cars pile up in the rear of the line, very few are moving forward at the front of that mess and some drivers just give up or get lost or have car problems etc. No forward movement! If people just slow down, spread out and put a few car lengths between each other, traffic would keep moving but at a much slower pace. This is what happens when you decrease the transmit buffer. SPREAD OUT! SLOW DOWN A LITTLE BIT and everyone will get to where they're going but just a few minutes (milliseconds in data speak) later.

Well, that's my theory right now and I'm trying out my fix. I'll let you know how it went in a few hours.
Failing that, I'll be in the market for a Ultra High Speed USB 3.0 hub. I don't need RUSH HOUR TRAFFIC ON MY COMPUTER! AAAAAAAAAHHHHHHHHHHHHHHHH!!!!!!!!!!!!!  

As always, try this stuff at your own risk! You fugg it up, you eat it!
Peace!
Wolfey2014

It sounds like testing the another hub/firmware could be a good start.

I noticed today that my controller is not displaying the hashrate (it says 0 khash/s), but it does register shares, and the pool also registers them. I hadn't had that before...

I have 10 miners per controller, 10 a USB hub. x2
Except for 2 that are troublesome, they all hash seperately for a while. So, they appear to work. (No idea if there is another test)
If i connect them miner after miner, it will hash and speedup.. Until the controller reboots and it never really picks up after that.

It still also seems the USB hub is at fault: i get only one YAY when connected from Windows. Tried my W8.1 desktop and my wifes W7 laptop. Both hubs the same result.
I might buy a new hub thats tested.. Orico or the Sipolar that's shipped with it usually.

I'm getting a gridseed kit soon for a friend, i'll try those hubs and controllers..

Hopefully Jack releases the final firmware today as promised, i'm told the reboots are gone, hashing speeds higher and overall stability.

Even if it hashes at lower speed as advertised, i would be OK.. The reboots, i don't like. This really takes the profits down..

I figured it out I got a power supply with more AMP's hooked it up and its now mining

I hooked up 1 directly to my PC powered off a basic 12V 1.5AMp transformer to test I get the CPU miner saying dispatching new work to GC3355 LTC core but never see YAY. Om the unit I have green light only blinking and the fan is running. Any ideas?

Ok, so you have 20 miners, which require around 10W each (in scrypt only mode), so 200W on 12V line, which is 16A. Every powersupply should be capable of delivering that, especially if it has 4x pci-e connector.

If the cables have been made correct (there are three +12V lines in a sata connector, I don't know if it works when only 1 or 2 are connected), they should work. The same for molex. I have my miners running of a molex connector.

Please remember this when troubleshooting:

- There have been multiple cases in which the controller couldn't handle the full 20 miners
- The controllers have a tendency to not hash at all, if something is wrong with one of the miners

So, I would say you need to try each miner separately, at a frequency of 600 to start with, and check if they all work. Then you need to check with a known working miner whether all USB and power cables work. If everything works, try adding miners one at a time, and see when they stop working.

Is there something missing from my thread?

I'll add whatever info you have that isn't addressed.

Yes, the work gets sent.. And that's it..
But, i might be onto something: as i told, i'm using an ATX PSU to power the miners, 20 of them..
It's an old PSU i had laying around, with 4 6 pin PCIE plugs, a few SATA and molex connectors.
I have aquired adapter cables from PCIE to 3x miner plug each, so 12 are powered directly by PCIE.
For the others i had an extra cable while i wait for a new PSU, from SATA to PCIE, i also have a molex PCIE if needed.

I only tested 1 miner directly from PCIE.. That one gets "YAY", the ones on SATA don't..
Is the power insufficiant? The miners do get power, because they run as expected.. (LEDs, fans are on)

Just tried the miners connected to PCIE directly: through the USB hub ONLY 1 hashes.. I'll reboot and test again, but it looks like the SATA and MOLEX are underpowered, but the USB hub seems to be no good either..

Sad..

Hopefully LightningAsic releases the new controller FW today (i was told it's ready), that does seem to hash a bit at least.. Speed not that great, but most units do SOMETHING for a while..

So what steps go wrong? Cpuminer tries to send work to the miners?

It's powered, 5V, 3.5A, Delock USB 2.0 External Hub 10 Port, p/n 87468.
The miners are powered by an ATX PSU. I just flip the switch and the miners, controllers, USB hubs are powered.

Is it a powered hub? Otherwise it probably will not work.

Also, did you maybe forget to power the miners? I did that once... I forgot to plug in the miners power supply, they responded fine on cpuminer commands, they just didn't start hashing :p

Hi,

I tried my USB hub connected miners on Windows 8.1.. None of them EVER says or call out "YAY".. Is this driver related? Or does the USB hub not work correctly?
I don't think ALL my miners are broken..

Whick hub is tested OK? I have seen the Orico and the Sipolar X-300 shipped with the set, any others?

Thanks,

Andre

good to know. I will test it maybe later

thanks for a link!


Cheers!

haha yah to bad all this info is separated into many different threads, makes of a very inefficient use of time scanning many threads... and then actually trying to implement what you read is even more time consuming lol but sorta fun kinda