Mining with multiple pools (round robin)

[qdot]

Hi!

I wonder if there is (it's so simple I must not be the first user to come up with it) a patch for any OpenCL mining software that enables switching to other pools in a round-robin fashion in the event pool becomes unreachable.


Incentives for miners:
 * Register at multiple pools, mine even when a pool is down
 * Stop worrying about down pools
 * 'Steady as she goes' thermal workload on mining rigs.. I wonder if cooling down the idle miner and reheating it, negatively impacts it's reliability.. it probably does.

Incentives for network:
 * No tragic loss of 40% hashing power should deepbit.net go down again.. which means no confirmations for 30mins etc..
 * Overall balanced use of pools - entering multiple pools, each of which dying randomly means the miners would distribute evenly on all pools entered in the client

Disincentives for miners:
 * Having to manage payouts from multiple pools.. not quite as important, since most pay automatically after certain threshold.
 * Hard to estimate earnings.

[keybaud]

I just run 2 miners on each GPU. They work at 50% to each pool and if one pool goes down the second pool's miner ramps up to 100%. I seem to get a higher total hash rate doing this than with a single pool as well.

I actually have 3 installations of Phoenix per GPU, each one setup for a different pool. I run 2 pools per GPU at the same time, but if one pool goes down long term, I fire up the 3rd pool's miners. I do this with 3 GPUs, so have 9 miners configured, but only run 6 at a time.

[russelljohnson]

I just run 2 miners on each GPU. They work at 50% to each pool and if one pool goes down the second pool's miner ramps up to 100%. I seem to get a higher total hash rate doing this than with a single pool as well.

Are there any negative effects for that?

[keybaud]

I just run 2 miners on each GPU. They work at 50% to each pool and if one pool goes down the second pool's miner ramps up to 100%. I seem to get a higher total hash rate doing this than with a single pool as well.

Are there any negative effects for that?

Other than managing them, no.

[russelljohnson]

I just run 2 miners on each GPU. They work at 50% to each pool and if one pool goes down the second pool's miner ramps up to 100%. I seem to get a higher total hash rate doing this than with a single pool as well.

Are there any negative effects for that?

Other than managing them, no.

Should we all put each GPU to mine at 2 pools then? I'd do it

Do you use guiminer? If you do, do you use different -f flags?

[TheSeven]

Incentives for miners:
 * Register at multiple pools, mine even when a pool is down
 * Stop worrying about down pools
 * 'Steady as she goes' thermal workload on mining rigs.. I wonder if cooling down the idle miner and reheating it, negatively impacts it's reliability.. it probably does.

Incentives for network:
 * No tragic loss of 40% hashing power should deepbit.net go down again.. which means no confirmations for 30mins etc..
 * Overall balanced use of pools - entering multiple pools, each of which dying randomly means the miners would distribute evenly on all pools entered in the client

Disincentives for miners:
 * Having to manage payouts from multiple pools.. not quite as important, since most pay automatically after certain threshold.
 * Hard to estimate earnings.

Disincentives for pool operators and the network:
 * If someone manages to DDoS two or three pools, the other ones will be taken down by the load of additional miners switching over to them as well?

[qdot]

I just run 2 miners on each GPU. They work at 50% to each pool and if one pool goes down the second pool's miner ramps up to 100%. I seem to get a higher total hash rate doing this than with a single pool as well.

That's a good hint! Mining is ALU-bound, so it should behave that way

Unfortunately, on my machine, it (sometimes??) crashes the other miner?

Perhaps it's just X11 that got in the way, I disabled it, and it should be chugging along just fine.

[russelljohnson]

I think if you simultaneously mine at two separate pools, you lose a % of your hashing power. My max hashing power for 1 rig was 720MH/s but running them at two different pools simultaneously give me only a max of 688MH/s. GPU usage was hovering around 92% instead 99%.

But I don't know if that's better than having random connection errors at one single pool.

[keybaud]

I think if you simultaneously mine at two separate pools, you lose a % of your hashing power. My max hashing power for 1 rig was 720MH/s but running them at two different pools simultaneously give me only a max of 688MH/s. GPU usage was hovering around 92% instead 99%.

But I don't know if that's better than having random connection errors at one single pool.

My combined hashing on 2 pools per GPU is higher than with one pool and my GPU usage is permanently at 100%.

What miner and settings are you using?

[qdot]

Disincentives for pool operators and the network:
 * If someone manages to DDoS two or three pools, the other ones will be taken down by the load of additional miners switching over to them as well?

A sensible pool should be able to handle growth in legitimate traffic.

Speaking with no experience in bitcoin pool operations in general, but with a lot of experience in designing scalable web applications, a typical DDoS is roughly 10x larger than a foreseeable legitimate traffic.

I don't think all legitimate miners could bring down a pool that is supposed to have a fighting chance against organized DDoS.

[russelljohnson]

I think if you simultaneously mine at two separate pools, you lose a % of your hashing power. My max hashing power for 1 rig was 720MH/s but running them at two different pools simultaneously give me only a max of 688MH/s. GPU usage was hovering around 92% instead 99%.

But I don't know if that's better than having random connection errors at one single pool.

My combined hashing on 2 pools per GPU is higher than with one pool and my GPU usage is permanently at 100%.

What miner and settings are you using?

guiminer poclbm

flags: -v -w256 -f1

[keybaud]

I think if you simultaneously mine at two separate pools, you lose a % of your hashing power. My max hashing power for 1 rig was 720MH/s but running them at two different pools simultaneously give me only a max of 688MH/s. GPU usage was hovering around 92% instead 99%.

But I don't know if that's better than having random connection errors at one single pool.

My combined hashing on 2 pools per GPU is higher than with one pool and my GPU usage is permanently at 100%.

What miner and settings are you using?

guiminer poclbm

flags: -v -w256 -f1

I'm using Phoenix with a command line batch file and I use separate installations so they run independently of each other.

You could try copying the GUIminer folder and running the second pool from that exe, to see if that improves things.

[russelljohnson]

I think if you simultaneously mine at two separate pools, you lose a % of your hashing power. My max hashing power for 1 rig was 720MH/s but running them at two different pools simultaneously give me only a max of 688MH/s. GPU usage was hovering around 92% instead 99%.

But I don't know if that's better than having random connection errors at one single pool.

My combined hashing on 2 pools per GPU is higher than with one pool and my GPU usage is permanently at 100%.

What miner and settings are you using?

guiminer poclbm

flags: -v -w256 -f1

I'm using Phoenix with a command line batch file and I use separate installations so they run independently of each other.

You could try copying the GUIminer folder and running the second pool from that exe, to see if that improves things.

appreciate the help. I'm gonna get some sleep now it's 3AM here in California. Hopefully things work out later.

[techman05]

Just curious how I manage a round robin with cgMiner.  I set it there thinking it would.be easier than using rotate but it hasn't.done as expected.  I missing how round robin is expected to work?

Sorry for resurecting but it seemed like a good description and the title matchex what I was thinking.

[os2sam]

Just curious how I manage a round robin with cgMiner.  I set it there thinking it would.be easier than using rotate but it hasn't.done as expected.  I missing how round robin is expected to work?

With Round Robin, when a pool fails for 5 minutes it will move to the next pool and mine on it until it fails and then moves to the next.  If your primary pool is a good one and never fails then it will never move to another pool.

What behavior are you seeing?

What behavior are you expecting that your not seeing?

[Aco]

I was told by a very experienced friend that pool hopping tends to be worse.

[os2sam]

I was told by a very experienced friend that pool hopping tends to be worse.

Was he talking about pool hopping?  Or was he talking about how it takes a while for your earnings to level out with payout systems like DGM and PPLNS?

If it's the latter you will still get paid for your shares over time so no real need to worry.

If its the former, there are only like two hoppable pools left so it's pretty much a non issue these days.

[techman05]

Just curious how I manage a round robin with cgMiner.  I set it there thinking it would.be easier than using rotate but it hasn't.done as expected.  I missing how round robin is expected to work?

With Round Robin, when a pool fails for 5 minutes it will move to the next pool and mine on it until it fails and then moves to the next.  If your primary pool is a good one and never fails then it will never move to another pool.

What behavior are you seeing?

What behavior are you expecting that your not seeing?

I was trying to maximize the amount of shares I got, that's why I was thinking round robin would share time for both pools so I wouldn't miss out on either with my hashing speed.  I'll just have to finish one and stick with the other. Both seem good except one was bitminter and if its donations don't get in the way it would be nice with more mining power out of my shares in it.

[os2sam]

Just curious how I manage a round robin with cgMiner.  I set it there thinking it would.be easier than using rotate but it hasn't.done as expected.  I missing how round robin is expected to work?

With Round Robin, when a pool fails for 5 minutes it will move to the next pool and mine on it until it fails and then moves to the next.  If your primary pool is a good one and never fails then it will never move to another pool.

What behavior are you seeing?

What behavior are you expecting that your not seeing?

I was trying to maximize the amount of shares I got, that's why I was thinking round robin would share time for both pools so I wouldn't miss out on either with my hashing speed.  I'll just have to finish one and stick with the other. Both seem good except one was bitminter and if its donations don't get in the way it would be nice with more mining power out of my shares in it.

Balance and load balance are probably more along the lines of what your thinking.

Maybe the new quota's would help?  Don't know much about what quota's are yet myself.