Topic: [4+ EH] Slush Pool (slushpool.com); Overt AsicBoost; World First Mining Pool - page 1152. (Read 4382755 times)
December 15, 2010, 11:36:32 PM
I'm regularly getting the error on the m0mchil's GPU miner: "IOError: [Errno socket error] [Errno 10060] A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond"
December 15, 2010, 11:36:15 PM
Is the source for your cooperative miner available?
December 15, 2010, 11:15:55 PM
Dunno because there is the doublec server using puddingcups code would be cool to add that extra 250M hash
AFAIK they use special protocol and different algo for counting shares. But they can change miners and connect us
. But I'm not affraid that there will not be enough power. I have to optimize code very quickly to be ready for many fast GPUs. 
December 15, 2010, 11:09:46 PM
Dunno because there is the doublec server using puddingcups code would be cool to add that extra 250M hash
December 15, 2010, 11:07:54 PM
How does rounding work does it get carried over.
All financial computation are made multiplied by 10^8, so with maximal precision. If there will be still some rests after dividing 50BTCs, the rest is added to random worker which participated in mining. But don't forget we are talking about 'rests' on eighth decimal place...
Quote
Can you speak with the other pooled mining server so we can all work together.
How?
December 15, 2010, 11:03:14 PM
ok I'll help you guys out for a little while. 
the blocks are two easy!
They need to be about twice as hard.You are right, dynamic targets are on the way. Don't forget it is beta. Current target is compromise for low-end and high-end machines. By the way, normal CPUs (say 1000khash/s) will compute current target about half an hour
. December 15, 2010, 10:58:35 PM
How does rounding work does it get carried over.
Can you speak with the other pooled mining server so we can all work together.
Can you speak with the other pooled mining server so we can all work together.
December 15, 2010, 10:51:24 PM
ok I'll help you guys out for a little while.
the blocks are two easy!
They need to be about twice as hard. December 15, 2010, 10:50:20 PM
ok I'll help you guys out for a little while.
December 15, 2010, 10:39:06 PM
Oh, found a bug. Please do not register logins with dots. As dots are delimiters between login and worker name, login with dots does not work correctly.
I will fix it tomorrow (probably by prohibiting dots in worker names and splitting 'login.worker' by the most right dot). Thanks!
I will fix it tomorrow (probably by prohibiting dots in worker names and splitting 'login.worker' by the most right dot). Thanks!
December 15, 2010, 08:36:50 PM
These aren't using the bitcoinr miners are they? It's using the standard JSON API, right?
Right, internally it works with standard bitcoin client.
December 15, 2010, 08:35:30 PM
now we just need to solve some blocks to see if it pays.
If m0mchil provide power which he promised me today, we will see it very soon :-).
December 15, 2010, 08:32:50 PM
These aren't using the bitcoinr miners are they? It's using the standard JSON API, right?
December 15, 2010, 08:28:10 PM
works nicely with jgarzik's CPU (only tested 0.2.1) and m0mchils GPU miner,
it seems so at least, workers share-count shows what miners tell.
hashcount is pretty much the same as on local getwork-servers, now we just need to solve some blocks to see if it pays.
it seems so at least, workers share-count shows what miners tell.
hashcount is pretty much the same as on local getwork-servers, now we just need to solve some blocks to see if it pays.
December 15, 2010, 07:02:15 PM
Hi all, I just started my own bitcoin mining server on http://mining.bitcoin.cz . Server was heavily tested on Testnet, but it is still beta stage software. But there is already ~600000khash/s of power and more will come tomorrow.
Server is based on principle which is discussed above and which satoshi accepted as possible. Server works with all CPU/GPU miners following standard getwork() interface - jgarzik's CPU miner, m0mchil's and Diablo's GPU miners. All needed information are on site homepage.
Sending rewards - when block found, 50BTCs are written to server's wallet, not on miner's wallets. Every miner can specify own 'sending treshold', which is minimum amount to send to his wallet. This should help with sending 0.0001 BTCs and so on. Currently the cron script for sending is disabled, I will process few first payments manually - just to be sure that everything is working fine.
If you are interested, please open account, enter wallet address, register own workers and start mining! Also report all troubles, I will work on them ASAP! Please be patient with short server outages in this stage. In the most cases, miners will solve that by restarting itself.
Server is based on principle which is discussed above and which satoshi accepted as possible. Server works with all CPU/GPU miners following standard getwork() interface - jgarzik's CPU miner, m0mchil's and Diablo's GPU miners. All needed information are on site homepage.
Sending rewards - when block found, 50BTCs are written to server's wallet, not on miner's wallets. Every miner can specify own 'sending treshold', which is minimum amount to send to his wallet. This should help with sending 0.0001 BTCs and so on. Currently the cron script for sending is disabled, I will process few first payments manually - just to be sure that everything is working fine.
If you are interested, please open account, enter wallet address, register own workers and start mining! Also report all troubles, I will work on them ASAP! Please be patient with short server outages in this stage. In the most cases, miners will solve that by restarting itself.
December 08, 2010, 05:11:52 AM
I will change my miner to check lower targets.
Glad to read that!
Quote
Keep in mind that GPUs are perhaps 30x+ faster than CPUs. Are you going to adjust the target to control the number of results you receive from specific client? Because with target suitable for getting 1 result per minute from CPU, you may receive 1 result per SECOND from a GPU.
You are right. Currently I have same target for all workers, but dynamic target depending on worker performance is in my TODO. I'm testing on Nvidia ION (~2000khash/s) and HD 5970 (~560mhash/s) and think target near 8 zeros should be fine for beginning. 8 zeros (32bit) gives one blocks every ~30 minutes on Nvidia and every tens seconds on Radeon, which is yet OK.
December 08, 2010, 05:02:03 AM
It's too slow to do a full 256-bit comparison for each hash generated. Taking the lead from mainline bitcoin, miners tend to "cheat" and simply look for a certain number of zero bits in the area that, when byte-swapped, holds the high-order bits of a 256-bit number. My cpuminer stops if the top 32 bits are all zero. Mainline bitcoin stops scanning for hashes if 16 bits are zero:
I'm glad I understood your code well
. I agree full check is too expensive. But what about compromise - check so many bits for zeros how many zeros is in target sent by getwork? Simply do not have hardcoded '32 bits' inside miner... This should be quite simple, fast and good enough!By the way, I'm using modified version of your scanhash_c for checking proof-of-work in my server. My reward from first block found by cooperative mining is yours.
December 08, 2010, 04:35:01 AM
I will change my miner to check lower targets. Keep in mind that GPUs are perhaps 30x+ faster than CPUs. Are you going to adjust the target to control the number of results you receive from specific client? Because with target suitable for getting 1 result per minute from CPU, you may receive 1 result per SECOND from a GPU.
December 08, 2010, 02:56:27 AM
It's too slow to do a full 256-bit comparison for each hash generated. Taking the lead from mainline bitcoin, miners tend to "cheat" and simply look for a certain number of zero bits in the area that, when byte-swapped, holds the high-order bits of a 256-bit number. My cpuminer stops if the top 32 bits are all zero. Mainline bitcoin stops scanning for hashes if 16 bits are zero:
The reason the mainline bitcoin only tests 15 bits is to support testnet where difficulty can be less than 1.00. 32 zero bits corresponds to difficulty 1.00.
December 07, 2010, 10:24:09 PM
It is working with jgarzik's cpuminer, m0mchil's python miner and Diablo's java miner pretty well. But I found one problem in comparing targets. As I understand their codes well, they don't fully compare found hashes with target. So if I set difficulty pretty low, because I want many low-difficulty blocks for fair accounting, miners simply does not return correct hashes (hashes corresponding current target).
It's too slow to do a full 256-bit comparison for each hash generated. Taking the lead from mainline bitcoin, miners tend to "cheat" and simply look for a certain number of zero bits in the area that, when byte-swapped, holds the high-order bits of a 256-bit number. My cpuminer stops if the top 32 bits are all zero. Mainline bitcoin stops scanning for hashes if 16 bits are zero:
Code:
if (((unsigned short*)phash)[14] == 0)
return nNonce;
return nNonce;
Ideally, the miners would all do something like
Code:
if (test for certain number of zero bits)
if (perform full target comparison)
// found a solution
return nNonce;
if (perform full target comparison)
// found a solution
return nNonce;
but we programmers can be a lazy bunch sometimes.
Jump to: