Build Faster Miners People :)

[5chdn]

An optimised miner not tied to any pool becomes very important now. Ypool has more than 50% of hashing power and that is not good for a decentralised currency!

I am about to release an update to my miner that will perform 150% better than the current version.  If you get 40HPM now you should get 100 HPM... I believe this will be competitive with ypool.

The new ptsminer 0.5 from beer performs better than the ypool jhMiner. Xolo adapted and improved some code. Maybe some people will switch now.

[bytemaster]

I just posted an open source getwork miner... this bounty is now closed.

[bytemaster]

An optimised miner not tied to any pool becomes very important now. Ypool has more than 50% of hashing power and that is not good for a decentralised currency!

I am about to release an update to my miner that will perform 150% better than the current version.  If you get 40HPM now you should get 100 HPM... I believe this will be competitive with ypool.   

[Proto]

An optimised miner not tied to any pool becomes very important now. Ypool has more than 50% of hashing power and that is not good for a decentralised currency!

[FreeTrade]

Agree, better is optimized standalone miner which can mine at pool and solo using getwork/getworktemplate from the wallet.

Agree too - it would be good to have a standard interface to the various pools so that users of the optimized miner could select between them.

[gigawatt]

I dunno who I should pester about this small tweak, but...

Given that even on a massive system, the collisionspermin metric is < 100, maybe this "hashmeter" line should be adjusted.

I think regular units would be fine here.  hashpermin is already a small number, so hashpersec would be tiny, khashpersec is unreadable.
printf("hashmeter %.6f hash/s\n", dHashesPerSec);

[bytemaster]

Agree, better is optimized standalone miner which can mine at pool and solo using getwork/getworktemplate from the wallet.

How about we go with the lowest common denominator: optimize the momentum_search function.
If you're doing anything with ProtoShares, it depends directly on momentum_search.  If you're optimizing anything else, it's application-specific.

I second this.  I need this algorithm for more than just ProtoShares so we will benchmark momentum_search()!   

[gigawatt]

Agree, better is optimized standalone miner which can mine at pool and solo using getwork/getworktemplate from the wallet.

How about we go with the lowest common denominator: optimize the momentum_search function.
If you're doing anything with ProtoShares, it depends directly on momentum_search.  If you're optimizing anything else, it's application-specific.

[bytemaster]

But just for fun, I will add a 1 BTC bounty paid to anyone that can build a faster miner than mine EVERYONE ELSE and keep it that way for over a week.   You must submit your miner as a patch to the  ProtoShares-Qt client so it is easy to distribute for all! 

Why you're requesting to submit miner code for pts-qt? I think its better to have standalone miner

Agree, better is optimized standalone miner which can mine at pool and solo using getwork/getworktemplate from the wallet.

Ok, as long as it works with the wallet locally and is not tied to a pool.

[testz]

But just for fun, I will add a 1 BTC bounty paid to anyone that can build a faster miner than mine EVERYONE ELSE and keep it that way for over a week.   You must submit your miner as a patch to the  ProtoShares-Qt client so it is easy to distribute for all! 

Why you're requesting to submit miner code for pts-qt? I think its better to have standalone miner

Agree, better is optimized standalone miner which can mine at pool and solo using getwork/getworktemplate from the wallet.

[gigawatt]

Made a few more tweaks.



Averaging ~6 cpm per GHz.

[gigawatt]

I'm testing on a 2.0 GHz single core VPN.
So yeah.  I'm doing pretty well I'd say.


The default setup is ~2 collisions/min and requires 50% more memory.

[bytemaster]

See the Coyote Miner / Pool thread for the latest on my speeds... I am getting 43 HPM on an 3.4Ghz i7 iMac.

Can we clarify if we're talking hashes per minute or collisions per minute?

For competing algorithms we absolutely should be talking collisions per minute. I can do a lot of hashes so long as I don't have to find any collisions

Sure, we can opt for collisions per minute, but what we really want is solutions per minute to blocks below difficulty X.

[FreeTrade]

See the Coyote Miner / Pool thread for the latest on my speeds... I am getting 43 HPM on an 3.4Ghz i7 iMac.

Can we clarify if we're talking hashes per minute or collisions per minute?

For competing algorithms we absolutely should be talking collisions per minute. I can do a lot of hashes so long as I don't have to find any collisions

[bytemaster]

Hmm... I do have an improved miner with only a few lines changed and ~20% more HPM.
Maybe try this bounty later, after I am paid the 1BTC for fixing crash (I am still waiting for that 1BTC )

Ok.. I was waiting on bitstamp withdraw to pay your bounty... coming now.

20% more than my pool miner?

Got the 1BTC, thank you!

No, it is just 20% more than the original pts-miner. But that's just a few line changes within a few minutes. I just wanted to show that there is much room to make improvements on the miner.

For the record, I spent more time implementing the getwork API than I did optimizing the code to get the 2-3x gains... low hanging fruit people...