SolidCoin v2.0 features new hashing algorithm, faster on CPUs

[CoinHunter]

http://solidcointalk.org/topic/220-solidcoin-v20-update-1/

So for those of you who don't pop into IRC too often you will have missed some of the latest news on SolidCoin v2.0

It will now be CPU mining based instead of GPU based. Why is this? Well the blockchain format has changed to be more efficient for pools and networking, and due to this new miners were needed. Since new miners were needed anyhow I thought I may as well change the way block headers are hashed so that CPUs are more efficient (currently and for the near future) processing hashes as this gives SolidCoin numerous benefits.

1) You can still mine other coins like Bitcoin whilst simultaneously mining SolidCoins on your CPU
2) The amount of coins generated will be the same as the network adjusts difficulty based on number of hashes
3) No more need to run a separate miner, you can connect to your own client and pools from SolidCoin v2.0 itself, much easier to get your friends mining which will bring balance to the network

[1713264395]

1713264395

[1713264395]

1713264395

[1713264395]

1713264395

[skEwb]

good stuff.

[Spacy]

Yeah, if the SC2 mining is CPU based, we can mine 2 blockchains at the same time 

[SuperTramp]

Interesting, bringing back the cpu mining. Wonder how my OC'd 3.9ghz 1090T will do?

[CoinHunter]

Interesting, bringing back the cpu mining. Wonder how my OC'd 3.9ghz 1090T will do?

Quite well since that's up there with the fastest. I haven't quite benchmarked the new algorithm extensively (because it's not completely optimized yet) but it's about 1000 times more intensive than the last method on CPU. And the changes should make GPUs not a viable option for some time. It's utilizing SHA256 and one of the new SHA-3 candidates "BLAKE" along with some other features which the CPU is better at handling than GPU.

[CoinHunter]

This coming from the guy who destroyed the SolidCoin 1.10 testnet... the testnet which didn't exist? We are all scared.

It wasn't an idle threat, I've lethally demonstrated it twice, Once on Geist Geld and Once on the SC testnet for CH on his ver 1.10 hence his rapid shutdown. Withdrawing the attack due to a negotiated settlement.

[worldinacoin]

I feel that CPU mining will help encourage more miners who do not want to spend money on expensive GPU upgrades.  Deeply appreciate such a change.

[CoinHunter]

Why is there a need for money to be put on it? If you think you can be a hero and take down the SC2.0 then why not just take it over and claim it?  

Insta-street-cred. Plus you don't have 10000 BTC, if you did you wouldn't have a Geforce2MX mining rig, so nice troll. But yes I'll take whatever BTC you have, put it in escrow.

[CoinHunter]

I feel that CPU mining will help encourage more miners who do not want to spend money on expensive GPU upgrades.  Deeply appreciate such a change.

Yes and it also means businesses can run a little mining on their servers to help cover costs too.

[kano]

LOL THIS IS SO FUNNY

Can someone please delete this thread - it's just so moronic.
I cannot believe you came up with this idea.
Seriously? This must be a joke thread? It's got to be. There is no other explanation I can think of.
Again seriously?

[omri]

Quite well since that's up there with the fastest. I haven't quite benchmarked the new algorithm extensively (because it's not completely optimized yet) but it's about 1000 times more intensive than the last method on CPU. And the changes should make GPUs not a viable option for some time. It's utilizing SHA256 and one of the new SHA-3 candidates "BLAKE" along with some other features which the CPU is better at handling than GPU.

 I've had a look at "BLAKE" and it seams o be using similar basic operation to the current SHA-256. How is it less efficient at GPU mining then the current SHA2/256 algorithm? It does not include the non-linear elements Maj and Ch from SHA2, but these were not the main selling points of the GPUs anyway.
 I am not a GPU expert, so there is probably something major that I missed here, can somebody please explain it.

[Cosbycoin]

I'm wondering what a parallel processor can't do that a linear processor can.

I'm not buying it until I see some technical explaination as to why GPUs would not be viable for mining with a particular algorithm.

Please spare us your nonsense talk and give us technical details.

[Spacy]

Let's wait for his paper and the code, then we can decide, how long (if efficient) it will take for the GPU-miners to join the game ;-)

[kano]

I'm wondering what a parallel processor can't do that a linear processor can.

I'm not buying it until I see some technical explaination as to why GPUs would not be viable for mining with a particular algorithm.

Please spare us your nonsense talk and give us technical details.

It's closed source so no one will know what the algorithm is so no one will be able to write a GPU miner ...

Of course that is a load of crap, it's not hard to reverse engineer what it is doing and it only takes one person with a grudge and a small amount time to do this and destroy the network ...
My only question is: will it be possible with a single rig?

[Cosbycoin]

I'm wondering what a parallel processor can't do that a linear processor can.

I'm not buying it until I see some technical explaination as to why GPUs would not be viable for mining with a particular algorithm.

Please spare us your nonsense talk and give us technical details.

It's closed source so no one will know what the algorithm is so no one will be able to write a GPU miner ...

Of course that is a load of crap, it's not hard to reverse engineer what it is doing and it only takes one person with a grudge and a small amount time to do this and destroy the network ...
My only question is: will it be possible with a single rig?

I have to agree with you there. reverse engineering software doesn't take long.

Should be fun to watch/participate.

[CoinHunter]

Quite well since that's up there with the fastest. I haven't quite benchmarked the new algorithm extensively (because it's not completely optimized yet) but it's about 1000 times more intensive than the last method on CPU. And the changes should make GPUs not a viable option for some time. It's utilizing SHA256 and one of the new SHA-3 candidates "BLAKE" along with some other features which the CPU is better at handling than GPU.

 I've had a look at "BLAKE" and it seams o be using similar basic operation to the current SHA-256. How is it less efficient at GPU mining then the current SHA2/256 algorithm? It does not include the non-linear elements Maj and Ch from SHA2, but these were not the main selling points of the GPUs anyway.
 I am not a GPU expert, so there is probably something major that I missed here, can somebody please explain it.

It's not just a straight hashing. There are pre and post processing of hashes using certain techniques which favor CPU over GPU. They have to be executed in order and are very hard to split up into smaller problems.

Of course this won't stop the ability for it to run on multiple GPU "threads" per se, but there are other techniques in there which don't favor current GPU technology. Comparatively more memory is used and in ways which aren't conducive to the best memory reading/writing of GPUs. All this being said I'm not going to guarantee CPU will always be faster, perhaps there are some OpenCL coders who can make some magic happen. I am not an OpenCL guru as I haven't had much experience with it to this point in time.... so we will have to wait and see how long CPU will be faster as there's no other way to know. I can also improve it going forward without needing to restart the chain so if anyone can offer suggestions after source is released they are welcome to.

[CoinHunter]

It's closed source so no one will know what the algorithm is so no one will be able to write a GPU miner ...

Of course that is a load of crap, it's not hard to reverse engineer what it is doing and it only takes one person with a grudge and a small amount time to do this and destroy the network ...
My only question is: will it be possible with a single rig?

Who told you it is closed source? It will be released..... please don't speculate and pass things off as fact when you don't know the truth.

[kano]

Well since you have now pointed out it is NOT closed source then you have proven what you said to start with is pure stupidity.

Since you have even stated you don't understand how the CL code works why have you made this absolutely stupid assumption that CPU mining will be all the things you have stated it to be?

What you have said is:
1: "CPUs are more efficient (currently and for the near future)"
 and
2: You don't know about CL coding

LOL is the only obvious response to that.

IF someone bothers to write a GPU version suddenly all your CPU miners are dead in the water.
And all the stupidity you have said about getting everyone's friends involved with their CPUs and not having to pay much to make a miner are suddenly all proven to be an absolute load of crap.

I was silly enough to believe you had a reason (though not a good one) to think that CPU mining would do better than GPU mining.
... and the ONLY reason I can think of that you might have been silly enough to believe was that no one could write a GPU miner for it due to not knowing WHAT to write.
Looks like your reasoning was even less knowledgeable than even that ...

Since the source will be available then there will be no issues with writing a GPU version ... as long as anyone gives a damn to bother doing it ... ... ...

[CoinHunter]

Well since you have now pointed out it is NOT closed source then you have proven what you said to start with is pure stupidity.

Since you have even stated you don't understand how the CL code works why have you made this absolutely stupid assumption that CPU mining will be all the things you have stated it to be?

What you have said is:
1: "CPUs are more efficient (currently and for the near future)"
 and
2: You don't know about CL coding

LOL is the only obvious response to that.

IF someone bothers to write a GPU version suddenly all your CPU miners are dead in the water.
And all the stupidity you have said about getting everyone's friends involved with their CPUs and not having to pay much to make a miner are suddenly all proven to be an absolute load of crap.

I was silly enough to believe you had a reason (though not a good one) to think that CPU mining would do better than GPU mining.
... and the ONLY reason I can think of that you might have been silly enough to believe was that no one could write a GPU miner for it due to not knowing WHAT to write.
Looks like your reasoning was even less knowledgeable than even that ...

Since the source will be available then there will be no issues with writing a GPU version ... as long as anyone gives a damn to bother doing it ... ... ...

Sigh, please try and read my post clearly. I know how GPU architectures work to some extent and the things "Not to do" with OpenCL if you want high performance. So I did the opposite. I said I'm not an OpenCL "guru", I have done some things with it, and also many things in the 3D world with shaders.

Whether or not that is enough what is in there we will have to wait and see, but I'm not sure why you have to start throwing insults and lies into this thread, it is quite immature.

[kano]

What are these things that a GPU miner cannot do that your hash algorithm will use?

If you can't answer that clearly and CORRECTLY then yes what you have stated is a load of crap.

The high performance comment is meaningless since even average performance from a GPU will be more than an order or magnitude faster than a CPU.