Quote
So, with only one cgminer instance, we can find Bitcoin/Namecoin and Litecoin blocks!
TL;DR: OP's request will require a lot of work.
Topic: Request for an Scrypt capable fork of cgminer - maybe within new CGMiner / RPC (Read 5502 times)
January 13, 2012, 06:32:08 AM
Ah, ok. I think this answers better OP's request (bold mine):
TL;DR: OP's request will require a lot of work.
TL;DR: OP's request will require a lot of work.
legendary
Activity: 1862
Merit: 1011
Reverse engineer from time to time
January 13, 2012, 06:02:33 AM
Mixing GPU mining and CPU-mining in the same process is not a good idea. Imagine that most people like me put cgminer with multiple GPUs on one underclocked core. Every thread will fight for CPU time.
Of course, the newest drivers seem to have fixed the 100% CPU usage bug, but this does not mean that everyone upgraded.
In order to implement such a thing, a massive recode would be needed, so that per-GPU(or per-thread) pools are allowed(each thread/GPU to mine on a different pool), and only then can the CPU threads work like that.
So the best thing here is to use one miner process for GPUs and one for CPU, doing it like this allows free control of which process which core of the CPU to use. CPU mining from the same process will impact GPU mining performance severely, so best not is my advice.
Of course, the newest drivers seem to have fixed the 100% CPU usage bug, but this does not mean that everyone upgraded.
In order to implement such a thing, a massive recode would be needed, so that per-GPU(or per-thread) pools are allowed(each thread/GPU to mine on a different pool), and only then can the CPU threads work like that.
So the best thing here is to use one miner process for GPUs and one for CPU, doing it like this allows free control of which process which core of the CPU to use. CPU mining from the same process will impact GPU mining performance severely, so best not is my advice.
January 13, 2012, 05:44:36 AM
As for the ability to switch algos for different pools, I find that unneeded since CPU bitcoin mining is extremely rare, plus some pools are banning miners which are requesting work, but their efficiency is low.
I don't see how you can avoid pinning some pools to the CPU and others to the GPU. I assume people mining bitcoin and a cpu chain at the same time will use different pools for them?
Also, this might not be a real problem worth working on, but it would be nice if there would be support for pool fallback/balancing for CPU too. So I guess there would be two sets of pools (CPU and GPU), each with their own balancing/failover rules. I don't know if it makes sense at this point or how much effort will be needed... it's just an idea.
January 12, 2012, 05:02:51 PM
I agree with you!! Not hate... Sorry...
I'll never talk about Scrypt in CGMiner again... Gotcha!! ^_^
Ckolivas wants to drop CPU support from CGMiner... I'm fine with that... Finally... lol
Nevertheless, I think that even if we don't believe in some alt-coin, why not help who believes?
People can even pay, with any cryptocurrency, for the job...
I'll never talk about Scrypt in CGMiner again... Gotcha!! ^_^
Ckolivas wants to drop CPU support from CGMiner... I'm fine with that... Finally... lol
Nevertheless, I think that even if we don't believe in some alt-coin, why not help who believes?
People can even pay, with any cryptocurrency, for the job...
January 12, 2012, 02:12:44 PM
I just being curious on why the two hate so much a new piece of code within CGMiner... I think it can not hurt cgminer, it can be a plus to it.
I won't speak for them but I don't think it is "hate". It is simply seeing not future in alt-coins (at least alt-coins so far). Ckolivas name is defacto attached to cgminer and Kano did the extensive work on RPC code.
You can understand that they wouldn't be enthusiastic about providing resources (maybe even resulting in user's assuming that means implicit support) for something they find to have no value. Right?
The good news ... this is why we support free software. It is GPLed so you have the freedom to do what you want with it. You could even work on a fork which mines (shudder)
legendary
Activity: 1862
Merit: 1011
Reverse engineer from time to time
January 12, 2012, 01:54:48 PM
Well...
I was able to make Scrypt work with cgminer which was the primary goal here.
I even was able to compile a 64-bit Windows CGminer thus the Scrypt miner speed from minerd is retained.
I have a few issues, the decay speed reported after a while turns into the word "inf"(but not as problematic as people may think).
And second, Mingw-w64 does not seem to like localtime_r, which was a problem when I was asked to compile the binaries from pooler's code.
Implementing scrypt into cgminer was basically a copy/paste job, so I do not know if that was enough for a reward, but I leave that for you to decide. As for the ability to switch algos for different pools, I find that unneeded since CPU bitcoin mining is extremely rare, plus some pools are banning miners which are requesting work, but their efficiency is low.
CGMiner used is latest one with API support.
As for why I have not released it, I am only waiting to see if "just" porting the algorithm was enough to complete the bounty.
I was able to make Scrypt work with cgminer which was the primary goal here.
I even was able to compile a 64-bit Windows CGminer thus the Scrypt miner speed from minerd is retained.
I have a few issues, the decay speed reported after a while turns into the word "inf"(but not as problematic as people may think).
And second, Mingw-w64 does not seem to like localtime_r, which was a problem when I was asked to compile the binaries from pooler's code.
Implementing scrypt into cgminer was basically a copy/paste job, so I do not know if that was enough for a reward, but I leave that for you to decide. As for the ability to switch algos for different pools, I find that unneeded since CPU bitcoin mining is extremely rare, plus some pools are banning miners which are requesting work, but their efficiency is low.
CGMiner used is latest one with API support.
As for why I have not released it, I am only waiting to see if "just" porting the algorithm was enough to complete the bounty.
January 04, 2012, 10:50:58 PM
Maybe Pooler can fork the CGMiner-CPU-only code and implement his minerd code into it ?
The cgminer-cpuonly code IS the cgminer code, just with a different configuration option (--disable-opencl) when I build it. January 04, 2012, 08:05:33 PM
Maybe Pooler can fork the CGMiner-CPU-only code and implement his minerd code into it ?
But ckolivas do not want that code within his git (main cgminer tree) so, who will maintain that fork?!
Anyway, I was thinking in having only one cgminer instance, with scrypt mining for pre-selected pools and, normal Bitcoin mining at GPUs... No only with cgminer-cpuonly but, It can be awesome too!
Ideas?!
January 03, 2012, 02:30:42 PM
Maybe Pooler can fork the CGMiner-CPU-only code and implement his minerd code into it ?
January 03, 2012, 02:24:32 PM
With Pooler's minerd being such a killer app for Intel CPUs, do you really think that CGMiner could compete without ALOT of optimizations ?
Why not Bounty-Request Pooler to add enhancements to his minerd for LTC mining ?
Why not Bounty-Request Pooler to add enhancements to his minerd for LTC mining ?
I like the CGMiner interface / RPC features and, the CGMiner with Scrypt support will (I guess) have the pooler's code. So, it will be basically the same.
Anyway... We already have the negative response from ckolivas and Kano, so, never mind...
I just being curious on why the two hate so much a new piece of code within CGMiner... I think it can not hurt cgminer, it can be a plus to it.
January 03, 2012, 02:04:19 PM
With Pooler's minerd being such a killer app for Intel CPUs, do you really think that CGMiner could compete without ALOT of optimizations ?
Why not Bounty-Request Pooler to add enhancements to his minerd for LTC mining ?
Why not Bounty-Request Pooler to add enhancements to his minerd for LTC mining ?
January 03, 2012, 12:42:44 PM
Kano,
I'm sorry to ask you this but... Why so much hate?!
It is just some lines of C code... What evil it can have?!
Litecoin is a good experience, don't you think?!
I am very curious to know your opinion in deep about this alt-chain (Litecoin) in specific.
Best!
Thiago
I'm sorry to ask you this but... Why so much hate?!
It is just some lines of C code... What evil it can have?!
Litecoin is a good experience, don't you think?!
I am very curious to know your opinion in deep about this alt-chain (Litecoin) in specific.
Best!
Thiago
January 03, 2012, 11:30:23 AM
ThiagoCMC,
I know ckolivas hates the alternate chains so I'm pretty sure he would NOT want the code in his git.
(and I do understand why)
Thus I'd not be interested in implementing it either.
I know ckolivas hates the alternate chains so I'm pretty sure he would NOT want the code in his git.
(and I do understand why)
Thus I'd not be interested in implementing it either.
December 30, 2011, 12:00:49 AM
Hi!
Guys, how much will cost to add support for Scrypt (Litecoin mining) to cgminer (maybe cgminer-cpuonly)?
Of course, based on: https://github.com/pooler/cpuminer
10 BTC?
So, with only one cgminer instance, we can find Bitcoin/Namecoin and Litecoin blocks!
But keeping the option to mine for Bitcoins using cgminer-cpuonly too, if user wish that...
We should be able to choose a different algorithm for a pool (which will be a Litecoin pool)...
Best!
Thiago
Guys, how much will cost to add support for Scrypt (Litecoin mining) to cgminer (maybe cgminer-cpuonly)?
Of course, based on: https://github.com/pooler/cpuminer
10 BTC?
So, with only one cgminer instance, we can find Bitcoin/Namecoin and Litecoin blocks!
But keeping the option to mine for Bitcoins using cgminer-cpuonly too, if user wish that...
We should be able to choose a different algorithm for a pool (which will be a Litecoin pool)...
Best!
Thiago
Jump to: