Gateless Gate Sharp 1.3.8: 30Mh/s (Ethash) on RX 480!

[zzzzzzzzzz]

Seems like claymore is in the same spot...wouldnt be surprised if his recent "eureka" moment is from the discussion happening in this thread.

Oh, I'm now pretty sure he reads this thread. It's really funny in an ironic sense, but it does make things a bit more challenging for me. It's like playing poker with you being the only player with open hands. Oh well.

I am not entirely sure if I would take his current approach, though. I'm not going to feed my competition this time around, so he has to figure things out himself...

I find these assumptions a bit odd, he has been fastest 95% of the time since launch, and people assume he uses methods from slower miners.  I think recent improvements are just to keep up with Optiminer, and timing is a coincidence.  He already had ASM working well many weeks ago for all cards.  I can't know for sure, I just find it odd that every time there is a CM speedup, some assume automatically it is from others.

Keep up the good work, just posting my thoughts.

So, if CM had "ASM working well many weeks ago for all cards", why did he post this 6 days ago:

https://bitcointalk.org/index.php?topic=1670733.msg17725004#msg17725004

And no, I'm not implying, in any way, that CM is doing anything unethical. I'm just saying, according to CM himself, your statement is completely false.

[1714193188]

1714193188

[1714193188]

1714193188

[zawawa]

Don't get me wrong. I truly admire Claymore's coding skills. To be very honest, I am rarely impressed by other programmers, but he is definitely top notch.

All I am saying is that he studies his competitions, be it potential or actual, and he studies them really well. This is not an automatic assumption but an educated, informed observation of his work as a long-time user. Also, it is often the case that different developers come up with different ideas for optimizations, so we are constantly learning from each other.

Anyway, all this talk is meaningless if I cannot come up with a faster miner than Claymore's. We shall see very, very soon.

[toptek]

Don't get me wrong. I truly admire Claymore's coding skills. To be very honest, I am rarely impressed by other programmers, but he is definitely top notch.

All I am saying is that he studies his competitions, be it potential or actual, and he studies them really well. This is not an automatic assumption but an educated, informed observation of his work as a long-time user. Also, it is often the case that different developers come up with different ideas for optimizations, so we are constantly learning from each other.

Anyway, all this talk is meaningless if I cannot come up with a faster miner than Claymore's. We shall see very, very soon.

I hope you can at least match it , either way to keep you at it, i plan to keep donating like i said in a PM . for a while ....

[xeridea]

Seems like claymore is in the same spot...wouldnt be surprised if his recent "eureka" moment is from the discussion happening in this thread.

Oh, I'm now pretty sure he reads this thread. It's really funny in an ironic sense, but it does make things a bit more challenging for me. It's like playing poker with you being the only player with open hands. Oh well.

I am not entirely sure if I would take his current approach, though. I'm not going to feed my competition this time around, so he has to figure things out himself...

I find these assumptions a bit odd, he has been fastest 95% of the time since launch, and people assume he uses methods from slower miners.  I think recent improvements are just to keep up with Optiminer, and timing is a coincidence.  He already had ASM working well many weeks ago for all cards.  I can't know for sure, I just find it odd that every time there is a CM speedup, some assume automatically it is from others.

Keep up the good work, just posting my thoughts.

So, if CM had "ASM working well many weeks ago for all cards", why did he post this 6 days ago:

https://bitcointalk.org/index.php?topic=1670733.msg17725004#msg17725004

And no, I'm not implying, in any way, that CM is doing anything unethical. I'm just saying, according to CM himself, your statement is completely false.

I mean he had good speed in ASM a while ago after limits of OpenCL alone were reached.  Recently for further optimization he figured it would be worth it to have a cleaner environment to work with.  He just said he got an efficient ASM environment for his miners, but we don't know what setup he is using. I am not saying anyone is right or wrong, I just see this repeated trend.

[xeridea]

Don't get me wrong. I truly admire Claymore's coding skills. To be very honest, I am rarely impressed by other programmers, but he is definitely top notch.

All I am saying is that he studies his competitions, be it potential or actual, and he studies them really well. This is not an automatic assumption but an educated, informed observation of his work as a long-time user. Also, it is often the case that different developers come up with different ideas for optimizations, so we are constantly learning from each other.

Anyway, all this talk is meaningless if I cannot come up with a faster miner than Claymore's. We shall see very, very soon.

I am not implying that anyone is condemning, I just though I would express my thoughts, I noticed a while ago.  I am a developer, and I know how learning and collaboration goes, it is good to at least be aware of different methods of doing things to help your creativity. You are the best hope for OS ZEC miner, so keep it up

[Stealth8368]

Seems like claymore is in the same spot...wouldnt be surprised if his recent "eureka" moment is from the discussion happening in this thread.

Oh, I'm now pretty sure he reads this thread. It's really funny in an ironic sense, but it does make things a bit more challenging for me. It's like playing poker with you being the only player with open hands. Oh well.

I am not entirely sure if I would take his current approach, though. I'm not going to feed my competition this time around, so he has to figure things out himself...

I find these assumptions a bit odd, he has been fastest 95% of the time since launch, and people assume he uses methods from slower miners.  I think recent improvements are just to keep up with Optiminer, and timing is a coincidence.  He already had ASM working well many weeks ago for all cards.  I can't know for sure, I just find it odd that every time there is a CM speedup, some assume automatically it is from others.

Keep up the good work, just posting my thoughts.

well I think we are all familiar with Claymores use of others code and from what I can see he is struggling to come up with any speed increases of late,there is no doubt he could learn a thing or two from this thread so perhaps time to keep the secret sauce to ur self for a bit zawawa,have followed this project from beginning and starting to get excited.

[FFI2013]

I think claymore and optiminer are both struggling claymore hasn't released a upgrade in a few and optiminer got a increase in speed because he dropped his dev fee I thing he had it a 10% at one point now it's either 2 or 2.5% so he gained 8% in speed I also think he's running a little over claymore.
Also of course claymore and optiminer both watch these forums to see who's doing what they are competing they'd be idiots not keeping tabs on each other and zawawa

[nerdralph]

Don't get me wrong. I truly admire Claymore's coding skills. To be very honest, I am rarely impressed by other programmers, but he is definitely top notch.

Look at Wolf's code (both OpenCL and gcn assembler).  He's much better than Claymore, and doesn't have the ego issues.  Claymore is more of a persistent but competent hack.  He can experiment with code to see what works, copy from others, but he lacks the fundamental understanding to know what is most likely to give the best performance.  An example of that is his recent comment about using "assembly tricks" to improve the performance of his other miners.  At least for Tonga and Polaris, you can't make an ethereum miner significantly faster than the open source options like Genoil's and Wolf's (in sgminer-gm).

p.s. and his inability to write a decent asm ZEC kernel for Tonga weeks after Optiminer already did it is another example.  So if you are going to try to follow a closed-source miner dev, go with Optiminer, not Claymore.

[nerdralph]

I think claymore and optiminer are both struggling claymore hasn't released a upgrade in a few and optiminer got a increase in speed because he dropped his dev fee I thing he had it a 10% at one point now it's either 2 or 2.5% so he gained 8% in speed I also think he's running a little over claymore.

Optiminer is finished for the Tonga, Pitcairn and Polaris kernels; there's virtually no more performance to be gained in software.  Hawaii can still be tweaked for another boost of about 10%, by splitting the row counters between L2 and GDS as I previously mentioned.

[zawawa]

Don't get me wrong. I truly admire Claymore's coding skills. To be very honest, I am rarely impressed by other programmers, but he is definitely top notch.

Look at Wolf's code (both OpenCL and gcn assembler).  He's much better than Claymore, and doesn't have the ego issues.  Claymore is more of a persistent but competent hack.  He can experiment with code to see what works, copy from others, but he lacks the fundamental understanding to know what is most likely to give the best performance.  An example of that is his recent comment about using "assembly tricks" to improve the performance of his other miners.  At least for Tonga and Polaris, you can't make an ethereum miner significantly faster than the open source options like Genoil's and Wolf's (in sgminer-gm).

p.s. and his inability to write a decent asm ZEC kernel for Tonga weeks after Optiminer already did it is another example.  So if you are going to try to follow a closed-source miner dev, go with Optiminer, not Claymore.

You may or may not be right. I cannot tell because I don't have the extra information you got through reverse-engineering, which I personally don't do. Character analysis of other devs is an interesting subject as it is close to what I do professionally, but I would rather code at this point. I am curious about Wolf's codes, though. I didn't have a chance to take a close look at his codes yet, but he seems to know what he is doing.

As for LLVM, it turned out that the most recent version in the repo has some serious issues with code generation, so I am reverting back to LLVM 3.9.0, which was presumably used for Crimson 16.9.2. Hope this one works better...

[zawawa]

So, I decided to keep modifications to LLVM to the minimum and use an external tool to modify the assembly source code for CLRX instead. This could be turned into a GCN assembly code analyzer/optimizer, which would be a fun project for the future.

[jstefanop]

So, I decided to keep modifications to LLVM to the minimum and use an external tool to modify the assembly source code for CLRX instead. This could be turned into a GCN assembly code analyzer/optimizer, which would be a fun project for the future.

Lol sounds like your building a ground up GCN software stack....quite the project from some custom ZEC ASM kernels

[nerdralph]

You may or may not be right. I cannot tell because I don't have the extra information you got through reverse-engineering, which I personally don't do.

I've tested Claymore's miners since his first eth miner, and already had a good idea about his coding skills and personality before I dumped his ZEC miner kernel.  The last time I dumped his ZEC miner kernel, it just confirmed what I already knew based on the performance; it wasn't using GDS.  In other words, you don't need to disassemble anyone's code to get a good idea of how it works when you understand what it needs to do (i.e. the mining algorithm), and you understand the fundamental capabilities of the hardware it is running on.

As for LLVM, it turned out that the most recent version in the repo has some serious issues with code generation, so I am reverting back to LLVM 3.9.0, which was presumably used for Crimson 16.9.2. Hope this one works better...

I've been using 3.9.1 since I started experimenting with inline asm.

Code:

clang --version
clang version 3.9.1-svn288847-1~exp1 (branches/release_39)
Target: x86_64-unknown-linux-gnu

[nerdralph]

I am curious about Wolf's codes, though. I didn't have a chance to take a close look at his codes yet, but he seems to know what he is doing.

Last year he posted about his private asm eth miner, and gave some technical details about LDS use and GLC memory reads that made it immediately evident to me he knew what he was talking about.  Also, despite writing a private miner, he was willing to share and discuss optimizations with other devs.  Shortly after that I started perusing his github account.
https://github.com/wolf9466/

[zawawa]

I can't wait to go back to miner programming... I was chasing down some pretty weird bugs in LLVM and libclc yesterday, which convinced me that I must have my own fork of LLVM. This is probably an overkill for a ZEC miner, but I always wanted to have my own compiler anyway. It is pretty stable now, so I should be able to go back to GG today,

[m1n1ngP4d4w4n]

I can't wait to go back to miner programming... I was chasing down some pretty weird bugs in LLVM and libclc yesterday, which convinced me that I must have my own fork of LLVM. This is probably an overkill for a ZEC miner, but I always wanted to have my own compiler anyway. It is pretty stable now, so I should be able to go back to GG today,

 Well before having good product you need good tools to build them, if that will help you getting things out faster & better it was the right thing to do and you probably learned alot of stuff , good luck !

[woodaxe]

ok the  good news is i can run your miner mining xmr and its close to claymores miner with the advantage its pulling 100w less at the wall with 6 x sapphire rx470 +nitro OC  8 gig  still a little unstable atm but getting there .

i still cannot mine ethereum this is my bat file  any help would be great as hoping it will also pull less watts at the wall

gatelessgate.exe --gpu-platform 0 -k ethash -o stratum+tcp://eth-eu.dwarfpool.com:8008 -u mywalletaddy -p x --gpu-threads 1 --worksize 256 --xintensity 1024 --no-extranonce
pause

also dont fully understand the worksize settings  intensity etc

would i be right in thinking that any improvements you make to the zcash miner will aslo show on the xmr  miner as well

many thanks for your work

[jstefanop]

would i be right in thinking that any improvements you make to the zcash miner will aslo show on the xmr  miner as well

many thanks for your work

No...each algorithm is different and any optimization to one has zero effect on the other. Keep in mind this miner is just a fork of sgminer, which is a GPU miner that has dozens of different algorithms bundled in it.

[zorvalth]

ok the  good news is i can run your miner mining xmr and its close to claymores miner with the advantage its pulling 100w less at the wall with 6 x sapphire rx470 +nitro OC  8 gig  still a little unstable atm but getting there .

i still cannot mine ethereum this is my bat file  any help would be great as hoping it will also pull less watts at the wall

gatelessgate.exe --gpu-platform 0 -k ethash -o stratum+tcp://eth-eu.dwarfpool.com:8008 -u mywalletaddy -p x --gpu-threads 1 --worksize 256 --xintensity 1024 --no-extranonce
pause

also dont fully understand the worksize settings  intensity etc

would i be right in thinking that any improvements you make to the zcash miner will aslo show on the xmr  miner as well

many thanks for your work

Ethereum mining works fine but its way slower then other ethereum miners as Claymore or genoil. On RX instead of 29mh i got 25.3mh, on r9 380 is worse proportion.

[elgi76]

Ethereum mining works fine but its way slower then other ethereum miners as Claymore or genoil. On RX instead of 29mh i got 25.3mh, on r9 380 is worse proportion.

for me :
sgminer or gateless 161 MH and 960 W
claymore eth miner 166 MH and 1200 W

4 x rx470 and 2 x r9 fury