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

[zawawa]

It's true that GG's Ethash kernel performs better on Windows.
I just started with optimizations anyway.

So, respectfully, what ever happened to Zcash? We still don't have a first-rate open source miner. All these other algos already have decent open source miners available to them. Just sayin'.

Oh, I haven'y given up on that yet.
You have to understand, though, that the competition for Zcash miners is much more intense than those for other miners, and my resources are quite limited. It seems to me that people assume that first-class open-source miners just automagically happen, but that's simply not the case.

[1714803160]

1714803160

[1714803160]

1714803160

[1714803160]

1714803160

[1714803160]

1714803160

[zawawa]

At any rate, I will have to rethink the financial aspects of open-source miner development.
Sadly, I was making much more money when I was developing private kernels just for myself.

[sp_]

At any rate, I will have to rethink the financial aspects of open-source miner development.
Sadly, I was making much more money when I was developing private kernels just for myself.

Claymore is making more than $50 000 a month... And you seem to be working for nothing.

[zawawa]

At any rate, I will have to rethink the financial aspects of open-source miner development.
Sadly, I was making much more money when I was developing private kernels just for myself.

Claymore is making more than $50 000 a month... And you seem to be working for nothing.

Is that what he's making? No wonder...
I'm doing pretty well competing against him, then. 

[WBF1]

This could be a matter of tweaking, but the new GG ASM kernels reported similar performance locally as Claymore, but at the pool the performance was not as good. Higher rate of stale shares as well. (I also realize are other factors that could have lead to this). I'm talking in terms of 24 hr averages. Lower power consumption than Claymore though, which is nice.

This was on pre4 with the initial ASM stuff dropped in on windows 10. I will test out the pre5 release this week sometime.

[zawawa]

So I had a chance to ask an AMD engineer questions about the Global Data Share:

http://gpuopen.com/using-sub-dword-addressing-on-amd-gpus-with-rocm/

If he could answer my questions, GG' Equihash kernel would get a huge boost in performance.
We will see...

[rickadone]

It's true that GG's Ethash kernel performs better on Windows.
I just started with optimizations anyway.

So, respectfully, what ever happened to Zcash? We still don't have a first-rate open source miner. All these other algos already have decent open source miners available to them. Just sayin'.

Oh, I haven'y given up on that yet.
You have to understand, though, that the competition for Zcash miners is much more intense than those for other miners, and my resources are quite limited. It seems to me that people assume that first-class open-source miners just automagically happen, but that's simply not the case.

“Automagically” I love that Word. No they probably got hyped, you managed to get so many miners and you look like a developer who can achieve this and they do not know the process of coding usually, so they automatically think you will do it but they have no idea how long it will take, they know you can but don’t know when basically.

So, they keep bothering developers, its a common thing in any development job not just miners. From basic html5 games to complex financial software its all the same.

[FFI2013]

I just got a few msi rx470 4gb and love them but I'm used to playing with r9s it seems you guys are running higher rates than I am I'm only running 20 with stock clocks and no mods what is the best way to get higher rates modding the straps than over clocking

[cryptominer420]

Results of pre 5 using the new default bat

Ethpool reports only 24.5MH/s not sure why.

[lexele]

At any rate, I will have to rethink the financial aspects of open-source miner development.
Sadly, I was making much more money when I was developing private kernels just for myself.

Well, you should add optional devfee to your miner( on by default), I'm sure a good percentage of miners would pay.

[zawawa]

At any rate, I will have to rethink the financial aspects of open-source miner development.
Sadly, I was making much more money when I was developing private kernels just for myself.

Well, you should add optional devfee to your miner( on by default), I'm sure a good percentage of miners would pay.

Yeah, somebody posted a configuration file for sgminer for dev fees in this thread a while back. Let me take that up now...

[h311m4n]

Hi Zawawa,

Thanks for pre5, on Win 10, I'm getting 28.5Mh/s on average for ETH on a 6x Sapphire 470 nitro+ 8GB rig, 1750 timing, 1100 core and 2150 memory. On Claymore I get 29 but with the dev fees, I think your miner is now about the same in terms of pure miner performance. What I mean by that, is that I'll have to see what the pool reports after a few hours.

I do get a lot of HW errors though. Also, consumption is through the roof compared to Claymore!

[zawawa]

I actually haven't done anything about power consumption yet.
No worries, I will get to that in no time.

[Tmdz]

HW Errors concern me.

Won't the hw errors cause the gpu to degrade over time?

[zawawa]

HW Errors concern me.

Won't the hw errors cause the gpu to degrade over time?

Yeah, I'm looking into that as well.

[nerdralph]

HW Errors concern me.

Won't the hw errors cause the gpu to degrade over time?

Yeah, I'm looking into that as well.

No, it just means you're pushing it a bit to hard.  It's heat that will cause the GPU (any semiconductor) to degrade over time.
To reduce HW errors, try dropping worksize (or start at xi1024).

[nerdralph]

Took a rig from 4.04 w/ 16.40 to 4.11 w/ 16.60, no speed difference when it comes to claymore (29.5 hynix and 30+ samsung on rx470 4g).

Will test gg next.

I got a small boost (<1%) with sgminer going from 4.8/16.40 to 4.10.5/16.6.  I haven't heard anyone getting a speed boost going to 4.11, so I'm sticking with 4.10.

[zawawa]

HW Errors concern me.

Won't the hw errors cause the gpu to degrade over time?

Yeah, I'm looking into that as well.

No, it just means you're pushing it a bit to hard.  It's heat that will cause the GPU (any semiconductor) to degrade over time.
To reduce HW errors, try dropping worksize (or start at xi1024).

It seems that GG's Ethask kernel is more susceptible to hardware errors, though.
I am trying to figure out how to reduce the work load on the GPU without sacrificing the performance.

[Dr_Voidberg]

This is a very interesting project! But what does S/s mean? How to convert to H/s?

[SunStruck]

This is a very interesting project! But what does S/s mean? How to convert to H/s?

Basically 1:1