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

[zawawa]

I run the same experiment on Ubuntu 16.04.1 LTC + Linux Kernel 4.10.2 with AMDGPU-Pro 16.60 and the difference was even stalker.

The assembly version:



The original ethash-new.cl:



I will go ahead and prepare binaries for the other cards.
I will probably include GCN1 as it really does not take too long.

[WBF1]

I dropped your new files in over top of version pre4 and am also seeing a 1-1.5 mhs boost on my RX 470s and 480. Did not try on R9 390, but it was already performing as good as Claymore ever had.

[laik2]

I run the same experiment on Ubuntu 16.04.1 LTC + Linux Kernel 4.10.2 with AMDGPU-Pro 16.60 and the difference was even stalker.

The assembly version:



The original ethash-new.cl:



I will go ahead and prepare binaries for the other cards.
I will probably include GCN1 as it really does not take too long.

Please, do share your linux xorg.conf and also I see a lot of hw errors for stock GPU...Do you run gatelessgate as root or as user?
I see gpu monitoring temp/fan is working for you...

EDIT: Here are the latest tests...
GG ethash-new.cl copied from binary-kernel dir

GG ethash-new.cl copied from kernel dir

GG ethash-new.cl + compiled binaries from binary-kernel dir

Claymore 8.0 Linux default settings

[jstefanop]

Look like im getting 28.8 vs 29.9 on claymore. Not sure if its running your ASM version though.

[zawawa]

I just pushed to the repo an optimized GCN assembly version of ethash-new.cl for RX 470/480.
Each card should get a 1Mh/s boost with it. If this actually works, then I will extend its support to GCN1/GCN3 devices.
(I sold all of my GCN2 cards a while back...)

That's not optimized - you flipped the SLC and GLC bits, which will likely make it a tad SLOWER; it did when I tried that.

I was expecting just SLC (bypass L2) to help, though I recall Wolf's comments about GLC (bypass L1) actually helping.  I'd even expect GLC to hurt performance if you weren't very careful to ensure data was read in 64-byte chunks.

p.s.  There's also some easy optimizations to do with instruction reordering (though they might not make much difference in performance).  For example:

Code:

/*d11c6a3e 01a9013c*/ v_addc_u32      v62, vcc, v60, 0, vcc
/*2a7e62b2         */ v_xor_b32       v63, 50, v49
/*dc5c0000 4000003d*/ flat_load_dwordx4 v[64:67], v[61:62] slc glc
/*dc5c0000 3b00003b*/ flat_load_dwordx4 v[59:62], v[59:60] slc glc
/*bf8c0171         */ s_waitcnt       vmcnt(1) & lgkmcnt(1)

The v_xor_b32 can be moved to after the flat_load_dwordx4.

That's a good catch. I was actually thinking about automating this kind of instruction reordering.
My compiler driver rewrites the output of LLVM/Clang, so it shouldn't be that difficult.
I really want to combine this feature with register usage analysis.

[zawawa]

Look like im getting 28.8 vs 29.9 on claymore. Not sure if its running your ASM version though.

The ASM version is only for Ellesmere for now. I will prepare binaries for other cards shortly.

[laik2]

Look like im getting 28.8 vs 29.9 on claymore. Not sure if its running your ASM version though.

The ASM version is only for Ellesmere for now. I will prepare binaries for other cards shortly.

Can you confirm that I'm doing it right?
There are 2 folders
kernels(default kernels of sgminer and equihash.cl)
binary-kernels( your opencl binary kernels + ethash-new.cl using asm)
I think copying kernels from binary-kernel folder to main gatelessgate install directory uses your last kernel...
I'm used to old sgminer style, all kernels in same folder with executable, but I might be wrong...

[jstefanop]

Look like im getting 28.8 vs 29.9 on claymore. Not sure if its running your ASM version though.

The ASM version is only for Ellesmere for now. I will prepare binaries for other cards shortly.

Yea I have Ellsemere cards but its loading the normal .bin kernels. I did remove the pre-built .bin and built new kernal binaries using ethash-new.cl in /binary-kernal by moving it to /kernal and that bumped up the speed to 29.2, but still a bit short from claymore.

[laik2]

Look like im getting 28.8 vs 29.9 on claymore. Not sure if its running your ASM version though.

The ASM version is only for Ellesmere for now. I will prepare binaries for other cards shortly.

Yea I have Ellsemere cards but its loading the normal .bin kernels. I did remove the pre-built .bin and built new kernal binaries using ethash-new.cl in /binary-kernal by moving it to /kernal and that bumped up the speed to 29.2, but still a bit short from claymore.

For me it's quite the opposite...using 2 threads gains additional +0.3/0.4 but too many hw errors.
As you can see in the picture 29.42(1 thread) does gg vs 29.8 on claymore vs 30 (2 threads gg).
EDIT: Did some timings mods and now hw errors are acceptable ~1/2 every minute.

[zawawa]

Look like im getting 28.8 vs 29.9 on claymore. Not sure if its running your ASM version though.

The ASM version is only for Ellesmere for now. I will prepare binaries for other cards shortly.

Yea I have Ellsemere cards but its loading the normal .bin kernels. I did remove the pre-built .bin and built new kernal binaries using ethash-new.cl in /binary-kernal by moving it to /kernal and that bumped up the speed to 29.2, but still a bit short from claymore.

The current version was working fine with Ubuntu, though. Strange, strange...
I just pushed a fix to the repo anyway.

[laik2]

Look like im getting 28.8 vs 29.9 on claymore. Not sure if its running your ASM version though.

The ASM version is only for Ellesmere for now. I will prepare binaries for other cards shortly.

Yea I have Ellsemere cards but its loading the normal .bin kernels. I did remove the pre-built .bin and built new kernal binaries using ethash-new.cl in /binary-kernal by moving it to /kernal and that bumped up the speed to 29.2, but still a bit short from claymore.

The current version was working fine with Ubuntu, though. Strange, strange...
I just pushed a fix to the repo anyway.

It is
It's currently hashing with constant 30Mh vs. 29.7 with claymore. But I still have to copy binary-kernel/* to $gatelessgate_install_dir/bin

I am still wondering how your temp/fans were working but mine are not...Did you start gg with root priv and mining GPU as main one?

[sp_]

If you manage to beat claymore zawawa, thats impressive work.

[jstefanop]

Look like im getting 28.8 vs 29.9 on claymore. Not sure if its running your ASM version though.

The ASM version is only for Ellesmere for now. I will prepare binaries for other cards shortly.

Yea I have Ellsemere cards but its loading the normal .bin kernels. I did remove the pre-built .bin and built new kernal binaries using ethash-new.cl in /binary-kernal by moving it to /kernal and that bumped up the speed to 29.2, but still a bit short from claymore.

The current version was working fine with Ubuntu, though. Strange, strange...
I just pushed a fix to the repo anyway.

It does load the .bins in binary-kernal, but how do you use the asm file? (ethash-new-gcn3-ocl20.asm)

[jstefanop]

Look like im getting 28.8 vs 29.9 on claymore. Not sure if its running your ASM version though.

The ASM version is only for Ellesmere for now. I will prepare binaries for other cards shortly.

Yea I have Ellsemere cards but its loading the normal .bin kernels. I did remove the pre-built .bin and built new kernal binaries using ethash-new.cl in /binary-kernal by moving it to /kernal and that bumped up the speed to 29.2, but still a bit short from claymore.

For me it's quite the opposite...using 2 threads gains additional +0.3/0.4 but too many hw errors.
As you can see in the picture 29.42(1 thread) does gg vs 29.8 on claymore vs 30 (2 threads gg).
EDIT: Did some timings mods and now hw errors are acceptable ~1/2 every minute.

Your running 4.10 kernel...im on the stock 4.4. I think thats where the difference is from.

[laik2]

Look like im getting 28.8 vs 29.9 on claymore. Not sure if its running your ASM version though.

The ASM version is only for Ellesmere for now. I will prepare binaries for other cards shortly.

Yea I have Ellsemere cards but its loading the normal .bin kernels. I did remove the pre-built .bin and built new kernal binaries using ethash-new.cl in /binary-kernal by moving it to /kernal and that bumped up the speed to 29.2, but still a bit short from claymore.

For me it's quite the opposite...using 2 threads gains additional +0.3/0.4 but too many hw errors.
As you can see in the picture 29.42(1 thread) does gg vs 29.8 on claymore vs 30 (2 threads gg).
EDIT: Did some timings mods and now hw errors are acceptable ~1/2 every minute.

Your running 4.10 kernel...im on the stock 4.4. I think thats where the difference is from.

https://drive.google.com/drive/folders/0B72yKpOokCMcVnV5LWNMS2ltYmM
I've uploaded some of my kernels. 4.10/4.11 are tested and working fine.
Just remember to update only opencl packages from amdgpu-pro 16.60.
Ditto about the asm ...

[SunStruck]

Look like im getting 28.8 vs 29.9 on claymore. Not sure if its running your ASM version though.

The ASM version is only for Ellesmere for now. I will prepare binaries for other cards shortly.

Yea I have Ellsemere cards but its loading the normal .bin kernels. I did remove the pre-built .bin and built new kernal binaries using ethash-new.cl in /binary-kernal by moving it to /kernal and that bumped up the speed to 29.2, but still a bit short from claymore.

The current version was working fine with Ubuntu, though. Strange, strange...
I just pushed a fix to the repo anyway.

got an update from the suprnova front.

just had a similar problems with claymore miner and he has a special mode to mine on suprnova

Code:

-esm	Ethereum Stratum mode. 0 - eth-proxy mode (for example, dwarpool.com), 1 - qtminer mode (for example, ethpool.org), 
	2 - miner-proxy mode (for example, coinotron.com), 3 - nicehash mode. 0 is default. 

And for suprnova it is the nicehash mode  (-esm 3)

Maybe that helps digging further why gg-miner doesnt work with it.

[zawawa]

Ah, I see... There is a patch for sgminer-gm and nicehash. That should work with suprnova, then.

[lupanar]

Hi, can I mine with GG some forks of ETH or only ETH is supported? And GG it's for both AMD & NVIDIA cards?

[zawawa]

Hi, can I mine with GG some forks of ETH or only ETH is supported? And GG it's for both AMD & NVIDIA cards?

Somebody reported in this thread it's possible.

[lupanar]

Hi, can I mine with GG some forks of ETH or only ETH is supported? And GG it's for both AMD & NVIDIA cards?

Somebody reported in this thread it's possible.

Probably not every ETH fork can be mine by GG, trying ti mine some fork with gtx1060 and getting an error

[16:17:33] Probing for an alive pool
[16:17:34] Startup GPU initialization... Using settings from pool 213.154.202.174.
[16:17:34] Startup Pool No = 0
[16:17:34] 213.154.202.174 difficulty changed to 0.466
[16:17:34] Rejected untracked stratum share from 213.154.202.174
[16:17:35] Initialising kernel ethash-new.cl with nfactor 10, n 1024
[16:17:35] Initialising kernel ethash-new.cl with nfactor 10, n 1024
[16:17:36] Initialising kernel ethash-new.cl with nfactor 10, n 1024
[16:17:36] Initialising kernel ethash-new.cl with nfactor 10, n 1024
[16:17:37] Initialising kernel ethash-new.cl with nfactor 10, n 1024
[16:17:37] Initialising kernel ethash-new.cl with nfactor 10, n 1024
[16:17:38] Initialising kernel ethash-new.cl with nfactor 10, n 1024
[16:17:39] Initialising kernel ethash-new.cl with nfactor 10, n 1024
[16:17:49] Waiting for work to be available from pools.
[16:17:55] Rejected untracked stratum share from 213.154.202.174
[16:18:58] Rejected untracked stratum share from 213.154.202.174

So what is the problem can be with it?