[ANN] sgminer v5 - optimized X11/X13/NeoScrypt/Lyra2RE/etc. kernel-switch miner - page 192.

Eastwind

hero member

Activity: 896

Merit: 1000

Quote from: badman74 on July 17, 2014, 09:18:34 PM

ok i think i have the fix
modded the groestl.cl file a little and set #define SPH_LUFFA_PARALLEL 1, and re-enabled #include "aes_helper.cl" in darkcoin-mod.cl
getting 6mh/s no real changes in the other algo's that i can see
darkcoin-mod.cl
https://www.dropbox.com/s/gqv2m7y62olfzbs/darkcoin-mod.cl
groestl.cl
https://www.dropbox.com/s/9ugaly2utnwbda2/groestl.cl
EDIT: there is still a little something different as i am only getting 5.96 instead of the 6.04 i was getting with the other kernel

I tried those two files.

There are 20% increase of nist5 hash with 10% increase of power consumption.
But for X11, there is only 1-3% increase of hash, with 10% increase of power consumption.

This is compared to the sgminer 4.1 by djm34.

I use 7990, 7970 and 7950.

badman74

hero member

Activity: 658

Merit: 500

Quote from: platinum4 on July 18, 2014, 05:13:46 AM

Quote from: phzi on July 18, 2014, 05:08:47 AM

Quote from: platinum4 on July 18, 2014, 04:55:23 AM

clone from mrdbro_testing or _develop, report back if you can

Just pulled and compiled from mrbrdo_testing and develop branches. ~3.1MH/s from both (although they do seem to both provide a very very marginal hashrate increase, in the order of a few KH/s).

Only other thing I can think of is that my 4.44MH/s bin was generated on an older version of sgminer. I'll start stepping back along the revisions to see if I can get the hashrate up again.

Interestingly, hashrate is dropping to ~2.8MH/s after running for a short period. I ran into this a lot (hashrate drops until you reboot) until I found the settings above which offered 4.5MH/s stable (with an occasional card seemingly randomly deciding to run at 4.6+MH/s).

When xI 256 and worksize 128 was giving me 4.5MH/s, changing xI or worksize slightly (up or down 1 or 2) or a lot (to say worksize 64) killed hashrate (below 4MH/s). Now, adjusting either barely affects hashrate (remains somewhere around 3MH/s).

idk, i use Windows binaries, wait til badman74 wakes up he may have some guidance for you.

My suggestion would be to try replacing your darkcoin-mod.cl and groestl.cl with the ones I posted last night and see if it will work
I am getting 5mh/s on my non x 290's on my pimp miner
If that doesn't work for you, try cloning aznboy84's miner from https://github.com/aznboy84/sgminer/tree/v5_0-x15 and see if it works
Edit: git clone --recursive -b v5_0-x15 git://github.com/aznboy84/sgminer/sgminer.git

platinum4

sr. member

Activity: 547

Merit: 250

Quote from: Bojcha on July 18, 2014, 08:30:06 AM

Why i cannot put this in my conf ?

"kernel" : "bitblock,bitblock,bitblockold,bitblockold"
or
"algorithm" : "bitblock,bitblock,bitblockold,bitblockold"

it crates wrong kernel/bin. >

Not available yet.
https://github.com/sgminer-dev/sgminer/issues/311

Bojcha

hero member

Activity: 848

Merit: 500

Why i cannot put this in my conf ?

"kernel" : "bitblock,bitblock,bitblockold,bitblockold"
or
"algorithm" : "bitblock,bitblock,bitblockold,bitblockold"

it crates wrong kernel/bin. >

yaamp

full member

Activity: 175

Merit: 100

Quote from: merc82 on May 30, 2014, 09:09:37 PM

Spouting off some random thinking here, how about a stratum protocol amendment -- introducing an algo definition on connection and allowing algo changes to be initiated from the remote stratum.. that'd be fantabulous for us multipools who run multiple algos Wink

That would be great to let the pool decide which algorithm is the best to mine without having to reconnect. Either using a new function like mining.set_algorithm or directly as a extra parameter to the mining.notify method.

platinum4

sr. member

Activity: 547

Merit: 250

Quote from: phzi on July 18, 2014, 05:08:47 AM

Quote from: platinum4 on July 18, 2014, 04:55:23 AM

clone from mrdbro_testing or _develop, report back if you can

Just pulled and compiled from mrbrdo_testing and develop branches. ~3.1MH/s from both (although they do seem to both provide a very very marginal hashrate increase, in the order of a few KH/s).

Only other thing I can think of is that my 4.44MH/s bin was generated on an older version of sgminer. I'll start stepping back along the revisions to see if I can get the hashrate up again.

Interestingly, hashrate is dropping to ~2.8MH/s after running for a short period. I ran into this a lot (hashrate drops until you reboot) until I found the settings above which offered 4.5MH/s stable (with an occasional card seemingly randomly deciding to run at 4.6+MH/s).

When xI 256 and worksize 128 was giving me 4.5MH/s, changing xI or worksize slightly (up or down 1 or 2) or a lot (to say worksize 64) killed hashrate (below 4MH/s). Now, adjusting either barely affects hashrate (remains somewhere around 3MH/s).

idk, i use Windows binaries, wait til badman74 wakes up he may have some guidance for you.

phzi

hero member

Activity: 700

Merit: 500

Quote from: platinum4 on July 18, 2014, 04:55:23 AM

clone from mrdbro_testing or _develop, report back if you can

Just pulled and compiled from mrbrdo_testing and develop branches. ~3.1MH/s from both (although they do seem to both provide a very very marginal hashrate increase, in the order of a few KH/s).

Only other thing I can think of is that my 4.5MH/s bin was generated on an older version of sgminer. I'll start stepping back along the revisions to see if I can get the hashrate up again.

Interestingly, hashrate is dropping to ~2.8MH/s after running for a short period. I ran into this a lot (hashrate drops until you reboot) until I found the settings above which offered 4.5MH/s stable (with an occasional card seemingly randomly deciding to run at 4.6+MH/s).

When xI 256 and worksize 128 was giving me 4.5MH/s, changing xI or worksize slightly (up or down 1 or 2) or a lot (to say worksize 64) killed hashrate (below 4MH/s). Now, adjusting either barely affects hashrate (remains somewhere around 3MH/s).

platinum4

sr. member

Activity: 547

Merit: 250

Quote from: phzi on July 18, 2014, 04:51:42 AM

Quote from: platinum4 on July 18, 2014, 04:48:28 AM

idk man, if you had something working before, then try that again.

if you have it all fucked up, save your conf files, and blast it all, then recompile and try again.

Ya, this is what's messing with me... I nuked sgminer, cloned from v5_0 git, re-compiled, and... 3.11MH/s. Exact same config as was giving me 4.5MH/s.

Boggling my mind... nothing else has changed that I can think of.

I dropped 2 cl files in to test, and each time I launched sgminer after the hashrate dropped (as if there were memleaks or something). Rebooting, re-compiling, hard reboot (shutdown, wait a while, start again)... still 3.11MH/s.

clone from mrdbro_testing or _develop, report back if you can

phzi

hero member

Activity: 700

Merit: 500

Quote from: platinum4 on July 18, 2014, 04:48:28 AM

idk man, if you had something working before, then try that again.

if you have it all fucked up, save your conf files, and blast it all, then recompile and try again.

Ya, this is what's messing with me... I nuked sgminer, cloned from v5_0 git, re-compiled, and... 3.11MH/s. Exact same config as was giving me 4.5MH/s.

Boggling my mind... nothing else has changed that I can think of.

I dropped 2 cl files in to test, started sgminer and got 3.11MH/s; and, each time I quit and launched sgminer after the hashrate dropped (as if there were memleaks or something). Rebooting, re-compiling, hard reboot (shutdown, wait a while, start again)... still 3.11MH/s.

The only thing I can think is that my 4.5MH/s bin file was generated on a slightly older sgminer codebase (but I was running it in the latest sgminer compile). Screw me for not copying the .bin before deleting it.

platinum4

sr. member

Activity: 547

Merit: 250

idk man, if you had something working before, then try that again.

if you have it all fucked up, save your conf files, and blast it all, then recompile and try again.

phzi

hero member

Activity: 700

Merit: 500

Quote from: platinum4 on July 18, 2014, 04:25:31 AM

Quote from: phzi on July 18, 2014, 04:07:38 AM

No idea what the heck I did...

I was pulling 4.5MH/s for X11 on each R9 290 before today (with fairly conservative clocks too). Tried some of the cl modifications in this thread, deleted existing bins, and my hashrate dropped.
Even cloned and compiled a fresh copy of sgminer. Hashrate is maxing out at 3.11MH/s now no matter what I do.

It's a Gentoo Linux rig with 14.6 ati-drivers.

This profile was working great for me before (stable 4.5MH/s ... ran for days without issue)

Code:

 {
                "name" : "X11",
                "algorithm" : "darkcoin-mod",
                "xintensity" : "256",
                "gpu-threads" : "1",
                "gpu-powertune" : "1",
                "worksize": "128",
                "gpu-engine": "980-980",
                "gpu-memclock": "1300",
                "gpu-fan": "60"
        },

What the heck did I do, lol?

gpu-threads:2
xinstensity:50

remove auto-gpu
set one frequency for gpu-engine

enable auto-fan
set fan range from 60-100

Worse. 2.7MH/s max with those settings. If I drop xI to 32 I can get 2.8MH/s. And... one card just went SICK VERY fast.

Code:

        {
                "name" : "X11",
                "algorithm" : "darkcoin-mod",
                "xintensity" : "50",
                "gpu-threads" : "2",
                "gpu-powertune" : "10",
                "worksize": "128",
                "gpu-engine": "980",
                "gpu-memclock": "1300",
                "gpu-fan": "60-100",
                "auto-fan" : true,
                "auto-gpu" : false
        },

It shouldn't be a configuration issue... I had 4.5MH/s rock solid with the settings above up until I had to re-compile sgminer...

Tried a few different GCC versions even, same thing.

---

Edit: these cards simply do not like gpu-threads: 2. It's ALWAYS worse performance for me. Using the exact settings your suggest, but gpu-threads: 1 results in 2.95MH/s.

Auto-fan is also unneccesary for me, as the cards never go over 70.0C at 60% fan (so, auto-fan will never ramp up the fans anyway).

platinum4

sr. member

Activity: 547

Merit: 250

Quote from: Eastwind on July 18, 2014, 03:03:08 AM

Quote from: ?? on ??

Working config for scrypt/nscrypt/x11/x13/x14/x15/keccak/nist5/qubit algos on 3x 290X (shaders 2816) rig on CGWatcher 1.4.1, W7x64 CCC 14.7RC
updated with badman74's advice on temperature targets; overclocks with lower intensities to avoid driver failures; removed redundant settings, globally specified "gpu-threads" : "2", and removed redundant gpu-threads settings on relevant profiles; disabled ULPS in MSI AB; substituted 65KB groestl.cl in place of 67KB groestl.cl; #define SPH_LUFFA_PARALLEL 1 in darkcoin-mod.cl, marucoin-mod.cl, x14.cl, bitblock.cl; #include "aes_helper.cl" in 39KB darkcoin-mod.cl; dropped worksize to 64 like bullus; recalibrated speedfactors for nicehash password field.

Do you use the 65k or 67kb version?

Is it possible to upload the modified files?

http://pastebin.com/0aia0VXU

or

https://bitcointalksearch.org/topic/m.7902608

badman has 2 loops changed at the end of his groestl.cl though

platinum4

sr. member

Activity: 547

Merit: 250

Quote from: phzi on July 18, 2014, 04:07:38 AM

No idea what the heck I did...

I was pulling 4.5MH/s for X11 on each R9 290 before today (with fairly conservative clocks too). Tried some of the cl modifications in this thread, deleted existing bins, and my hashrate dropped.
Even cloned and compiled a fresh copy of sgminer. Hashrate is maxing out at 3.11MH/s now no matter what I do.

It's a Gentoo Linux rig with 14.6 ati-drivers.

This profile was working great for me before (stable 4.5MH/s ... ran for days without issue)

Code:

 {
                "name" : "X11",
                "algorithm" : "darkcoin-mod",
                "xintensity" : "256",
                "gpu-threads" : "1",
                "gpu-powertune" : "1",
                "worksize": "128",
                "gpu-engine": "980-980",
                "gpu-memclock": "1300",
                "gpu-fan": "60"
        },

What the heck did I do, lol?

gpu-threads:2
xinstensity:50

remove auto-gpu
set one frequency for gpu-engine

enable auto-fan
set fan range from 60-100

phzi

hero member

Activity: 700

Merit: 500

No idea what the heck I did...

I was pulling 4.5MH/s for X11 on each R9 290 before today (with fairly conservative clocks too). Tried some of the cl modifications in this thread, deleted existing bins, and my hashrate dropped.
Even cloned and compiled a fresh copy of sgminer. Hashrate is maxing out at 3.11MH/s now no matter what I do.

It's a Gentoo Linux rig with 14.6 ati-drivers.

This profile was working great for me before (stable 4.5MH/s ... ran for days without issue)

Code:

 {
                "name" : "X11",
                "algorithm" : "darkcoin-mod",
                "xintensity" : "256",
                "gpu-threads" : "1",
                "gpu-powertune" : "1",
                "worksize": "128",
                "gpu-engine": "980-980",
                "gpu-memclock": "1300",
                "gpu-fan": "60"
        },

What the heck did I do, lol?

Eastwind

hero member

Activity: 896

Merit: 1000

Quote from: ?? on ??

Working config for scrypt/nscrypt/x11/x13/x14/x15/keccak/nist5/qubit algos on 3x 290X (shaders 2816) rig on CGWatcher 1.4.1, W7x64 CCC 14.7RC
updated with badman74's advice on temperature targets; overclocks with lower intensities to avoid driver failures; removed redundant settings, globally specified "gpu-threads" : "2", and removed redundant gpu-threads settings on relevant profiles; disabled ULPS in MSI AB; substituted 65KB groestl.cl in place of 67KB groestl.cl; #define SPH_LUFFA_PARALLEL 1 in darkcoin-mod.cl, marucoin-mod.cl, x14.cl, bitblock.cl; #include "aes_helper.cl" in 39KB darkcoin-mod.cl; dropped worksize to 64 like bullus; recalibrated speedfactors for nicehash password field.

Do you use the 65k or 67kb version?

Is it possible to upload the modified files?

poiuty

full member

Activity: 196

Merit: 100

badman74, thx!
Now my R9 290 non-x ~5600Kh/s

Quote

"kernel": "darkcoin-mod",
"xintensity" : "50",
"gpu-threads" : "2",
"lookup-gap" : "2",
"worksize" : "128",
"gpu-powertune" : "0",
"gpu-engine" : "1100",
"gpu-memclock" : "1500",

platinum4

sr. member

Activity: 547

Merit: 250

So I got bored, and I read through all of the comments on each .cl file that was part of x11/x13/x14/x15.

Here's what I found.

hamsi-expand-big '1' performs similar to hamsi-expand-big '8' with hamsi-short: true
keccak unroll 0 turns into keccak unroll 8 by default, so tried specifying keccak unroll 2 which is for smaller machines

And pretty much no change in hashrate, NO LOSS, but... no gain.

Intensity 17 seems to be the best setting, which is I:17 = 131072 threads

Later on I think I may experiment with xintensities that are between both 16 (drop in hash) and 17, and 17 and 18 (drop in hash)

For right now, no hung rigs on hamsi-expand-big '8', then again, no real difference noticed either.

platinum4

sr. member

Activity: 547

Merit: 250

Quote from: badman74 on July 17, 2014, 09:18:34 PM

ok i think i have the fix
modded the groestl.cl file a little and set #define SPH_LUFFA_PARALLEL 1, and re-enabled #include "aes_helper.cl" in darkcoin-mod.cl
getting 6mh/s no real changes in the other algo's that i can see
EDIT: there is still a little something different as i am only getting 5.96 instead of the 6.04 i was getting with the other kernel

Yeah I've topped out, with bullus's bins I can hit 5.88 stable on my clocks, without having to deal with hung rigs; applying every one of those hacks plus KECCAK UNROLL 1 in marucoin-mod, x14, & bitblock leaves me at near the same speeds I had posted with my configs, so I think once hamsi is added in, the hamsi-short and hamsi_helper take over in terms of performance.

Your groestl.cl changes were kind of cool to try, but the original 2 loops at the end of groestl.cl seem to work better as they are, instead of shortened. I think the key here is to keep the .cl file as small and as short as possible; I wonder if editing out the copyright contents would make for a shorter file? Grin

Can those 2 last loops you altered be pragma unrolled?

platinum4

sr. member

Activity: 547

Merit: 250

Quote from: Undland on July 17, 2014, 11:19:10 PM

Every day my miner is turning off one or two times. Both Windows and Linux versions. Sometimes i losing profit for 6 hours, before i detect offline rigs. Sad

Every day I'm hustlin' hustlin' every every day I'm hustlin'...

Undland

legendary

Activity: 1120

Merit: 1055

Every day my miner is turning off one or two times. Both Windows and Linux versions. Sometimes i losing profit for 6 hours, before i detect offline rigs. Sad

Topic: [ANN] sgminer v5 - optimized X11/X13/NeoScrypt/Lyra2RE/etc. kernel-switch miner - page 192. (Read 877859 times)