Claymore's CryptoNote AMD GPU Miner v11.3

[IojkinKot]

anyone running this miner stable?

Yes, very stable
AMD drivers 17.11.1

Which kind of card do you use ? RX470 ?

Sapphire Nitro+ RX470 8G Samsung

[IojkinKot]

anyone running this miner stable?

Yes, very stable
AMD drivers 17.11.1
http://i98.fastpic.ru/big/2017/1116/aa/a3e944dd3f975b5f1599203aabf8e9aa.png

For your RX 470 used 1750 timing to 2000 ? Or what is best option for this very nice results?

Uber-Mix Extreme v3.1 timings

[Ursul0]

just grab the latest release: https://github.com/genesismining/sgminer-gm/releases
this is what I use for 4*470+1*370

Code:

     "rawintensity": "512,512,512,512,384",
            "worksize": "4,4,4,4,10"

there were no cryptonight improvements for polaris and older cards for a while now... also unreasonable to expect any...

How much hashes from your 470? I managed to squeeze 800 with 650 rawint and worksize 4.

800 is really good. custom straps? at what reject rate?  
I'm getting on 470 refs up to 715 with below 3% rejects and 0 hw errors.
These are hot and like to cook wires:) ... so my top volts are 923 - never above it. normally below 900. some could do just fine on 854 on eth (1080:2000)

I have lots molexes from psu chains from shitty Thermaltakes making crispy noises on move and loosing plastic dust  cheap fucking thieves

[BCMan]

How to fix devfee - connection lost, retry in 20 sec? Also having just connection lost, retry in 20 sec. ALL THE TIME! This is slowing down mining, right? How to fix this crap? Mining at nicehash using v9.7.

I tried this fix, but it doesnt work.
https://github.com/nicehash/NiceHashMinerLegacy/issues/424

[garytheasshole]

How to fix devfee - connection lost, retry in 20 sec? Also having just connection lost, retry in 20 sec. ALL THE TIME! This is slowing down mining, right? How to fix this crap? Mining at nicehash using v9.7.

I tried this fix, but it doesnt work.
https://github.com/nicehash/NiceHashMinerLegacy/issues/424

-nofee 1

[BCMan]

How to fix devfee - connection lost, retry in 20 sec? Also having just connection lost, retry in 20 sec. ALL THE TIME! This is slowing down mining, right? How to fix this crap? Mining at nicehash using v9.7.

I tried this fix, but it doesnt work.
https://github.com/nicehash/NiceHashMinerLegacy/issues/424

-nofee 1

But this will cut down mining speed. How to fix devfee pool connection problem?

[noobminer001]

Haven't been able to get Claymore stable 24hr+ with my RX580 8G rigs. Settings were 1200/2000/1100mV with power stage 5. Stilll reseting.

Since some of you are talking about SGminer...trying to config sgminer 5.5.5 in SMOS for my rigs default is:

--no-submit-stale --kernel cryptonight -o stratum+tcp://cryptonight.usa.nicehash.com:3355 -u [yourwallet].$rigName -p x --rawintensity 1000 -w 8 --text-only

I would like to change the rawintensity, worksize etc to be optimized for my miners. I have two different setups, either RX 570 4G or RX580 8G.

For the RX570 4G I'm trying to pass a --rawintensity 512 --worksize 4. For the RX580 8G, I'm trying to pass --rawintensity 1152 --worksize 8.

I've also seen people change the gpu-threads to 2, though I haven't played with this setting. Has anyone seen a change in performance by changing # GPU threads?

When I do this, I see no change in performance, even when I adjust values. It just shows:

[21:54:08] Startup GPU initialization... Using settings from pool cryptonight.usa.nicehash.com.
[21:54:08] Startup Pool No = 0
[21:54:08] Building binary cryptonightEllesmeregw8l8.bin
Initialising kernel cryptonight.cl with nfactor 10, n 1024.

Right now I'm only hashing at 480H/s on the 570 and 677H/s on the RX580. With Claymore I was getting 630H/s and 740H/s respectively.

Thanks.

[songjuhee]

Haven't been able to get Claymore stable 24hr+ with my RX580 8G rigs. Settings were 1200/2000/1100mV with power stage 5. Stilll reseting.

Since some of you are talking about SGminer...trying to config sgminer 5.5.5 in SMOS for my rigs default is:

--no-submit-stale --kernel cryptonight -o stratum+tcp://cryptonight.usa.nicehash.com:3355 -u [yourwallet].$rigName -p x --rawintensity 1000 -w 8 --text-only

I would like to change the rawintensity, worksize etc to be optimized for my miners. I have two different setups, either RX 570 4G or RX580 8G.

For the RX570 4G I'm trying to pass a --rawintensity 512 --worksize 4. For the RX580 8G, I'm trying to pass --rawintensity 1152 --worksize 8.

I've also seen people change the gpu-threads to 2, though I haven't played with this setting. Has anyone seen a change in performance by changing # GPU threads?

When I do this, I see no change in performance, even when I adjust values. It just shows:

[21:54:08] Startup GPU initialization... Using settings from pool cryptonight.usa.nicehash.com.
[21:54:08] Startup Pool No = 0
[21:54:08] Building binary cryptonightEllesmeregw8l8.bin
Initialising kernel cryptonight.cl with nfactor 10, n 1024.

Right now I'm only hashing at 480H/s on the 570 and 677H/s on the RX580. With Claymore I was getting 630H/s and 740H/s respectively.

Thanks.

leave your voltage automatically... it works for me without reset

[Ed4252]

Does anyone know the fix for the WATCHDOG: GPU 0 hangs in OpenCL call, exit?

using this version with a sapphire r9 380x

Also, my screen will turn black for like 1-2 secs and go back to normal and that's when my h/s goes to zero and has to start up again.

[songjuhee]

Does anyone know the fix for the WATCHDOG: GPU 0 hangs in OpenCL call, exit?

using this version with a sapphire r9 380x

Also, my screen will turn black for like 1-2 secs and go back to normal and that's when my h/s goes to zero and has to start up again.

so did i, before i run claymore from nicehash. you can overclock but leave the voltage.... and everything fine (for me)

[halker2010]

Does anyone know the fix for the WATCHDOG: GPU 0 hangs in OpenCL call, exit?

using this version with a sapphire r9 380x

Also, my screen will turn black for like 1-2 secs and go back to normal and that's when my h/s goes to zero and has to start up again.

I had this issue with asus version of the same GPU the way i fixed it was to lower the intensity and disable and re enable the gpu from device manager.

[chschlumpf]

Guys, i  need your help.

I have 8x Sapphire RX480 8GB Micron GPU with custom BIOS. Dualmining with 198MH/s ETH, 3950H/s SIA. Settings: 1150;875 2000;910. Virtual Memory: 60GB.
Now I tried do start ETN with this settings. But only 3 Cards are working.

00:23:28:915   1d50   XMR - Total Speed: 2286 H/s, Total Shares: 32, Rejected: 0, Time: 00:00
00:23:28:917   1d50   XMR: GPU0 767 H/s, GPU1 773 H/s, GPU2 746 H/s, GPU3 0 H/s, GPU4 0 H/s, GPU5 0 H/s, GPU6 0 H/s

What I have to change, that this will work with 7 Cards?

[Ed4252]

Does anyone know the fix for the WATCHDOG: GPU 0 hangs in OpenCL call, exit?

using this version with a sapphire r9 380x

Also, my screen will turn black for like 1-2 secs and go back to normal and that's when my h/s goes to zero and has to start up again.

I had this issue with asus version of the same GPU the way i fixed it was to lower the intensity and disable and re enable the gpu from device manager.

how do you lower the intensity?

[mofofofo]

I disabled a single GPU in my config (I have 5 GPUs that will be on Claymore).  My configs only seem to apply to my first GPU (#0); all others GPUs run at stock.  I've tried comma-separated values for each GPU and this does not work either.  Is there another way of applying configs to all GPUs (obviously not the disabled one)?

Applicable configs...

Code:

;disabling GPU3
-di 01245

-ftime 10,10,10,10,10
-tstop 85,85,85,85,85
-tt 65,65,65,65,65
-fanmin 60,60,60,60,60
-fanmax 100,100,100,100,100
-cclock 1250,1250,1250,1250,1250
-mclock 2250,2250,2250,2250,2250
-cvddc 875,875,875,875,875
-mvddc 850,850,850,850,850

-dmem 0

;-h 1800

#-allpools 1
#-tt 75

[sir4o]

just grab the latest release: https://github.com/genesismining/sgminer-gm/releases
this is what I use for 4*470+1*370

Code:

     "rawintensity": "512,512,512,512,384",
            "worksize": "4,4,4,4,10"

there were no cryptonight improvements for polaris and older cards for a while now... also unreasonable to expect any...

How much hashes from your 470? I managed to squeeze 800 with 650 rawint and worksize 4.

800 is really good. custom straps? at what reject rate?  
I'm getting on 470 refs up to 715 with below 3% rejects and 0 hw errors.
These are hot and like to cook wires:) ... so my top volts are 923 - never above it. normally below 900. some could do just fine on 854 on eth (1080:2000)

I have lots molexes from psu chains from shitty Thermaltakes making crispy noises on move and loosing plastic dust  cheap fucking thieves

0 hw errors for me also. Watts are quite high though (measured from the wall ofc), compared to 10.2 it's 23% more. And both (sgminer and 10.2) are running with exactly the same frequencies. Haven't noticed any rejects, but as cryptonight is not so heavily dependant on mem freq., I'm not overclocking the memory too much (1850). Reverted back to 10.2 as better hashrate at less wattage. Even with the devfee is still better for me.

[Call_Me_Bambi]

How to fix devfee - connection lost, retry in 20 sec? Also having just connection lost, retry in 20 sec. ALL THE TIME! This is slowing down mining, right? How to fix this crap? Mining at nicehash using v9.7.

I tried this fix, but it doesnt work.
https://github.com/nicehash/NiceHashMinerLegacy/issues/424

-nofee 1

But this will cut down mining speed. How to fix devfee pool connection problem?

It may cut down the mining speed a little but this is could be offset by the time lost during the switch from your pool to the dev pool and back again to your pool.
Someone, way back, did a calculation (it was on the Claymore Ethash miner) and worked out that you did not lose any overall hashrate by using '-nofee 1' as it took so long to load/reload the DAG that it was just as well not to pay the fee, it could be the same for the Cryptonote miner but I don't know.

[Darkbane]

Sooo I have a bit of an unusual situation, almost nobody uses workstation GPU's to mine with, my project is over and now I have a "Radeon Pro Duo (polaris)" sitting idle doing nothing at all, it gets recognized as an RX 480/580 Ellesmere GPU which is basically what it is, but my problem is the largest intensity I can set is 2040 because any number larger just gets changed to 2040 and I am wondering if this is a software limitation? The GPU has 32GB total split into two 16GB chunks, one for each chip on the card, but it seems I can't use a full 2304 so every shader has its 2MB plus double memory enabled.

I don't suppose anyone has insight to this, I realize this is the 0.01% question but I just don't see anything stopping me from using 2304 intensity with double memory since that's only 9216MB which is well below the 16GB per chip on the card, it just changes it to 2040 when I use higher numbers.

Claymore 10.2
Windows 10 Pro
Blockchain Drivers
24GB System RAM (pagefile sizes up to 128GB tried makes no difference with or without one)

With intensity 2040 and a -20 power limit I get 1300 H/s and I can't help but wonder, what could I get if I was able to use 2304 instead, but it just won't let me set anything higher, it always changes itself back to 2040 on me, is this an internal limit since nobody expected a workstation card to be used?

[vega64miner]

Has any one had a situation where the O/S and miner both freeze after an hour of operating?

Set up:
Windows 10 1079 Build 16299.64
AsRock H110 Mobo
5x Vega 64 no mod
2x Rx 580 founders edition
16gb DDR4 ram
100gb virtual memory
M.2 SSD 250gb
I3-7350k
Amd 17.11.2

everything runs good but things just seem to freeze after a while...

Please help...

[carlo_0000]

Sooo I have a bit of an unusual situation, almost nobody uses workstation GPU's to mine with, my project is over and now I have a "Radeon Pro Duo (polaris)" sitting idle doing nothing at all, it gets recognized as an RX 480/580 Ellesmere GPU which is basically what it is, but my problem is the largest intensity I can set is 2040 because any number larger just gets changed to 2040 and I am wondering if this is a software limitation? The GPU has 32GB total split into two 16GB chunks, one for each chip on the card, but it seems I can't use a full 2304 so every shader has its 2MB plus double memory enabled.

I don't suppose anyone has insight to this, I realize this is the 0.01% question but I just don't see anything stopping me from using 2304 intensity with double memory since that's only 9216MB which is well below the 16GB per chip on the card, it just changes it to 2040 when I use higher numbers.

Claymore 10.2
Windows 10 Pro
Blockchain Drivers
24GB System RAM (pagefile sizes up to 128GB tried makes no difference with or without one)

With intensity 2040 and a -20 power limit I get 1300 H/s and I can't help but wonder, what could I get if I was able to use 2304 instead, but it just won't let me set anything higher, it always changes itself back to 2040 on me, is this an internal limit since nobody expected a workstation card to be used?

try amd xmr stack
than you double or triple  the gpu thread and try different king of intensities

start with this (470 4G)
    { "index" : 1, "intensity" :940, "worksize" : 8, "affine_to_cpu" : false },
    { "index" : 1, "intensity" :940, "worksize" : 8, "affine_to_cpu" : false },

then try 940+  until you find the best setting , and check gpu memory usage

the vega 56 use  for 8G of memeory
   { "index" : 0, "intensity" : 2000, "worksize" : 8, "affine_to_cpu" : false },
    { "index" : 0, "intensity" : 1600, "worksize" : 8, "affine_to_cpu" : false },

after try with 3 x  or 4x (but not sure gpu if fast enough to do it)

[Darkbane]

Sooo I have a bit of an unusual situation, almost nobody uses workstation GPU's to mine with, my project is over and now I have a "Radeon Pro Duo (polaris)" sitting idle doing nothing at all, it gets recognized as an RX 480/580 Ellesmere GPU which is basically what it is, but my problem is the largest intensity I can set is 2040 because any number larger just gets changed to 2040 and I am wondering if this is a software limitation? The GPU has 32GB total split into two 16GB chunks, one for each chip on the card, but it seems I can't use a full 2304 so every shader has its 2MB plus double memory enabled.

I don't suppose anyone has insight to this, I realize this is the 0.01% question but I just don't see anything stopping me from using 2304 intensity with double memory since that's only 9216MB which is well below the 16GB per chip on the card, it just changes it to 2040 when I use higher numbers.

Claymore 10.2
Windows 10 Pro
Blockchain Drivers
24GB System RAM (pagefile sizes up to 128GB tried makes no difference with or without one)

With intensity 2040 and a -20 power limit I get 1300 H/s and I can't help but wonder, what could I get if I was able to use 2304 instead, but it just won't let me set anything higher, it always changes itself back to 2040 on me, is this an internal limit since nobody expected a workstation card to be used?

try amd xmr stack
than you double or triple  the gpu thread and try different king of intensities

start with this (470 4G)
    { "index" : 1, "intensity" :940, "worksize" : 8, "affine_to_cpu" : false },
    { "index" : 1, "intensity" :940, "worksize" : 8, "affine_to_cpu" : false },

then try 940+  until you find the best setting , and check gpu memory usage

the vega 56 use  for 8G of memeory
   { "index" : 0, "intensity" : 2000, "worksize" : 8, "affine_to_cpu" : false },
    { "index" : 0, "intensity" : 1600, "worksize" : 8, "affine_to_cpu" : false },

after try with 3 x  or 4x (but not sure gpu if fast enough to do it)

I've already tried this, tried numerous miners, tried various multiple thread tricks, they don't improve performance on the Radeon Pro Duo, so far claymore 10.2 at this 2040 setting it falls back to when I set higher numbers is the best H/s overall no matter what I do in the other miners, which is why I am trying to figure out if this is some hard coded limit because he didn't expect anyone to have larger than 8GB cards, it would be nice to see what 2304 with double memory enabled will do for this card and miner. no other intensity settings performed better, maxed was best, which indicates to me I have a lot of head room to go with this card, but I'm simply not able to capitalize on it yet, normally folks clog their cards and have to lower intensity, I seem to be in the opposite scenario where I am hitting some software limited wall, at least thats my assumption, I just don't know, which is why I am asking if folks know if there is a limit like this, its not like we can view source to find out so all I can do is ask and hope heh.

P.S. and yes I have tried 3 thread even 4 threads with those other miners, they always seem to choke on it, when I tried copying files to run two completely separate instances the program would kill itself, so I've tried everything I can possibly think of and have run out of ideas. I realize this is a workstation card so the odds of me finding someone with direct experience is almost zero, but hey its the internet, maybe I'll get lucky right.