Pages:
Author

Topic: [ANN] Kryptohash | Brand new PoW algo | 320bit hash | ed25519 | PID algo for dif - page 5. (Read 149396 times)

full member
Activity: 140
Merit: 100
sr. member
Activity: 261
Merit: 250
sr. member
Activity: 329
Merit: 250
Current testing environment has  ATI Radeon™ HD 5670 Graphics : http://www.amd.com/en-us/products/graphics/desktop/5000/5670
I also have other 3 R7 200 series to mine, but first I want to make it work on this one

Bitcoin, Litecoin, Doge, all are working with this card on GPU so Kryptohash should be okey from my point of view

I have tried the following scenarios :

--shaders 1 --shaders-mul 1
--shaders 400 --shaders-mul 1
--shaders 400 --shaders-mul 7
--shaders 128 --shaders-mul 7

everything seems to output error -54 Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)

I found that the param -w would solve the problem  so right now I tried

 -I 16 -g 2 -w 128 --shaders 400 --shaders-mul 7

in this case I get 3.5 Kh/s Smiley

for scrypt (litecoing) there is a param : --thread-concurrency , there is something similar here ?  or any ideas to get that 3.5 value increased ?  or where can I find a list with all accepted params for this version ?

Thanks


For Kryptohash, the thread concurrency is the combination of --shaders and --shaders-mul parameters.

Try something like this:

 -w 128 --shaders 448 --shaders-mul 2

Then, increase the shaders-mul value until you find the max Kh/s

Also, try with -w 64  and see if you get better performance.



newbie
Activity: 3
Merit: 0
Current testing environment has  ATI Radeon™ HD 5670 Graphics : http://www.amd.com/en-us/products/graphics/desktop/5000/5670
I also have other 3 R7 200 series to mine, but first I want to make it work on this one

Bitcoin, Litecoin, Doge, all are working with this card on GPU so Kryptohash should be okey from my point of view

I have tried the following scenarios :

--shaders 1 --shaders-mul 1
--shaders 400 --shaders-mul 1
--shaders 400 --shaders-mul 7
--shaders 128 --shaders-mul 7

everything seems to output error -54 Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)

I found that the param -w would solve the problem  so right now I tried

 -I 16 -g 2 -w 128 --shaders 400 --shaders-mul 7

in this case I get 3.5 Kh/s Smiley

for scrypt (litecoing) there is a param : --thread-concurrency , there is something similar here ?  or any ideas to get that 3.5 value increased ?  or where can I find a list with all accepted params for this version ?

Thanks
sr. member
Activity: 329
Merit: 250
That was the problem ! You should add an observation for new users about this on the first page Smiley .

Now I get another error : No suitable long pool found for http://127.0.0.1 and (see image)



I have an ATI Radeon™ HD 5670 Graphics



Hmm, the "No suitable long pool found" is not the problem, it is the error after that which is related to OpenCL.

Your GPU has only 400 stream processors.   What values are you using for --shaders and --shaders-mul ?
newbie
Activity: 3
Merit: 0
That was the problem ! You should add an observation for new users about this on the first page Smiley .

Now I get another error : No suitable long pool found for http://127.0.0.1 and (see image)

http://s15.postimg.org/iofd2d523/image.png

I have an ATI Radeon™ HD 5670 Graphics



sr. member
Activity: 329
Merit: 250
I am not a beginner in mining (6 months experience), but my first attempt to mine Kryptohash
Just Downloaded Wallet v0.6.2   - installed into Windows :
C:\Program Files (x86)\Kryptohash Core 32-bit\kryptohash.conf
and the data is default at
C:\Users\UserX\AppData\Roaming\Kryptohash

running latest version cgminer-khc and 100%(hope) correct config on kryptohash.conf

server=1
rpcuser=test
rpcpassword=test
rpcport=38912
rpcconnect=127.0.0.1

after command cgminer --kryptohash -o http://127.0.0.1:38912 -u test -p test
i receive:
 [2015-07-31 13:21:58] Started cgminer 3.7.8
 [2015-07-31 13:21:58] Started cgminer 3.7.8
 [2015-07-31 13:21:58] Probing for an alive pool
 [2015-07-31 13:21:59] HTTP request returned 7 "Failed to connect to 127.0.0.1 p
ort 38912: Connection refused"
 [2015-07-31 13:22:00] HTTP request returned 7 "Failed to connect to 127.0.0.1 p
ort 38912: Connection refused"
 [2015-07-31 13:22:01] Pool 0 slow/down or URL or credentials invalid

"getwork" on console works

any ideas ?


Did you copy the kryptohash.conf to the C:\Users\UserX\AppData\Roaming\Kryptohash folder?   

newbie
Activity: 3
Merit: 0
I am not a beginner in mining (6 months experience), but my first attempt to mine Kryptohash
Just Downloaded Wallet v0.6.2   - installed into Windows :
C:\Program Files (x86)\Kryptohash Core 32-bit\kryptohash.conf
and the data is default at
C:\Users\UserX\AppData\Roaming\Kryptohash

running latest version cgminer-khc and 100%(hope) correct config on kryptohash.conf

server=1
rpcuser=test
rpcpassword=test
rpcport=38912
rpcconnect=127.0.0.1

after command cgminer --kryptohash -o http://127.0.0.1:38912 -u test -p test
i receive:
 [2015-07-31 13:21:58] Started cgminer 3.7.8
 [2015-07-31 13:21:58] Started cgminer 3.7.8
 [2015-07-31 13:21:58] Probing for an alive pool
 [2015-07-31 13:21:59] HTTP request returned 7 "Failed to connect to 127.0.0.1 p
ort 38912: Connection refused"
 [2015-07-31 13:22:00] HTTP request returned 7 "Failed to connect to 127.0.0.1 p
ort 38912: Connection refused"
 [2015-07-31 13:22:01] Pool 0 slow/down or URL or credentials invalid

"getwork" on console works

any ideas ?




sr. member
Activity: 329
Merit: 250
New Wallet version 0.6.2 available.

Changes since 0.6.1

- Fixed an issue that could cause the PID algorithm to calculate the wrong difficulty, if an old block arrived out of order.
- Added checkpoint for block 100,000
- Updated OpenSSL to version 1.0.1p for Windows binaries.

Download from GitHub:
https://github.com/kryptohash/kryptohash/releases
full member
Activity: 140
Merit: 100
how does nvidia compare with ati in mining KHC ?

For KHC mining, AMD GPUs are 4 to 5 times faster than nVidia GPUs.  This is because the KShake320 algorithm is Hash-based and requires only Integer Arithmetic and Logic operations. The algorithm benefits from the high number of ALUs provided by the AMD architecture.  
nVidia architecture on the other hand, is much stronger on the Floating Point calculations but, the KShake320 algorithm does not use the FPU.
Good ,I love this GPU algo.
sr. member
Activity: 329
Merit: 250
how does nvidia compare with ati in mining KHC ?

For KHC mining, AMD GPUs are 4 to 5 times faster than nVidia GPUs.  This is because the KShake320 algorithm is Hash-based and requires only Integer Arithmetic and Logic operations. The algorithm benefits from the high number of ALUs provided by the AMD architecture.  
nVidia architecture on the other hand, is much stronger on the Floating Point calculations but, the KShake320 algorithm does not use the FPU.
sr. member
Activity: 261
Merit: 250
how does nvidia compare with ati in mining KHC ?
sr. member
Activity: 329
Merit: 250
I've installed the wallet + setup the .conf file but when I start cgminer i get this.


Any help? Thanks.

Be sure you are editing the kryptohash.conf in the right location.  

If you didn't pick a custom directory the first time you executed kryptohash-qt.exe, Windows places all the files in C:\Users\{YourUserName}\AppData\Roaming\Kryptohash\

Edit the kryptohash.conf file and ensure you to have these lines:

server=1
rpcuser={myRPCusername}
rpcpassword={myRPCpassword}
rpcconnect=127.0.0.1

Then, restart the wallet.

Create a solo.bat file inside the cgminer-khc folder and add these lines:

setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
cgminer --kryptohash --cl-opt-disable -o http://127.0.0.1:38912 -u {myRPCusername} -p {myRPCpassword} --shaders {Num_shaders_in_my_GPU} --shaders-mul 8 --expiry 10


Edit:  That error could also happen if the Wallet is still synchronizing with the network.   Try the 'getwork' command from the Wallet's Debug console to confirm.
newbie
Activity: 28
Merit: 0
I've installed the wallet + setup the .conf file but when I start cgminer i get this.
http://i.imgur.com/gd8zB1b.png

Any help? Thanks.
sr. member
Activity: 329
Merit: 250
I just tested the new AMD 15.7 official drivers on my 290x and the KH/s dropped by 5%.  

I also noticed that the --cl-opt-disable switch that I added to cgminer-khc no longer makes a difference with these new drivers.

Edit:  Actually, the performance is the same.  I forgot to re-adjust the Power Limit setting in the AMD Catalyst Control Center after I upgraded the drivers.

I use the r9 280x and for me cl opt disable still makes a noticeable difference, using the 15.7 windows 10 drivers


You're correct sir.  I forgot that cgminer-khc has to re-compile the OpenCL kernel after removing the --cl-opt-disable switch.  Doh!

But, I'm back to the 14.12 drivers because, I was having stability problems with the 15.7 drivers. 
member
Activity: 143
Merit: 10
I just tested the new AMD 15.7 official drivers on my 290x and the KH/s dropped by 5%.  

I also noticed that the --cl-opt-disable switch that I added to cgminer-khc no longer makes a difference with these new drivers.

Edit:  Actually, the performance is the same.  I forgot to re-adjust the Power Limit setting in the AMD Catalyst Control Center after I upgraded the drivers.

I use the r9 280x and for me cl opt disable still makes a noticeable difference, using the 15.7 windows 10 drivers
sr. member
Activity: 329
Merit: 250
I just tested the new AMD 15.7 official drivers on my 290x and the KH/s dropped by 5%.  

I also noticed that the --cl-opt-disable switch that I added to cgminer-khc no longer makes a difference with these new drivers.

Edit:  Actually, the performance is the same.  I forgot to re-adjust the Power Limit setting in the AMD Catalyst Control Center after I upgraded the drivers.
sr. member
Activity: 329
Merit: 250
member
Activity: 79
Merit: 10
Could we have a full-funtional block exploerer?

We do have a new explorer but, I need to fix a bug that makes it stop responding after a while.


Yes,no responding wastes my power fee.
hero member
Activity: 690
Merit: 500
Surprisingly, today my miner worked through the day without freezes. Let's see if the problem appears again.
Pages:
Jump to: