Bitcoin Forum
March 19, 2024, 02:51:22 AM *
News: Latest Bitcoin Core release: 26.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: « 1 2 3 4 5 6 7 [8] 9 10 11 »  All
  Print  
Author Topic: NVIDIA Kepler (K20) from 134MHash/s to 330MHash/s with CUDA  (Read 73284 times)
charliemaggot
Member
**
Offline Offline

Activity: 79
Merit: 10



View Profile
April 11, 2013, 04:33:46 AM
 #141

i have win8 and geforce710m on my laptotp

can someone give me the instructions for make it?

Start here, https://github.com/cdmackie/rpcminer-mod

Feel free to ask if you have any questions.
1710816682
Hero Member
*
Offline Offline

Posts: 1710816682

View Profile Personal Message (Offline)

Ignore
1710816682
Reply with quote  #2

1710816682
Report to moderator
1710816682
Hero Member
*
Offline Offline

Posts: 1710816682

View Profile Personal Message (Offline)

Ignore
1710816682
Reply with quote  #2

1710816682
Report to moderator
"In a nutshell, the network works like a distributed timestamp server, stamping the first transaction to spend a coin. It takes advantage of the nature of information being easy to spread but hard to stifle." -- Satoshi
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction.
1710816682
Hero Member
*
Offline Offline

Posts: 1710816682

View Profile Personal Message (Offline)

Ignore
1710816682
Reply with quote  #2

1710816682
Report to moderator
1710816682
Hero Member
*
Offline Offline

Posts: 1710816682

View Profile Personal Message (Offline)

Ignore
1710816682
Reply with quote  #2

1710816682
Report to moderator
1710816682
Hero Member
*
Offline Offline

Posts: 1710816682

View Profile Personal Message (Offline)

Ignore
1710816682
Reply with quote  #2

1710816682
Report to moderator
jeprokzki
Newbie
*
Offline Offline

Activity: 49
Merit: 0


View Profile
April 11, 2013, 07:49:44 AM
 #142

thanks sir it's working now Smiley but my computer gets really slow it is normal? my old hashrate is 101.5mh/s but now it's 179mh/s im using gtx 660ti oc ed.

Is its right that you now faster than before? 101->179 MHash/s
With which program you get 101 MHash/s, is it up to date?

cgminer v.2,11,4 sir
Skami
Member
**
Offline Offline

Activity: 103
Merit: 10



View Profile WWW
April 11, 2013, 08:32:33 AM
 #143

Once I start it I get this, any suggestions?

Starting parameters are:

rpcminer-mod-cuda.exe -url="http://api.bitcoin.cz:8332" -user="WORKER" -password="PASS"

Quote
Client will start 1 miner threads
Work will be refreshed every 4000 ms
1 GPU GUDA device(s) found
Setting GPU GUDA device 0
CUDA initialized
Search Configuration for gpu named: GeForce GTX 670
Your device:
   - 7 streaming multiprocessors (SM)
   - can run sm version 300
   - bitcoin miner is optimized for 5 gpugrids per SM
   - max value for gpugrid parameter is 2147483647
max nonce 3FFFFFFF
Autotuning is on because no gridsize is given
Target = 00000000ffffffffffffffffffffffffffffffffffffffffffffffffffffff
Done allocating CUDA resources for (131040,256)
iterations on gpu=32 blocks=131040 threads=256
Iterations: 1,hashs per iteration 1073479680
<src\cuda\bitcoinminercuda.cpp>:193[CUDA] Error: unknown error
<src\cuda\bitcoinminercuda.cpp>:101[CUDA] Error: unknown error

charliemaggot
Member
**
Offline Offline

Activity: 79
Merit: 10



View Profile
April 11, 2013, 02:44:46 PM
 #144

Once I start it I get this, any suggestions?

Starting parameters are:

rpcminer-mod-cuda.exe -url="http://api.bitcoin.cz:8332" -user="WORKER" -password="PASS"


Add "-gpugrid=512" or try "-gpugrid=256"

rpcminer-mod-cuda.exe -url="http://api.bitcoin.cz:8332" -user="WORKER" -password="PASS" -gpu -gpugrid=512



Skami
Member
**
Offline Offline

Activity: 103
Merit: 10



View Profile WWW
April 11, 2013, 03:30:31 PM
 #145

Once I start it I get this, any suggestions?

Starting parameters are:

rpcminer-mod-cuda.exe -url="http://api.bitcoin.cz:8332" -user="WORKER" -password="PASS"


Add "-gpugrid=512" or try "-gpugrid=256"

rpcminer-mod-cuda.exe -url="http://api.bitcoin.cz:8332" -user="WORKER" -password="PASS" -gpu -gpugrid=512




Getting around 99 Mhash/s while I get around 104 Mhash/s with CGMiner

gateway
Hero Member
*****
Offline Offline

Activity: 552
Merit: 500


View Profile
April 11, 2013, 04:59:59 PM
 #146

So the thread says going from 134 - 330.. I have yet to see any sort of leap from using other miners or rpcminer at default.  I realize that people are working on functions to try to improve that and their efforts are greatly appreciated but, I think all of us would like to see a nice bump in our mining with our nvidia cards.

thoughts?
psychocoder (OP)
Newbie
*
Offline Offline

Activity: 49
Merit: 0


View Profile
April 11, 2013, 07:22:06 PM
 #147

So the thread says going from 134 - 330.. I have yet to see any sort of leap from using other miners or rpcminer at default.  I realize that people are working on functions to try to improve that and their efforts are greatly appreciated but, I think all of us would like to see a nice bump in our mining with our nvidia cards.

thoughts?

134 MHash/s to 330 MHash/s is for a Kepler K20 or the consumer version Titan. A big jump with old NVIDIA GPUs is not possible. CGMIner is very near to the maximum power of old NVIDIA GPUs. Al GPUs before Titan has no funnel opertor and need 3 operation for a rotate, the new one need one operation.
gateway
Hero Member
*****
Offline Offline

Activity: 552
Merit: 500


View Profile
April 11, 2013, 07:40:49 PM
 #148

So the thread says going from 134 - 330.. I have yet to see any sort of leap from using other miners or rpcminer at default.  I realize that people are working on functions to try to improve that and their efforts are greatly appreciated but, I think all of us would like to see a nice bump in our mining with our nvidia cards.

thoughts?

134 MHash/s to 330 MHash/s is for a Kepler K20 or the consumer version Titan. A big jump with old NVIDIA GPUs is not possible. CGMIner is very near to the maximum power of old NVIDIA GPUs. Al GPUs before Titan has no funnel opertor and need 3 operation for a rotate, the new one need one operation.

Ahh ok sorry my bad, thanks for the clarification.. so your saying the 690's and below will still be somewhat limited or same speed with out much boost?

I have a gtx 680..
Evan
Hero Member
*****
Offline Offline

Activity: 507
Merit: 500



View Profile
April 12, 2013, 01:33:54 AM
 #149

So the thread says going from 134 - 330.. I have yet to see any sort of leap from using other miners or rpcminer at default.  I realize that people are working on functions to try to improve that and their efforts are greatly appreciated but, I think all of us would like to see a nice bump in our mining with our nvidia cards.

thoughts?

134 MHash/s to 330 MHash/s is for a Kepler K20 or the consumer version Titan. A big jump with old NVIDIA GPUs is not possible. CGMIner is very near to the maximum power of old NVIDIA GPUs. Al GPUs before Titan has no funnel opertor and need 3 operation for a rotate, the new one need one operation.

Ahh ok sorry my bad, thanks for the clarification.. so your saying the 690's and below will still be somewhat limited or same speed with out much boost?

I have a gtx 680..

GTX 680 IS a Kepler..

I am poor, but i do work for Coin Smiley
1PtHcavXoakgNkQfEQdvnvEksEY2NvwaLM
gateway
Hero Member
*****
Offline Offline

Activity: 552
Merit: 500


View Profile
April 12, 2013, 02:08:31 AM
 #150

So the max I have been able to push my 680 with out oc , is between 120-130.. mostly lows 120's.  I would love to even add 100 more Tongue but its not my main mining rig just add's to my pool from my other cards..
psychocoder (OP)
Newbie
*
Offline Offline

Activity: 49
Merit: 0


View Profile
April 12, 2013, 05:38:48 AM
 #151

So the thread says going from 134 - 330.. I have yet to see any sort of leap from using other miners or rpcminer at default.  I realize that people are working on functions to try to improve that and their efforts are greatly appreciated but, I think all of us would like to see a nice bump in our mining with our nvidia cards.

thoughts?

134 MHash/s to 330 MHash/s is for a Kepler K20 or the consumer version Titan. A big jump with old NVIDIA GPUs is not possible. CGMIner is very near to the maximum power of old NVIDIA GPUs. Al GPUs before Titan has no funnel opertor and need 3 operation for a rotate, the new one need one operation.

Ahh ok sorry my bad, thanks for the clarification.. so your saying the 690's and below will still be somewhat limited or same speed with out much boost?

I have a gtx 680..

GTX 680 IS a Kepler..

Onyl Kepler equal or higher than titan GPUs. GTX680 is K10 architecture, you can say it is a better fermi. K10 has no funnel operator :-(. Thay have the same amount of 32Bit registers and a lot of opration have the same speed  than a k20 but the importend operations are not available.
ssateneth
Legendary
*
Offline Offline

Activity: 1344
Merit: 1004



View Profile
April 12, 2013, 10:50:04 AM
Last edit: April 12, 2013, 11:14:08 AM by ssateneth
 #152

So did this thread ever amount to anything? Will I ever see a "significant" mhash from my gtx 680? I use win7 x64. Linux not an option for me. I tried "-gpu=0  -aggression=8 -gpugrid=2048 -gputhreads=256" but I only got 112 mhash.

Evan
Hero Member
*****
Offline Offline

Activity: 507
Merit: 500



View Profile
April 12, 2013, 08:23:48 PM
 #153

So the thread says going from 134 - 330.. I have yet to see any sort of leap from using other miners or rpcminer at default.  I realize that people are working on functions to try to improve that and their efforts are greatly appreciated but, I think all of us would like to see a nice bump in our mining with our nvidia cards.

thoughts?

134 MHash/s to 330 MHash/s is for a Kepler K20 or the consumer version Titan. A big jump with old NVIDIA GPUs is not possible. CGMIner is very near to the maximum power of old NVIDIA GPUs. Al GPUs before Titan has no funnel opertor and need 3 operation for a rotate, the new one need one operation.

Ahh ok sorry my bad, thanks for the clarification.. so your saying the 690's and below will still be somewhat limited or same speed with out much boost?

I have a gtx 680..

GTX 680 IS a Kepler..

Onyl Kepler equal or higher than titan GPUs. GTX680 is K10 architecture, you can say it is a better fermi. K10 has no funnel operator :-(. Thay have the same amount of 32Bit registers and a lot of opration have the same speed  than a k20 but the importend operations are not available.

I am waiting for the 700 GPUs to come out, or get a deal on another titan.


Psycho So none of the Kepler Base Quadros can do this either?

I am poor, but i do work for Coin Smiley
1PtHcavXoakgNkQfEQdvnvEksEY2NvwaLM
psychocoder (OP)
Newbie
*
Offline Offline

Activity: 49
Merit: 0


View Profile
April 13, 2013, 10:49:35 AM
 #154


I am waiting for the 700 GPUs to come out, or get a deal on another titan.
Psycho So none of the Kepler Base Quadros can do this either?

I think so, but I think there is no Quadro out with K20 architecture. At the moment the GTX Titan is also to expensive to buy.
slon_ru
Newbie
*
Offline Offline

Activity: 24
Merit: 0


View Profile
April 16, 2013, 08:57:47 AM
Last edit: April 16, 2013, 10:26:25 AM by slon_ru
 #155

Hi charliemaggot !

I have nv670 - 100 m.hash/sec with GUIminer.

If i run rpcminer-mod-cuda.exe with GUIminer i get:

Faulting application name: rpcminer-mod-cuda.exe, version: 0.0.0.0, time stamp: 0x5163a5d3
Faulting module name: MSVCR100.dll, version: 10.0.40219.325, time stamp: 0x4df2bcac
Exception code: 0x40000015
Faulting application path: D:\bitcoin\miner\guiminer\miners\puddinpop\bk\rpcminer-mod-cuda.exe
Faulting module path: D:\bitcoin\miner\guiminer\miners\puddinpop\MSVCR100.dll
Report Id: b1683d16-a64f-11e2-b5bb-5404a61bb659

Display driver nvlddmkm stopped responding and has successfully recovered.



With command line i get:

"Could not retrieve work from RPC server.
CURL return value = 22"

I set TdrDelay and TdrDdiDelay = 10,60.

Can you help me to run rpcminer-mod-cuda.exe ?

br

Amph
Legendary
*
Offline Offline

Activity: 3206
Merit: 1069



View Profile
April 16, 2013, 05:28:46 PM
 #156

nothing changed, still 100mh with my 670
charliemaggot
Member
**
Offline Offline

Activity: 79
Merit: 10



View Profile
April 16, 2013, 09:54:07 PM
 #157

@slon_ru can you post the manual command line (hide your username/password) that causes the crash, and the output you are getting.

CURL error 22 is a HTTP error from the server, is your pool working ok?

If you reduce -gpugrid down to 256 or 512, do you still get the error?

Lastly, please try downloading and running the files from the debug branch, which will hopefully give more information on the crash:

https://github.com/cdmackie/rpcminer-mod/tree/Debug/bin

Again, post any errors.
slon_ru
Newbie
*
Offline Offline

Activity: 24
Merit: 0


View Profile
April 17, 2013, 05:35:27 AM
Last edit: April 17, 2013, 10:29:36 AM by slon_ru
 #158

@charliemaggot Thx!
It was my fault, i wrote without "-password",i have no pass.

Now i have 90 m.hash/s with rpcminer-mod-cuda.

br.
Jimmy2011
Hero Member
*****
Offline Offline

Activity: 589
Merit: 500



View Profile
April 17, 2013, 07:54:13 AM
 #159

It seems that no additional hashrate for GTX 580 here.
Lacan82
Sr. Member
****
Offline Offline

Activity: 247
Merit: 250


View Profile
April 17, 2013, 11:13:15 AM
 #160

I can't even get it to run. I get unable to load cuda module error. When I try compile the module myself per the instructions...I get this error:

LIBCMT.lib(crt0.obj) : error LNK2019: unresolved external symbol _main reference
d in function ___tmainCRTStartup
a.exe : fatal error LNK1120: 1 unresolved externals

No one can tell me how to fix it. I'm not much for a programmer..

Pages: « 1 2 3 4 5 6 7 [8] 9 10 11 »  All
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!