Any plan for stratum support?
miningpoolhub would like to support new stratum protocol and open siacoin pool.
Topic: [ANN] Sia GPU Miner - page 2. (Read 10889 times)
I'm curious what this does on CUDA devices:
In the cl kernel, replace the ror64 with this and change the calls to ror64_2 to ror64:
I was on AWS to try it but then realized I needs Compute 3.5 at least..
Is there any reasonably priced Maxwell in the cloud yet?
In the cl kernel, replace the ror64 with this and change the calls to ror64_2 to ror64:
Code:
static uint2 ror64(const uint2 a, const int offset) {
uint2 result;
if (offset < 32) {
asm("shf.r.wrap.b32 %0, %1, %2, %3;" : "=r"(result.x) : "r"(a.x), "r"(a.y), "r"(offset));
asm("shf.r.wrap.b32 %0, %1, %2, %3;" : "=r"(result.y) : "r"(a.y), "r"(a.x), "r"(offset));
}
else {
asm("shf.r.wrap.b32 %0, %1, %2, %3;" : "=r"(result.x) : "r"(a.y), "r"(a.x), "r"(offset));
asm("shf.r.wrap.b32 %0, %1, %2, %3;" : "=r"(result.y) : "r"(a.x), "r"(a.y), "r"(offset));
}
return result;
}
uint2 result;
if (offset < 32) {
asm("shf.r.wrap.b32 %0, %1, %2, %3;" : "=r"(result.x) : "r"(a.x), "r"(a.y), "r"(offset));
asm("shf.r.wrap.b32 %0, %1, %2, %3;" : "=r"(result.y) : "r"(a.y), "r"(a.x), "r"(offset));
}
else {
asm("shf.r.wrap.b32 %0, %1, %2, %3;" : "=r"(result.x) : "r"(a.y), "r"(a.x), "r"(offset));
asm("shf.r.wrap.b32 %0, %1, %2, %3;" : "=r"(result.y) : "r"(a.x), "r"(a.y), "r"(offset));
}
return result;
}
I was on AWS to try it but then realized I needs Compute 3.5 at least..
Is there any reasonably priced Maxwell in the cloud yet?
Hey,
I got bored with ETH
and decided to give this a spin. Forked the SIA GPU Miner written in C and added a few things:
- MSVC project files for those who want to hack on it using Windows.
- pool support using -Q parameter (works similar to Go Miner)
- took the cl kernel from the GO miner to improve performance (roughly +18% on a 7950)
- x64 Windows binaries zipped up in the bin folder.
Planning to improve nvidia opencl performance using some inline PTX asm. This should bring performance close(r) to the native CUDA miners around.
Perhaps improve the CL performance a bit, but that's a crowded market these days.
If you like poking around in CL, there's an undocumented command line parameter (-k filename) that allows you to provide a custom CL kernel.
https://github.com/Genoil/Sia-GPU-Miner
Have fun!
[edit]
it looks like the optimized kernel using uitn2 needs some work. the client reports found blocks, but my hashrate doesn't show any longer on nanopool. I'll wait a little while, may need to revert back and pay closer attention to what results it gives. stay tuned.
I got bored with ETH
and decided to give this a spin. Forked the SIA GPU Miner written in C and added a few things:- MSVC project files for those who want to hack on it using Windows.
- pool support using -Q parameter (works similar to Go Miner)
- took the cl kernel from the GO miner to improve performance (roughly +18% on a 7950)
- x64 Windows binaries zipped up in the bin folder.
Planning to improve nvidia opencl performance using some inline PTX asm. This should bring performance close(r) to the native CUDA miners around.
Perhaps improve the CL performance a bit, but that's a crowded market these days.
If you like poking around in CL, there's an undocumented command line parameter (-k filename) that allows you to provide a custom CL kernel.
https://github.com/Genoil/Sia-GPU-Miner
Have fun!
[edit]
it looks like the optimized kernel using uitn2 needs some work. the client reports found blocks, but my hashrate doesn't show any longer on nanopool. I'll wait a little while, may need to revert back and pay closer attention to what results it gives. stay tuned.
is cuda that unfriendly to code with, that it's better to go with opencl first?
No quite the opposite. But I don't have any CUDA GPUs at the moment. Sold my 970 for a good price just before RX 480 hit the market, with the idea of exchanging it for a 1070. But now with 1060 coming, I'll wait for that as I'm only really interested in the architecture, not the performance.
With ethash I learned that some inline PTX assembly in OpenCL can do a lot to match Nvidia OpenCL with CUDA. It means a whole lot of host and device code less to maintain.
What do you think of it?
I *think* of it, just like everybody
. Problem with the shares seems fixed. The kernel is fine, it had something to do with the query string and ampersands...
Hey,
I got bored with ETH and decided to give this a spin. Forked the SIA GPU Miner written in C and added a few things:
- MSVC project files for those who want to hack on it using Windows.
- pool support using -Q parameter (works similar to Go Miner)
- took the cl kernel from the GO miner to improve performance (roughly +18% on a 7950)
- x64 Windows binaries zipped up in the bin folder.
Planning to improve nvidia opencl performance using some inline PTX asm. This should bring performance close(r) to the native CUDA miners around.
Perhaps improve the CL performance a bit, but that's a crowded market these days.
If you like poking around in CL, there's an undocumented command line parameter (-k filename) that allows you to provide a custom CL kernel.
https://github.com/Genoil/Sia-GPU-Miner
Have fun!
[edit]
it looks like the optimized kernel using uitn2 needs some work. the client reports found blocks, but my hashrate doesn't show any longer on nanopool. I'll wait a little while, may need to revert back and pay closer attention to what results it gives. stay tuned.
I got bored with ETH
and decided to give this a spin. Forked the SIA GPU Miner written in C and added a few things:- MSVC project files for those who want to hack on it using Windows.
- pool support using -Q parameter (works similar to Go Miner)
- took the cl kernel from the GO miner to improve performance (roughly +18% on a 7950)
- x64 Windows binaries zipped up in the bin folder.
Planning to improve nvidia opencl performance using some inline PTX asm. This should bring performance close(r) to the native CUDA miners around.
Perhaps improve the CL performance a bit, but that's a crowded market these days.
If you like poking around in CL, there's an undocumented command line parameter (-k filename) that allows you to provide a custom CL kernel.
https://github.com/Genoil/Sia-GPU-Miner
Have fun!
[edit]
it looks like the optimized kernel using uitn2 needs some work. the client reports found blocks, but my hashrate doesn't show any longer on nanopool. I'll wait a little while, may need to revert back and pay closer attention to what results it gives. stay tuned.
is cuda that unfriendly to code with, that it's better to go with opencl first?
No quite the opposite. But I don't have any CUDA GPUs at the moment. Sold my 970 for a good price just before RX 480 hit the market, with the idea of exchanging it for a 1070. But now with 1060 coming, I'll wait for that as I'm only really interested in the architecture, not the performance.
With ethash I learned that some inline PTX assembly in OpenCL can do a lot to match Nvidia OpenCL with CUDA. It means a whole lot of host and device code less to maintain.
What do you think of it?
Hey,
I got bored with ETH and decided to give this a spin. Forked the SIA GPU Miner written in C and added a few things:
- MSVC project files for those who want to hack on it using Windows.
- pool support using -Q parameter (works similar to Go Miner)
- took the cl kernel from the GO miner to improve performance (roughly +18% on a 7950)
- x64 Windows binaries zipped up in the bin folder.
Planning to improve nvidia opencl performance using some inline PTX asm. This should bring performance close(r) to the native CUDA miners around.
Perhaps improve the CL performance a bit, but that's a crowded market these days.
If you like poking around in CL, there's an undocumented command line parameter (-k filename) that allows you to provide a custom CL kernel.
https://github.com/Genoil/Sia-GPU-Miner
Have fun!
[edit]
it looks like the optimized kernel using uitn2 needs some work. the client reports found blocks, but my hashrate doesn't show any longer on nanopool. I'll wait a little while, may need to revert back and pay closer attention to what results it gives. stay tuned.
I got bored with ETH
and decided to give this a spin. Forked the SIA GPU Miner written in C and added a few things:- MSVC project files for those who want to hack on it using Windows.
- pool support using -Q parameter (works similar to Go Miner)
- took the cl kernel from the GO miner to improve performance (roughly +18% on a 7950)
- x64 Windows binaries zipped up in the bin folder.
Planning to improve nvidia opencl performance using some inline PTX asm. This should bring performance close(r) to the native CUDA miners around.
Perhaps improve the CL performance a bit, but that's a crowded market these days.
If you like poking around in CL, there's an undocumented command line parameter (-k filename) that allows you to provide a custom CL kernel.
https://github.com/Genoil/Sia-GPU-Miner
Have fun!
[edit]
it looks like the optimized kernel using uitn2 needs some work. the client reports found blocks, but my hashrate doesn't show any longer on nanopool. I'll wait a little while, may need to revert back and pay closer attention to what results it gives. stay tuned.
is cuda that unfriendly to code with, that it's better to go with opencl first?
No quite the opposite. But I don't have any CUDA GPUs at the moment. Sold my 970 for a good price just before RX 480 hit the market, with the idea of exchanging it for a 1070. But now with 1060 coming, I'll wait for that as I'm only really interested in the architecture, not the performance.
With ethash I learned that some inline PTX assembly in OpenCL can do a lot to match Nvidia OpenCL with CUDA. It means a whole lot of host and device code less to maintain.
Hey,
I got bored with ETH and decided to give this a spin. Forked the SIA GPU Miner written in C and added a few things:
- MSVC project files for those who want to hack on it using Windows.
- pool support using -Q parameter (works similar to Go Miner)
- took the cl kernel from the GO miner to improve performance (roughly +18% on a 7950)
- x64 Windows binaries zipped up in the bin folder.
Planning to improve nvidia opencl performance using some inline PTX asm. This should bring performance close(r) to the native CUDA miners around.
Perhaps improve the CL performance a bit, but that's a crowded market these days.
If you like poking around in CL, there's an undocumented command line parameter (-k filename) that allows you to provide a custom CL kernel.
https://github.com/Genoil/Sia-GPU-Miner
Have fun!
[edit]
it looks like the optimized kernel using uitn2 needs some work. the client reports found blocks, but my hashrate doesn't show any longer on nanopool. I'll wait a little while, may need to revert back and pay closer attention to what results it gives. stay tuned.
I got bored with ETH
and decided to give this a spin. Forked the SIA GPU Miner written in C and added a few things:- MSVC project files for those who want to hack on it using Windows.
- pool support using -Q parameter (works similar to Go Miner)
- took the cl kernel from the GO miner to improve performance (roughly +18% on a 7950)
- x64 Windows binaries zipped up in the bin folder.
Planning to improve nvidia opencl performance using some inline PTX asm. This should bring performance close(r) to the native CUDA miners around.
Perhaps improve the CL performance a bit, but that's a crowded market these days.
If you like poking around in CL, there's an undocumented command line parameter (-k filename) that allows you to provide a custom CL kernel.
https://github.com/Genoil/Sia-GPU-Miner
Have fun!
[edit]
it looks like the optimized kernel using uitn2 needs some work. the client reports found blocks, but my hashrate doesn't show any longer on nanopool. I'll wait a little while, may need to revert back and pay closer attention to what results it gives. stay tuned.
is cuda that unfriendly to code with, that it's better to go with opencl first?
Hey,
I got bored with ETH and decided to give this a spin. Forked the SIA GPU Miner written in C and added a few things:
- MSVC project files for those who want to hack on it using Windows.
- pool support using -Q parameter (works similar to Go Miner)
- took the cl kernel from the GO miner to improve performance (roughly +18% on a 7950)
- x64 Windows binaries zipped up in the bin folder.
Planning to improve nvidia opencl performance using some inline PTX asm. This should bring performance close(r) to the native CUDA miners around.
Perhaps improve the CL performance a bit, but that's a crowded market these days.
If you like poking around in CL, there's an undocumented command line parameter (-k filename) that allows you to provide a custom CL kernel.
https://github.com/Genoil/Sia-GPU-Miner
Have fun!
[edit]
it looks like the optimized kernel using uitn2 needs some work. the client reports found blocks, but my hashrate doesn't show any longer on nanopool. I'll wait a little while, may need to revert back and pay closer attention to what results it gives. stay tuned.
Fixed
I got bored with ETH
and decided to give this a spin. Forked the SIA GPU Miner written in C and added a few things:- MSVC project files for those who want to hack on it using Windows.
- pool support using -Q parameter (works similar to Go Miner)
- took the cl kernel from the GO miner to improve performance (roughly +18% on a 7950)
- x64 Windows binaries zipped up in the bin folder.
Planning to improve nvidia opencl performance using some inline PTX asm. This should bring performance close(r) to the native CUDA miners around.
Perhaps improve the CL performance a bit, but that's a crowded market these days.
If you like poking around in CL, there's an undocumented command line parameter (-k filename) that allows you to provide a custom CL kernel.
https://github.com/Genoil/Sia-GPU-Miner
Have fun!
[edit]
Fixed
Think it's best sto write here instead open new topic.
I downloaded the gnu-miner.exe and relative dependency.
Everything works fine (i see the Mhs and block) my question is how i can take the (eventually) sia coin for blocking reward.
Also, some instructions to install this plugin? https://github.com/droghio/Sia-GPU-Miner/releases/tag/v2-ui
I downloaded the gnu-miner.exe and relative dependency.
Everything works fine (i see the Mhs and block) my question is how i can take the (eventually) sia coin for blocking reward.
Also, some instructions to install this plugin? https://github.com/droghio/Sia-GPU-Miner/releases/tag/v2-ui
Can I mine by network using few PC's? I don't want to make two wallets.
HI Taek !
I will not compile, how to begin.
I will not compile, how to begin.
Unfortunately:
A critical security vulnerability has been discovered.
This is a vulnerability that has existed in all versions prior to the current version. Even if you downloaded 0.3.3.2, you should download it again. The original download for 0.3.3.2 was still affected by the security vulnerability.
For obvious reasons, we will not be disclosing the nature of the vulnerability for the next few weeks, we wish to give everyone a chance to upgrade before opening the doors to exploitation. We will say however that it is not related to the blockchain code, it is related to the frontend code.
A critical security vulnerability has been discovered.
This is a vulnerability that has existed in all versions prior to the current version. Even if you downloaded 0.3.3.2, you should download it again. The original download for 0.3.3.2 was still affected by the security vulnerability.
For obvious reasons, we will not be disclosing the nature of the vulnerability for the next few weeks, we wish to give everyone a chance to upgrade before opening the doors to exploitation. We will say however that it is not related to the blockchain code, it is related to the frontend code.
Didn't read this until now... Hope you will not disclosing it in awhile.
May I ask how the exploit was discovered, by internal review or external usage of it?
Any windows verison of the gpu miner?
Unfortunately:
A critical security vulnerability has been discovered.
This is a vulnerability that has existed in all versions prior to the current version. Even if you downloaded 0.3.3.2, you should download it again. The original download for 0.3.3.2 was still affected by the security vulnerability.
For obvious reasons, we will not be disclosing the nature of the vulnerability for the next few weeks, we wish to give everyone a chance to upgrade before opening the doors to exploitation. We will say however that it is not related to the blockchain code, it is related to the frontend code.
A critical security vulnerability has been discovered.
This is a vulnerability that has existed in all versions prior to the current version. Even if you downloaded 0.3.3.2, you should download it again. The original download for 0.3.3.2 was still affected by the security vulnerability.
For obvious reasons, we will not be disclosing the nature of the vulnerability for the next few weeks, we wish to give everyone a chance to upgrade before opening the doors to exploitation. We will say however that it is not related to the blockchain code, it is related to the frontend code.
git cloned both repo in MinGV (win7 x64 - I always compile miner in such way)
but "make" for gpu-miner/"make release" for sia - doesn`t work.
but "make" for gpu-miner/"make release" for sia - doesn`t work.
Code:
xxxx@rap ~/sia
$ make release
go install -a ./...
make: go: Command not found
make: *** [release] Error 127
To make Sia you are going to need a golang compiler, this error means you don't have a go compiler.
Code:
xxxx@rap ~/sia-gpu-miner
$ make
gcc gpu-miner.o network.o -o gpu-miner -lOpenCL -lcurl
gpu-miner.o:gpu-miner.c:(.text+0x13e): undefined reference to `_imp__curl_easy_i
nit'
network.o:network.c:(.text+0x16a): undefined reference to `_imp__curl_easy_reset
'
network.o:network.c:(.text+0x18d): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1b0): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1d2): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1df): undefined reference to `_imp__curl_easy_perfo
rm'
network.o:network.c:(.text+0x1f5): undefined reference to `_imp__curl_easy_strer
ror'
network.o:network.c:(.text+0x32b): undefined reference to `_imp__curl_easy_reset
'
network.o:network.c:(.text+0x34e): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x371): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x39a): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x3bc): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x3c9): undefined reference to `_imp__curl_easy_perfo
rm'
network.o:network.c:(.text+0x3df): undefined reference to `_imp__curl_easy_strer
ror'
c:/mingw/bin/../lib/gcc/mingw32/4.8.1/../../../../mingw32/bin/ld.exe: network.o:
bad reloc address 0x20 in section `.eh_frame'
c:/mingw/bin/../lib/gcc/mingw32/4.8.1/../../../../mingw32/bin/ld.exe: final link
failed: Invalid operation
collect2.exe: error: ld returned 1 exit status
make: *** [gpu-miner] Error 1
any suggestions?
This makes it look like you don't have libCurl, which is how the miner talks to the daemon.
so how is one to compile sia miner using go? ...
apart from the golang and dependencies that get installed with yum - what are the commands and procedures to compile sia gpuminer? ...
#crysx
Er... you don't. It's not in Go; it's in C.
well - thats what im asking wolf ...
how on earth do you compile this thing ...
it seems WAY different to any of the compiles i have had to do before ...
and reading earlier - taek was saying that you needed the golang compiler to do so ...
so now im confused as ever ...
#crysx
On Linux? A simple make should do it. As long as libcurl and the opencl libs aren't in stupid places, it'll work.
wolf ... this is fedora ...
OF COURSE they are in stupid places! ...
-------
[root@ace-test-cm Sia-GPU-Miner] $ make
gcc -c -s -std=c11 -Wall -pedantic gpu-miner.c -o gpu-miner.o
gpu-miner.c:20:19: fatal error: CL/cl.h: No such file or directory
#include
-------
thats error everytime ...
#crysx
Right, well, you need to find it, then. If you can build SGMiner, then you will be able to build this.
You'll need to find it, though. First, if OpenCL headers aren't installed, you'll need 'em. They're only needed for building OpenCL - not running it - so you may not have it already. Install them, do a "find /usr -name cl.h", and add that to the include path in the CFLAGS var in the makefile. If you have another problem after this at the link step, it's just another find command away, then add that to the LFLAGS variable in the makefile to make the linker search there.
tanx wolf ...
all good
...#crysx
Good to hear; can be a minor pain without configure around to find stuff automatically.
exactly ...
if ./configure was there and active - it may have made the whole process a lot easier ...
eitherway - its a learning experience ...
i will be working on setting up a wallet / app on the storage server some time soon - too much on my plate at the moment ... this server will have at least 5TB of RAIDed storage for immediate use / testing ...
so i hope i wont have to go through the same compilation issues as this miner ... its a production system ( that is being decommissioned in our network ) - so i dont want to stuff around with it too much ...
scientific linux x64 - and rock solid ...
may even setup a separate network for this storage facility altogether ...
#crysx
git cloned both repo in MinGV (win7 x64 - I always compile miner in such way)
but "make" for gpu-miner/"make release" for sia - doesn`t work.
but "make" for gpu-miner/"make release" for sia - doesn`t work.
Code:
xxxx@rap ~/sia
$ make release
go install -a ./...
make: go: Command not found
make: *** [release] Error 127
To make Sia you are going to need a golang compiler, this error means you don't have a go compiler.
Code:
xxxx@rap ~/sia-gpu-miner
$ make
gcc gpu-miner.o network.o -o gpu-miner -lOpenCL -lcurl
gpu-miner.o:gpu-miner.c:(.text+0x13e): undefined reference to `_imp__curl_easy_i
nit'
network.o:network.c:(.text+0x16a): undefined reference to `_imp__curl_easy_reset
'
network.o:network.c:(.text+0x18d): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1b0): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1d2): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1df): undefined reference to `_imp__curl_easy_perfo
rm'
network.o:network.c:(.text+0x1f5): undefined reference to `_imp__curl_easy_strer
ror'
network.o:network.c:(.text+0x32b): undefined reference to `_imp__curl_easy_reset
'
network.o:network.c:(.text+0x34e): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x371): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x39a): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x3bc): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x3c9): undefined reference to `_imp__curl_easy_perfo
rm'
network.o:network.c:(.text+0x3df): undefined reference to `_imp__curl_easy_strer
ror'
c:/mingw/bin/../lib/gcc/mingw32/4.8.1/../../../../mingw32/bin/ld.exe: network.o:
bad reloc address 0x20 in section `.eh_frame'
c:/mingw/bin/../lib/gcc/mingw32/4.8.1/../../../../mingw32/bin/ld.exe: final link
failed: Invalid operation
collect2.exe: error: ld returned 1 exit status
make: *** [gpu-miner] Error 1
any suggestions?
This makes it look like you don't have libCurl, which is how the miner talks to the daemon.
so how is one to compile sia miner using go? ...
apart from the golang and dependencies that get installed with yum - what are the commands and procedures to compile sia gpuminer? ...
#crysx
Er... you don't. It's not in Go; it's in C.
well - thats what im asking wolf ...
how on earth do you compile this thing ...
it seems WAY different to any of the compiles i have had to do before ...
and reading earlier - taek was saying that you needed the golang compiler to do so ...
so now im confused as ever ...
#crysx
On Linux? A simple make should do it. As long as libcurl and the opencl libs aren't in stupid places, it'll work.
wolf ... this is fedora ...
OF COURSE they are in stupid places! ...
-------
[root@ace-test-cm Sia-GPU-Miner] $ make
gcc -c -s -std=c11 -Wall -pedantic gpu-miner.c -o gpu-miner.o
gpu-miner.c:20:19: fatal error: CL/cl.h: No such file or directory
#include
-------
thats error everytime ...
#crysx
Right, well, you need to find it, then. If you can build SGMiner, then you will be able to build this.
You'll need to find it, though. First, if OpenCL headers aren't installed, you'll need 'em. They're only needed for building OpenCL - not running it - so you may not have it already. Install them, do a "find /usr -name cl.h", and add that to the include path in the CFLAGS var in the makefile. If you have another problem after this at the link step, it's just another find command away, then add that to the LFLAGS variable in the makefile to make the linker search there.
tanx wolf ...
all good
...#crysx
git cloned both repo in MinGV (win7 x64 - I always compile miner in such way)
but "make" for gpu-miner/"make release" for sia - doesn`t work.
but "make" for gpu-miner/"make release" for sia - doesn`t work.
Code:
xxxx@rap ~/sia
$ make release
go install -a ./...
make: go: Command not found
make: *** [release] Error 127
To make Sia you are going to need a golang compiler, this error means you don't have a go compiler.
Code:
xxxx@rap ~/sia-gpu-miner
$ make
gcc gpu-miner.o network.o -o gpu-miner -lOpenCL -lcurl
gpu-miner.o:gpu-miner.c:(.text+0x13e): undefined reference to `_imp__curl_easy_i
nit'
network.o:network.c:(.text+0x16a): undefined reference to `_imp__curl_easy_reset
'
network.o:network.c:(.text+0x18d): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1b0): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1d2): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1df): undefined reference to `_imp__curl_easy_perfo
rm'
network.o:network.c:(.text+0x1f5): undefined reference to `_imp__curl_easy_strer
ror'
network.o:network.c:(.text+0x32b): undefined reference to `_imp__curl_easy_reset
'
network.o:network.c:(.text+0x34e): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x371): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x39a): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x3bc): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x3c9): undefined reference to `_imp__curl_easy_perfo
rm'
network.o:network.c:(.text+0x3df): undefined reference to `_imp__curl_easy_strer
ror'
c:/mingw/bin/../lib/gcc/mingw32/4.8.1/../../../../mingw32/bin/ld.exe: network.o:
bad reloc address 0x20 in section `.eh_frame'
c:/mingw/bin/../lib/gcc/mingw32/4.8.1/../../../../mingw32/bin/ld.exe: final link
failed: Invalid operation
collect2.exe: error: ld returned 1 exit status
make: *** [gpu-miner] Error 1
any suggestions?
This makes it look like you don't have libCurl, which is how the miner talks to the daemon.
so how is one to compile sia miner using go? ...
apart from the golang and dependencies that get installed with yum - what are the commands and procedures to compile sia gpuminer? ...
#crysx
Er... you don't. It's not in Go; it's in C.
well - thats what im asking wolf ...
how on earth do you compile this thing ...
it seems WAY different to any of the compiles i have had to do before ...
and reading earlier - taek was saying that you needed the golang compiler to do so ...
so now im confused as ever ...
#crysx
On Linux? A simple make should do it. As long as libcurl and the opencl libs aren't in stupid places, it'll work.
wolf ... this is fedora ...
OF COURSE they are in stupid places! ...
-------
[root@ace-test-cm Sia-GPU-Miner] $ make
gcc -c -s -std=c11 -Wall -pedantic gpu-miner.c -o gpu-miner.o
gpu-miner.c:20:19: fatal error: CL/cl.h: No such file or directory
#include
-------
thats error everytime ...
#crysx
git cloned both repo in MinGV (win7 x64 - I always compile miner in such way)
but "make" for gpu-miner/"make release" for sia - doesn`t work.
but "make" for gpu-miner/"make release" for sia - doesn`t work.
Code:
xxxx@rap ~/sia
$ make release
go install -a ./...
make: go: Command not found
make: *** [release] Error 127
To make Sia you are going to need a golang compiler, this error means you don't have a go compiler.
Code:
xxxx@rap ~/sia-gpu-miner
$ make
gcc gpu-miner.o network.o -o gpu-miner -lOpenCL -lcurl
gpu-miner.o:gpu-miner.c:(.text+0x13e): undefined reference to `_imp__curl_easy_i
nit'
network.o:network.c:(.text+0x16a): undefined reference to `_imp__curl_easy_reset
'
network.o:network.c:(.text+0x18d): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1b0): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1d2): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1df): undefined reference to `_imp__curl_easy_perfo
rm'
network.o:network.c:(.text+0x1f5): undefined reference to `_imp__curl_easy_strer
ror'
network.o:network.c:(.text+0x32b): undefined reference to `_imp__curl_easy_reset
'
network.o:network.c:(.text+0x34e): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x371): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x39a): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x3bc): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x3c9): undefined reference to `_imp__curl_easy_perfo
rm'
network.o:network.c:(.text+0x3df): undefined reference to `_imp__curl_easy_strer
ror'
c:/mingw/bin/../lib/gcc/mingw32/4.8.1/../../../../mingw32/bin/ld.exe: network.o:
bad reloc address 0x20 in section `.eh_frame'
c:/mingw/bin/../lib/gcc/mingw32/4.8.1/../../../../mingw32/bin/ld.exe: final link
failed: Invalid operation
collect2.exe: error: ld returned 1 exit status
make: *** [gpu-miner] Error 1
any suggestions?
This makes it look like you don't have libCurl, which is how the miner talks to the daemon.
so how is one to compile sia miner using go? ...
apart from the golang and dependencies that get installed with yum - what are the commands and procedures to compile sia gpuminer? ...
#crysx
Er... you don't. It's not in Go; it's in C.
well - thats what im asking wolf ...
how on earth do you compile this thing ...
it seems WAY different to any of the compiles i have had to do before ...
and reading earlier - taek was saying that you needed the golang compiler to do so ...
so now im confused as ever ...
#crysx
git cloned both repo in MinGV (win7 x64 - I always compile miner in such way)
but "make" for gpu-miner/"make release" for sia - doesn`t work.
but "make" for gpu-miner/"make release" for sia - doesn`t work.
Code:
xxxx@rap ~/sia
$ make release
go install -a ./...
make: go: Command not found
make: *** [release] Error 127
To make Sia you are going to need a golang compiler, this error means you don't have a go compiler.
Code:
xxxx@rap ~/sia-gpu-miner
$ make
gcc gpu-miner.o network.o -o gpu-miner -lOpenCL -lcurl
gpu-miner.o:gpu-miner.c:(.text+0x13e): undefined reference to `_imp__curl_easy_i
nit'
network.o:network.c:(.text+0x16a): undefined reference to `_imp__curl_easy_reset
'
network.o:network.c:(.text+0x18d): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1b0): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1d2): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x1df): undefined reference to `_imp__curl_easy_perfo
rm'
network.o:network.c:(.text+0x1f5): undefined reference to `_imp__curl_easy_strer
ror'
network.o:network.c:(.text+0x32b): undefined reference to `_imp__curl_easy_reset
'
network.o:network.c:(.text+0x34e): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x371): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x39a): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x3bc): undefined reference to `_imp__curl_easy_setop
t'
network.o:network.c:(.text+0x3c9): undefined reference to `_imp__curl_easy_perfo
rm'
network.o:network.c:(.text+0x3df): undefined reference to `_imp__curl_easy_strer
ror'
c:/mingw/bin/../lib/gcc/mingw32/4.8.1/../../../../mingw32/bin/ld.exe: network.o:
bad reloc address 0x20 in section `.eh_frame'
c:/mingw/bin/../lib/gcc/mingw32/4.8.1/../../../../mingw32/bin/ld.exe: final link
failed: Invalid operation
collect2.exe: error: ld returned 1 exit status
make: *** [gpu-miner] Error 1
any suggestions?
This makes it look like you don't have libCurl, which is how the miner talks to the daemon.
so how is one to compile sia miner using go? ...
apart from the golang and dependencies that get installed with yum - what are the commands and procedures to compile sia gpuminer? ...
#crysx
Still can't start this new miner release "as is", Sia_GPU-Miner-v1.0.1-win64. Why does it seems that Wolf0 is smarter than developers of this miner?..
Win7 x64, 14.6 RC catalyst driver.
With the last version you can specify -p1 : the opencl platform is usually 0 but in my case it was 1 : I had the same error than you, and now it works. Quote
D:\Distrib\Siacoin\Sia_GPU-Miner-v1.0.1-win64>sia_GPU-Miner.exe
Initializing...failed to get Device IDs: -1
Initializing...failed to get Device IDs: -1
Win7 x64, 14.6 RC catalyst driver.
Still can't start this new miner release "as is", Sia_GPU-Miner-v1.0.1-win64. Why does it seems that Wolf0 is smarter than developers of this miner?..
Because Wolf0 has been doing gpu mining for a long time, and knows his way around miners much better than we do. When it comes to mining, Wolf0 definitely knows more than we do.
On that note, the master branch of the official Sia miner has some recent changes + optimizations, some of which were made by Wolf0 himself: https://github.com/NebulousLabs/Sia-GPU-Miner
Jump to: