New demonstration CPU miner available

[jgarzik]

Long polling was just added to git, and I posted a test Windows version here:
     http://yyz.us/bitcoin/cpuminer-installer-0.7.2-lp1.zip

Test feedback appreciated, on Linux (build your own from git) or Windows.  http://deepbit.net/ is known to support long polling, and slush's pool should be able to do it soon, too.

This should increase people's CPU mining efficiency quite a bit, while also reducing network usage (and load on pool servers).

[Arceny]

Hello

I have a problems with compilation 0.7.2 under FreeBSD 8.1, gcc version 4.2.1 20070719  [FreeBSD]

Code:

gcc -DHAVE_CONFIG_H -I. -pthread -fno-strict-aliasing -I./compat/jansson     -I/usr/local/include -MT sha256_4way.o -MD -MP -MF .deps/sha256_4way.Tpo -c -o sha256_4way.o sha256_4way.c
sha256_4way.c: In function 'ROTR':
sha256_4way.c:51: error: shift must be an immediate
sha256_4way.c:51: error: shift must be an immediate
sha256_4way.c: In function 'SHR':
sha256_4way.c:55: error: shift must be an immediate

Any ideas

[Arceny]

Hello

I have a problems with compilation 0.7.2 under FreeBSD 8.1, gcc version 4.2.1 20070719  [FreeBSD]

gcc 4.5 will fix this problem.

But speed is to slow for 2x Xeon E5420, only 133 khash per thread (8 threads total). In linux this will be at least ~1000 khash/thread speed.

[TurdHurdur]

Sometimes I'll get a "Floating point exception" which kills it on my Debian 6 VM. Anything build-wise I should be doing to avoid that?

[jgarzik]

Sometimes I'll get a "Floating point exception" which kills it on my Debian 6 VM. Anything build-wise I should be doing to avoid that?

Which algorithm are you using?

Which cpuminer version?  Longpolling/git or a release version?

[TurdHurdur]

Which algorithm are you using?

Which cpuminer version?  Longpolling/git or a release version?

I'm using sse2_64 with the Longpolling/git version.

[jgarzik]

Which algorithm are you using?

Which cpuminer version?  Longpolling/git or a release version?

I'm using sse2_64 with the Longpolling/git version.

Does it happen with any other algorithm, such as 4way?

[jgarzik]

Sometimes I'll get a "Floating point exception" which kills it on my Debian 6 VM. Anything build-wise I should be doing to avoid that?

I spotted one potential problem source, and pushed a fix out to git.

[jgarzik]

Version 0.8 released.  See top of thread for URLs.

Changes:
- Support long polling (beta): http://deepbit.net/longpolling.php  By default, cpuminer will enable long-polling automatically, if the X-Long-Polling HTTP header is present.  To disable this autodetection, pass option --no-longpoll.

- Adjust max workload based on --scantime (default 5 seconds, or 60 seconds for longpoll)

- Standardize program output, and support syslog on Unix platforms

- Suport --user/--pass options (and "user" and "pass" in config file), as an alternative to the current --userpass

SHA1: 2b89588a99912d3980a9d5871b439fdbbd1d806b  cpuminer-installer-0.8.zip
MD5: 12464cf941dee3e317d98c643dfaee5a  cpuminer-installer-0.8.zip

[Arceny]

Why sse2_amd64 algorithm did not compiling? My system is ubuntu 10.04 64 bit and Core2Duo P8400 cpu.

As I can see in miner.h there is a definition WANT_X8664_SSE2, with checking sse2 support. But checking don't passing. When i comment this checking and try to 'make' there is an error:

Code:

gcc  -g -O2 -pthread  -o minerd cpu-miner.o util.o sha256_generic.o sha256_4way.o sha256_via.o sha256_cryptopp.o sha256_sse2_amd64.o -lcurl compat/jansson/libjansson.a -lpthread  
sha256_sse2_amd64.o: In function `scanhash_sse2_64':
sha256_sse2_amd64.c:98: undefined reference to `CalcSha256_x64'
sha256_sse2_amd64.c:99: undefined reference to `CalcSha256_x64'
collect2: ld returned 1 exit status

 

[chromicant]

Why sse2_amd64 algorithm did not compiling? My system is ubuntu 10.04 64 bit and Core2Duo P8400 cpu.

As I can see in miner.h there is a definition WANT_X8664_SSE2, with checking sse2 support. But checking don't passing. When i comment this checking and try to 'make' there is an error:

Don't do that. There's a reason the check is failing.

If you looked at config.log, it probably says your version of yasm is 0.8.0. You need 1.1.0. For Ubuntu, you either compile it yourself or get the right version from Debian's repository.

[Arceny]

Thanks, yasm 1.1 fix my problem.

[adv]

- Suport --user/--pass options (and "user" and "pass" in config file), as an alternative to the current --userpass

Minor typo in usage help:

Code:

        { "pass PASSWORD",
          "(-p USERNAME) Password for bitcoin JSON-RPC server "
          "(default: " DEF_RPC_PASSWORD ")" },

must be "(-p PASSWORD)".

BTW, may be you want get manpage SGML template from my source deb and put it into your source package? It's under GPL.
For build man from SGML i use docbook-to-man:

Code:

docbook-to-man debian/btc-cpuminer.sgml > debian/btc-cpuminer.1

Link to source deb: http://home.agalakov.spb.ru/debian/pool/main/b/btc-cpuminer/btc-cpuminer_0.8-2.debian.tar.gz
Or I can send it to you in some other way.

[FnuGk]

for a complete newb how do i go around to make this work on mac OSx 10.6?

[Patches]

for a complete newb how do i go around to make this work on mac OSx 10.6?

As far as I know no one has been able to easily compile and get this to run on OS X natively. However, you can run the miner (unfortunately without ufasoft's sse2 optimizations) using the following steps:

1) Install MacPorts (here:http://www.macports.org/install.php). This software will allow you to easily install various linux packages on your Mac.
2) Open up the terminal, and run 'sudo port -v selfupdate' to get the latest package info and then 'sudo port install wine'. This will install wine, an application that will let you run windows programs.
3) Download the windows installer of the miner (from here: http://bitcointalk.org/index.php?topic=1925.msg24217#msg24217) and unzip it. Run it in the terminal with 'wine cpuminer-installer.exe' and make sure to choose a directory you will remember (I just made one on the desktop). If you aren't sure the full path to cpuminer-installer.exe, just drag the file from the finder window to the terminal window and terminal will add it.
4) Run the miner in the terminal with 'wine /yourdirectory/minerd.exe --help' or with whatever command line options you decide. I found the cryptopp_asm32 algorithm to give my MacPro the highest hashing rates.

Happy mining.

[chromicant]

As far as I know no one has been able to easily compile and get this to run on OS X natively. However, you can run the miner (unfortunately without ufasoft's sse2 optimizations) using the following steps:

Urm. Have you tried to compile it natively using XCode on OSX 10.6 on a machine that can handle SSE2? I'd like to see some bug reports as to why it won't compile. I don't think there's anything super specific that would prevent a build from happening. I've seen nothing on what sort of compile errors one gets on OSX. Also remember, I crafted the SSE2 code in jgarzik's miner to require the 64-bit mode of Intel/AMD chips. So if you're running a 32-bit kernel/userspace, you won't be able to use it.

(I'd try myself, but my Mac is in little bits...and it's just a CoreDuo, so I couldn't test the SSE2 code if I wanted to)

I do know my SSE2 WILL NOT WORK in win64, since the calling ABI is different.

The one place where there probably is a problem is I used ELF64 as the output for yasm. That should be Mach-O for OSX...but it may also just work.

[Patches]

As far as I know no one has been able to easily compile and get this to run on OS X natively. However, you can run the miner (unfortunately without ufasoft's sse2 optimizations) using the following steps:

Urm. Have you tried to compile it natively using XCode on OSX 10.6 on a machine that can handle SSE2? I'd like to see some bug reports as to why it won't compile. I don't think there's anything super specific that would prevent a build from happening. I've seen nothing on what sort of compile errors one gets on OSX. Also remember, I crafted the SSE2 code in jgarzik's miner to require the 64-bit mode of Intel/AMD chips. So if you're running a 32-bit kernel/userspace, you won't be able to use it.

(I'd try myself, but my Mac is in little bits...and it's just a CoreDuo, so I couldn't test the SSE2 code if I wanted to)

I do know my SSE2 WILL NOT WORK in win64, since the calling ABI is different.

The one place where there probably is a problem is I used ELF64 as the output for yasm. That should be Mach-O for OSX...but it may also just work.

I have only spent a couple minutes playing with it but I got stuck at the same byteswap.h problem described previously in this thread (see: http://bitcointalk.org/index.php?topic=1925.msg48679#msg48679). I am sure it is possible so I'll give it another go when I have some time.

[jgarzik]

I have only spent a couple minutes playing with it but I got stuck at the same byteswap.h problem described previously in this thread (see: http://bitcointalk.org/index.php?topic=1925.msg48679#msg48679). I am sure it is possible so I'll give it another go when I have some time.

Those with the ability to test can add

     AC_CHECK_HEADERS(byteswap.h)

to configure.ac, and then check

     #ifdef HAVE_BYTESWAP_H
     ... current code ...
     #else
     ... OSX code ...
     #endif

after running ./autogen.sh with proper autoconf/libcurl/etc. build dependencies, to regenerate configure.

[chromicant]

I have only spent a couple minutes playing with it but I got stuck at the same byteswap.h problem described previously in this thread (see: http://bitcointalk.org/index.php?topic=1925.msg48679#msg48679). I am sure it is possible so I'll give it another go when I have some time.

Hmm. Looks like it'll be a 5 line patch to fix that. Something like (the 5 lines plus the rest of the define):

Code:

#if defined(__APPLE__)
#include <libkern/OSByteOrder.h>
#define bswap_16 OSSwapInt16
#define bswap_32 OSSwapInt32
#define bswap_64 OSSwapInt64
#elif ((__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3))
#define WANT_BUILTIN_BSWAP
#else
#include <byteswap.h>
#endif

Taken from XOrg, which had the same problem on OSX.

...and I think jgarzik's solution needs to be massaged into the OSX code above for maximum cleanliness!

[jgarzik]

I think byte swapping is simple enough that we can afford to create compat/byteswap.h, and include inline versions of byteswap macros for the case where the OS does not provide.  This seems like an area of breakage that we could solve permanently.

#ifdef __APPLE__ is fine, but we should provide fallbacks nonetheless.

Volunteers?