OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.0 - page 829.

ancow

full member

Activity: 373

Merit: 100

Latest git, on my 32bit Kubuntu 11.04 segfaults when executing "cgminer --help". GDB's backtrace returns:

Code:

#0  0x0024701a in strncpy () from /lib/i386-linux-gnu/libc.so.6
#1  0x08049e8d in show_algo (buf=0xbffff938 "\324\371\377\277t\332\022", algo=0x80714f8) at /usr/include/bits/string3.h:121
#2  0x0806c996 in opt_usage ()
#3  0x0806b509 in opt_usage_and_exit ()
#4  0x0806c5c5 in parse_one ()
#5  0x0806c1dc in opt_parse ()
#6  0x08050206 in main (argc=2, argv=0xbffffcb4) at main.c:3260

smurfix

newbie

Activity: 23

Merit: 0

Quote from: ancow on July 23, 2011, 11:55:32 AM

As ckolivas already mentioned

Bah. I was sure I've read the whole thread.

Thanks.

ancow

full member

Activity: 373

Merit: 100

Quote from: smurfix on July 23, 2011, 11:49:37 AM

Hi,

has anybody gotten cgminer to run on the console (i.e., no running X)?

Tracing the cgminer process, it gets as far as opening /dev/ati/card0 and running a few ioctl calls (and incidentally _not_ opening /dev/ati/card1, despite the fact that I have two of them), but then clGetDeviceIDs() returns -1. Ouch.

As ckolivas already mentioned, you need X for the drivers to communicate with the graphics card. See http://forum.bitcoin.org/index.php?topic=28402.msg387438#msg387438 and answers.

smurfix

newbie

Activity: 23

Merit: 0

Hi,

has anybody gotten cgminer to run on the console (i.e., no running X)?

Tracing the cgminer process, it gets as far as opening /dev/ati/card0 and running a few ioctl calls (and incidentally _not_ opening /dev/ati/card1, despite the fact that I have two of them), but then clGetDeviceIDs() returns -1. Ouch.

twmz

hero member

Activity: 737

Merit: 500

Quote from: -ck on July 23, 2011, 09:28:58 AM

I was aware phatk was incompatible with quite a few things, which is why I kept the poclbm kernel in there. Some of the detection changes might select phatk now if they hadn't previously, so perhaps that's the issue. Try explicitly setting the kernel (new feature in 1.4.0) to poclbm with -k poclbm

That didn't make a difference. Same error as before, so apparently not phatk kernel related. Let me know what other information I can provide.

P.S. My error is the same as what ancow is seeing, and it looks like he was able to provide a stacktrace of the segfault.

twmz

hero member

Activity: 737

Merit: 500

Quote from: -ck on July 23, 2011, 09:28:58 AM

I was aware phatk was incompatible with quite a few things, which is why I kept the poclbm kernel in there. Some of the detection changes might select phatk now if they hadn't previously, so perhaps that's the issue. Try explicitly setting the kernel (new feature in 1.4.0) to poclbm with -k poclbm

I'll try that, later today.

ancow

full member

Activity: 373

Merit: 100

I get a segfault, too (on OS X 10.6.8 ).
Here's the output + command line:

Code:

cgminer-1.4.0> gdb --args ./cgminer --url http://pit.deepbit.net:8332 -u XXX -p XXX --url http://swepool.net-u XXX -p XXX -T
[2011-07-23 17:35:59] Long-polling activated for http://pit.deepbit.net:8332/listenChannel                    
[2011-07-23 17:35:59] Creating Command Queue. (clCreateCommandQueue)                    
[2011-07-23 17:35:59] Failed to init GPU thread 0                    
[2011-07-23 17:35:59] Creating Command Queue. (clCreateCommandQueue)                    
[2011-07-23 17:35:59] Failed to init GPU thread 0                    
 GPU 0: [0.0 Mh/s] [Q:0  A:0  R:0  HW:0  E:0%  U:0.00/m]

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_INVALID_ADDRESS at address: 0x0000000000000038
text_print_status [inlined] () at main.c:738
738             printf(" %sPU %d: [%.1f Mh/s] [Q:%d  A:%d  R:%d  HW:%d  E:%.0f%%  U:%.2f/m]\n",

And the trace:

Code:

(gdb) backtrace
#0  text_print_status [inlined] () at main.c:738
#1  print_status [inlined] () at /Users/XXX/cgminer-1.4.0/main.c:804
#2  0x00000001000041ef in main (argc=1, argv=) at main.c:3516

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: twmz on July 23, 2011, 09:25:10 AM

1.4.0 seg faults on OS X (Lion)

1.3.1 worked fine (with -v 1)

With 1.4.0, part of the screen draws before the segfault, so here is a screen grab in case the location that "Segmentation Fault 11" draws might be an indication of where the crash is happening (there are also some errors in the message log):

http://i.imgur.com/8EW8T.png

Note, if you are not aware, the phatk kernel in the latest poclbm and phoenix don't work with Apples Open CL implementation (and never have). But your kernel did work until 1.4.0. Not sure if your kernel changes in 1.4.0 triggered this problem or if it was changes in the C code that triggered the problem. Let me know what you'd like me to do to help debug this, because cgminer is (was) one of the only miners working in OS X Lion.

I was aware phatk was incompatible with quite a few things, which is why I kept the poclbm kernel in there. Some of the detection changes might select phatk now if they hadn't previously, so perhaps that's the issue. Try explicitly setting the kernel (new feature in 1.4.0) to poclbm with -k poclbm

twmz

hero member

Activity: 737

Merit: 500

1.4.0 seg faults on OS X (Lion)

1.3.1 worked fine (with -v 1)

With 1.4.0, part of the screen draws before the segfault, so here is a screen grab in case the location that "Segmentation Fault 11" draws might be an indication of where the crash is happening (there are also some errors in the message log):

http://i.imgur.com/8EW8T.png

Note, if you are not aware, the phatk kernel in the latest poclbm and phoenix don't work with Apples Open CL implementation (and never have). But your kernel did work until 1.4.0. Not sure if your kernel changes in 1.4.0 triggered this problem or if it was changes in the C code that triggered the problem. Let me know what you'd like me to do to help debug this, because cgminer is (was) one of the only miners working in OS X Lion.

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: zaytsev on July 23, 2011, 08:32:51 AM

Re. config format: JSON + arrays is absolutely awesome + you don't have to parse it, because you are using a library, why even think of changing?

Uh yeah, about that. I haven't got the foggiest how to talk [to] that library yet... There was so much I wanted to code for cgminer first that I knew I could do before looking at it (as you can probably see from the hundreds of hours of code I've put in already).

ffwd

newbie

Activity: 17

Merit: 0

Quote from: -ck on July 23, 2011, 05:21:05 AM

Specifically it searches for SSE4.1 and I don't know how that differs from sse4.2 and sse4a

looking at the asm code - it uses "movntdqa" which is a 4.1 instruction - that means intel (only) processors starting with core 2 and newer (core i3, i5, i7)
it also looks like its the only instruction in the code that requires sse4. that explains the marginal difference with sse2 code

perhaps someone more familiar with asm will find a way to use the higher sse sets much more effectively, but for now, the difference between sse2 and sse4(.1) implementation is not worth switching cpu's )

just for reference:

sse 4.2 is intel (only) core i3, i5, i7
sse 4a is amd only starting with barcelona arch

zaytsev

newbie

Activity: 59

Merit: 0

Quote from: Sukrim on July 23, 2011, 08:37:18 AM

Could somebody who uses Arch Linux please update this pkgbuild for cgminer on AUR?

Try this:

Quote

pkgname=cgminer
pkgver=1.4.0
pkgrel=1
pkgdesc="Con Kolivas's fork of cpuminer. http://forum.bitcoin.org/index.php?topic=28402.0"
arch=('i686' 'x86_64')
url='https://github.com/ckolivas/cgminer'
license=('GPL')
depends=('curl')
makedepends=('yasm')
source=("http://ck.kolivas.org/apps/cgminer/cgminer-${pkgver}.tar.bz2")
md5sums=('725d9e499aa4bdbe307250180259d920')

build() {
cd cgminer-${pkgver}

./configure \
--prefix=/usr

make CFLAGS="-O3 -march=native -mtune=native"
}

package() {
cd cgminer-${pkgver}
make DESTDIR=${pkgdir} install
install -Dm600 "example-cfg.json" "${pkgdir}/etc/cpuminer.json"
}

I've never actually seen Arch, but should work BTC =)

P.S. RPM for EL6

xcooling

member

Activity: 145

Merit: 10

Quote from: zaytsev on July 23, 2011, 08:32:51 AM

Quote from: xcooling on July 23, 2011, 08:20:35 AM

your a hardcore c programmer ;-P

You know, it's the same guy that is known for Linux kernel schedulers outside of this forum... No wonder.

Re. config format: JSON + arrays is absolutely awesome + you don't have to parse it, because you are using a library, why even think of changing?

I agree 100% on json, use it for all my php scripts. However you can dump the varibles into a binary database and load them, it can be easier if options are not formalized.

Yeah the CK kernel patch is legend.

Sukrim

legendary

Activity: 2618

Merit: 1007

Could somebody who uses Arch Linux please update this pkgbuild for cgminer on AUR?
https://aur.archlinux.org/packages.php?ID=50710

ffwd

newbie

Activity: 17

Merit: 0

Quote from: xcooling on July 23, 2011, 08:15:15 AM

visualstudio doesnt play nice with mingw

try code::blocks ide. been using it since my first C course

zaytsev

newbie

Activity: 59

Merit: 0

Quote from: xcooling on July 23, 2011, 08:20:35 AM

your a hardcore c programmer ;-P

You know, it's the same guy that is known for Linux kernel schedulers outside of this forum... No wonder.

Re. config format: JSON + arrays is absolutely awesome + you don't have to parse it, because you are using a library, why even think of changing?

xcooling

member

Activity: 145

Merit: 10

your a hardcore c programmer ;-P

i code php like that. but not c/c++

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

I don't use an IDE. Just a text editor (kate) on linux.

xcooling

member

Activity: 145

Merit: 10

you could save it in binary format ;-P

what IDE are you using for dev ? might take a look at it. visualstudio doesnt play nice with mingw

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

I was considering that long term. It's just that with the code and features changing so rapidly, deciding on a format for the config file that's easy to use and parse is the real issue. Currently it accepts some basic json formatted config file and working on parsing code is not remotely my strong point Undecided

Topic: OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.0 - page 829. (Read 5805677 times)