[ANN] ptsweb.beeeeer.org - Protoshares mining sub-pool

[tony29]

OK,thanks  jabbaah, I will send you 1 pts.

[skaffen]

can someone on a i7 with SSE4 but -NO- AVX check if replacing
-march=native
with
-march=corei7
in
makefile.unix

then compile, then run, helps?

- xolokram

ps. sorry, i'm currently not able to test is

I'm on a VPS that disables avx.  I also had to comment out the selection of the avx assembly code in sha512.c:

   if (pc->proc_type == PROC_X64_INTEL || pc->proc_type == PROC_X64_AMD) {
      /*if (pc->avx_level > 0) {
         sha512_update_func = sha512_avx;
      } else*/ if (pc->sse_level >= 4) {

[jabbaah]

I am running a CentOS 5.7 64 on cloud server , does it matter?

Just before you stated to run CentOS 6.2.

I guess CentOS 5.7 is way too old... you won't have all necessary dependencies available.

[tony29]

I am running a CentOS 5.7 64 on cloud server , does it matter?

[tony29]

yes i have installed yasm ,it give me this information when i execute :

[root@AY131122190016007b5aZ src]# yasm
yasm: No input files specified

just now I found a patch from google, and added belowing code to sha512.c:
#include <sys/param.h>
 #ifdef _BSD
 #include <sys/endian.h>
 #elif defined(__linux__)
 #include <byteswap.h>
 # if __BYTE_ORDER == __LITTLE_ENDIAN
 # ifndef htobe64
 # define htobe64(x) bswap_64 (x)
 # endif
 # ifndef be64toh
 # define be64toh(x) bswap_64 (x)
 # endif
 # else
 # ifndef htobe64
 # define htobe64(x) (x)
 # endif
 # ifndef be64toh
 # define be64toh(x) (x)
 # endif
 # endif
 #endif

  #ifdef HAVE_CONFIG_H
  #include "config.h"
  #endif

but when i make it againg ,
 it shows me a lot of  "*** has not been declared" and
"*** was not declared in this scope"

[jabbaah]

hi, i am not familar with linux, could anyone help me plz. when i make the linux source code on centos6.2-64 ,it give me this information:


[root@AY131122190016007b5aZ src]# make -f makefile.unix
g++ -c -O2  -fpermissive -o obj/sha512.o sha512.c
sha512.c: In function ‘void _final(SHA512_Context*, uint8_t*, int, int)’:
sha512.c:257: error: ‘be64toh’ was not declared in this scope
make: *** [obj/sha512.o] Error 1

how can i make it work?

I'm not absolutely sure, but I guess to remember that this error message was displayed before I installed 'yasm'. Did you install 'yasm' before compilation?

Yasm is a new dependency for the 0.7 miner.

[tony29]

hi, i am not familar with linux, could anyone help me plz. when i make the linux source code on centos6.2-64 ,it give me this information:


[root@AY131122190016007b5aZ src]# make -f makefile.unix
g++ -c -O2  -fpermissive -o obj/sha512.o sha512.c
sha512.c: In function ‘void _final(SHA512_Context*, uint8_t*, int, int)’:
sha512.c:257: error: ‘be64toh’ was not declared in this scope
make: *** [obj/sha512.o] Error 1

how can i make it work?

[xolokram]

i had 1-2% on one machine and 4-6% on other machines yesterday

i would say 15% is too much

can you PM me a part of your console output of the miner with some information about REJECTED shares?

- xolokram

[soulhouser]

15 % Rejected shares is it normal or too many?

[ryen123]

Just my feedback on the v0.7. I compiled myself in ubuntu 13.10, it seems to run much slower on i7-4770k compared to v0.6 (about 1.1sh/m slower). But v0.7 runs faster on my i7-2600k. So weird.

Xolokram my own pc running native ubuntu 13.10 running v0.6 get about 190c/m & 3.9sh/m. After running v0.7, its about 160+c/m & 2.8sh/m. At the beginning of the miner it says "using avx/sse". It's quite a big difference.

Just to put things in perspective my other pc i7-2600k also running native ubuntu13.10 in v0.6 about 160c/m, in v0.7 about 190c/m. Much improvement but sh/m drops.

[xolokram]

from v0.6 to v0.7? OS? VM or native?
does it state "using sphlib" or "using AVX/SSE" at the beginning of the miner?

i have a wild guess that the changes to the makefile (esp. -march=native) are not suitable for all distros/systems/configs

- xolokram

ps. @all: don't compare sh/m, use col/m

[JohnD]

Just my feedback on the v0.7. I compiled myself in ubuntu 13.10, it seems to run much slower on i7-4770k compared to v0.6 (about 1.1sh/m slower). But v0.7 runs faster on my i7-2600k. So weird.

Have exactly the same issue. My I7-4770 dropped from 85 c/m to 70 c/m running on 2 cores.

[Aikuro]

Managed to compile it under FreeBSD 9.2 x64.
Just had to change '#include <endian.h>' to '#include <machine/endian.h>' and comment out this low-priority stuff.
Mining @167c/m using an i7 2600k, 7 threads, 2048MB/thread. Pretty acceptable I guess.

[xolokram]

can someone on a i7 with SSE4 but -NO- AVX check if replacing
-march=native
with
-march=corei7
in
makefile.unix

then compile, then run, helps?

- xolokram

ps. sorry, i'm currently not able to test is

[cholik]

0.7 AVX version works fine on 3570K & 4670K CPU's, but doesn't work on 3570 (without "K")

my i5 3570 (without "K" ) works fine for me with avx/sse. running win7 64b