Quote from: AcidCrash on November 26, 2013, 04:24:53 AM
PS: any news about compiling ptsminer on Macos Xor? Thank you for your job guys!

Sorry, i got hung up in personal stuff and evil work crap and haven't had a chance to update. I ran into a couple snags on mavericks that i don't get in 10.7/10.8, but i have a little time tonight to work on it. I'll update as soon as i know more.

I'll be on irc too if you want to troubleshoot live.

~Xornof

Quote from: AcidCrash on November 24, 2013, 11:47:22 PM
Thank you so much xornof! Waiting for your feedback!

Just got home,  I'll start looking at it in a few minutes... Will let everyone know what i find.


@xolo: FYI - I did get a chance to put the "newer" .7 RC2 windows binaries through the paces on most of my windows servers this afternoon, no issues so far. Thanks!


~Xornof

Quote from: Silver_Pharaoh on November 25, 2013, 12:44:30 AM
EDIT: can I use the older version of the pts miner still? Or will the server not count the collisions it finds? (like version 0.5 or 0.4)

I may be speaking out of turn, but AFAIK, there was no problems continuing to use the .5a Hotfix1 version, it just doesnt have the AVX/SSE4 instruction set code. I go back and use it when i am troubleshooting an issue on the .7 miner(to eliminate hardware as a cause).  I generally don't see the need to go back to the .4 or earlier versions.

Hope that helps!
X

Quote from: xolokram on November 24, 2013, 07:39:40 PM
i updated the windows binaries --- see the v0.7 post (& OP)
download: here

Testing the "newerer" binaries now.

Quote from: xolokram on November 24, 2013, 07:39:40 PM
i think the pool will switch to a score-based payout system like this soon

Something similar to slush's? That's might work out pretty well, interested to hear what you have in store for us next.

X

Quote from: AcidCrash on November 24, 2013, 05:25:45 PM
Hello, I'm trying to compile ptsminer on MacOS Maverick and I get this error:

Code Select Expand


AcidCrash ~/mining/ptsminer/src $ make -f makefile.unix
yasm -f elf64 -o obj/sha512_avx.o intel/sha512_avx.asm
yasm -f elf64 -o obj/sha512_sse4.o intel/sha512_sse4.asm
c++ -c -O2  -fpermissive -o obj/sha512.o sha512.c
clang: warning: treating 'c' input as 'c++' when in C++ mode, this behavior is deprecated
clang: warning: argument unused during compilation: '-fpermissive'
c++ -c -O1  -fpermissive -o obj/sph_sha2.o sph_sha2.c
clang: warning: treating 'c' input as 'c++' when in C++ mode, this behavior is deprecated
clang: warning: argument unused during compilation: '-fpermissive'
sph_sha2.c:33:32: warning: pragma diagnostic expected option name (e.g. "-Wundef") [-Wunknown-pragmas]
#pragma GCC diagnostic ignored "-fpermissive"
                               ^
sph_sha2.c:628:5: error: assigning to 'sph_sha224_context *' from incompatible type 'void *'
        sc = cc;
           ^ ~~
sph_sha2.c:643:5: error: assigning to 'sph_sha256_context *' (aka 'sph_sha224_context *') from incompatible type 'void *'
        sc = cc;
           ^ ~~
In file included from sph_sha2.c:655:
./sph_md_helper.c:79:32: warning: pragma diagnostic expected option name (e.g. "-Wundef") [-Wunknown-pragmas]
#pragma GCC diagnostic ignored "-fpermissive"
                               ^
./sph_md_helper.c:136:5: error: assigning to 'sph_sha224_context *' from incompatible type 'void *'
        sc = cc;
           ^ ~~
./sph_md_helper.c:186:5: error: assigning to 'sph_sha224_context *' from incompatible type 'void *'
        sc = cc;
           ^ ~~
./sph_md_helper.c:208:3: error: no matching function for call to 'sha2_round'
                RFUN(data, SPH_VAL);
                ^~~~
sph_sha2.c:652:16: note: expanded from macro 'RFUN'
#define RFUN   sha2_round
               ^~~~~~~~~~
sph_sha2.c:615:1: note: candidate function not viable: cannot convert argument of incomplete type 'const void *' to 'const unsigned char *'
sha2_round(const unsigned char *data, sph_u32 r[8])
^
In file included from sph_sha2.c:655:
./sph_md_helper.c:250:5: error: assigning to 'sph_sha224_context *' from incompatible type 'void *'
        sc = cc;
           ^ ~~
2 warnings and 6 errors generated.
make: *** [obj/sph_sha2.o] Error 1


Anyone got ptsminer working on MacOS?

There is a link in the OP to some mac compilaton instructions, IIRC they may not be specific to ptsminer...
I haven't tried to compile ptsminer for mac osx yet but i own a few, I'll give it a try on my MacBookPro on Mavericks tonight and I'll see if I can find/fix the issue you are describing.
I'll also give it a try on my OSX10.6.8 iMac too for those still behind a few releases.

X

Quote from: gsrcrxsi on November 24, 2013, 02:10:29 PM
I noticed something odd. The memory option doesn't seem to be working. On my q9550 box, which has 8Gb ram (DDR2-800), using 27 or 28 results in the same 2GB of ram usage.

See my post above, if you add the sse4 option to the end of your command(for the sse4 binary), it will then correctly pick up the memory option. I thought it was me at first, it drove me nuts until i discovered the workaround. xolokram is working on the bugfix.

In the meantime, To use 4 cores and 4GB RAM on the windows SSE4 binary,  use:
ptsminer_sse4.exe %walletaddress% 4 28 sse4

Hope that helps!

-Xornof

Quote from: xolokram on November 24, 2013, 08:58:01 AM
@xornof:
thank you for the bug report, will fix this asap
like i said these are not the final v0.7 binaries :D

No, thank you sharing your work with the community, we all really do appreciate it.

I have some baseline averages now that it has been a few days, pretty positive across the board.  I am seeing somewhere between 10-20% increase(in c/m) to the older SSE3 machines since .7 dropped. My SSE4.1 and AVX/SSE4 based machines are seeing between 20-40% (right at 40% in the case of my newer Xeon E5-4650 based 32core monsters)
DISCLAIMER: These numbers are just my observations, your results may vary...

Anyhow, i am rambling again.  :P I hope you save some personal time for you and yours, cheers! :)

Quote from: xolokram on November 24, 2013, 01:01:31 AM
new windows version out (it's not final yet, still testing)
http://www.mediafire.com/download/52fdhw2c3ny81ti/ptsminer_v07a_x64_bundle_new.zip
now containing several binaries: generic (amd / intel), sse4 and avx
for more information please read: http://bitsharestalk.org/index.php?topic=234.msg9657#msg9657

Found a minor bug(or maybe just a documentation issue) with the new win64 binaries you provided. Specifically, the SSE4 binary.
If you just specify cores and ram (and not the instruction set), the SSE binary ignores the memory setting. For example:
ptsminer_sse4.exe %walletaddress% 4  = 4core, 512MB a core
ptsminer_sse4.exe %walletaddress% 4 28 = still does 4cores, 512MB a core
ptsminer_sse4.exe %walletaddress% 4 28 sse4 = 4 cores, 1024MB a core

The AVX binary seems to work fine with just cores and ram settings in my testing. I haven't noticed anything with the intel binary, since in my testing I was using the sph switch already.

This was tested on mainly non-AVX machines running Win7 x64 Ultimate/Pro, Server 2008R2, and 1 Server 2003x64 box.
Proc wise, here are the non-avx ones: 2 C2Q Q6600(SSE3), 1 C2Q Q9550, 1 C2E X7900(SSE3), 2 Xeon DP 5150's(SSE3), and 4 Xeon DP X5365's.

All of the above seem to exhibit the same issue, not a major deal, by any stretch. Note: I only tested using 26/27/28 memory settings and thread values of 2,3,4,6,8.

Side note: Interestingly enough, i do get decent gains using the sse4 binary over intel/sph on the SSE3 based procs. It doesn't seem to be as significant as the gains on the SSE4.1 based procs, but that may just be other factors such as less memory bandwidth, older chipset, etc.

Thanks again!

Quote from: xolokram on November 23, 2013, 08:32:59 PM
pull the latest github update  :D

Thanks, new compile went without a hitch, looks good so far.
I had the counter showing <10c/m this morning, i initially sobbed :'( , and then i checked my overnight totals and they were up 10% from the night before, woot... :)

Thanks for all your efforts!

Quote from: xolokram on November 21, 2013, 05:57:48 PM
KNOWN ISSUES:
1) on some machines the windows binary crashs right after the first share submission (see next posts), looking into that issue

I am having the same issue across most of my hardware, I know it's already known, but if you need any assistance with testing, drop me a line and I will be happy to help.  I can test on workstations with i7's, i5's, C2Qs, and Xeon servers with dual 51xx, E54xx, X53xx, X55xx, and X56xx series procs. (most running windows 64bit OS's) Just shy of 200 cores total.

Figure it's the least i can contribute.  :)

Keep up the great work!

-Xornof