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

[xornof]

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

[Silver_Pharaoh]

@anybody who know the answer

What was the verdict on the miner not submitting the collisions to the server?

btw, it is still happening on the latest release of 0.7 miner. I'm using the AVX version using 4 threads and 1 GB memory/thread. Though it also happens with the sse4 version too.

Is there a version I can use to mine overnight? I'd rather not leave the miner on all night only to see lots of unsubmitted collisions...


Sorry if it sounds like Im complaining xolo, I just want to let you know that it's still happening. Keep up the good work, best pts/xpm pool ever made!

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)

[lordfirefox]

Cute, I get to 0.32 and the pool stops confirming shares

[GoldTiger69]

Hey Xolo, have you seen this?:

The pool, as listed in the topic: "Mining pool list - Updating",
PTSPool:
web:54.238.185.113

Its miner on github goes to the miner of beeeeer pool. Is it strange?
I try to compile it, then when running, the parameter of PTSPool can't be used.
It's beeeeer's miner.

So where is PTSPool's miner?
Thank you!

author=Clark link=topic=1024.msg10553#msg10553 date=1385272289

That's odd

[AcidCrash]

Hello, I'm trying to compile ptsminer on MacOS Maverick and I get this error:

Code: [Select]

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

Thank you so much xornof! Waiting for your feedback!

[xornof]

i updated the windows binaries --- see the v0.7 post (& OP)
download: here

Testing the "newerer" binaries now.

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

[xornof]

Hello, I'm trying to compile ptsminer on MacOS Maverick and I get this error:

Code: [Select]

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

[xolokram]

Hi,

i updated the windows binaries --- see the v0.7 post (& OP)
download: here

• fixed the memory parameter bug
• adjustments to the mingw64 makefile
• split avx binary into intel version & amd version

/edit: STOP, AVX VERSION IS NOT USING AVX AUTOMATICALLY @ WINDOWS, FIX INCOMING *fixed*

i think the pool will switch to a score-based payout system like this soon

- xolokram

ps. github update will follow soon
pps. i have the strange feeling, that the avx code is slower that the sse4 code?!? ok, seems to be just fluctuacting values, will check that with a longer testing period

[AcidCrash]

Hello, I'm trying to compile ptsminer on MacOS Maverick and I get this error:

Code: [Select]

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?

[xornof]

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:

Code: [Select]

ptsminer_sse4.exe %walletaddress% 4 28 sse4

Hope that helps!

-Xornof

[xornof]

@xornof:
thank you for the bug report, will fix this asap
like i said these are not the final v0.7 binaries

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.  I hope you save some personal time for you and yours, cheers!

[gsrcrxsi]

Thanks for the windows update. I'm seeing good gains on the sse4 version vs v0.5a on my Q9650 and Q9550 machines.

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.

[xolokram]

thank you joele

and thank you crz & flashoftheblades <-- give them some hugs (@irc)

@memory settings:
it's highly depending on your system and the bottlenecks that arise from tweaking this parameter; if memory & cpu is fast enough 1024 can be the best solution for you. my tests showed that decent systems are working best with the 512 option. other & older systems may switch to 256 mb (=older) or 2048 (=high-end) or even higher/lower.

@xornof:
thank you for the bug report, will fix this asap
like i said these are not the final v0.7 binaries

the advantage of the sse4 binary for sse3 cpus is caused by the auto-optimization which is enabled in the 4 binaries.
ptsminer_amd.exe & ptsminer_intel.exe include MMX, (3DNow), SSE and SSE2
ptsminer_sse4.exe includes MMX, SSE, SSE2, SSE3 and SSE4.x
ptsminer_avx.exe includes MMX, SSE, SSE2, SSE3, SSE4.x and AVX
thus it looks like we're lucky and the code for the sse4 binary has been optimized using up to SSE3, giving you a small advantage over the _amd and _intel version (using SPHLIB)
if it would use SSE4 instructions (which is possible, because i compiled it with SSE4 support) too, it would crash on your SSE3 cpus

@miner: in general
there's still enough room for optimization, e.g. i found a sse3 version for the sha computation (but i'll have to check the code first). i'm not sure about the avx binary and bulldozer cpus, i think i should split this binary into avx-intel and avx-bulldozer to give the binaries a small extra boost.

also i'll have to re-enable the auto-detection for sse4/avx for the linux build.

but first i'll create the "payout abandoned addresses"-script today (later today, it's sunday, pts has to wait a little )

- xolokram

[xornof]

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!

[btc2084]

I also was trying to figure out what's optimal.  It would be nice if the app could auto-tune itself but my guess is that the defaults are already right for you: 4 cores, multithreaded is 8 threads; 8 x 512MB is 4GB ram required.  If you took 1GB ram per thread you'd be swapping all the time.  It's possible 4 threads at 1 GB might be better.  Why not try those two and let us know what works better?  i.e. appending one of these in your command line:

a) 8 27 or
b) 4 28

I found that I can use my machine okay if I just choose "7" threads (on my i7 4770 that has 8 threads avail.)

Hello 

I'm running the v0.7 64 Bit binary, but how many threads and memory should I use for optimal performance? I have a i5-4670K with 8gigs of RAM