[XPM]unofficial jhPrimeminer thread - page 28.

neo101one

member

Activity: 61

Merit: 10

Quote from: rwessels on August 16, 2013, 03:02:09 PM

I have quite a few computers mining primecoin now - from Dual Core Intel chips to 3rd gen i5/i7's, and even older dual processor quad core intel servers. I thought I would add a new AMD 8320 to the mix. 8 cores running at least at 3.5 ghz should work out fairly well, right?

Well my first problem is that the 7.1 avx version crashes. No big deal, I can just run the regular version. Problem 2 is that each time it finds a share, it gives the message Invalid Share - Invalid data time overflow. Then I moved to the original one from ypool - same problem, but in addition to the data time overflow messages, some times I get other ones like duplicate share. So I downgraded to the v6 version. Although I can run both the AVX and normal versions of v6 - and my shares are actually accepted - my latest problem is that it seems VERY slow. I am getting about 1,600-2,000 4 chains/hr. My other computers are averaging 4,000 to 6,000 4 chains an hour.

The only setting I am using is the number of threads and it matches the processor cores - 8. Is anyone else using this processor - FX-8320 and is having better luck than I am? When my value per hour is around 2 and I have other computers that are over 20, something just doesn't seem right.

-Randy

Im actually getting that cpu later on when I get a new mb. I dont know exactly what causes the crashes or invalid shares but I hope its fixed in v8. Id check out v5 with AVX back on page 10 and see if it runs faster.

belltown

sr. member

Activity: 301

Merit: 250

still can't change my profile pic

Mumus, great job! I love your miner. Will be sending you tip once my XPM is confirmed.

rwessels

member

Activity: 102

Merit: 10

I have quite a few computers mining primecoin now - from Dual Core Intel chips to 3rd gen i5/i7's, and even older dual processor quad core intel servers. I thought I would add a new AMD 8320 to the mix. 8 cores running at least at 3.5 ghz should work out fairly well, right?

Well my first problem is that the 7.1 avx version crashes. No big deal, I can just run the regular version. Problem 2 is that each time it finds a share, it gives the message Invalid Share - Invalid data time overflow. Then I moved to the original one from ypool - same problem, but in addition to the data time overflow messages, some times I get other ones like duplicate share. So I downgraded to the v6 version. Although I can run both the AVX and normal versions of v6 - and my shares are actually accepted - my latest problem is that it seems VERY slow. I am getting about 1,600-2,000 4 chains/hr. My other computers are averaging 4,000 to 6,000 4 chains an hour.

The only setting I am using is the number of threads and it matches the processor cores - 8. Is anyone else using this processor - FX-8320 and is having better luck than I am? When my value per hour is around 2 and I have other computers that are over 20, something just doesn't seem right.

-Randy

clintar

full member

Activity: 212

Merit: 100

Quote from: bengtåke on August 15, 2013, 03:47:15 PM

Quote from: clintar on August 15, 2013, 03:30:27 PM

I was running the 64 bit wine one, but i am just going by what pps says and it does fluctuate a lot. This one was running around the 15k-20k range, and the 7.1 seemed to hover around 30k, but i really don't know much about how this works, so maybe that's invalid. Running on an AMD A6-3650.

I see. Did you compile it yourself or use one of my binaries (which one)? I might remove the binaries since I didn't recompile gmp every time. I'd recommend compiling it yourself.

I tried my own compiled one and the a8-ss38 one you posted. Maybe it's because the wine64 runs pretty good compared to regular wine? Well, don't use me as a basis, cause I'm new to the primecoin mining, too. I don't know what's good and what's bad.

Maybe I need to compile my own version of libgpm. I'm just running the one the system came with.

Thanks, btw and good work. I'm sure a lot of people would love to run this in linux. Maybe when I get some xpm confirmed, I can tip you.

whitepaint0909

newbie

Activity: 46

Merit: 0

Quote from: mhps on August 16, 2013, 08:49:40 AM

Quote from: Sterling on August 08, 2013, 01:24:38 PM

Uhhhh, what's going on?

I got this on an E5620 with -t 4 -s 3000000, about 8 hours after starting the miner

v7.1 win8

I woke up to one of my miners doing the same thing.. My I5 running 7.1AVX. All other versions have not had this problem. 7.1 non avx

mhps

hero member

Activity: 516

Merit: 500

CAT.EX Exchange

Quote from: Sterling on August 08, 2013, 01:24:38 PM

Uhhhh, what's going on?

I got this on an E5620 with -t 4 -s 3000000, about 8 hours after starting the miner

v7.1 win8

bengtåke

sr. member

Activity: 309

Merit: 250

confused developer

Quote from: Sy on August 16, 2013, 05:05:04 AM

Quote from: bengtåke on August 15, 2013, 02:02:18 PM

Hey! Thanks for your diff! Silly me had forgotten to remove the boost dependency in one file. I've made some changes to the makefile and source files so it should compile without those changes now *knock on wood*.

I've compiled a few binaries for linux and freebsd here: http:/(FORUM WARNING: BE CAREFUL WHEN CLICKING UNKNOWN LINKS!) (FORUM WARNING: BE CAREFUL WHEN CLICKING UNKNOWN LINKS!) (FORUM WARNING: BE CAREFUL WHEN CLICKING UNKNOWN LINKS!) goo.gl /kamnG2 they're all staticly linked so shouldn't require any extra libraries

Just compiled on ubuntu64 server, had to install libgmp-dev and change the lib file to

Quote

LIBS = -lgmp -lgmpxx -lcrypto -lssl -lpthread -lboost_system -lboost_timer -lboost_chrono

it compiled fine afterwards and its running around 20% faster than wine so far.

Odd, I just tried on a fresh ubuntu64 install and I didn't need to link boost to compile it (need to add -lrt though). Nice to hear that someone has gotten an improvement out of it anyway.

Edit: got rid of the manual -lrt

Sy

legendary

Activity: 1484

Merit: 1003

Bounty Detective

Quote from: bengtåke on August 15, 2013, 02:02:18 PM

Hey! Thanks for your diff! Silly me had forgotten to remove the boost dependency in one file. I've made some changes to the makefile and source files so it should compile without those changes now *knock on wood*.

I've compiled a few binaries for linux and freebsd here: http:/(FORUM WARNING: BE CAREFUL WHEN CLICKING UNKNOWN LINKS!) goo.gl /kamnG2 they're all staticly linked so shouldn't require any extra libraries

Just compiled on ubuntu64 server, had to install libgmp-dev and change the lib file to

Quote

LIBS = -lgmp -lgmpxx -lcrypto -lssl -lpthread -lboost_system -lboost_timer -lboost_chrono

it compiled fine afterwards and its running around 20% faster than wine so far.

tadakaluri

hero member

Activity: 616

Merit: 500

Quote from: Sterling on August 15, 2013, 08:30:22 PM

Quote from: Jabroni on August 14, 2013, 01:50:50 PM

Im testing the v7.1 inside a ESXi Windows 2008 64bit VM (Running on a i7 920) w/ 4 cores assigned. However it seems that this wont produce any share at all. PPS values are ~15k. Anyone experiencing the same probs??

Heres an output example after leaving it running like 40 mins

Code:

---------------------------------------------------------------
---- New Block: 117030 - Diff: 9.750861 / 6.000000
---- Total/Valid shares: [ 0 / 13 ] - Max diff: 7.24141
---- 6-chain count: 14 - 6ch/h: 18.896 - Share Value: 1.400
---- 7-chain count: 1 - 7ch/h: 1.350 - Share Value: 1.000
---- 8-chain count: 0 - 8ch/h: 0.000 - Share Value: 0.000
---- 9-chain count: 0 - 9ch/h: 0.000 - Share Value: 0.000
---- 10-chain count: 0 - 10ch/h: 0.000 - Share Value: 0.000
---- Total Share Value submitted to the pool: 0.000000
---------------------------------------------------------------
Val/h 3.231 - PPS: 14099 - 4ch/h: 2503.87 - 5ch/h: 250.39 - 6ch/h: 18.85 - Last 4ch/h: 3142.93 - Last 5ch/h: 0.00
Val/h 3.221 - PPS: 12557 - 4ch/h: 2508.46 - 5ch/h: 249.64 - 6ch/h: 18.79 - Last 4ch/h: 4040.91 - Last 5ch/h: 0.00

And if i check on the ypool webpage the worker hasnt submitted a single valid share

v7.1 has submitted nothing but invalid shares for me (same errors etc).

Switched to v6 and instantly the problem is gone.

But with my i5 2500 it runs smoothly........ But there is no difference between v7 and v7.1

Sterling

newbie

Activity: 39

Merit: 0

Quote from: Jabroni on August 14, 2013, 01:50:50 PM

Im testing the v7.1 inside a ESXi Windows 2008 64bit VM (Running on a i7 920) w/ 4 cores assigned. However it seems that this wont produce any share at all. PPS values are ~15k. Anyone experiencing the same probs??

Heres an output example after leaving it running like 40 mins

Code:

---------------------------------------------------------------
---- New Block: 117030 - Diff: 9.750861 / 6.000000
---- Total/Valid shares: [ 0 / 13 ] - Max diff: 7.24141
---- 6-chain count: 14 - 6ch/h: 18.896 - Share Value: 1.400
---- 7-chain count: 1 - 7ch/h: 1.350 - Share Value: 1.000
---- 8-chain count: 0 - 8ch/h: 0.000 - Share Value: 0.000
---- 9-chain count: 0 - 9ch/h: 0.000 - Share Value: 0.000
---- 10-chain count: 0 - 10ch/h: 0.000 - Share Value: 0.000
---- Total Share Value submitted to the pool: 0.000000
---------------------------------------------------------------
Val/h 3.231 - PPS: 14099 - 4ch/h: 2503.87 - 5ch/h: 250.39 - 6ch/h: 18.85 - Last 4ch/h: 3142.93 - Last 5ch/h: 0.00
Val/h 3.221 - PPS: 12557 - 4ch/h: 2508.46 - 5ch/h: 249.64 - 6ch/h: 18.79 - Last 4ch/h: 4040.91 - Last 5ch/h: 0.00

And if i check on the ypool webpage the worker hasnt submitted a single valid share

v7.1 has submitted nothing but invalid shares for me (same errors etc).

Switched to v6 and instantly the problem is gone.

clintar

full member

Activity: 212

Merit: 100

Quote from: dudeguy on August 15, 2013, 03:59:48 PM

Quote from: clintar on August 15, 2013, 03:30:27 PM

I was running the 64 bit wine one, but i am just going by what pps says and it does fluctuate a lot. This one was running around the 15k-20k range, and the 7.1 seemed to hover around 30k, but i really don't know much about how this works, so maybe that's invalid. Running on an AMD A6-3650.

Are you using stock settings? I have a A6-3400m which actually is quite similar to your chip. Mine likes less than 1M sieve size. Trying to tune it.

Nope, haven't been around much. Not sure what to change. I'll look into <1M sieve. Edit: I mean yes, I was using stock.

Quirlorimbo

member

Activity: 89

Merit: 10

since v.7.1 im on a luck streak
24 hours with 4x 9-chain and 1x 10-chain ...
val ~ 70/h ... but it is still luck ...

best miner ever, though.

bitrich

member

Activity: 109

Merit: 10

Wondering if there is a windows 32bit V7 or 7.1 yet? I want to be able to adjust my L1 cache size on my older laptop. Still running v6 on it. Im still playing with settings but I got my i5 2410M 2.3ghz running v7.1 a fairly consistent 13-15 val/h. Found 2 9-chains in the last 24 hours as well.

18RATTT

sr. member

Activity: 282

Merit: 250

Quote from: mumus on August 13, 2013, 02:16:02 PM

Hi
Several users complained about some memory leaks in v5+ so finally I catch them and hopefully it's fixed in this minor update.
V7.1 64bit - https://dl.dropboxusercontent.com/u/1189851/jhPrimeminer-GMP-v7.1.zip
v7.1 64bit AVX - https://dl.dropboxusercontent.com/u/1189851/jhPrimeminer-GMP-v7.1-AVX.zip
Sorry no 32bit yet. I still didn't managed to understand why the same code compiled for 32bit is crashing. Its something related to std::map (if somebody has more experience in this territory).

Some minor features is also included:
-Printing the total share value that was submitted and confirmed by the pool
-The current settings can be printed by pressing the 's' key

good stuff mumu, cheers.

dudeguy

member

Activity: 182

Merit: 10

Quote from: clintar on August 15, 2013, 03:30:27 PM

I was running the 64 bit wine one, but i am just going by what pps says and it does fluctuate a lot. This one was running around the 15k-20k range, and the 7.1 seemed to hover around 30k, but i really don't know much about how this works, so maybe that's invalid. Running on an AMD A6-3650.

Are you using stock settings? I have a A6-3400m which actually is quite similar to your chip. Mine likes less than 1M sieve size. Trying to tune it.

bengtåke

sr. member

Activity: 309

Merit: 250

confused developer

Quote from: clintar on August 15, 2013, 03:30:27 PM

I was running the 64 bit wine one, but i am just going by what pps says and it does fluctuate a lot. This one was running around the 15k-20k range, and the 7.1 seemed to hover around 30k, but i really don't know much about how this works, so maybe that's invalid. Running on an AMD A6-3650.

I see. Did you compile it yourself or use one of my binaries (which one)? I might remove the binaries since I didn't recompile gmp every time. I'd recommend compiling it yourself.

clintar

full member

Activity: 212

Merit: 100

I was running the 64 bit wine one, but i am just going by what pps says and it does fluctuate a lot. This one was running around the 15k-20k range, and the 7.1 seemed to hover around 30k, but i really don't know much about how this works, so maybe that's invalid. Running on an AMD A6-3650.

bengtåke

sr. member

Activity: 309

Merit: 250

confused developer

Quote from: clintar on August 15, 2013, 03:15:40 PM

Quote from: bengtåke on August 15, 2013, 02:02:18 PM

I've compiled a bunch of binaries with various cpu optimizations for a few platforms here: http://(FORUM WARNING: BE CAREFUL WHEN CLICKING UNKNOWN LINKS!) (FORUM WARNING: BE CAREFUL WHEN CLICKING UNKNOWN LINKS!) goo.gl /kamnG2 they're all staticly linked so shouldn't require any extra libraries

Cool, so it seems this runs slower than the 7.1 version. Were you planning on making changes up to that revision? probably mumus will i guess.

Really? It's slower than 7.1 in 32 bit wine for you? What system and cpu do you use? I guess I could update it when I have the time

clintar

full member

Activity: 212

Merit: 100

Quote from: bengtåke on August 15, 2013, 02:02:18 PM

I've compiled a bunch of binaries with various cpu optimizations for a few platforms here: http://(FORUM WARNING: BE CAREFUL WHEN CLICKING UNKNOWN LINKS!) goo.gl /kamnG2 they're all staticly linked so shouldn't require any extra libraries

Cool, so it seems this runs slower than the 7.1 version. Were you planning on making changes up to that revision? probably mumus will i guess.

bengtåke

sr. member

Activity: 309

Merit: 250

confused developer

Quote from: clintar on August 15, 2013, 12:55:17 PM

I got it to compile in linux.
here's a git diff

Code:

diff --git a/Makefile b/Makefile
index 7bd34ef..21b72f7 100644
--- a/Makefile
+++ b/Makefile
@@ -2,7 +2,7 @@ CXX = g++
CFLAGS = -mtune=native -O3 -Wall -Wextra

# on linux you will probably need to add -lrt and change -pthread to -lpthread
-LIBS = -lgmp -lgmpxx -lcrypto -lssl -pthread
+LIBS = -lgmp -lgmpxx -lcrypto -lssl -lpthread -lboost_system -lboost_timer -lboost_chrono

# You might need to edit these paths too
LIBPATHS = -L/usr/local/lib -L/usr/lib
diff --git a/src/primecoinMiner/ticker.h b/src/primecoinMiner/ticker.h
index bcb846d..b96f9e0 100644
--- a/src/primecoinMiner/ticker.h
+++ b/src/primecoinMiner/ticker.h
@@ -4,7 +4,7 @@
#if (defined(__MACH__) && defined(__APPLE__))
#include
#elif defined(__linux__)
-#include
+//#include
#else
#include
#endif

Hey! Thanks for your diff! Silly me had forgotten to remove the boost dependency in one file. I've made some changes to the makefile and source files so it should compile without those changes now *knock on wood*.

I've compiled a few binaries for linux and freebsd here: http:/goo.gl/kamnG2 they're all staticly linked so shouldn't require any extra libraries

Topic: [XPM]unofficial jhPrimeminer thread - page 28. (Read 180246 times)