btc_artist
Full Member
Offline
Activity: 154
Merit: 102
Bitcoin!
|
|
December 19, 2011, 06:16:48 PM |
|
I went from 6.20 MH/s to 6.52 MH/s on my AMD box. Not a *huge* improvement, but I'll definitely take it. I'll test on an Intel box next.
|
BTC: 1CDCLDBHbAzHyYUkk1wYHPYmrtDZNhk8zf LTC: LMS7SqZJnqzxo76iDSEua33WCyYZdjaQoE
|
|
|
Schwede65
|
|
December 19, 2011, 06:17:47 PM |
|
Yeah, on Win7 64, I can't find libeay32.dll. I grabbed one from a version of OpenSSL, but it complains about missing functions, etc. Thank you for reporting this, I have added the DLL to the package. this works fantastic now with that DLL: core i7 w7-64 / 6 threads / old 2.6 KH/s/thread / new 5.3 KH/s/thread
|
|
|
|
Intention
|
|
December 19, 2011, 06:21:45 PM |
|
Awesome: Intel i7 2600k @ 4.2ghz - 4 threads - Old: 2.8khash/s New: 5.3-5.5khash/s
|
|
|
|
pooler (OP)
|
|
December 19, 2011, 06:29:54 PM |
|
In case you're wondering why the SSE2 version sucks on K8 and K10 ... reason is rather simple. the salsa20 function is a long string of data dependent 4*32-bit vector integer operations (i.e. output of one operation is used as input to the next). And the execution latencies for the most used instructions in the salsa20 core (shift r/l immediate, add, xor) are all 2 clocks on K8/K10, all 1 clock on Atom/Core/Core2/Nehalem/SB. End result ... sse2 salsa20 needs roughly twice the clocks/round on AMD compared to any modern intel.
Thank you for your insight, ArtForz! Yes, I think I have read somewhere that since the Core architecture Intel CPUs can actually handle SSE registers 128 bits at a time. I have never been too fond of Intel, but it's nice to see that sometimes you get what you pay for!
|
BTC: 15MRTcUweNVJbhTyH5rq9aeSdyigFrskqE · LTC: LTCPooLqTK1SANSNeTR63GbGwabTKEkuS7
|
|
|
GenTarkin
Legendary
Offline
Activity: 2450
Merit: 1002
|
|
December 19, 2011, 07:09:15 PM |
|
POOLER!!! YOU FUCKIN ROCK DUDE! I knew you could do it and I had a feeling assembly miner would freakin own! Fuckin awesome dude! Thanks a ton! If anyone is wondering - I updated my service installer w/ the new binaries and both 32bit and 64bit now work properly!
|
|
|
|
btc_artist
Full Member
Offline
Activity: 154
Merit: 102
Bitcoin!
|
|
December 19, 2011, 07:19:47 PM |
|
Twice as many hashes per second would by 100% better, not 50% better, no?
|
BTC: 1CDCLDBHbAzHyYUkk1wYHPYmrtDZNhk8zf LTC: LMS7SqZJnqzxo76iDSEua33WCyYZdjaQoE
|
|
|
xurious
|
|
December 19, 2011, 07:21:15 PM |
|
This is the shit. I can only test at work on my work laptop, but my mobile i7 @ 1.6ghz went from .7kh/s per thread to 1.68kh/s per thread. That's around %130 increase. I can't wait to try out my i7 at home (running at 4.8ghz with HT.)
|
SiaMining.com -- First PPS SiaMining Pool! 3%, VarDiff, Stratum Support
|
|
|
jjiimm_64
Legendary
Offline
Activity: 1876
Merit: 1000
|
|
December 19, 2011, 07:23:17 PM |
|
Pooler... you rock.. EXCELLENT work. so much for 'high level languages'.
Is there any chance you can compile a version for us linuxcoin users?
donation sent!
|
1jimbitm6hAKTjKX4qurCNQubbnk2YsFw
|
|
|
pooler (OP)
|
|
December 19, 2011, 07:37:58 PM |
|
Thank you for reporting this, I have added the DLL to the package.
In w7x64: Faulting application name: minerd.exe, version: 0.0.0.0, time stamp: 0x4eef5c94 Faulting module name: pthreadGC2.dll, version: 2.8.0.0, time stamp: 0x4be6d174 Exception code: 0xc0000005 Fault offset: 0x00000000000036a3 Faulting process id: 0xeb8 Faulting application start time: 0x01ccbe78f0cabebe BTW, this was with updated binaries. I have asked diki, who prepared the binaries, on IRC. He told me that pthreads-w64 is still experimental (see http://sourceforge.net/apps/trac/mingw-w64/wiki/Compile%20pthreads) so it may not always work. He suggests you might want to try a different version of the library, like this one: http://code.google.com/p/mingw-w64-dgn/downloads/detail?name=mingw_dgn_basic-w64-bin-x86_64-20110225.7z&can=2&q=pthread
|
BTC: 15MRTcUweNVJbhTyH5rq9aeSdyigFrskqE · LTC: LTCPooLqTK1SANSNeTR63GbGwabTKEkuS7
|
|
|
Come-from-Beyond
Legendary
Offline
Activity: 2142
Merit: 1010
Newbie
|
|
December 19, 2011, 08:23:53 PM |
|
I heard rumours about so called "dark pool" that (probably) made 51% attack. Now with improved miner we have more chances to survive such kind of attack. That's good news. But I have a question:
Why the hell we have been using unoptimized miners for 2 months?
Laziness? Conspiracy?
Price of LTC fell below the floor. Someone mined and sold a lot of coins...
|
|
|
|
btc_artist
Full Member
Offline
Activity: 154
Merit: 102
Bitcoin!
|
|
December 19, 2011, 08:26:08 PM |
|
I heard rumours about so called "dark pool" that (probably) made 51% attack. Now with improved miner we have more chances to survive such kind of attack. That's good news.
Irrelevant. The "dark pool" can use the improved miner as well. Why the hell we have been using unoptimized miners for 2 months?
Laziness? Conspiracy?
Umm, nobody had taken it upon themselves to further optimize the miner? No conspiracy theories here.
|
BTC: 1CDCLDBHbAzHyYUkk1wYHPYmrtDZNhk8zf LTC: LMS7SqZJnqzxo76iDSEua33WCyYZdjaQoE
|
|
|
Come-from-Beyond
Legendary
Offline
Activity: 2142
Merit: 1010
Newbie
|
|
December 19, 2011, 08:44:26 PM |
|
I think that dark pool already used optimized miner.
No conspiracy? Ok. I hope u are 100% right. But who knows...
|
|
|
|
wknight
Legendary
Offline
Activity: 889
Merit: 1000
Bitcoin calls me an Orphan
|
|
December 19, 2011, 08:50:28 PM |
|
It could be very well that someone else optimized a miner and kept it to themselves.. but at the same time its great that Pooler took the time to work on the cpu miner and share his code with all of us!
|
Mining Both Bitcoin and Litecoin.
|
|
|
localhost
|
|
December 19, 2011, 08:59:27 PM |
|
Barely 5% improvement on my AMD laptop but indeed about 40% better on my Intel server. Pretty neat
|
-
|
|
|
Schwede65
|
|
December 19, 2011, 09:23:53 PM |
|
hey pooler, really great job you have done! the results for my intel-cpu's are: ~ 100 % more hashing-power ~ 15 % less energy amount (for doubled mining-power!) now we know, why the intel-cpu's are more expensive i'll set up the donation at your pool to 30 %, mining with 235 KH/s now
|
|
|
|
kronosvl
|
|
December 19, 2011, 09:50:46 PM |
|
AMD Athlon 3800+ (2.4Ghz) single core: old: 1.07Kh/s new: ~1.5Kh/s ~1.65Kh/s
I'm using mining section from litecoin client to start mining and speed is displayed as: 'Speed ~nan Khash/sec'.
Not a big deal as actual hashrate is displayed in debug mode however though you should know.
Thanks for this improved miner
|
|
|
|
pooler (OP)
|
|
December 19, 2011, 10:20:51 PM Last edit: December 19, 2011, 10:33:36 PM by pooler |
|
I have added some details for the curious in the original post, check them out. hey pooler, really great job you have done! the results for my intel-cpu's are: ~ 100 % more hashing-power ~ 15 % less energy amount (for doubled mining-power!) now we know, why the intel-cpu's are more expensive i'll set up the donation at your pool to 30 %, mining with 235 KH/s now Hey, thank you!
|
BTC: 15MRTcUweNVJbhTyH5rq9aeSdyigFrskqE · LTC: LTCPooLqTK1SANSNeTR63GbGwabTKEkuS7
|
|
|
exahash
|
|
December 19, 2011, 10:35:27 PM |
|
My Xeon E5335's (clovertown) jumped 50% from 1.97 KH/s/thread to 2.97.
My Xeon X3430's jumped 58% from 2.28 to 3.62
Very nice. Props to Pooler!
My Opteron 8218's moved about 5% from 2.54 to 2.68
My Sempron 140's moved about 8% from 2.65 to 2.88
Those are nice improvements too.
BTW - I used CFLAGS="-march=native -O3" I tested with and without the sse extensions and it didn't matter (like Pooler said earlier in this thread).
|
|
|
|
coblee
Donator
Legendary
Offline
Activity: 1654
Merit: 1351
Creator of Litecoin. Cryptocurrency enthusiast.
|
|
December 19, 2011, 10:48:11 PM |
|
I have added some details for the curious in the original post, check them out. hey pooler, really great job you have done! the results for my intel-cpu's are: ~ 100 % more hashing-power ~ 15 % less energy amount (for doubled mining-power!) now we know, why the intel-cpu's are more expensive i'll set up the donation at your pool to 30 %, mining with 235 KH/s now Hey, thank you! Pooler, you are amazing. Thanks for your hard work. I have sent you a little something: b345dbfe9ae6e37dcd560075c7899787dfd99fae81afcd964b1a96157ccbe362
|
|
|
|
pooler (OP)
|
|
December 19, 2011, 11:42:12 PM |
|
Pooler, you are amazing. Thanks for your hard work.
Coblee, hearing that from you is such a huge compliment and honor... Thank you!
|
BTC: 15MRTcUweNVJbhTyH5rq9aeSdyigFrskqE · LTC: LTCPooLqTK1SANSNeTR63GbGwabTKEkuS7
|
|
|
|