daily by working for 2 hours Does this require any work - no |
|
|
|
|
1AdaTU1q3DGaHg5eRagpHD6Nns7wacaUgB
|
|
|
|
I tested on a i7-6700K with 8 threads and saw no difference between multi with or without -flto.
One thing to keep in mind is that you are using a desktop cpu, I am using a server cpu. Those xeons are not multiple times more expensive for no reason (sure partly it's just branding, more reliability etc. etc., but there are also some functional differences). Or it (the fact that -flto doesn't do anything to you) could simply be the compiler. Yours is 2 years old, mine 2 months. But the crucial information is that you can actually compile with -flto. Could you please give me the exact flags that work for you? I will take the effort and find out what's the problem on my end, I just need the starting (compilable) point. I have no idea whether he improved the performance after you took the code from him. The very last commit is 3 days old but which one is the last one that could have had any real impact on Lyra2RE speed, or the overall speed, I am not going to investigate. But if you could remember, at least roughly, when did you take his code I can revert his tree to that date and try that version. Hi experts, I'm trying to build in my openSUSE linux. But I've got an error:
./build.sh
make: /lib64/libc.so.6: version `GLIBC_2.14' not found (required by make)
make: /lib64/libc.so.6: version `GLIBC_2.17' not found (required by make)
strip: 'cpuminer': No such file
Regards
This problem has nothing to do with cpuminer, your building environment is messed up. Glibc is the very core linux library, if make can't find the version it likes then there is something wrong with either. But if glibc was messed up the system would hardly even boot properly. Maybe try to update? |
|
|
|
Things are looking up. I solved the alignment problem and made progress with performance. I've added 3%
to Lyra2 so far and I have a few functions left to convert. So far I've only implemented AVX2, I still have to do
AVX implementations of al functions. The improvements are to the lyra2 core so they should also help lyra2v2.
You are talking about Lyra2RE, right? So it will get sped up? Because so far tpruvot-cpuminer-multi's Lyra2RE is faster than yours by some 3.8% on my servers. I guess it might be because of -flto that I use with his but can't use with yours (doesn't compile)  I think you are doing something wrong, lyra2RE in cpuminer-opt v3.3.7 was improved 7% faster than cpuminer-multi. In the next release it will another 3% faster. If you have very old CPUs (ie core2) you won't get the benefits of my optimisations. I know you didn't mean to but what you said was very amusing. The whole time I am talking about servers, the real servers in data centers, not some desktop pc in your home that you call "a server", and you come back at me with "core2" - a 10 years old super-obsolete desktop cpu. Hilarious! But maybe you are right, maybe I am doing something wrong but it's not the hardware. I am using the up-to-date Ubuntu 16.04, if I messed something up then it must be the flags in the build.sh. Or maybe you did something wrong actually. Maybe you didn't compile tpruvot's cpuminer with -flto so now you are competing with a crippled sw because this flag does make the difference and it's not on by default, it's commented out in the build.sh so you have to enable it. Either way, instead of accusing each other let's try to make things better. In this spirit I made a little test for you. I picked the most powerful server AWS has to offer, the x1.32xlarge ( https://aws.amazon.com/ec2/instance-types/x1/) with 128 cores and 1952 GB memory. Here are the specs: root@xxx:~/# grep -e name -e flags /proc/cpuinfo | head -n2
model name : Intel(R) Xeon(R) CPU E7-8880 v3 @ 2.30GHz
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq monitor est ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm fsgsbase bmi1 hle avx2 smep bmi2 erms invpcid rtm xsaveopt ida
root@xxx:~/# grep processor /proc/cpuinfo | wc -l
128
root@xxx:~/# free -h
total used free shared buff/cache available
Mem: 1.9T 3.3G 1.9T 9.1M 624M 1.9T
Swap: 0B 0B 0B
Now I fresh recompiled tpruvot and here is the result: (I noticed that with the many-cores machines it is sometimes actually more powerful to use just half the threads; here on x1.32xlarge the difference is marginal, on g2.8xlarge it's quite significant; the spike at the beginning of full cores I attribute to some throttling done by Amazon, it's just a vps after all, not a dedi) root@xxx:~/tpruvot-cpuminer-multi# ./cpuminer -a lyra2re --benchmark --threads=64 | grep Total
[2016-08-04 02:44:55] Total: 8903 kH/s
[2016-08-04 02:44:59] Total: 8833 kH/s
[2016-08-04 02:45:04] Total: 8735 kH/s
[2016-08-04 02:45:09] Total: 8728 kH/s
[2016-08-04 02:45:14] Total: 8727 kH/s
root@xxx:~/tpruvot-cpuminer-multi# ./cpuminer -a lyra2re --benchmark --threads=128 | grep Total
[2016-08-04 02:45:21] Total: 11661 kH/s
[2016-08-04 02:45:25] Total: 11568 kH/s
[2016-08-04 02:45:30] Total: 8731 kH/s
[2016-08-04 02:45:35] Total: 8720 kH/s
[2016-08-04 02:45:40] Total: 8722 kH/s
[2016-08-04 02:45:45] Total: 8703 kH/s
Now let's look at you: CPU: Intel(R) Xeon(R) CPU E7-8880 v3 @ 2.30GHz
CPU features: SSE2 AES AVX AVX2
SW built on Aug 4 2016 with GCC 5.4.0
SW features: SSE2 AES AVX AVX2
Algo features: SSE2 AES
Start mining with AES-AVX optimizations...
root@xxx:~/cpuminer-opt-3.3.9# ./cpuminer -a lyra2re --benchmark --threads=64 | grep Total
[2016-08-04 02:47:57] Total: 4128.77 kH, 8660.05 kH/s
[2016-08-04 02:48:01] Total: 35.19 MH, 8668.56 kH/s
[2016-08-04 02:48:06] Total: 43.34 MH, 8590.78 kH/s
[2016-08-04 02:48:11] Total: 42.95 MH, 8600.93 kH/s
[2016-08-04 02:48:16] Total: 43.00 MH, 8621.90 kH/s
root@xxx:~/cpuminer-opt-3.3.9# ./cpuminer -a lyra2re --benchmark --threads=128 | grep Total
[2016-08-04 02:48:23] Total: 8323.07 kH, 11.96 MH/s
[2016-08-04 02:48:26] Total: 11.68 MH, 12.03 MH/s
[2016-08-04 02:48:31] Total: 32.93 MH, 8544.43 kH/s
[2016-08-04 02:48:36] Total: 39.36 MH, 8531.49 kH/s
[2016-08-04 02:48:41] Total: 42.52 MH, 8531.90 kH/s
[2016-08-04 02:48:46] Total: 42.33 MH, 8536.75 kH/s
[2016-08-04 02:48:51] Total: 41.88 MH, 8536.12 kH/s
[2016-08-04 02:48:56] Total: 42.41 MH, 8526.90 kH/s
Here tpruvot is faster by over 2%. So let's look at the build.sh. Tpruvot's: root@xxx:~/tpruvot-cpuminer-multi# cat build.sh
#!/bin/bash
if [ "$OS" = "Windows_NT" ]; then
./mingw64.sh
exit 0
fi
# Linux build
make clean || echo clean
rm -f config.status
./autogen.sh || echo done
# Ubuntu 10.04 (gcc 4.4)
extracflags="-O3 -march=native -D_REENTRANT -funroll-loops -fvariable-expansion-in-unroller -fmerge-all-constants -fbranch-target-load-optimize2 -fsched2-use-superblocks -falign-loops=16 -falign-functions=16 -falign-jumps=16 -falign-labels=16"
# Debian 7.7 / Ubuntu 14.04 (gcc 4.7+)
extracflags="$extracflags -Ofast -flto -fuse-linker-plugin -ftree-loop-if-convert-stores"
if [ ! "0" = `cat /proc/cpuinfo | grep -c avx` ]; then
# march native doesn't always works, ex. some Pentium Gxxx (no avx)
extracflags="$extracflags -march=native"
fi
./configure --with-crypto --with-curl CFLAGS="-O3 $extracflags -march=native -DUSE_ASM -pg"
make -j $(grep processor /proc/cpuinfo | wc -l)
strip -s cpuminer
Yours ("customized" by me, but all I actually did was taking most of the flags from tpruvot's as long as it was compilable; maybe I messed it up?): root@xxx:~/cpuminer-opt-3.3.9# cat build.sh
#!/bin/bash
#if [ "$OS" = "Windows_NT" ]; then
# ./mingw64.sh
# exit 0
#fi
# Linux build
make clean || echo clean
rm -f config.status
./autogen.sh || echo done
# Ubuntu 10.04 (gcc 4.4)
extracflags="-O3 -march=native -D_REENTRANT -funroll-loops -fvariable-expansion-in-unroller -fmerge-all-constants -fbranch-target-load-optimize2 -fsched2-use-superblocks -falign-loops=16 -falign-functions=16 -falign-jumps=16 -falign-labels=16"
# Debian 7.7 / Ubuntu 14.04 (gcc 4.7+)
extracflags="$extracflags -Ofast -fuse-linker-plugin -ftree-loop-if-convert-stores"
CFLAGS="-O3 $extracflags -march=native -DUSE_ASM" CXXFLAGS="$CFLAGS -std=gnu++11" ./configure --with-crypto --with-curl
make -j $(grep processor /proc/cpuinfo | wc -l)
strip -s cpuminer
You don't have -flto and -pg, neither is compilable, he doesn't have -std=gnu++11 (doesn't compile either). Now let's see what all the -flto fuzz is about. What happens if I compile tpruvot without it: root@xxx:~/tpruvot-cpuminer-multi# ./cpuminer -a lyra2re --benchmark --threads=64 | grep Total
[2016-08-04 02:56:24] Total: 356.42 kH/s
[2016-08-04 02:56:29] Total: 352.18 kH/s
[2016-08-04 02:56:34] Total: 352.00 kH/s
[2016-08-04 02:56:39] Total: 352.03 kH/s
[2016-08-04 02:56:44] Total: 352.09 kH/s
root@xxx:~/tpruvot-cpuminer-multi# ./cpuminer -a lyra2re --benchmark --threads=128 | grep Total
[2016-08-04 02:57:29] Total: 358.48 kH/s
[2016-08-04 02:57:34] Total: 357.76 kH/s
[2016-08-04 02:57:39] Total: 357.96 kH/s
It turned into a snail. As if it couldn't manage the multiplexing or something. So let's try just 1 thread: root@xxx:~/tpruvot-cpuminer-multi# ./cpuminer -a lyra2re --benchmark --threads=1 | grep Total
[2016-08-04 02:57:45] Total: 124.22 kH/s
[2016-08-04 02:57:50] Total: 124.66 kH/s
[2016-08-04 02:57:55] Total: 125.81 kH/s
[2016-08-04 02:58:00] Total: 126.19 kH/s
[2016-08-04 02:58:05] Total: 126.84 kH/s
And yours: root@xxx:~/cpuminer-opt-3.3.9# ./cpuminer -a lyra2re --benchmark --threads=1 | grep Total
[2016-08-04 02:58:25] Total: 65.54 kH, 136.75 kH/s
[2016-08-04 02:58:30] Total: 683.77 kH, 138.37 kH/s
[2016-08-04 02:58:35] Total: 691.85 kH, 140.37 kH/s
[2016-08-04 02:58:40] Total: 701.86 kH, 140.28 kH/s
[2016-08-04 02:58:45] Total: 701.42 kH, 140.33 kH/s
[2016-08-04 02:58:50] Total: 701.68 kH, 140.74 kH/s
You are clearly faster if he doesn't use -flto. But if I again turn -flto back on and recompile: root@xxx:~/tpruvot-cpuminer-multi# ./cpuminer -a lyra2re --benchmark --threads=1 | grep Total
[2016-08-04 02:59:54] Total: 138.24 kH/s
[2016-08-04 02:59:58] Total: 139.78 kH/s
[2016-08-04 03:00:03] Total: 140.41 kH/s
[2016-08-04 03:00:08] Total: 140.39 kH/s
[2016-08-04 03:00:13] Total: 141.38 kH/s
[2016-08-04 03:00:18] Total: 141.65 kH/s
[2016-08-04 03:00:23] Total: 141.64 kH/s
[2016-08-04 03:00:28] Total: 142.04 kH/s
[2016-08-04 03:00:33] Total: 141.98 kH/s
He is clearly faster after all. Now how do you do with 8 threads? root@xxx:~/cpuminer-opt-3.3.9# ./cpuminer -a lyra2re --benchmark --threads=8 | grep Total
[2016-08-04 03:00:53] Total: 524.29 kH, 1128.89 kH/s
[2016-08-04 03:00:58] Total: 5644.46 kH, 1130.25 kH/s
[2016-08-04 03:01:03] Total: 5651.25 kH, 1130.72 kH/s
[2016-08-04 03:01:08] Total: 5653.59 kH, 1130.44 kH/s
[2016-08-04 03:01:13] Total: 5652.21 kH, 1130.53 kH/s
[2016-08-04 03:01:18] Total: 5652.64 kH, 1130.45 kH/s
And him with -flto? root@xxx:~/tpruvot-cpuminer-multi# ./cpuminer -a lyra2re --benchmark --threads=8 | grep Total
[2016-08-04 03:01:29] Total: 1143 kH/s
[2016-08-04 03:01:34] Total: 1144 kH/s
[2016-08-04 03:01:39] Total: 1144 kH/s
[2016-08-04 03:01:44] Total: 1144 kH/s
[2016-08-04 03:01:49] Total: 1144 kH/s
[2016-08-04 03:01:54] Total: 1145 kH/s
And him without -flto: root@xxx:~/tpruvot-cpuminer-multi# ./cpuminer -a lyra2re --benchmark --threads=8 | grep Total
[2016-08-04 03:03:13] Total: 637.08 kH/s
[2016-08-04 03:03:17] Total: 611.28 kH/s
[2016-08-04 03:03:22] Total: 605.97 kH/s
[2016-08-04 03:03:27] Total: 605.81 kH/s
[2016-08-04 03:03:32] Total: 605.90 kH/s
I support very much your optimizing effort and whenever you need I will gladly do tests for you on various machines. |
|
|
|
|
Ask them, not me. Many of my nodes stopped working with Nicehash, their servers kill the connection immediately.
Someone suggested that it might be because all my servers use the same BTC address of mine and that, for some illogical inexplicable insane unknown reason, is bad so Nicehash stopped accepting connections from any of my servers' IPs. So I will just go elsewhere, what else to say..
|
|
|
|
Things are looking up. I solved the alignment problem and made progress with performance. I've added 3%
to Lyra2 so far and I have a few functions left to convert. So far I've only implemented AVX2, I still have to do
AVX implementations of al functions. The improvements are to the lyra2 core so they should also help lyra2v2.
You are talking about Lyra2RE, right? So it will get sped up? Because so far tpruvot-cpuminer-multi's Lyra2RE is faster than yours by some 3.8% on my servers. I guess it might be because of -flto that I use with his but can't use with yours (doesn't compile) |
|
|
|
yes its "more than a dozen *different* ips", same ip but different miner should not get blocked
Yes, all my machines have (obviously) each a different IP and they all connect to Nicehash using the same BTC address of mine. So that's the problem? So it would be ok if all those connections went from the same IP address (if I for example NATed them all through one single IP) but it's not ok if they all use their own real IPs and my real single BTC? So I should give each machine a different BTC address? But why Nicehash does that? They have no reason to do such a blockage, it doesn't prevent anyone from doing anything, and if I actually did give each of my machine a different BTC address then I would be having many incoming little transactions and I would be loosing money on transaction fees. The Nicehash fee is proportional to the amount so that wouldn't change but when they send BTC they also have to pay a transaction fee (which is not proportional to the amount but fixed) for each transaction and that, I suppose, they deduce from the amount being sent (i.e. I am paying it) so with many little transactions the sum of their fees may stack up. Well I think the best solution will be the stratum proxy. If I won't get that to work then I will just tunnel all those machines through one single IP. |
|
|
|
it works for me, but i had a similar problem when i pointed multiple miners to nicehash, asked support, they block your address if more then a dozen ip's connect to their service for this address
if thats the case, use multiple addresses for each location
Yes you are absolutely right, I figured that they probably block IPs. I think I will try to install a stratum proxy on a clear IP, then all my nodes will connect to the proxy and only the proxy will make one single connection to Nicehash. Btw if anyone is interested, here I made a simple one-liner to test all Nicehash's ports on all their IPs (requires netcat => apt-get install netcat): for port in $(seq 3333 3354); do echo "port $port"; for ip in 37.58.117.214 159.8.42.123 159.253.151.98 198.11.195.136 192.155.218.242 119.81.240.198 161.202.120.197; do echo -n " $ip: "; result=$(echo '{"id": 1, "method": "mining.subscribe", "params": ["cpuminer-multi/1.3-dev"]}' | nc -w 2 $ip $port); [ -n "$result" ] && echo $result || echo "no response"; done; echo; doneNicehash is the only one I found where I can mine Lyra2RE, it seems to be way most profitable CPU algo. I mine ether on GPUs and Lyra2RE on CPUs. I guess everybody else is just leaving all their CPUs idle when they mine with GPUs  Mining Lyra2RE makes 20-25% additional profit on my machines. |
|
|
|
Hey, A new minor upgrade of NiceHash Miner has been released; NVIDIA Lyra2RE algorithm added (very profitable currently). Best regards, NiceHash team. Lyra2RE isn't working for several days now on any of NiceHash servers. EDIT: kind of false alarm. They just block IPs if you actually use them. I guess I have too many machines? Should take my business elsewhere then |
|
|
|
|
Coincidentally Nicehash's Lyra2RE stopped working for me since the weekend, I am getting "stratum_recv_line failed" from all 4 (usa,eu,hk,jp) of their servers. Can anybody please check it because I am going crazy:
cpuminer -a lyra2re -o stratum+tcp://lyra2re.usa.nicehash.com:3342
|
|
|
|
I bought a few of these, the emails work. From what I tested so far: you can get a github pack for students in a few minutes, AWS educate does not seem to work, you get unlimited storage in google drive (that in itself is well worth it).
Thanks for your vouch. However 1 .edu may not cover everything all the time which is really hard to achieve Absolutely. And I think you should mention in the OP that this email enables access to the Google Apps for Education which among others include unlimited storage for Google Drive, Gmail, and Google Photos with individual files up to 5TB in size. |
|
|
|
|
I bought a few of these, the emails work. From what I tested so far: you can get a github pack for students in a few minutes, AWS educate does not seem to work, you get unlimited storage in google drive (that in itself is well worth it).
|
|
|
|
|
How many of those can be applied in the same account?
EDIT: the code works and I suppose only 1 can go.
|
|
|
|
|
Well then get a U.S. bank account or just CC/VCC with U.S. IBAN and withdraw the money yourself. For those $22 that you are willing to spend to get your money you should be able to get multiple of those. That way you will get your money, you will keep your PayPal account and as a bonus you will have a U.S. bank account with the card.
|
|
|
|
|
I am sure we all are sorry for your situation (if that's the truth) but you have to realize that nobody can withdraw money from someone else's account, it would be a criminal offense of fraud/theft. You can't officially sell your PayPal account so it would still be your account and anyone doing anything to your account, even just logging into it, will be violating the rules and regulations and possibly the law, and if they try to touch your money they become criminals. Nobody is going to risk all the legal trouble for lousy $92 (for which they had to pay $70 beforehand) so just drop it.
$92 won't save you from death nor help you in any real way but it's an convenient amount to scam people for.
|
|
|
|
I strongly discourage anyone from falling for this. Just think about it: if he, the rightful owner of the account (or so he claims), can't withdraw his own money how you, the fake owner who obtained the account illegally, possibly could? There is nothing you can do that he couldn't do himself. Here is the official PayPal information regarding PayPal accounts in Egypt and it clearly states that an account holder can withdraw money either to a Visa card issued in Egypt or a bank account in the U.S.: https://www.paypal.com/eg/webapps/mpp/setup-to-receive-withdraw. Now if he claims he can't withdraw to his Visa card issued in Egypt whom are you going to believe? An anonymous person or PayPal? Furthermore, even this suspicious text with laughably incorrect grammar and capitalization (allegedly from some PayPal "supporter", or so you are led to believe) states again that he in fact can withdraw money to his Visa XXX16:   If you buy this account from him it will be against the PayPal terms, you can't rename it so it will still be in his name and if you try to link it with your bank account (which is in your name) and withdraw into it you will be committing a felony according to the U.S. Penal Code. Let alone the fact that once you log into this account from a totally different IP address from a different country and possibly even a wholly different continent only hours after him the account may get locked just because of that as it's physically impossible that you could be the account owner. It just makes you wonder why would anyone open a PayPal account he can't withdraw money from and then receive some money into it..
|
|
|
|
Just purchased, the code works and indeed does go together with the other codes I've already got. Thanks. There is only 1 code left so hurry, guys  |
|
|
|
I don't think this is a matter of having or not having "enough trust" as Angell mentioned, rather this is a matter of having a common sense. Once you have the private keys that's it, nothing is stopping you from having these keys for ever and ever and ever and thus anytime in the future being able to take all the money from that address. Whatever you may possibly say, whatever promises, assurances etc., that's all just empty words, just blah blah blah. Trust and money are two totally incompatible concepts. There could be exceptions on case-to-case basis but this would require a lobotomy. Furthermore, comparing the ability to anytime, for ever, take any amount of money from that address to a limited ability of one single particular chargeback of one single particular amount, that's so incomparable that the very fact that you are actually trying to make them seem similar is a red flag in itself. I understand if you just want to make a few extra bucks but if you are really honest then just by trying this you are risking a permanent damage to your reputation and your reputation is in fact the only thing that proves your honesty. For a few bucks you are basically reducing yourself down to the level of this straightforward scammer Symeave who opens an account, obtains some virtual credit cards with IBANs and then tries to sell those VCCs to the people here, of course without telling them that the VCCs are issued in his account and he has, and forever will have, full access to whatever money they ever send to the cards' IBANs. When I pointed that out ( https://bitcointalk.org/index.php?topic=1542649.msg15709256#msg15709256) he silently created the same thread again ( https://bitcointalk.org/index.php?topic=1565393.0), this time as a self-moderated topic so that he can delete any post that he doesn't like. Do you really want to be seen as the same lowlife scammer? The choice is yours. |
|
|
|
|
You are selling a BTC address to which you have, and will keep, full access so that you can anytime and forever take any money stored in that address at your will, and you actually expect somebody to pay you $65 for it?
|
|
|
|
|
Show Posts
