New pure-python CPU miner, for fun and testing

tarzen

newbie

Activity: 10

Merit: 0

i'm trying this with Python 2.7 on Linux and get a an error on line ~158 at

Code:

return (nonce + 1, None)

Python complains that nonce has not been defined - which is true since this call is made outside of the loop

Code:

for nonce in xrange(self.max_nonce):

which defines nonce

I added a class variable self._nonce in the __init__ function of class Miner and use this t keep track of the nonce value.

A question:
Is it expected behaviour for the hash rate to halve on each iteration of the work function?

Unthinkingbit

hero member

Activity: 935

Merit: 1015

By condensing code in the nonce loop, pyminer gets around 6% more speed on my machine. The hash part of the loop ends up being:

Code:

# hash final 4b, the nonce value
hash1_o = static_hash.copy()

# encode 32-bit nonce value
hash1_o.update(struct.pack("
# sha256 hash of sha256 hash
hash = hashlib.sha256(hash1_o.digest()).digest()

Below is a diff of the changes.

Code:

--- pyminer.py 2011-06-11 14:47:44.000000000 -0700
+++ pyminer_test.py 2011-06-11 14:47:04.000000000 -0700
@@ -121,18 +121,14 @@

for nonce in xrange(self.max_nonce):

- # encode 32-bit nonce value
- nonce_bin = struct.pack("-
# hash final 4b, the nonce value
hash1_o = static_hash.copy()
- hash1_o.update(nonce_bin)
- hash1 = hash1_o.digest()
+
+ # encode 32-bit nonce value
+ hash1_o.update(struct.pack("
# sha256 hash of sha256 hash
- hash_o = hashlib.sha256()
- hash_o.update(hash1)
- hash = hash_o.digest()
+ hash = hashlib.sha256(hash1_o.digest()).digest()

# quick test for winning solution: high 32 bits zero?
if hash[-4:] != '\0\0\0\0':

tcash21

newbie

Activity: 1

Merit: 0

Quote from: dsp77 on June 03, 2011, 06:21:21 PM

Quote from: dsp77 on June 03, 2011, 05:59:06 PM

...
Sorry, not in a patch form.

OK, fixed that. Here is a diff of it.

Code:

diff --git pyminer.py pyminer.py
index 051abb3..b203aa8 100755
--- pyminer.py
+++ pyminer.py
@@ -97,8 +97,9 @@ def wordreverse(in_buf):
return ''.join(out_words)

class Miner:
- def __init__(self, id):
+ def __init__(self, id, settings=None):
self.id = id
+ self.settings=settings
self.max_nonce = MAX_NONCE

def work(self, datastr, targetstr):
@@ -181,11 +182,11 @@ class Miner:
time_diff = time_end - time_start

self.max_nonce = long(
- (hashes_done * settings['scantime']) / time_diff)
+ (hashes_done * self.settings['scantime']) / time_diff)
if self.max_nonce > 0xfffffffaL:
self.max_nonce = 0xfffffffaL

- if settings['hashmeter']:
+ if self.settings['hashmeter']:
print "HashMeter(%d): %d hashes, %.2f Khash/sec" % (
self.id, hashes_done,
(hashes_done / 1000.0) / time_diff)
@@ -194,16 +195,16 @@ class Miner:
self.submit_work(rpc, work['data'], nonce_bin)

def loop(self):
- rpc = BitcoinRPC(settings['host'], settings['port'],
- settings['rpcuser'], settings['rpcpass'])
+ rpc = BitcoinRPC(self.settings['host'], self.settings['port'],
+ self.settings['rpcuser'], self.settings['rpcpass'])
if rpc is None:
return

while True:
self.iterate(rpc)

-def miner_thread(id):
- miner = Miner(id)
+def miner_thread(id, settings):
+ miner = Miner(id, settings)
miner.loop()

if __name__ == '__main__':
@@ -248,7 +249,7 @@ if __name__ == '__main__':

thr_list = []
for thr_id in range(settings['threads']):
- p = Process(target=miner_thread, args=(thr_id,))
+ p = Process(target=miner_thread, args=(thr_id, settings,))
p.start()
thr_list.append(p)
time.sleep(1) # stagger threads

I'm still getting the BadStatusLine error even after adding these fixes to the script. I'm trying to run on Ubuntu 10.04.2 LTS 64-bit.
Any ideas?

elmom

newbie

Activity: 21

Merit: 0

Running it on pypy (recent nightly build), it leaks and fills my memory fast, any idea if the code relies too much on cpython reference counting, or should I report this to pypy devs?

Otherwise, about the slowness of pypy is propalby related to the fact that the hashing is relegated to optimized c-code, but pypy uses a pure python or non-optimized rpython implementation. Has someone done profiling of the latest code?

dsp77

newbie

Activity: 8

Merit: 0

Quote from: psychocoder on June 06, 2011, 12:26:52 PM

I also get the error if I use btcguild. I use phython 2.5.3 on linux debian lenny (is this a problem). I run The program with one thread.

...

I ran it under Ubuntu Lucid Lynx with Python 2.6.5 and under Windows with Python 2.7. In both cases I get the error.

It looks like the problem occurs when after the second request another one is started. When adding debug output I see it run in the miner.loop() the self.iterate(rpc) call twice. After it finishes the second loop and starts the 3rd call to self.iterate(rpc) the response with an empty header is returned.

Breaking with the 3rd rpc call is independent of the amount of threads. So with one thread it breaks with the 3rd call and with two threads running it breaks each thread with the 3rd call as well. Undecided

One interesting thing to note is, that the max_nonce value goes up. I thought it would go down in consecutive runs. In the first run it is set to 1,000,000. In the second run it goes up to 4,885,595.

psychocoder

newbie

Activity: 49

Merit: 0

I also get the error if I use btcguild. I use phython 2.5.3 on linux debian lenny (is this a problem). I run The program with one thread.

Quote

Process Process-1:
Traceback (most recent call last):
  File "/var/lib/python-support/python2.5/multiprocessing/process.py", line 237, in _bootstrap
   self.run()
  File "/var/lib/python-support/python2.5/multiprocessing/process.py", line 93, in run
   self._target(*self._args, **self._kwargs)
  File "pyminer.py", line 214, in miner_thread
   miner.loop()
  File "pyminer.py", line 210, in loop
   self.iterate(rpc)
  File "pyminer.py", line 174, in iterate
   work = rpc.getwork()
  File "pyminer.py", line 83, in getwork
   return self.rpc('getwork', data)
  File "pyminer.py", line 63, in rpc
   resp = self.conn.getresponse()
  File "/usr/lib/python2.5/httplib.py", line 928, in getresponse
   response.begin()
  File "/usr/lib/python2.5/httplib.py", line 385, in begin
   version, status, reason = self._read_status()
  File "/usr/lib/python2.5/httplib.py", line 349, in _read_status
   raise BadStatusLine(line)
BadStatusLine

dsp77

newbie

Activity: 8

Merit: 0

I ran the latest git-version with the patch I posted earlier under windows and linux using an account at btcguild.com and got the following error message:

Code:

G:\Dokumente und Einstellungen\anna\Eigene Dateien\pj\pyminer>python pyminer.py btcguild-config.cfg
2 mining threads started
Sat Jun 04 00:41:52 2011 Miner Starts - btcguild.com:8332
HashMeter(0): 1000000 hashes, 136.18 Khash/sec
HashMeter(1): 1000000 hashes, 136.46 Khash/sec
HashMeter(0): 8171047 hashes, 135.41 Khash/sec
Process Process-1:
Traceback (most recent call last):
File "G:\Python27\lib\multiprocessing\process.py", line 232, in _bootstrap
self.run()
File "G:\Python27\lib\multiprocessing\process.py", line 88, in run
self._target(*self._args, **self._kwargs)
File "G:\Dokumente und Einstellungen\anna\Eigene Dateien\pj\pyminer\pyminer.py", line 208, in miner_thread
miner.loop()
File "G:\Dokumente und Einstellungen\anna\Eigene Dateien\pj\pyminer\pyminer.py", line 204, in loop
self.iterate(rpc)
File "G:\Dokumente und Einstellungen\anna\Eigene Dateien\pj\pyminer\pyminer.py", line 168, in iterate
work = rpc.getwork()
File "G:\Dokumente und Einstellungen\anna\Eigene Dateien\pj\pyminer\pyminer.py", line 76, in getwork
return self.rpc('getwork', data)
File "G:\Dokumente und Einstellungen\anna\Eigene Dateien\pj\pyminer\pyminer.py", line 56, in rpc
resp = self.conn.getresponse()
File "G:\Python27\lib\httplib.py", line 1025, in getresponse
response.begin()
File "G:\Python27\lib\httplib.py", line 401, in begin
version, status, reason = self._read_status()
File "G:\Python27\lib\httplib.py", line 365, in _read_status
raise BadStatusLine(line)
BadStatusLine: ''
HashMeter(1): 8187772 hashes, 135.47 Khash/sec
Process Process-2:
...
Sat Jun 04 00:43:00 2011 Miner Stops - btcguild.com:8332

Running it with two threads I took out the trace back from the second process.

Anyone knows why there is an empty status line returned?

Thanks for your help.

dsp77

newbie

Activity: 8

Merit: 0

Quote from: dsp77 on June 03, 2011, 05:59:06 PM

...
Sorry, not in a patch form.

OK, fixed that. Here is a diff of it.

Code:

diff --git pyminer.py pyminer.py
index 051abb3..b203aa8 100755
--- pyminer.py
+++ pyminer.py
@@ -97,8 +97,9 @@ def wordreverse(in_buf):
return ''.join(out_words)

class Miner:
- def __init__(self, id):
+ def __init__(self, id, settings=None):
self.id = id
+ self.settings=settings
self.max_nonce = MAX_NONCE

def work(self, datastr, targetstr):
@@ -181,11 +182,11 @@ class Miner:
time_diff = time_end - time_start

self.max_nonce = long(
- (hashes_done * settings['scantime']) / time_diff)
+ (hashes_done * self.settings['scantime']) / time_diff)
if self.max_nonce > 0xfffffffaL:
self.max_nonce = 0xfffffffaL

- if settings['hashmeter']:
+ if self.settings['hashmeter']:
print "HashMeter(%d): %d hashes, %.2f Khash/sec" % (
self.id, hashes_done,
(hashes_done / 1000.0) / time_diff)
@@ -194,16 +195,16 @@ class Miner:
self.submit_work(rpc, work['data'], nonce_bin)

def loop(self):
- rpc = BitcoinRPC(settings['host'], settings['port'],
- settings['rpcuser'], settings['rpcpass'])
+ rpc = BitcoinRPC(self.settings['host'], self.settings['port'],
+ self.settings['rpcuser'], self.settings['rpcpass'])
if rpc is None:
return

while True:
self.iterate(rpc)

-def miner_thread(id):
- miner = Miner(id)
+def miner_thread(id, settings):
+ miner = Miner(id, settings)
miner.loop()

if __name__ == '__main__':
@@ -248,7 +249,7 @@ if __name__ == '__main__':

thr_list = []
for thr_id in range(settings['threads']):
- p = Process(target=miner_thread, args=(thr_id,))
+ p = Process(target=miner_thread, args=(thr_id, settings,))
p.start()
thr_list.append(p)
time.sleep(1) # stagger threads

dsp77

newbie

Activity: 8

Merit: 0

Quote from: Kiv on February 27, 2011, 03:53:42 PM

I'm trying to run this and getting an exception:

Code:

Process Process-1:
Traceback (most recent call last):
File "C:\Python26\lib\multiprocessing\process.py", line 232, in _bootstrap
self.run()
File "C:\Python26\lib\multiprocessing\process.py", line 88, in run
self._target(*self._args, **self._kwargs)
File "C:\Users\Chris\workspace\pyminer\pyminer.py", line 197, in miner_thread
miner.loop()
File "C:\Users\Chris\workspace\pyminer\pyminer.py", line 187, in loop
rpc = BitcoinRPC(settings['host'], settings['port'],
KeyError: 'host'
1 mining threads started
Sun Feb 27 15:51:18 2011 Miner Starts - 127.0.0.1:8332
Sun Feb 27 15:51:18 2011 Miner Stops - 127.0.0.1:8332

It seems that the settings dictionary is empty, because when we're using separate processes the children don't have access to the parent's variables. Probably the settings should be passed in as constructor arguments. I might put this up on GitHub and hack on it a bit if you don't mind

I did some changes to make it work under Windows:

Code:

class Miner:
def __init__(self, id, settings=None):
self.id = id
self.settings=settings
self.max_nonce = MAX_NONCE
...

def iterate(self, rpc):
...
self.max_nonce = long(
(hashes_done * self.settings['scantime']) / time_diff)
if self.max_nonce > 0xfffffffaL:
self.max_nonce = 0xfffffffaL
...
if self.settings['hashmeter']:
print "HashMeter(%d): %d hashes, %.2f Khash/sec" % (
self.id, hashes_done,
(hashes_done / 1000.0) / time_diff)

...

def loop(self):
rpc = BitcoinRPC(self.settings['host'], self.settings['port'],
self.settings['rpcuser'], self.settings['rpcpass'])
if rpc is None:
return

...

def miner_thread(id, settings):
miner = Miner(id, settings)
miner.loop()

...

thr_list = []
for thr_id in range(settings['threads']):
p = Process(target=miner_thread, args=(thr_id, settings))
p.start()
thr_list.append(p)
time.sleep(1) # stagger threads

Sorry, not in a patch form.

fix

newbie

Activity: 4

Merit: 0

Really, really useful. Thank you very much.

But what I don't understand is the meaning of the RPC work data (data, hash1, midstate, target values) and why when a proof-of-work isn't found this isn't replied to the pool. I know, I've got some confusion in my mind.
There's some documentation about this?

jgarzik

legendary

Activity: 1596

Merit: 1091

Updated the git repo with many bug fixes. pyminer.py successfully generated a block on testnet, and a share in slush's pool.

TehRainbowGuy

newbie

Activity: 8

Merit: 0

Thanks!

jgarzik

legendary

Activity: 1596

Merit: 1091

Sure, here's a git repo: https://github.com/jgarzik/pyminer

TehRainbowGuy

newbie

Activity: 8

Merit: 0

I am loving the idea of this miner, much simpler than all of the others.

Just one question, any chance of a git repo?

rasputin

newbie

Activity: 14

Merit: 0

Quote from: jgarzik on February 17, 2011, 08:11:01 PM

Quote from: kseistrup on February 17, 2011, 07:13:23 PM

Quote from: jgarzik on February 17, 2011, 05:52:40 PM

I'm not sure you caught my point.

You're right, I didn't catch your point. I have to chew on that, though. What we need is a binary buffer with the right endianness, not a python long integer, right?

No, we really do need a 256-integer, because that is the fundamental proof-of-work test in the bitcoin system, comparing two 256-bit integers:

hash < target

You may just as well compare the two arrays of binary data:

Code:

>>> a = "aaaa".decode('hex')
>>> b = "bbbb".decode('hex')
>>> a > b
False
>>> b > a
True

Got to love high level languages.

Considered putting this on github as well, btw? Being smaller and higher level, I find it a lot easier to grasp than the C miner. Too bad about all the byte order manipulation, but I guess that can't be helped.

rasputin

newbie

Activity: 14

Merit: 0

Quote from: kseistrup on February 19, 2011, 06:10:33 AM

I wonder if this is a bug:

Code:

# hash final 4b, the nonce value
   hash1_o = hash1_precalc_o
   hash1_o.update(nonce_bin)
   hash1 = hash1_o.digest()

Do I interpret your meaning right when I say that the code — for each value of nonce — intends to calculate the SHA256 digest of the precalculated value (intended to stay constant throughout the loop) updated with the nonce? If so, that is not what the code does. Please see this simplified example:

Code:

>>> from hashlib import sha256
>>> pre = sha256('abc')
>>> post = pre
>>> post.update('xyz')
>>> pre.digest() == post.digest()
True
>>>

So pre and post (like hash1_precalc_o and hash1_o) both points to the same object

Instead of assignment, you'd use hashlib's copy() method to duplicate the hash state.

This code should work as intended:

Code:

# hash final 4b, the nonce value
hash1_o = hash1_precalc_o.copy()
hash1_o.update(nonce_bin)
hash1 = hash1_o.digest()

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: Kiv on February 27, 2011, 04:26:13 PM

If everyone else is running on Linux but I am running on Windows, this explains why it would work for others but not for me.

It sure does.

Cheers,

Kiv

full member

Activity: 162

Merit: 100

This is probably a Windows-only bug. From the multiprocessing docs:

Since Windows lacks os.fork() it has a few extra restrictions:
Global variables

Bear in mind that if code run in a child process tries to access a global variable, then the value it sees (if any) may not be the same as the value in the parent process at the time that Process.start() was called.

If everyone else is running on Linux but I am running on Windows, this explains why it would work for others but not for me.

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: Kiv on February 27, 2011, 04:03:21 PM

I suspect that the program worked fine at one point and I have a more recent version with a bug in it.

The current version runs fine with this config file (username and password edited for obvious reasons):

Code:

host = mining.bitcoin.cz
port = 8332
rpcuser = myRPCUser
rpcpass = myRPCPass
logdir = /tmp
threads = 1
hashmeter = 1

Cheers,

Kiv

full member

Activity: 162

Merit: 100

Thank you, but I did supply the argument. In fact the program will not start without the config file. My config file is the right format as well and has "host=127.0.0.1" in it.

I suspect that the program worked fine at one point and I have a more recent version with a bug in it.

Quote from: kseistrup on February 27, 2011, 03:57:49 PM

Quote from: Kiv on February 27, 2011, 03:53:42 PM

Code:

File "C:\Users\Chris\workspace\pyminer\pyminer.py", line 187, in loop
rpc = BitcoinRPC(settings['host'], settings['port'],
KeyError: 'host'

Pyminer takes as its single argument the name of the config file. Please see message #10 in this thread.

Cheers,

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: Kiv on February 27, 2011, 03:53:42 PM

Code:

File "C:\Users\Chris\workspace\pyminer\pyminer.py", line 187, in loop
rpc = BitcoinRPC(settings['host'], settings['port'],
KeyError: 'host'

Pyminer takes as its single argument the name of the config file. Please see message #10 in this thread.

Cheers,

Kiv

full member

Activity: 162

Merit: 100

Quote from: bitjet on February 19, 2011, 03:43:22 PM

Are you guys getting near triple efficiency with this vs over poclbm?

Definitely not, I get about 900 KHash per core on poclbm and only 182 KHash per core on pure-python.

Kiv

full member

Activity: 162

Merit: 100

I'm trying to run this and getting an exception:

Code:

Process Process-1:
Traceback (most recent call last):
File "C:\Python26\lib\multiprocessing\process.py", line 232, in _bootstrap
self.run()
File "C:\Python26\lib\multiprocessing\process.py", line 88, in run
self._target(*self._args, **self._kwargs)
File "C:\Users\Chris\workspace\pyminer\pyminer.py", line 197, in miner_thread
miner.loop()
File "C:\Users\Chris\workspace\pyminer\pyminer.py", line 187, in loop
rpc = BitcoinRPC(settings['host'], settings['port'],
KeyError: 'host'
1 mining threads started
Sun Feb 27 15:51:18 2011 Miner Starts - 127.0.0.1:8332
Sun Feb 27 15:51:18 2011 Miner Stops - 127.0.0.1:8332

It seems that the settings dictionary is empty, because when we're using separate processes the children don't have access to the parent's variables. Probably the settings should be passed in as constructor arguments. I might put this up on GitHub and hack on it a bit if you don't mind

bitjet

hero member

Activity: 696

Merit: 500

Are you guys getting near triple efficiency with this vs over poclbm?

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: mail2345 on February 19, 2011, 02:54:30 PM

Psyco should speed things up.

Psyco only runs on “obsolete” versions of Python (and it doesn't work on any 64-bit systems at all).

Cheers,

mail2345

newbie

Activity: 16

Merit: 0

Psyco should speed things up.

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: jgarzik on February 19, 2011, 01:54:04 PM

No, just the URL.

Okay, no pull-request for you, then. Wink

Please consider the following patch for inclusion:

Code:

--- pyminer.py 2011-02-19 19:05:29.794843964 +0100
+++ pyminer-opt.py 2011-02-19 19:20:06.136093826 +0100
@@ -113,28 +113,22 @@
targetbin_str = targetbin.encode('hex')
target = long(targetbin_str, 16)

- hashes_done = 1
- for nonce in range(MAX_NONCE):
+ for nonce in xrange(MAX_NONCE):

# encode 32-bit nonce value
nonce_bin = struct.pack("
# hash final 4b, the nonce value
- hash1_o = hashlib.sha256()
- hash1_o.update(blk_hdr)
+ hash1_o = hashlib.sha256(blk_hdr)
hash1_o.update(nonce_bin)
hash1 = hash1_o.digest()

# sha256 hash of sha256 hash
- hash_o = hashlib.sha256()
- hash_o.update(hash1)
+ hash_o = hashlib.sha256(hash1)
hash = hash_o.digest()

- hashes_done += 1
-
# quick test for winning solution: high 32 bits zero?
- H = struct.unpack('- if H:
+ if hash[-4:] != '\0\0\0\0':
continue

# convert binary hash to 256-bit Python long
@@ -147,12 +141,12 @@
# proof-of-work test: hash < target
if l < target:
print time.asctime(), "PROOF-OF-WORK found: %064x" % (l,)
- return (hashes_done,
+ return (nonce+1,
static_data[:76] + nonce_bin)
else:
print time.asctime(), "PROOF-OF-WORK false positive %064x" % (l,)

- return (hashes_done, None)
+ return (nonce+1, None)

def iterate(self, rpc):
work = rpc.getwork()

The patch eliminates 2 calls to hashlib.sha256(), 1 call to struct.unpack() and 1 one variable increment per work loop.

Cheers,

jgarzik

legendary

Activity: 1596

Merit: 1091

Quote from: kseistrup on February 19, 2011, 01:52:54 PM

Quote from: jgarzik on February 19, 2011, 01:44:36 PM

Updated.

Do you happen to have a git repository of the code?

No, just the URL.

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: jgarzik on February 19, 2011, 01:44:36 PM

Updated.

Do you happen to have a git repository of the code?

Cheers,

jgarzik

legendary

Activity: 1596

Merit: 1091

Quote from: kseistrup on February 19, 2011, 06:10:33 AM

Do I interpret your meaning right when I say that the code — for each value of nonce — intends to calculate the SHA256 digest of the precalculated value (intended to stay constant throughout the loop) updated with the nonce? If so, that is not what the code does. Please see this simplified example:

Well, that is disappointing. If true, yes, that is a bug. Updated.

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

I wonder if this is a bug:

Code:

# hash final 4b, the nonce value
hash1_o = hash1_precalc_o
hash1_o.update(nonce_bin)
hash1 = hash1_o.digest()

Do I interpret your meaning right when I say that the code — for each value of nonce — intends to calculate the SHA256 digest of the precalculated value (intended to stay constant throughout the loop) updated with the nonce? If so, that is not what the code does. Please see this simplified example:

Code:

>>> from hashlib import sha256
>>> pre = sha256('abc')
>>> post = pre
>>> post.update('xyz')
>>> pre.digest() == post.digest()
True
>>>

So pre and post (like hash1_precalc_o and hash1_o) both points to the same object, and so I guess we should really write:

Code:

# hash final 4b, the nonce value
hash1_o = hashlib.sha256(blk_hdr + nonce_bin)
hash1 = hash1_o.digest()

or even

Code:

# hash final 4b, the nonce value
hash1 = hashlib.sha256(blk_hdr + nonce_bin).digest()

Or did I misunderstand the whole concept?

Cheers,

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Oh, and one more thing:

In the work loop, the line “hashes_done += 1” is effectively useless since nonce is already running from 0 to MAX_NONCE. So at the end of the loop — either because a winning solution has been found, or because MAX_NONCE has been reached — hashes_done equals nonce+1. Dropping the “hashes_done += 1” line gives a few extra khash/sec since this line was executed so many times.

Cheers,

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: jgarzik on February 17, 2011, 08:27:15 PM

Ok, implemented the H==0 test shortcut (hey, everyone uses, might as well document it) and pyminer.py now gets 250 Khash/sec for one thread, on my box.

That's cool! Cool

If you drop the unpack stuff and compare the last 4 bytes directly you can gain another 10 khash/sec or so:

Code:

if hash[-4:] != '\0\0\0\0':
continue

Cheers,

jgarzik

legendary

Activity: 1596

Merit: 1091

Ok, implemented the H==0 test shortcut (hey, everyone uses, might as well document it) and pyminer.py now gets 250 Khash/sec for one thread, on my box.

jgarzik

legendary

Activity: 1596

Merit: 1091

Quote from: kseistrup on February 17, 2011, 07:13:23 PM

Quote from: jgarzik on February 17, 2011, 05:52:40 PM

I'm not sure you caught my point.

You're right, I didn't catch your point. I have to chew on that, though. What we need is a binary buffer with the right endianness, not a python long integer, right?

No, we really do need a 256-integer, because that is the fundamental proof-of-work test in the bitcoin system, comparing two 256-bit integers:

hash < target

Almost every practical miner simplifies this test to simply verify that the final, most-significant 32 bits are zero.

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: jgarzik on February 17, 2011, 05:52:40 PM

I'm not sure you caught my point.

You're right, I didn't catch your point. I have to chew on that, though. What we need is a binary buffer with the right endianness, not a python long integer, right?

Cheers,

jgarzik

legendary

Activity: 1596

Merit: 1091

Quote from: kseistrup on February 17, 2011, 05:08:02 PM

Quote from: jgarzik on February 17, 2011, 03:46:16 PM

And remember, that bufreverse/wordreverse/encode/long sequence exists solely to build a 256-bit integer.

If you accept the patch that basically makes “hash = hash[::-1]” instead of a call to bufreverse() followed by a call to wordreverse(), there is no longer any use for wordreverse(), plus bufreverse() is ever only used in one place.

The “hash[::-1]” (i.e., “reverse the entire buffer”) could be explained in a comment above the invocation.

I'm not sure you caught my point. If one (a) iterates over each 4-byte sub-string, (b) uses struct.unpack to perform endian conversion, and (c) uses Python integer math to build a 256-bit long, there should be no need for hash=hash[::-1] or bufreverse/wordreverse.

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: jgarzik on February 17, 2011, 03:46:16 PM

And remember, that bufreverse/wordreverse/encode/long sequence exists solely to build a 256-bit integer.

If you accept the patch that basically makes “hash = hash[::-1]” instead of a call to bufreverse() followed by a call to wordreverse(), there is no longer any use for wordreverse(), plus bufreverse() is ever only used in one place.

The “hash[::-1]” (i.e., “reverse the entire buffer”) could be explained in a comment above the invocation.

Cheers,

jgarzik

legendary

Activity: 1596

Merit: 1091

I don't mind optimizing bufreverse/wordreverse, but I would like to avoid optimizing Miner.work() so heavily that it cannot be read. That's the function that must be most-readable to other humans.

And remember, that bufreverse/wordreverse/encode/long sequence exists solely to build a 256-bit integer. It would probably be more optimal to simply build a 256-bit integer using a per-word loop and shifts, such as

Code:

s = 'binary string...'
r = 0L
for i in range(8):
w32 = struct.unpack('>I', s[i:i+4])
r = r | (w32 << (i * 32))

According to the docs, we can specify the byte order to struct.unpack()

nono2357

newbie

Activity: 8

Merit: 0

More optimized functions:

Code:

def bufreverse(ch):
return "".join([ch[i:i+4][::-1] for i in range(0,len(ch),4)])

def wordreverse(ch):
return "".join([ch[i:i+4] for i in range(0,len(ch),4)][::-1])

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

The way I see it, these two lines have the effect of reversing the entire data string:

Code:

hash = bufreverse(hash)
hash = wordreverse(hash)

and could be replaced by

Code:

hash = hash[::-1]

If I am right, then the following small patch will triple the efficiency (from ~32 khash/sec to more than 100 khash/sec):

Code:

--- pyminer.py 2011-02-17 19:37:34.254845344 +0100
+++ pyminer-opt.py 2011-02-17 19:47:44.794843788 +0100
@@ -76,18 +76,10 @@
def getwork(self, data=None):
return self.rpc('getwork', data)

-def uint32(x):
- return x & 0xffffffffL
-
-def bytereverse(x):
- return uint32(( ((x) << 24) | (((x) << 8) & 0x00ff0000) |
- (((x) >> 8) & 0x0000ff00) | ((x) >> 24) ))
-
def bufreverse(in_buf):
out_words = []
for i in range(0, len(in_buf), 4):
- word = struct.unpack('@I', in_buf[i:i+4])[0]
- out_words.append(struct.pack('@I', bytereverse(word)))
+ out_words.append(in_buf[i:i+4][::-1])
return ''.join(out_words)

def wordreverse(in_buf):
@@ -115,10 +107,7 @@

hash_o = hashlib.sha256()
hash_o.update(hash1)
- hash = hash_o.digest()
-
- hash = bufreverse(hash)
- hash = wordreverse(hash)
+ hash = hash_o.digest()[::-1]

hash_str = hash.encode('hex')
l = long(hash_str, 16)

A few other optimizations can be made within the loop to raise the efficiency to approx. 110 khash/sec on my machine, but this will make the code less legible and is probably not worth it.

Cheers,

jgarzik

legendary

Activity: 1596

Merit: 1091

Quote from: kseistrup on February 17, 2011, 08:41:40 AM

Quote from: jgarzik on February 16, 2011, 06:57:19 PM

http://yyz.us/bitcoin/pyminer.py

It is computational expensive to repeatedly concatenate strings in Python. I suggest using [].append() in the bufreverse() and wordreverse() methods:

Updated the source code with your suggestions. Thanks -- this is my second python program, so I guarantee there are other improvements that expert Python programmers can discover.

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: epicenter on February 17, 2011, 09:08:46 AM

required args?
not sure of the format for files in python

Pyminer takes as its single argument the name of the config file.

The config file consists of lines of the form

Code:

key = value

Keys can be host, port, rpcuser, rpcpass, threads, logdir and hashmeter.

Cheers,

epicenter

newbie

Activity: 10

Merit: 0

required args?
not sure of the format for files in python

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Furthermore, you could bytereverse a 4-byte buf using it's own builtin medthod:

Code:

>>> x = 'abcd'
>>> x[::-1]
'dcba'
>>>

No need to use the pack/unpack stuff…

Cheers,

Cdecker

hero member

Activity: 489

Merit: 504

Great stuff, finally I can wrap my head around how mining works without having to crawl through additional stuff

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: jgarzik on February 16, 2011, 06:57:19 PM

http://yyz.us/bitcoin/pyminer.py

It is computational expensive to repeatedly concatenate strings in Python. I suggest using [].append() in the bufreverse() and wordreverse() methods:

Code:

diff --git a/pyminer.py b/pyminer.py
index cfd79a2..f972a75 100644
--- a/pyminer.py
+++ b/pyminer.py
@@ -84,21 +84,18 @@ def bytereverse(x):
(((x) >> 8) & 0x0000ff00) | ((x) >> 24) ))

def bufreverse(in_buf):
- out_words = ""
+ out_words = []
for i in range(0, len(in_buf), 4):
word = struct.unpack('@I', in_buf[i:i+4])[0]
- out_words += struct.pack('@I', bytereverse(word))
- return out_words
+ out_words.append(struct.pack('@I', bytereverse(word)))
+ return ''.join(out_words)

def wordreverse(in_buf):
out_words = []
for i in range(0, len(in_buf), 4):
out_words.append(in_buf[i:i+4])
out_words.reverse()
- out_buf = ""
- for word in out_words:
- out_buf += word
- return out_buf
+ return ''.join(out_words)

class Miner:
def __init__(self, id):

Cheers,

Kiv

full member

Activity: 162

Merit: 100

This is really cool

It'll definitely help me understand better how the mining works.

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: kseistrup on February 16, 2011, 07:31:28 PM

I wonder if this will run faster under PyPy?

I just had to try… Wink

Running pyminer on Python2.7 yields ~30 khash/sec, under PyPy less than 3 khash/sec. Oh, well…

Cheers,

kseistrup

hero member

Activity: 566

Merit: 500

Unselfish actions pay back better

Quote from: jgarzik on February 16, 2011, 06:57:19 PM

This is single-threaded, and pulls a whopping 43 Khash/sec on my box.

Cool!

I wonder if this will run faster under PyPy (pystone.main() is more than 10 times faster on PyPy on my machine)?

Cheers,

Cryptoman

hero member

Activity: 726

Merit: 500

Quote from: jgarzik on February 16, 2011, 06:57:19 PM

This is single-threaded, and pulls a whopping 43 Khash/sec on my box.

What type of system are you running? Hamsters in a box with pencil and paper? Grin

jgarzik

legendary

Activity: 1596

Merit: 1091

Here is a new high performance, highly optimized CPU miner:

https://github.com/jgarzik/pyminer

This pulls a whopping 256 Khash/sec per thread on my box.

The main purpose is to demonstrate the mining algorithm, in a small amount of portable code. And to have a bit of fun.

Topic: New pure-python CPU miner, for fun and testing (Read 25535 times)