Pages:
Author

Topic: Ufasoft Miner - Windows/Linux, x86/x64, SSE2/OpenCL, Open Source - page 53. (Read 630787 times)

sr. member
Activity: 404
Merit: 251
Thanks for your help, but unfortunately it looks like I'll try to use your code as a guide to re-implement it as part of jgarzik's miner instead of trying to build it myself.

That is the best way. BTW I have reimplemented entire SHA-256 SSE algorithm on Assembly Language. It saves at least 10% cycles. So I think you need only this .asm module.
newbie
Activity: 34
Merit: 0
OK, so... your framework is not compiling for me. It seems to be looking for POSIX-style headers on a windows machine. The whole thing is a massive mess of ifdefs, and I have no idea what the actual problem is. Thanks for your help, but unfortunately it looks like I'll try to use your code as a guide to re-implement it as part of jgarzik's miner instead of trying to build it myself.
sr. member
Activity: 404
Merit: 251
By the way, which libraries are you using?
Our own libs. All necessary Sources of libraries you can find at http://ufasoft.com/files/ufasoft_lisp-4.30.tar.lzma
It is other project, but uses the same framework.

Additionally: bitcoin-miner uses BigInteger to compare with target. Really it is not necessary, memcmp() function is enough
newbie
Activity: 34
Merit: 0
By the way, which libraries are you using? The and the <../libext/ext-http.h> are creating problems for me when trying to compile, and after some googling I still can't figure out where these libraries are coming from. I also don't have definitions for a large number of the classes referenced, so I'm guessing that there is some dependency that is missing here. I'm using Visual Studio 2010 as my Windows build environment.
sr. member
Activity: 411
Merit: 250

FYI, I downloaded the newest version (Windows .exe) from the link at the head of this topic. No timestamp appears. This is the result I get:

It should be version 0.3. Probably your browser cache gives you old file.

Actually, I had the file in two different locations without realizing it, and was launching the wrong file. The correct one I had put in my "Bitcoin Mining" folder, but when I first tried out the program and created the .bat file to easily launch it, I had put it in a generic "New Folder". Since the .bat was still pointing there and I hadn't remembered the folder existed, I was launching the old version of the miner.

Problem solved, and my "average" performance based on eyeballing the numbers for a few seconds has increased again, with less variance when doing other tasks, from 5.5MHash/s to 6.5MHash/s. I'm liking this miner more and more.
sr. member
Activity: 404
Merit: 251

FYI, I downloaded the newest version (Windows .exe) from the link at the head of this topic. No timestamp appears. This is the result I get:

It should be version 0.3. Probably your browser cache gives you old file.
sr. member
Activity: 411
Merit: 250
Feature Request: a timestamp for an accepted string. 
Implemented

FYI, I downloaded the newest version (Windows .exe) from the link at the head of this topic. No timestamp appears. This is the result I get:

bitcoin-miner 0.1  Copyright 2011 Ufasoft http://ufasoft.com/open/bitcoin
2 threads         Using SSE2
5.88 MHash/s     Result FOUND: EED1711LKJASNX>?<>?<#@*(yBLAH BLAH BLAH
5.7 MHash/s                Accepted
5.84 MHash/s [<--live update of hash speed]
sr. member
Activity: 404
Merit: 251
Feature Request: a timestamp for an accepted string. 
Implemented
sr. member
Activity: 404
Merit: 251

Is the MHash/s rate that displays on an Accepted the average for that acceptance, or the hash rate at the instant it was solved?

At the instant it was solved. "Rate" here is just last printed line, because after sending solution to server the NewLine character printed.
sr. member
Activity: 411
Merit: 250
Feature Request: a timestamp for an accepted string. It's nice to be able to look back and see what my average time was between solutions.

Is the MHash/s rate that displays on an Accepted the average for that acceptance, or the hash rate at the instant it was solved?
sr. member
Activity: 411
Merit: 250
Thanks for the advice here. I had the URL as http://mining.bitcoin.cz:8332, when it needed to be http://mining.bitcoin.cz:8332/. Getting .7-.75Mhash/s.
sr. member
Activity: 373
Merit: 250
If you're getting 0 Mhash, make sure you've got the full URL (including port number) and that the username and password are spelled correctly including correct capitalization for both.  Also with slush's pool make sure you're using two different workers for your two instances of the miner.

This is really interesting.  While with jgarzik's CPU miner I can get just about 1 Mhash/s per core, with this one I'm getting closer to 2.2 Mhash/s per core.  This is on a Core 2 Duo T9300 @2.5 GHz.  I have not compared to any other 4way miner though for comparison.
sr. member
Activity: 411
Merit: 250
When using your program on another computer, it comes back with 0Mhash/s. When using the generator in bitcoin, or another CPU miner, I get between 400-800khash/s, or .8Mhash/s. This computer has an Intel Core Duo (not Core2) T2050 @ 1.6Ghz.

I sent .5 BTC your way ufasoft, thank you very much for your program - it's helping me out on at least one other computer!
sr. member
Activity: 404
Merit: 251
Quote
I'd love to bro. I'll take a look at his control code and see where it differs from yours, and then see what I can do with integrating the two systems for maximum positive effect.

most important differences are in file bitcoin-sha256.h:
1. I precalculate first 3 SHA-256 rounds, because their results don't change on every iteration. and I skip last 3 rounds, because we need not them to calc last 32-bit word of hash [h word].

2. I use _mm_andnot_si128()  to calc expresssions (~a & b)

3. I don't unfold SHA-256 loop. I think CPU can predict loop jumps very good.

foo
sr. member
Activity: 409
Merit: 250
Seems to not work so well for AMD processors?

Results on an AMD Sempron 3100+ (1.8 GHz):

jgarzik's miner: about 770 khash/sec
ufasoft's miner: "average Rate: 0.62 MHash/s"
newbie
Activity: 34
Merit: 0
According to the license, the code is public domain, so perhaps somebody might want to integrate this into cpuminer itself ;-)

I'd love to bro. I'll take a look at his control code and see where it differs from yours, and then see what I can do with integrating the two systems for maximum positive effect.

EDIT: Someone else had better do this. I know very little about cryptography, so I'm not sure how to change it without screwing up the program flow. I'll try it, but it's by no means guaranteed to work.
sr. member
Activity: 406
Merit: 250
Yes, thank you! I went from ~1.8MHash/sec to ~4.1MHash/sec on one of my old spare computers.

It's running a Core2 Duo 2.2Ghz.
sr. member
Activity: 411
Merit: 250
ufasoft, thank you very much for this! I went from 1.5-1.9MHash/s, to 5.5-5.9MHash/s. I have a Core2Duo E7600 3.06GHz, and doubling the hash rate without any performance impact (so far) is incredible. Do you have an address I could send a BTC to as a thank you?
sr. member
Activity: 404
Merit: 251
If you could post the rest of the code,

More  sources was added to archive:
http://ufasoft.com/files/open/bitcoin-miner.zip

It still don't include some framework classes, like String, DateTime.... they have obvious implementation, but uploading them will require me to check many dependencies between my sources.
legendary
Activity: 1596
Merit: 1091
According to the license, the code is public domain, so perhaps somebody might want to integrate this into cpuminer itself ;-)
Pages:
Jump to: