Ufasoft Miner - Windows/Linux, x86/x64, SSE2/OpenCL, Open Source

[muqali]

nice, looks like its working now! 0.4 MHash/s
but works...

thnks

lol what CPU?

edit - I ran it on an Atom N2600(the NEW ones) and got about 0.6KH/s
I knew they were weak CPUs, but that weak. wow.

[1714030495]

1714030495

[420]

Any parameters to improve gpu hash rate? Radeon 5970

Testing all the miners

[420]

I want to download the old verison 0.28 for windows 64-bit but cannot find it

[ufasoft]

I want to download the old verison 0.28 for windows 64-bit but cannot find it

Something wrong with last version?

[muqali]

I'm getting a 404 trying to download the source code using the URL link on the first post.

edit - though it can be downloaded from the webpage.

[ufasoft]

I'm getting a 404 trying to download the source code using the URL link on the first post.

fixed

[muqali]

Are there any plans to add FMA(3 or 4) support to the miner? AMD has CPUs in the market now, I just bought a system with an A10 APU in it this morning. I'm not sure if the fused multiply add would help or not. I glanced at the x86-64 asm file, but it was mostly greek to me. Looked more like adds, moves and bit shifting as opposed to MAD operations. I could be wrong though.

[ufasoft]

Are there any plans to add FMA(3 or 4) support to the miner? AMD has CPUs in the market now, I just bought a

FMA useless for SHA
Future Intel AVX2 should double the performance.

[muqali]

Are there any plans to add FMA(3 or 4) support to the miner? AMD has CPUs in the market now, I just bought a

FMA useless for SHA
Future Intel AVX2 should double the performance.

Yea, but that won't be until Haswell. I'll be waiting for whatever Haswell chipset replaces the X79 to upgrade. Unless there is a really awesome Ivy Bridge-E CPU released, but given their poor overclocking performance I don't see that happening.

Do you know if FMA would help with scrypt at all?

[ufasoft]

Do you know if FMA would help with scrypt at all?

No , In the scrypt RAM-latency is main factor

[420]

I want to download the old verison 0.28 for windows 64-bit but cannot find it

Someone claimed double the hashrate of a cpu by using that version on mining hardware comparison

[ufasoft]

Someone claimed double the hashrate of a cpu by using that version on mining hardware comparison

No changes in CPU algorithms since 0.28

[420]

Someone claimed double the hashrate of a cpu by using that version on mining hardware comparison

No changes in CPU algorithms since 0.28

so they are full of shit or i'd have to be under linux or tweak some thing which i doubt would get me 2x the hashing

since its already utilizing 100% of my 12 threads

Somethings strange. 12 threads: 33mhash, 6 threads: 26mhash

[-ck]

since its already utilizing 100% of my 12 threads

Somethings strange. 12 threads: 33mhash, 6 threads: 26mhash

Learn the difference between cores and threads on your CPU.

[420]

since its already utilizing 100% of my 12 threads

Somethings strange. 12 threads: 33mhash, 6 threads: 26mhash

Learn the difference between cores and threads on your CPU.

grrrrr

[ufasoft]

Somethings strange. 12 threads: 33mhash, 6 threads: 26mhash

Nothing strange here. Read about Hyper-Threading technology.
Intel claims up to a 30% performance improvement

[420]

Somethings strange. 12 threads: 33mhash, 6 threads: 26mhash

Nothing strange here. Read about Hyper-Threading technology.
Intel claims up to a 30% performance improvement

i guess right, about 30% improvement

i must have thought the back of my head that new hyperthreading was better than old P4 HT

[ufasoft]

The pooler-cpuminer indeed has better algorithm for cpus with AVX (intel i7).  Probably 50% improvement.

[420]

The pooler-cpuminer indeed has better algorithm for cpus with AVX (intel i7).  Probably 50% improvement.

did try that and I got 42mhash

[d3m0n1q_733rz]

Hey, I'm having difficulty getting this to compile under Linux.
When using INSTALL:

Code:

./INSTALL: line 1: Installation: command not found
./INSTALL: line 2: acinclude.m4: command not found
./INSTALL: line 4: If: command not found
./INSTALL: line 5: 1.: command not found
./INSTALL: line 6: 2.: command not found
./INSTALL: line 7: 3.: command not found
configure: error: cannot find install-sh, install.sh, or shtool in "." "./.." "./../.."
./INSTALL: line 11: If: command not found
configure: error: cannot find install-sh, install.sh, or shtool in "." "./.." "./../.."
./INSTALL: line 15: Then: command not found
make: *** No targets specified and no makefile found.  Stop.
./INSTALL: line 17: or: command not found
make: *** No targets specified and no makefile found.  Stop.
./INSTALL: line 21: Optionally: command not found
make: *** No rule to make target `install'.  Stop.
./INSTALL: line 25: unexpected EOF while looking for matching `''
./INSTALL: line 29: syntax error: unexpected end of file

When trying to configure:

Code:

configure: error: cannot find install-sh, install.sh, or shtool in "." "./.." "./../.."

I've installed shtool, so something else seems to be problematic.
Perhaps including a list of required dependencies in the README file would be helpful?  This way, everyone can compile without having to download every little library they can think of to find the right one.  Also, the Linux binary was asking for the ATI library files when I was doing CPU mining for Litecoin.  Considering I didn't even have an ATI card in that computer, it appears to be a slight problem.  It's even more of a problem because I put the library files in the same folder as the binary and it didn't even recognize they were there.  And so, there's a few issues that definitely need to be worked out with the compilation process.