Topic: OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.0 - page 810. (Read 5805728 times)

restart GPU works even if the GPU is hung due to things like overheat-lockup? 

I simply have a monitoring script that checks for serious lockups and coldreboots the rig if it is triggered, not pretty, but all GPU are at least up and mining again within a minute.
Will probably add that corespeed controlled by temperature levels script(it down clocks the core if temperature gets too high) I saw in some thread on the forum a few days ago.

Summertime is a bitch when it comes to mining   

Not for me. It won't restart either gpu on my 5970.

+1
But this is not that easy. You probably need to add a way to pull in JSON stats from pools and let the miner make decisions based on that information.

In my testing, under Fedora14, SDK2.4, restart GPU works in 100% of my 5xxx series cards, and 0% of my 6xxx series cards.

I was looking through the code, but I'm a little tired right now and can't figure out if this would be an easy change or not...

I want to set it so that it will do a pool rotation, but start it over at the beginning when a new block is detected via LP.

Just ran into a serious bug with cgminer tonight. I noticed when you only mine at one pool, and it goes down temporarily the miner will open over 4,000 connections (with a TCP timeout of 120 set on my router!) to attempt to reconnect.

Having multiple machines doing this just brought my router to its knees and it couldn't do shit until I unplugged all my rigs and waited for cgminer to close itself, which is another bug that occurs if internet access is temporarily lost.

This happened in Windows 7 SP1 32-bit with cgminer 1.5.1.

Me too. LinuxCoin beta final

A C-50 gives ~6,5 MHash/s for the GPU (Codename Loveland ^^).

Dia

I can't personally say that I'm having this error, what are your settings?

I'm having a problem with cgminer. It runs perfectly for 3-4 (random) hours and then it finishes itself without showing any error. It just shows the same report as if I would stopped it vountaryly pressing the quit key.

Anybody has had the same problem ?

Restart dead GPUs does not work for me too. Win7 64bit, Catalyst 11.7

e350 gives ~11mh
cpu is less than 1mh
also with 11.7cata, cpu usage is seriously high on that machine
EDIT: with 11.6 cata cpu usage dropped drastically

best i got on the e350 is poclbcm with latest (your last version that works with it) phatk kernel - 11.5mh

ps. not overclocked.

The -T option might also have worked. I prefer that for debugging as ncurses doesn't interfere that way.

I got it, you have to use . Guess now I can debug that C-50 .
Anyone tried CGMINER on an E-Series APU?

Dia

That idea is not working, because the screen clears the error after a few seconds or sooner and the app crashes, so there is no window, where I could scroll. Even a cgminer.exe blabla > Test.txt does not work. Is there a built-in function to write the output to a file?

Dia

Actually, when you open up your command prompt in windows, right click on the top bar and go to properties.  You can increase your buffer size from the default 50 lines to something larger.  Assuming the whole window doesn't crash, you can just scroll back through all the lines.

How can I log the CGMINER output to a file on Windows? I have a Netbook with AMD C-50 APU, which should work with phatk, but only compiles for poclbm kernel.
CGMINER just crashes, which is bad and I'm not able to read the output fast enough ^^.

Thanks,
Dia

While I do agree with you to some degree about Ufasoft being faster, with a little tweaking and changes of the C-code to take advantage of more optimizations, some of the bottlenecks in work passing and hashing could be removed from cgminer to reach beyond Ufasoft's hashing rate.  On my own PC, my Core2 Quad will reach up to 16.1 with Ufasoft and then drop to the upper 14's.  In cgminer, each core will reach up to 4.7 Mhash/s but will slowly level out to 3.9 due to the rate at which data is passed to the hashing algorithm.  With that said, Ufasoft's hashing likes to fluctuate drastically while mining whereas cgminer's will flux initially with rather high numbers and then level out as the bottlenecks in some of the code is reached.  This is because of its large dependence upon only SSE2 instructions which are not implemented so well on many (or most) processors.  To demonstrate this, the C code utilizes xmmintrin.h which only allows for up to SSE2 instructions to be used regardless of the CPUs capabilities and is written only to use them.  I don't know what Ufasoft uses, but cgminer can be much more than it is if rewritten to be.  For example, once the C-code is compiled, over half of the computations are on double-words instead of putting them all into double-quads before-hand and computing up to 4 double's at the same time.  Though it is the eventual result, the result comes too little too late and slows the entire process down.

I will say this for cgminer though, its efficiency is off the charts compared to other miners.  But without efficiently implementing the efficiency, it comes at too high of a cost.

I will edit this post with the disassembled code I'm referencing once I return home to my own PC in 2 or 3 days.

my 8 btc number is incorrect...been visiting silk road to much ))

The price decline really sucks for silk road purchases  

Such a beautiful place...you all should visit!

1. Advertise payment for running software with the stipulation that the owner must have a Radeon 5xxx or 6xxx video card.

2. Once the user has been set up, make use of CPU processing in addition to GPU.

3. Pay users 30-50% of what you gain per BTC generated.

4. User-authorized Botnet.

There are more details, but that's the gist of it.