I think so.  The reason why I use BAMT is because have you seen the # of steps to get everything working right on Linux?  Brutal.  It can be done but BAMT is "flash a drive", plug it in, and up and mining after editing a couple of text files via SSH.  If I ever get serious about remote config it will be even easier/faster.

If BAMT became a collection of programs then it would have dependency issues.  So it becomes grab a linux distro, but not this x,y, or z, and not b after version 1.23 or c before version 3.27.  Some people say d works but some people say it doesn't.  Then make sure you have pkg blah blah installed, it has to be version after 1.28 but watch out you can't use that with kernel 2.9 or later.  Using version 1.27 or earlier causes a critical bug but only on some distros.  Then install SDK & drivers, (insert some cryptic note about incompatibilities).   Install BAMT pkg, check the readme for a list of known incompatibilities, and if it isn't there you are on your own.  Make sure to export display, make sure xhost is running, track down bizare error messages on google.  The install BAMT package and pray nothing else goes wrong. 


"BAMT ... up in running in 3-5 days after teaching yourself Linux troubleshooting .... hopefully."  

I think a lot of the "magic" will be gone.  Hopefully we can get some Linux image guru to take that aspect off your shoulders.  I know enough to know it isn't me.

I have no initial problem with using any given linux version if it is proven reliable and does not negatively effect performance of mining.   As BAMT is a very special purpose Linux with only one real function, all choices are made based on how they effect that single role.

BAMT is almost exclusively used on USB keys, although it can be used from HDD.  Using a HDD is wasteful of power, using SSD for a mining rig would just be silly.  So.. support for SATA is not important.  I am not aware of anyone reporting their network hardware isn't supported by the current kernel, nor am I aware of any issues with motherboard support that have been attributed to the kernel, but who knows.

For the mining software, it seems to be general consensus that ATI driver 11.6 and SDK 2.4 is best fit/performance for a majority of GPUs.  Mining performance with 12.x drivers suffers, as it does with newer SDK.  Some cards like SDK 2.1, but its only a few from my impression.

If this is incorrect, someone please point to reliable evidence of the contrary.  I am happy to put whatever versions of driver/sdk/etc, I just try to guess what will be best for the most.  Of course you can install any alternate driver or SDK you prefer at any time anyway.

You seem to have forgotten that Linuxcoin comes with SDK 2.4 NOT 2.1 !

Cheers !

I think it works like that:

1) You promise a great product of great performance at a good price to attract buyers.
2) You let the pre-order list build on, that way, you receive money to build your product. It also give you the option of starting without any money.
3) With the money, you build the product you advertise, using what competitors already offer. From what I see, it's probably this board (https://bitcointalk.org/index.php?topic=49180) four times in a nice box.
4) The cost of building the product is probably a lot higher than his sale price. In that case, it cost around 1600$ and sells at 599$.
5) When you build your first batch, you ship them so the first buyers can give feedback.
6) Shipping MUST be slow. It's important later on.
7) The first buyers are indirectly a part of the scam. They are honest people who are going to say to others :"Hey, this thing WORKS! It's awesome, blablabla". The feedback is almost garanteed to be positive.
In the meantime, you can adjust the price if you want. It let you get more money without problems because of the feedback.
9) People who didn't bought into the first batch want to be a part of it now. The price is good, the performance is great, and looking at the feedback of the first buyers, everything is as promised. So, they pre-order too on the next batch.
10) First buyers, getting their promise fulfilled (this box is awesome!) will probably buy another one too.
11) The shipping took a long time the first time, so, on the second time, people will accept the long delay between payment and the reception of their product. People waited two months the first time, they can wait another two months.
12) Wait for the pre-orders, take the money and run. The longer the "shipping" takes, the more time you have to GTFO.

Alternate version:
- You can redo the scam multiple times. You simply need to redo the steps 7 to 11. The second batch buyers will also give great feedback, creating a new wave of batch buyers.
- The scammer will be able to create multiple waves of buyers and increasing his revenue. It's more risky though, because he could find it hard to stop, and it's better if he can control when the scam stop (instead of being discovered). When he decides when to stop, he can use his "shipping" delay more effectively to get out and hide.

Well, I'm interested to see how this will end. Have fun!

I've not had luck thus far on an xubuntu 11.04 install getting better than 430mh/s on a 5870 at 950. Downclocking below 300 on SDK 2.4 lowers hash rate, on SDK 2.1 it remains the same. I may mess around trying to get 2.1 reinstalled and give it another shot with the magic numbers mentioned in this thread. It may just be a linux thing, seems people on windows are getting much better mh on 5870s at 950.

Missed that. But I doubt AMD would lift this limit in the foreseeable future, people using more than 2-4 GPUs are already a niche.

So always use intensity 9 to avoid this bug then ?

Thanks !

So who / what is holding up the move to 64 bits ?

I would love 32 bits but I don't wanna hold up that awesomeness that is 0.5 !

gmaxwell: bulanula there is little evidence that anything is hacked.
majorminer: question
gmaxwell: s/little/none/ really.
gmaxwell: Someone could have made that user list by iterating https://btc-e.com/profile/1
majorminer: if btc goes to ..lets say 2$, which intern will drive the price of ltc up?
bulanula: ok
m3ta: there is only the evidence that it's easy to just scrape the fuck out of /profile/ and get a list of users. boo hoo. big deal.
bulanula: but still dodgy
gmaxwell: Whats still dodgy?
majorminer: anyone could clearify this for me?
bulanula: list of users no pass
gmaxwell: bulanula: anyone could generate that list trivally
m3ta: dude, even Facebook can be scraped to get usernames. so what?

If this is true then I am in heaven

Sadly, I still think we have to put up with AMD's BS drivers and bugs

Yeah, come to think of it I still think they need to implement something like bitalign and BFI_INT to be able to compete with 5870s and 7970s etc.

I think in this game the more shaders you have the more MHash/s you get.



Nvidia fail again.