Are you going to implement a 24 hour rewards stat? That was always my favorite..

Very nice pool, always been my favorite. Keep up the good work!

The bios pages are sometimes the most useful. On most boards you can disable 'quick boot' or spashscreens and see what bios detects. My one msi board will even tell you which slot and what bus speed it gives each device. So in there I can see if it sees all 4 cards, and what slot they are in. If bios only sees 3 cards, no OS will.

[322.339 MH/s (~291 MH/s)] [Rej: 73/7908 (0%)]

Is the display output from one of my poclbm miners, and the (~ MH) is one of the newer, and my favorite stat that I basically look at as local luck. As I understand it, the first number is how many hashes hashed from the miners point of view. The estimate is a timed based average depending on how many shares have been accepted (maybe + rejected? Not enough to matter it seems). So sometimes its a bit higher than the actual hash rate, and sometimes lower, like now.
These 'lucks' get added up in a pools mining stats, and basically multiplies by how many you have running. If all your miners on a lucky streak, your reported mash rate in the pool stats will be more than your actual.

Out of pure curiosity, has anyone tried some old sun sparc procs and mining ? Even though Sun is dead (god rest her soul, don't bother to try to tel me otherwise I know there's lots of boxes still hiding in datacenters and basements.. so has anyone thrown a miner on one just to see the mash rate? The whole sparc arch is a risc processor, so it does work differently than x86 chips.. just wonder how they stacked up to today's intel procs.

I noticed the same thing, until I did the powertune +20% thing in AMDOverDriveControl. After setting that up, it works as expected. Not sure whats happening there.. but that fixed it.

No, just with the x16 lanes. It's meant to move the speeds around depending on how many cards you have plugged in, as I can tell. It's one of the MSI 890 (? might be wrong on the number..) with 5 x16 slots, and can handle 4x16 cards. It is also spaced to allow 3 cards plugged in with a space inbetween the three of them. If memory serves, with three cards in they are all actually working at x8, and it disables the one x16 slot that isn't covered with a gfx card. With 4 cards plugged in they all run at x4, and it turns the last one off but its covered by a dual slot gfx card anyway. In the manual it has a fancy table with which slots get which speeds depending on whats plugged in. I wish you could force it in bios, then I'd just force all the slots to x1 =]
Not sure why, but with any card plugged into any of the x16 slots, I could not get it to see a card with a x16->x1 extender in the single x1 slot that the mobo does have. With no other cards installed the one on the x1 extender would work. Wierd. Got her stable with three cards in it, I need to get some more extenders to experiment more.

Prolly by easier to grab your run commands and start with a new copy of linux coin... write to USB, boot and start mining.

Maybe a BIOS update for the xtreme 5850? Maybe the card doesn't know how to handle being in a x1 slot? Just an idea..
Have you tried it with the extender on a diff mobo ?

simple things first, are you pointing your miners at different cards? ie, when you run just polcbm it will tell you that you have to give it a device, usually 0 and 1. You then have to start each miner with a -d0 and a -d1 to use the second card. I like to open a separate terminal for each miner so I can watch them, and usually have 1 or 2 more open to watch temps, etc.
Also, if it's a dedicated miner (not a part time gamer) take out the xfire bridge, it doesn't help ( maybe even hinders.. ) mining.

edit:
now that I've thought about it and come back, I'm thinkin it is the xfire setup.. I'd start there by taking it out.

try a ' lspci | grep VGA ' you should have a VGA entry for each card that the machine sees. Or just ' lspci ' to see all pci devices.
If you only have one here, then the system isn't seeing it... check BIOS settings, try diff slots on the mobo if your using extenders. I've run into a board that will turn off x16 slots if it sees cards in certain slots, you'll have to go thu your mobo manual to be sure there.
If it does see both cards, then it's something in aticonfig .. one command I run all the time is ' aticonfig --odgt --adapter=all ' which should give you the temps of all the cards. If you have two in lspci, and only one using --adapter=all, then I'd double check the drivers and SDK.
I'd also try each card one at a time, to make sure you don't have a dead card..

They work great, if your motherboard doesn't do screwy things like mine. Well, maybe not screwy, just anoying. When it sees cards in certain slots, it turns off others. So I had 3 cards plugged into the x16 slots on the board, and thought, "hey! a extender will get another card in there!" But after some research, i found that the mobo disables that slot when it sees cards in the three i was using. =[
I havn't tried it with all extenders, so all the slots would be detected at x1.. maybe that would leave them all on. Not sure, threw the card in another box.

TL;DR: yeah, they work great.

I disagree, one thing I also check for is resale value. A 5770 is a good miner per cost, but you won't make anything back selling it. A 6970 will have awsome resale value. I sincerely  hope that bitcoins really take off and I'll never sell my mining gear, but if I have to my 6950's will still sell at decent prices.

Do you get better temps with the plastic off the Gfx card? I've been wondering if that will help, or hinder since I would think the designers meant for it to channel the ait thu the heatsinks..

As long as your controller has the bandwidth..

Not bad .. but if you use all reference cards, they intake on that one fan. Since I only have anoying cards that exhaust out the front and back, I havn't been able to test a setup with those cards. I would think you'd get heat moving better just moving air front to back, ie bunch of fans blowing in the 'front' of the case (where the power connectors are on the cards) out the back. Would also help if you stacked these on top of each other, no need to draw air in from the top of each one. You could even put a shroud over the entire front and feed an air conditioner into it or something to that effect..
Could also save a little space by stacking the PSUs verticaly also.

the 11.6 drivers will let you use aticonfig to set clocks outside the normal config range.

I've been tracking down why my one box crashes, and it seems to be that if the miners go idle (not looking for the 'mine at two pools' suggestion here..) they must automatically down clock (and up clock the mem) themselves to the lower level settings, and then back up to the mining clocks once the work comes in. I'm thinking that the clocks bouncing around pisses it off and then locks her up. AMDOverdrive control goes part of the way, but with the 11.6 drivers can't (or doesn't?) clock it outside of the stock clocks, which aticonfig can.
So after some googling and reading through aticonfig docs, I can't see how to set the other sets of clocks. I would like to set all the clocks to the same, over/under clocked timings so that it doesn't switch around ..

This is mostly true .. in newer versions of Xen there is options for PCI passthrough, where it maps the physical device to a specific VM. So it is possible, but not very easy, and I wouldn't count on it not taking the whole box down if something went wrong.

Try different miners, and old versions too. My gts450 runs best on old versions of poclbm, the newer the worse hashing. All  the work ever done to mining code is to optimize for AMD cards, so sometimes it makes it worse for nvidia cards. Most don't care, except for the few squeezing out a couple MH on the gaming card.

Hmm.. i never tried to compare, i run w128 cause it gives me more hashes .. and stale shares have never been high enough to warrant investigation for me.