You dont know "far enough" then, because it doesnt. There is an equal chance of your shares ending up in a shift that doesnt get paid (because no block is found within the next 10 shifts) as ending up in a shift that gets paid multiple times (because several blocks are found within the next 10 shifts).

Perhaps looking at these graphs helps:
https://bitminter.com/stats/rewards

The top chart show shifts and their payouts. If you mine intermittently, you shares could end up in any of those shifts randomly, some are paid zero PPS, some are paid 1x PPS (yellow line), some are paid multiple times. On average, assuming average pool luck, you will end up on the yellow line and obtain 100% PPS payout.

Of course you will have a higher variability than with PPS, and probably a higher variability than a non stop miner, but on average you will make exactly the same per share as a 24/7 miner and exactly the same as with any other hop proof payout scheme.

 

In a proportional pool, hoppers will get, say 120% PPS and non hoppers 80% PPS (assuming equal average hash rate between hoppers and non hoppers)
In a hop-proof pool, everyone will get  100% PPS, there is no difference between hoppers and non hoppers.
Duh.

More like someone calibrating a price fixing bot. Or Mt Gox messing up:

No one has ever tried, because it doesnt make a lot of sense. It might work, it might fail miserably.
One concern is that adding 2 identical PCI cards of anything often doesnt work. I once made a crazy "budget SSD" based on a RAID array with >30 USB drives (company gifts) and since I didnt have enough hubs or bandwidth,  I tried putting in 3 identical PCI to USB controllers I had laying around, and it just wouldnt work. Only 1 PCI card would work at a time, no matter what I tried, the other 2 werent seen.
This was both on windows and linux.

OT, but here is part of the setup:

Above results were obtained on linux. FWIW, at 950/300 I got 433. 950/160 I got 439. SDK 2.1. 2.4

In the last 24 hours, price dropped 7%. If we extrapolate that for the next year, that means a BTC value of $0,000000000020. All the bitcoins in the world wouldnt buy you a can of coke.

Yep, thats the one I had in mind, thanks. Except his graph stops at 300,  while I tested everything under 300, so, phew, I didnt waste my time

No, how did he make that graph? He must have a more efficient way than me.

A long time ago I saw a super nice graph plotting hashrate vs memory speed. I cant find it anymore, so I made my own. I made a little script to run cgminer with various core and memory settings and let it run until it had found 100 shares.

This is the result so far (ignore the right Y axis, left one is correct):








On hindsight 100 shares wasnt enough, but a few things are already clear;
- Differences are relatively mild. I havent been able to find ckolivas' 20 MH boost.
- its the ratio between gpu and memory clock that matters
- 300 and 150 are both lousy for a ~900-1000Mh 5870

Tested with:
- Linuxcoin (SDK 2.4)
- cgminer 2.2.1

There is plenty of opportunity for alternative clients.
Check out Armory:
https://bitcointalk.org/index.php?topic=56424.0

If any of them one day proves to be better, people will switch.

IMO Armory looks very promising, but at this point is still lightyears away from being a proven stable client like what Gavin made. Still, feel free to donate to the developer

It doesnt work out to a nice round number. But I just tried.
5870 @ 950/300 : 432.4
5870 @ 950/180 : ~441-442
5870 @ 950/170 : ~442

Not a quantum leap, but still quite a nice boost!

So I tried another 5870 that runs at 1 GHz applying the same memory clock and same ratio and a few other speeds:

5870 @1000/300: 454.7
5870 @1000/200: ~459
5870 @1000/190: ~460
5870 @1000/180: ~460
5870 @1000/150: ~454

Smaller boost than at 950. Not too sure what to make of it, but there is nothing wrong with getting a few extra MH while consuming less power

Note these results were obtained testing just a few minutes on each setting, dont take it as gospel

So I guess CH is really telling us its time to quickly buy some litecoins, because difficulty is about to explode .

If that came from anyone else I would have laughed at it. But since its you, Ill have to believe it, but is it not more likely the reason is not so much the clockspeed as such, as the ratio between GPU and memory clock? IOW, rather than reducing your GPU clock, perhaps try increasing the memory clock a little bit?

Not exactly per share. Each time a block is found, you get paid for your hashrate over the last 10 shifts, a shift taking as long as 1/10th an average block. Its comparable to proportional payout, except it cant be hopped profitably, so you dont fork over 20% to hoppers.

However, your revenue is dependent on pool luck. You can have days with no payout at all if no blocks are found, and days where you make 4x as much as you would on a PPS pool.

In the end you should get ~1.5% more than on ABC pool, since there are no fees, but you will have to average that over weeks.

Yes, you upgraded your video drivers. Thats the root cause. The reason your old copy of cgminer still works faster, is because it contains a cached/compiled copy of the .BIN file generated with your old drivers and sdk. If you were to reinstall 2.0.8 you will likely get the same lousy hashrate.

See here:
https://bitcointalk.org/index.php?topic=63383.0

Coin received, gracias.
BTW, its not my pool, I just mine there and Im a fanboy .

Hmm. you cant set 2 different worksizes with cgminer then? Didnt know that, sounds like worthy of a bug report.

As for the bounty; tbh, I think next time you would do better not offering one instantly in the title to avoid lots of bounty hunters giving sometimes questionable advice hoping to get it right. You can always offer one later should you not get adequate help, or just offer it to whomever helped you without promising it up front. Im not saying that for this thread in specific, just something Ive observed a few times when bounties are involved..

As for my share of it; since i believe you have a large farm, consider pointing some of it to bitminter rather than sending me coins. We could use some extra hashing power, and with PPLNS, zero fees, you dont have much to lose (unless you are hopping or renting your rigs).

If you cant or dont want to, then I wont refuse a coin sent here:
1P99tan9oX16aFgSaLWR8N8pZR4pndhxvx

Looks I was wrong.
Try this, no space:
"worksize" : "128,256",

Or paste your config file.

You are missing 2 quotes there:

"worksize" : "128", "256",