Come to the #litecoinpool channel on Freenode IRC if you want to take part in the testing. I see from the logs that you joined the channel earlier today, but you only stayed connected for a couple of minutes. The channel is small and people are idling most of the time, so you have to stay much longer than that if you want to get a response.

I believe that what you wrote is misleading. This pool has always worked fine with miners of any hash rate. In fact, KnC's allegation that some pools are not "up to speed" with newer, faster, mining hardware is a downright lie spread to try to justify the numerous issues that Titans have. There's a simple reason for this: if a pool uses vardiff, a fast miner is no different from a slow miner: because of the difficulty adjustment, both will submit about the same amount of shares in a given time period. I am perfectly aware that Titans seem to behave differently on different pools; what I am saying is that this is KnC's fault, and not the pools' fault.

What is really disconcerting is that two different Titans with the same configuration often behave differently on the same pool. There are Titans on this pool hashing away happily at over 340 MH/s, while some others, like yours, apparently work better at other pools.

Weeks ago I wrote an email to KnC to ask them for more information about these miners, and I did get an initial reply from one of their devs, but sadly when I started posing more technical questions I got no more replies. During that same period, I worked with a Titan owner to try to maximize the compatibility of the pool with these machines, and this collaboration brought very good results.
However, it looks like the newer Titan firmwares (1.10 and later) have different requirements to work optimally. A few users brought this to my attention in the IRC channel yesterday, and we are now testing some new server configuration to try, yet again, to maximize compatibility. This is not easy because, as I've mentioned above, Titans do not seem to be created equal.

Sounds like a typical case of hardware errors.
Please send an email to support with a brief description of your setup and a screenshot of your mining software (cgminer/bfgminer/...) after it's been running for at least 10 minutes.

Happy 3rd Birthday, LitecoinPool.org!

https://www.litecoinpool.org/news?id=61

There is no fee.
The nominal PPS fee was superseded by the PPS ratio when merged mining was introduced. You can easily convert between the two, but it makes little sense to talk about a fee when the ratio is greater than 100%, as the corresponding fee is negative. For instance, a PPS ratio of 110% corresponds to a -10% fee.

By the way, I've adjusted the PPS ratio down to 110% because of the recent correction in the ULTC market. Yes, we are still paying more than all other pools I've checked, including profit-switching pools.

Just met neilt in the IRC channel, and it turned out the difference was indeed due to hardware errors.

A couple thoughts:

• The A (accepted) and R (rejected) counters in cgminer are expressed in difficulty-1 shares, while the HW count is in hashes, so they are not directly comparable, and not comparable at all if the share difficulty is not fixed.
• If the work utility displayed by cgminer matches the hash rate estimate displayed on the pool, then the problem is on the client side. Since WU is expressed in difficulty-1 shares per minute, you have to multiply by 65536/60 to get the corresponding speed in hashes per second.

(In the above, when I say "difficulty 1" I'm of course referring to the Stratum/scrypt definition of difficulty, not to the traditional one.)

First, make sure you have read this FAQ entry.
What kind of hardware are you using? Does cgminer report hardware errors (HW)? How long has it been running? A screenshot of cgminer running may be useful (you can send it to support via email if you wish).

What BFGminer displays conforms to the correct definition of difficulty (which is the one that has always been used for the network hash rate). A difficulty of 1 corresponds to what scrypt miners usually call 65536.
Since share difficulties can be much smaller than 1, BFGminer uses SI prefixes to display them. So, m=10^-3 and u=µ=10^-6.

Le me make an example.
976u = 976 x 10^-6 = 0.000976
That's a difficulty by the good old definition. To get the share difficulty that scrypt miners are used to, you multiply that by 65536, and get 63.96. Keeping in mind that the difficulty displayed by BFGminer is truncated, that probably means 64.

Another example:
15m = 15 x 10^-3 = 0.015
Multiplying by 65536 gives 983.04. Due to the aforementioned truncation, that is of course just an approximation; since powers of two are normally used, the actual difficulty when you see 15m is probably 1024.

If you have issues setting up your miner, I suggest that you write an email to support including your exact configuration.

No conversion is needed, because this is a purely PPS pool. The PPS ratio is is computed taking into account the expected rewards from merged mining. This is effectively the same as having a negative PPS "fee".

The profitability of Dogecoin is currently floating around 25% that of Litecoin. This is why the current PPS ratio is 125%, which means that you get 25% more than with a 0-fee PPS pool that doesn't do merged mining.

I hope I'll be able to, but to be honest I doubt the ratio will ever reach 125% again. The difficulty of the Dogecoin network keeps rising as more and more pools jump on the merged mining train, and you can see for yourself that it touched 10,000 just a few minutes ago.
I want miners to know that I'll do my best to keep the PPS ratio as fair as possible. Since merged mining started, I've been monitoring the relative profitability of Dogecoin with a script I wrote, and I may soon implement something similar within the pool code so that the ratio is updated automatically.

EDIT (Sep 15): Looks like I didn't take into account the possibility of a jump in the price of Dogecoin. Now at 130% PPS!

Ah, OK, I see what you mean now. Basically you're proposing proportional payouts for Dogecoin. That could work, I suppose, even if it would be subject to the issues of proportional systems. Another way could be to have a separate PPS reward system for Dogecoin.
Either way, things would get substantially more complicated for us, because we would have to implement and maintain two separate accounting systems. Also consider that we have a number of security devices in place which would have to be carefully modified.
Not to mention that the current solution of paying everything in litecoins makes things much easier and nicer for all those miners who are note interested in Dogecoin.
All in all, I'd prefer to keep things as simple and manageable as possible, so I think we're going to stick to this system for the time being. (This may of course be reconsidered in the future.)

It is not killing it. On the contrary, it's bringing more security to its network. The fact that the mining profitability of Dogecoin alone is much lower compared to when Doge was not merged-mined is perfectly normal and expected, and causes no harm whatsoever.

As I've said before, the difficulty of the Dogecoin network is rapidly increasing, as expected.

I'm not sure I understand what you mean there by "as same percent of LTC", sorry.

Let me point out that Litecoin and Dogecoin still retain independent network difficulties and market prices, and therefore independent mining profitability. In particular, after the introduction of merged mining the difficulty of the Dogecoin network has started rising rapidly, and this has caused the profitability of Dogecoin to sink from about 100% of Litecoin to the ~18% where it is at the time of this writing. I expect this figure to continue to decline as more pools start doing merged mining.

Unfortunately that would complicate our accounting considerably, so for the moment we are only going to offer Litecoin payouts.

Merged mining has started and is proceeding smoothly. We have already found our first Dogecoin block!
The PPS ratio is currently 125%, meaning that miners earn 25% more than with a 0-fee PPS system. Enjoy the extra rewards!

The software is already in place. We should be ready to go shortly after the Dogecoin network starts accepting auxiliary proofs of work, i.e. in a few hours.
Payouts will still be made in litecoins only (this is litecoinpool.org after all ), but the PPS ratio will be adjusted to take into account the expected rewards coming from merged mining. This will basically be the same as having a negative PPS "fee". Of course, after merged mining starts the PPS ratio may need to be adjusted frequently based on factors such as the DOGE/LTC exchange price and the difficulty of the Dogecoin network, which is expected to increase rapidly.
I am going to update the website to document this change, and will definitely post a news item when merged mining goes live.

If you already have a working C implementation of the algorithm (and it looks like you do) then it should be straightforward to fork cpuminer and add the algorithm to it, as has been done dozens of time already for other algorithms.
If what you are looking for is an optimized implementation of the algorithm, that's a different kettle of fish. Unfortunately I'm extremely busy with other projects, and have no time to work on this, not even as a paid job.

cpuminer has no API. You must be confusing it with cgminer or with some fork of it.

Working on it