Bad profit

[Chris Sokolowski]

To clarify, the work restart time is calculated not by the average but by the minimum time your miner has taken between a work restart being sent and the first share being received, within the last two hours. The more work restarts your miner has, the more accurate the work restart delay will be, since you will have a higher probability of getting lucky and submitting a share immediately after receiving work.

If work restart times are too high across the pool I may set the monitoring period to be longer so that we have a higher probability of getting a quick share. The two hour period is only present to conserve memory and CPU time on the mining server.

I'm investigating stale shares right now. I am testing to be sure that we are sending work restarts when appropriate and not sending them when they are not needed.

[GenTarkin]

Ah, thanks for explanation. Why use minimum time instead of an average?

Also, regards to the stales. Did you change something already? Im sitting here watching bfgminer and pool and now even when new blocks are storming all over the place from different coins ... my reject rate(stales) is staying at 2% or lower. Which is a significant improvement.

Im monitoring the actual coin switching(diff comparison) vs the live worker status on my account page right now too. =)

[GenTarkin]

Well, on second thought. I guess minimum detected work restart delay makes sense. Long as the work restart issuance then subsequent share detection is able to accurately detect if the next detected share is for that particular work restart. If so, that would signify the *earliest* a miner could get work back to the pool.
I think with the Titan, if the bfgminer *verbose* output (which shows work flush then work update for each die) is accurate, timewise. It seems the flushes all happen under a second for all dies. Then the work update happens to take about 5 seconds in total. But, it finishes the work update per die in under a second but seems to do them sequentially, so of course it adds up to 5 seconds in total by the last die is working on new work.
I wonder if there is a way to issue the work update in parallel for these dies... no idea. LOL!

This would explain why a share *could* be done on Titan in under a second for a particular work restart. Because like one of the first 2 dies would have had their work update completed by then and luck would have it that they just happened to find a share in that time frame.

Thats all my guestimate anyways LOL!

[Chris Sokolowski]

Yesterday we changed some pool configuration parameters to make the server assign less hashrate to small networks. Thanks to your advice I noticed that the pool could get into periods where it would assign too much hashrate to small networks. As a result, quick-restarting would find blocks and slower miners would never get any shares. My goal is to make sure that everyone is assigned to networks long enough to get a few shares in, which may require a bit more tweaking in the upcoming days

[CritterDog]

Almost finished with second test and I think part of this this is hashrate issue. I finished the 24hr mining litecoins only on prohash yesterday and results are:
Prohash paid 1.58 LTC and I was watching my hashrate close and it was hanging around 160-165Mhs
I will be done with the 24 hour test on Litecoin pool at 11pm but I looked at there charts and it shows the lowest my hashrate ever hit was 169Mhs. But average was 170mhs-175Mhs. Over 5Mhs more then I got on Prohash. Over 24hours would make a noticable diff. Will update at 11pm tonight with final total.

[Chris Sokolowski]

I'm not concerned about a slightly lower hashrate. Our goal is to have better overall earnings than a litecoin-only pool. By definition a coin-switching pool will provide miners lower hashrate than a litecoin-only pool. The increased profitability should make up for the lower hashrate though. If it does not, then there is no reason to use a coin-switching pool, which means that we have even more work to do.

[CritterDog]

Almost finished with second test and I think part of this this is hashrate issue. I finished the 24hr mining litecoins only on prohash yesterday and results are:
Prohash paid 1.58 LTC and I was watching my hashrate close and it was hanging around 160-165Mhs
I will be done with the 24 hour test on Litecoin pool at 11pm but I looked at there charts and it shows the lowest my hashrate ever hit was 169Mhs. But average was 170mhs-175Mhs. Over 5Mhs more then I got on Prohash. Over 24hours would make a noticable diff. Will update at 11pm tonight with final total.

Here is the final results. BTW the litecoinpool's payout calculator was crazy close in showing me what to expect to make in a 24 hour period. So at least for me something is still very wrong on prohash. I hope this can get fixed.

Prohash = 1.58 Litecoins
Litecoinpool = 1.82 Litecoins

Difference .24 LTC

[GenTarkin]

Almost finished with second test and I think part of this this is hashrate issue. I finished the 24hr mining litecoins only on prohash yesterday and results are:
Prohash paid 1.58 LTC and I was watching my hashrate close and it was hanging around 160-165Mhs
I will be done with the 24 hour test on Litecoin pool at 11pm but I looked at there charts and it shows the lowest my hashrate ever hit was 169Mhs. But average was 170mhs-175Mhs. Over 5Mhs more then I got on Prohash. Over 24hours would make a noticable diff. Will update at 11pm tonight with final total.

Here is the final results. BTW the litecoinpool's payout calculator was crazy close in showing me what to expect to make in a 24 hour period. So at least for me something is still very wrong on prohash. I hope this can get fixed.

Prohash = 1.58 Litecoins
Litecoinpool = 1.82 Litecoins

Difference .24 LTC

So, you made 86% pps @ prohashing, 14% less than 100% ltc pps.
What was your reject % on prohash vs litecoinpool?
Now, lets run some quick math:
Lets assume prohashing pulled off paying miners even(100%pps ltc)
Prohashing takes 5% in fees(drops to 95%pps ltc)
If your reject rate is 3-5% like mine has been, then thats another 3-5% loss(drops to 90%)
Add the 3% bonus you get at litecoinpool currently and that equates to another 3% loss at prohashing(drops to 87%)
So....
That 86% is awfully close to the 87% we just guestimated.
Prohashing needs to A: fix their high reject rate(if thats indeed happening) B: continue to work on their multi algo switching - Clevermining manages to always payout more than prohashing(as ive experimented with in the past month or so).
The nice thing about prohashing tho, is the ability to be paid in any coin you want, which saves on service fees when using something like shapeshift.io

[Steve Sokolowski]

Almost finished with second test and I think part of this this is hashrate issue. I finished the 24hr mining litecoins only on prohash yesterday and results are:
Prohash paid 1.58 LTC and I was watching my hashrate close and it was hanging around 160-165Mhs
I will be done with the 24 hour test on Litecoin pool at 11pm but I looked at there charts and it shows the lowest my hashrate ever hit was 169Mhs. But average was 170mhs-175Mhs. Over 5Mhs more then I got on Prohash. Over 24hours would make a noticable diff. Will update at 11pm tonight with final total.

Here is the final results. BTW the litecoinpool's payout calculator was crazy close in showing me what to expect to make in a 24 hour period. So at least for me something is still very wrong on prohash. I hope this can get fixed.

Prohash = 1.58 Litecoins
Litecoinpool = 1.82 Litecoins

Difference .24 LTC

So, you made 86% pps @ prohashing, 14% less than 100% ltc pps.
What was your reject % on prohash vs litecoinpool?
Now, lets run some quick math:
Lets assume prohashing pulled off paying miners even(100%pps ltc)
Prohashing takes 5% in fees(drops to 95%pps ltc)
If your reject rate is 3-5% like mine has been, then thats another 3-5% loss(drops to 90%)
Add the 3% bonus you get at litecoinpool currently and that equates to another 3% loss at prohashing(drops to 87%)
So....
That 86% is awfully close to the 87% we just guestimated.
Prohashing needs to A: fix their high reject rate(if thats indeed happening) B: continue to work on their multi algo switching - Clevermining manages to always payout more than prohashing(as ive experimented with in the past month or so).
The nice thing about prohashing tho, is the ability to be paid in any coin you want, which saves on service fees when using something like shapeshift.io

I spent some time this morning investigating the rejection rate. The rate of shares that are completely rejected are 0.7%.

Chris is running a query right now to investigate some other possibilities. One possibility is that many shares are being submitted stale, and therefore only the merge mining portion of those shares are being credited. Chris will have more information on that soon.

We ran a query and determined that coin selection can't be the cause of this problem because with the existing coins, only 92% of expected is being earned. Even after we penalized those "low luck" miners (who apparently fixed their systems, because there are no more forfeitures occurring), 8% is not being accounted for. We thought that the "low luck" miners were responsible for the entire difference, but there must be another issue that we have yet to be able to determine.

[CritterDog]

Here is the final results. BTW the litecoinpool's payout calculator was crazy close in showing me what to expect to make in a 24 hour period. So at least for me something is still very wrong on prohash. I hope this can get fixed.

Prohash = 1.58 Litecoins
Litecoinpool = 1.82 Litecoins

Difference .24 LTC

So, you made 86% pps @ prohashing, 14% less than 100% ltc pps.
What was your reject % on prohash vs litecoinpool?
Now, lets run some quick math:
Lets assume prohashing pulled off paying miners even(100%pps ltc)
Prohashing takes 5% in fees(drops to 95%pps ltc)
If your reject rate is 3-5% like mine has been, then thats another 3-5% loss(drops to 90%)
Add the 3% bonus you get at litecoinpool currently and that equates to another 3% loss at prohashing(drops to 87%)
So....
That 86% is awfully close to the 87% we just guestimated.
Prohashing needs to A: fix their high reject rate(if thats indeed happening) B: continue to work on their multi algo switching - Clevermining manages to always payout more than prohashing(as ive experimented with in the past month or so).
The nice thing about prohashing tho, is the ability to be paid in any coin you want, which saves on service fees when using something like shapeshift.io

I spent some time this morning investigating the rejection rate. The rate of shares that are completely rejected are 0.7%.

Chris is running a query right now to investigate some other possibilities. One possibility is that many shares are being submitted stale, and therefore only the merge mining portion of those shares are being credited. Chris will have more information on that soon.

We ran a query and determined that coin selection can't be the cause of this problem because with the existing coins, only 92% of expected is being earned. Even after we penalized those "low luck" miners (who apparently fixed their systems, because there are no more forfeitures occurring), 8% is not being accounted for. We thought that the "low luck" miners were responsible for the entire difference, but there must be another issue that we have yet to be able to determine.

My reject rate on the A2 is about 0.12 on both pools..I think its around the same on my gblacks but I reset them so cant say for sure