Meanwhile, a smaller pool in that same time frame may be much lower, or much higher than neutral.
Which sometimes is part of the "fun" EMC had a block that took 1 day 13 hrs that solved yesterday. Since then 6 in about 18 hours.
-Dave
Topic: [CLOSED] BTC Guild - Pays TxFees+NMC, Stratum, VarDiff, Private Servers - page 82. (Read 903150 times)
May 07, 2014, 01:55:28 PM
Which sometimes is part of the "fun" EMC had a block that took 1 day 13 hrs that solved yesterday. Since then 6 in about 18 hours.
-Dave
May 07, 2014, 01:47:48 PM
I feel that without being pushed you would not have "outed" this freerider, you kept claiming luck, until a number of people really pushed it.
I also feel that this "freerider" has cost me over $50,000 while I was being short changed for the blocks I was solving on your pool.
We have the right to know who this person it and what is wrong with his machines so that we can protect ourselves from them.
EDIT: Personally, I find it VERY interesting that you turn PPS off two weeks into this "bad luck" run and thus protecting yourself (financially) against any prolonged run of bad luck, but you had to be pushed to do anything to protect us miners that were being ripped off by this "bad luck" FREERIDER.
I also feel that this "freerider" has cost me over $50,000 while I was being short changed for the blocks I was solving on your pool.
We have the right to know who this person it and what is wrong with his machines so that we can protect ourselves from them.
EDIT: Personally, I find it VERY interesting that you turn PPS off two weeks into this "bad luck" run and thus protecting yourself (financially) against any prolonged run of bad luck, but you had to be pushed to do anything to protect us miners that were being ripped off by this "bad luck" FREERIDER.
May 07, 2014, 01:03:38 PM
As of right now, this doesn't appear to be affecting publicly sold miners, at least not to the same degree.
thanks for sharing .. I guess user 522067 got some explaining to do? (are they unlucky or certain behavior/limits are happening? and you probably see worse examples in the full data)
the brightside is if a use case is figured out, then perhaps you could automate sending out tests to miners on a periodic semi-random basis.. hopefully that would make a good safety net instead of manually playing wack-a-mole
EDIT after reading eleuthria: 5222067 is a newer miner during such higher diff so perhaps those numbers are fine.. i guess the ones that don't show up on the leaderboard are the real issue
As you edited, yes, 522067 isn't in any way showing signs of a withholding attack/bug. They started mining at much higher difficulty than many other miners, so obviously their solved blocks vs shares ratio is much worse than some of the other top users who were mining at difficulties significantly lower than it is when 522067 started.
May 07, 2014, 12:41:45 PM
As of right now, this doesn't appear to be affecting publicly sold miners, at least not to the same degree.
thanks for sharing .. I guess user 522067 got some explaining to do? (are they unlucky or certain behavior/limits are happening? and you probably see worse examples in the full data)
the brightside is if a use case is figured out, then perhaps you could automate sending out tests to miners on a periodic semi-random basis.. hopefully that would make a good safety net instead of manually playing wack-a-mole
EDIT after reading eleuthria: 5222067 is a newer miner during such higher diff so perhaps those numbers are fine.. i guess the ones that don't show up on the leaderboard are the real issue
May 07, 2014, 12:15:38 PM
So my biggest question right now is if it turns out to be errant miners, what can be done about it? For all I know I might have one such miner, but how to tell? I'm a very small player and have never solved a block. How do you weed out the problem miners without shitting on the little guys? And then on the flip side what incentive is there for a manufacturer to give a shit? They only "guarantee" a given hash rate, not that those hashes will ever solve a block. An unscrupulous manufacturer could deliberately exploit this.
I'm getting out for the time being anyway. Not because of this but my electricity costs are too high and I need to invest in my house and not my miners this summer. I might be back when winter comes around again so will be following this with interest.
I'm getting out for the time being anyway. Not because of this but my electricity costs are too high and I need to invest in my house and not my miners this summer. I might be back when winter comes around again so will be following this with interest.
As of right now, this doesn't appear to be affecting publicly sold miners, at least not to the same degree.
If you look at the ghash #'s over the last day they have had three 2+ hour times w/o a block and seven 1+ hour times w/o a block. Since they have about 3x the hashing power as BTC Guild that would that be the same as three 6+ hour blocks here and seven 3+ hour blocks?
-Dave
-Dave
Yes, basically. Luck is measured by shares vs network difficulty, but given constant speeds, the luck on each individual block would be comparable (6 hour block on one pool would be a 2 hour block on a pool that's 3x faster). This is why faster pools level out to neutral luck faster. They spend less time on bad blocks, but also less time on good ones. As a result they simply have more blocks in a given time frame, which means they're more likely to be close to neutral in that time frame compared to a smaller pool. Meanwhile, a smaller pool in that same time frame may be much lower, or much higher than neutral.
May 07, 2014, 11:07:52 AM
So my biggest question right now is if it turns out to be errant miners, what can be done about it? For all I know I might have one such miner, but how to tell? I'm a very small player and have never solved a block. How do you weed out the problem miners without shitting on the little guys? And then on the flip side what incentive is there for a manufacturer to give a shit? They only "guarantee" a given hash rate, not that those hashes will ever solve a block. An unscrupulous manufacturer could deliberately exploit this.
I'm getting out for the time being anyway. Not because of this but my electricity costs are too high and I need to invest in my house and not my miners this summer. I might be back when winter comes around again so will be following this with interest.
I'm getting out for the time being anyway. Not because of this but my electricity costs are too high and I need to invest in my house and not my miners this summer. I might be back when winter comes around again so will be following this with interest.
May 07, 2014, 10:01:34 AM
If you look at the ghash #'s over the last day they have had three 2+ hour times w/o a block and seven 1+ hour times w/o a block. Since they have about 3x the hashing power as BTC Guild that would that be the same as three 6+ hour blocks here and seven 3+ hour blocks?
-Dave
-Dave
May 07, 2014, 12:16:51 AM
There's 2 possible break points. There's ~2.1b and ~4.2b. One is for signed 32-bit, one is for unsigned. BTC Guild's luck took a sharp turn starting ~6 weeks ago. That is right in line with the network difficulty passing 4.2b (March 24).
Luck for BTC Guild and Eligius both started trending negative a little earlier than that. Roughly February. However, this wasn't bad enough on either pool to simply rule out variance for that time frame. For BTC Guild it was ~6% negative in March. For Eligius it was ~5% negative. However, February is when 2.1b break point (Jan 24th adjustment put diff over that amount) would have triggered.
Right now the largest users that had definite anomalies in their shares vs blocks solved have had accounts frozen, and one group has contacted me and is looking into it. If they come back and it is a definite 'Yes, something was broken in our software/firmware/hardware', then at least the problem is identified and will be fixed.
Luck for BTC Guild and Eligius both started trending negative a little earlier than that. Roughly February. However, this wasn't bad enough on either pool to simply rule out variance for that time frame. For BTC Guild it was ~6% negative in March. For Eligius it was ~5% negative. However, February is when 2.1b break point (Jan 24th adjustment put diff over that amount) would have triggered.
Right now the largest users that had definite anomalies in their shares vs blocks solved have had accounts frozen, and one group has contacted me and is looking into it. If they come back and it is a definite 'Yes, something was broken in our software/firmware/hardware', then at least the problem is identified and will be fixed.
Interesting that when the Chinese A1 chips miners were shipped in bulk from multiple sources. Probably some Chinese farms were being built with these miners as well.
Its also bugging me that CK asked them to release their source code and they completed ignored. I've been warning others in hardware section about this but noone seems to care... " Oh look miner reports 1TH/s, thats all i need"
May 06, 2014, 11:54:28 PM
That is some great sleuth work.
I have to say the pool rate dropped by almost 2000 in the last 48 hrs but our blocks have been coming faster and faster like before in your time frame.
I have to say the pool rate dropped by almost 2000 in the last 48 hrs but our blocks have been coming faster and faster like before in your time frame.
There's 2 possible break points. There's ~2.1b and ~4.2b. One is for signed 32-bit, one is for unsigned. BTC Guild's luck took a sharp turn starting ~6 weeks ago. That is right in line with the network difficulty passing 4.2b (March 24).
Luck for BTC Guild and Eligius both started trending negative a little earlier than that. Roughly February. However, this wasn't bad enough on either pool to simply rule out variance for that time frame. For BTC Guild it was ~6% negative in March. For Eligius it was ~5% negative. However, February is when 2.1b break point (Jan 24th adjustment put diff over that amount) would have triggered.
Right now the largest users that had definite anomalies in their shares vs blocks solved have had accounts frozen, and one group has contacted me and is looking into it. If they come back and it is a definite 'Yes, something was broken in our software/firmware/hardware', then at least the problem is identified and will be fixed.
Luck for BTC Guild and Eligius both started trending negative a little earlier than that. Roughly February. However, this wasn't bad enough on either pool to simply rule out variance for that time frame. For BTC Guild it was ~6% negative in March. For Eligius it was ~5% negative. However, February is when 2.1b break point (Jan 24th adjustment put diff over that amount) would have triggered.
Right now the largest users that had definite anomalies in their shares vs blocks solved have had accounts frozen, and one group has contacted me and is looking into it. If they come back and it is a definite 'Yes, something was broken in our software/firmware/hardware', then at least the problem is identified and will be fixed.
May 06, 2014, 11:08:54 PM
There's 2 possible break points. There's ~2.1b and ~4.2b. One is for signed 32-bit, one is for unsigned. BTC Guild's luck took a sharp turn starting ~6 weeks ago. That is right in line with the network difficulty passing 4.2b (March 24).
Luck for BTC Guild and Eligius both started trending negative a little earlier than that. Roughly February. However, this wasn't bad enough on either pool to simply rule out variance for that time frame. For BTC Guild it was ~6% negative in March. For Eligius it was ~5% negative. However, February is when 2.1b break point (Jan 24th adjustment put diff over that amount) would have triggered.
Right now the largest users that had definite anomalies in their shares vs blocks solved have had accounts frozen, and one group has contacted me and is looking into it. If they come back and it is a definite 'Yes, something was broken in our software/firmware/hardware', then at least the problem is identified and will be fixed.
Luck for BTC Guild and Eligius both started trending negative a little earlier than that. Roughly February. However, this wasn't bad enough on either pool to simply rule out variance for that time frame. For BTC Guild it was ~6% negative in March. For Eligius it was ~5% negative. However, February is when 2.1b break point (Jan 24th adjustment put diff over that amount) would have triggered.
Right now the largest users that had definite anomalies in their shares vs blocks solved have had accounts frozen, and one group has contacted me and is looking into it. If they come back and it is a definite 'Yes, something was broken in our software/firmware/hardware', then at least the problem is identified and will be fixed.
May 06, 2014, 11:06:25 PM
2.1 Billion was when the hash rate was 15,000 PH/s the last week of January. We had November Jupiters solving blocks after that.
would be nice to have some type of comprehensive test to run locally on a miner for sanity sake
Hashfast started shipping end of Jan early Feb. I can offer one Feb deliverd Babyjet for testing.
May 06, 2014, 11:04:19 PM
[/quote]
would be nice to have some type of comprehensive test to run locally on a miner for sanity sake
[/quote] Maybe Nwools could point you in the right direction for something like that.
May 06, 2014, 10:58:22 PM
2.1 Billion was when the hash rate was 15,000 PH/s the last week of January. We had November Jupiters solving blocks after that.
would be nice to have some type of comprehensive test to run locally on a miner for sanity sake
May 06, 2014, 10:52:43 PM
It would be very evil for a hardware manufacturer to neuter their own hardware via a firmware update, but based on the way KNC has been treating their customers lately, I wouldn't put it past them. 
actually a flaw with KNC firmware would be par for the course if anyone remembers how bad their flushwork and handling on/off cores were for so long... they did their own drivers and version of cgminer when releasedeleuthria, you mentioned that 2.1 bil diff was a distinct change of behavior... is there a way to query your Team KNC (most of them should only be KNC miners), dump the data and see if any pattern shows itself?
That would connect a lot of dot
I can point my 1.7 TH of oct jupiters to a test pool if someone has a plan with test data to prove anything
May 06, 2014, 10:42:45 PM
It would be very evil for a hardware manufacturer to neuter their own hardware via a firmware update, but based on the way KNC has been treating their customers lately, I wouldn't put it past them.
actually a flaw with KNC firmware would be par for the course if anyone remembers how bad their flushwork and handling on/off cores were for so long... they did their own drivers and version of cgminer when released
eleuthria, you mentioned that 2.1 bil diff was a distinct change of behavior... is there a way to query your Team KNC (most of them should only be KNC miners), dump the data and see if any pattern shows itself?
That would connect a lot of dots
I can point my 1.7 TH of oct jupiters to a test pool if someone has a plan with test data to prove anything
May 06, 2014, 08:17:56 PM
It would be very evil for a hardware manufacturer to neuter their own hardware via a firmware update, but based on the way KNC has been treating their customers lately, I wouldn't put it past them.
+1
I moved to another pool until we know what the problem is. I've been mining since early 2011 and i agree that its not a normal bad luck streak.
May 06, 2014, 07:34:56 PM
It would be very evil for a hardware manufacturer to neuter their own hardware via a firmware update, but based on the way KNC has been treating their customers lately, I wouldn't put it past them.
May 06, 2014, 07:26:30 PM
I have an Antminer S1 that hit a block within the last two weeks, and a November Jupiter that hit a block within the last two months.
May 06, 2014, 07:22:12 PM
Hardware is possible, but doesn't quite make sense given how barebones a mining ASIC *should* be.
-Dave
Personally, my bet is on any product where people should be solo mining yet are on the pools. Has that guy up in the PacNW hit a bunch of blocks lately?
May 06, 2014, 07:15:37 PM
Hardware is possible, but doesn't quite make sense given how barebones a mining ASIC *should* be.
https://twitter.com/ID_AA_Carmack/status/128988479114854401
-Dave
Jump to: