[1200 TH] EMC: 0 Fee DGM. Anonymous PPS. US & EU servers. No Registration!

[LazyOtto]

Makes sense to me. 1diff is the benchmark.

Might be nice to have on the "My Workers" page another column showing something like "Average diff over time period".

Abbreviated in some way, of course.

--

edit: What algorithm are you using to 'allocate' a diff level to a given worker / work request?

[1714311957]

1714311957

[Inaba]

Right now, I have it set to target 8 getworks per minute returned.  So if you submit more, you are bumped up in difficulty.  I do like the average diff over time idea, I will have to figure out how to make that happen.

[Askit2]

I like the sound of diffculty 1 equivilents of DOE. It is a good common ground for compairison. If you went variable by the lowest common denominator it is possible that people disconnecting would alter the equivelence. So I would say Difficulty 1 for now and some static equivilent later if needed.

[Inaba]

Ok, if anyone wants to try out the variable difficulty server, you can connect to:

http://us3.eclipsemc.com:8437

That will give you share difficulty based off your hashrate.  Let me know if anyone finds any problems.  I've added an "average difficulty" column to the My Workers page.

[dave3]

Ok, if anyone wants to try out the variable difficulty server, you can connect to:

http://us3.eclipsemc.com:8437

That will give you share difficulty based off your hashrate.  Let me know if anyone finds any problems.  I've added an "average difficulty" column to the My Workers page.

I'm trying it out now.  Here's a screenshot after it's had a chance to settle for about 15 minutes.

Code:

 cgminer version 2.7.5 - Started: [2012-09-04 13:44:23]
--------------------------------------------------------------------------------
 (5s):1622.0 (avg):1616.9 Mh/s | Q:36  A:176  R:0  HW:0  E:489%  U:10.1/m
 TQ: 0  ST: 3  SS: 0  DW: 5  NB: 2  LW: 710  GF: 1  RF: 0  WU: 26.3
 Connected to http://us3.eclipsemc.com:8437 with LP as user ---
 Block: 0000028ac1f60ffee5bd6bf3983c224b...  Started: [13:44:24]
--------------------------------------------------------------------------------
 [P]ool management [S]ettings [D]isplay options [Q]uit
 BFL 0:  58.0C         | 811.0/810.5Mh/s | A:92 R:0 HW:0 U: 5.29/m
 BFL 1:  59.6C         | 811.0/810.5Mh/s | A:86 R:0 HW:0 U: 4.94/m
--------------------------------------------------------------------------------

 [2012-09-04 13:57:54] Accepted 10268b48.ad38c5d7 BFL 0 pool 0
 [2012-09-04 13:57:54] Accepted 2e41fd1d.ec388744 BFL 1 pool 0
 [2012-09-04 13:57:59] Accepted 08496ec2.e110f28d BFL 0 pool 0
 [2012-09-04 13:58:00] Accepted 1f0b686e.dc00bb7e BFL 1 pool 0
 [2012-09-04 13:58:00] Accepted 06285863.e4528ac9 BFL 1 pool 0
 [2012-09-04 13:58:05] Accepted 2c89b5a4.b0c6387f BFL 0 pool 0
 [2012-09-04 13:58:10] Accepted 3a478ad4.61b0409c BFL 0 pool 0
 [2012-09-04 13:58:21] Accepted 1d34adf4.7cf74461 BFL 1 pool 0
 [2012-09-04 13:58:31] Accepted 0c0595d1.b1b6cebb BFL 1 pool 0
 [2012-09-04 13:58:58] Accepted 2bbac7f9.617672f6 BFL 0 pool 0
 [2012-09-04 13:59:03] Accepted 086b8829.f1232580 BFL 0 pool 0
 [2012-09-04 13:59:19] Accepted 2848d96b.ae49cbc9 BFL 0 pool 0
 [2012-09-04 13:59:19] Accepted 08a78648.026863f7 BFL 0 pool 0
 [2012-09-04 13:59:30] Accepted 0a0a4ee0.726e102e BFL 0 pool 0
 [2012-09-04 13:59:51] Accepted 171214fe.9d263b90 BFL 0 pool 0
 [2012-09-04 14:00:01] Accepted 3c8e91e0.a835c0c6 BFL 0 pool 0
 [2012-09-04 14:00:07] Accepted 2bc0c3c7.c147a6e6 BFL 1 pool 0
 [2012-09-04 14:00:07] Accepted 2d3e8319.3bf4b7bc BFL 1 pool 0
 [2012-09-04 14:00:07] Accepted 0cac4cfb.dbc96264 BFL 1 pool 0
 [2012-09-04 14:00:12] Accepted 3cb08837.0a28d981 BFL 0 pool 0
 [2012-09-04 14:00:17] Accepted 28d9a195.7e3b2c46 BFL 1 pool 0
 [2012-09-04 14:00:23] Accepted 45baf6e8.05d1ed00 BFL 1 pool 0
 [2012-09-04 14:00:38] Accepted 10e7ec4c.dc54a2b8 BFL 0 pool 0
 [2012-09-04 14:00:49] Accepted 09805306.4e95b86c BFL 0 pool 0
 [2012-09-04 14:00:49] Accepted 42ca7b3f.c3829ead BFL 1 pool 0
 [2012-09-04 14:01:00] Accepted 2c154b71.af0fb45c BFL 0 pool 0
 [2012-09-04 14:01:00] Accepted 34736018.6a2edb21 BFL 1 pool 0
 [2012-09-04 14:01:00] Accepted 3e6d4f29.acdd2dcf BFL 1 pool 0
 [2012-09-04 14:01:00] Accepted 382c6dd6.9b6084c8 BFL 1 pool 0
 [2012-09-04 14:01:10] Accepted 13b01a70.49631273 BFL 1 pool 0
 [2012-09-04 14:01:15] Accepted 0b8ab068.8b9aa206 BFL 0 pool 0
 [2012-09-04 14:01:26] Accepted 19ec869e.e8722e90 BFL 0 pool 0
 [2012-09-04 14:01:26] Accepted 063bd9d0.9222c128 BFL 1 pool 0
 [2012-09-04 14:01:37] Accepted 2e9d2a8f.995c76f7 BFL 0 pool 0
 [2012-09-04 14:01:37] Accepted 564191df.9c186cb5 BFL 1 pool 0
 [2012-09-04 14:01:47] Accepted 50c88ba2.ebce6aba BFL 0 pool 0
 [2012-09-04 14:01:47] Accepted 3d74fc33.bcccbae9 BFL 1 pool 0

[kano]

As per cgminer and the cgminer API, yep Diff1 is of course the most sensible base to use so comparisons can be made.
The WU: number on cgminer is Diff1 Work done per minute (it also includes all work, not just accepted shares)
The API has 'Diff1 Work' - only pools currently show it, but the next version will show it for devices also
(2.7.5 and before says 'Diff1 Shares' but I've changed that to 'Diff1 Work' for the next release since it includes all work)

Next thing I want to do is actually tally pool getworks under 3 headings: DiffA, DiffB, DiffOther
So you can see what the first 2 difficulties used were and also pool getwork totals for those and all others.
This will allow you to see what most of the work you are doing is for each pool.
I've no idea if keeping it in 3 groups like that is best or not, but that's just the idea.

[Inaba]

That sounds good to me, Kano.  What do you think about difficulty being doled out... should it be variable as it is now, or variable as powers of 2?  Do you think it matters? 

Right now I have the difficulty being adjusted every 3 minutes based on the getworks received in that time frame, with a target of 8 per minute.  I'm not sure this is optimal yet, but I'm open to suggestions.

Dave3: Those look in line with my testing.  Your speed reads a little high, but it's hard to gauge with only 15 minutes of testing.  How does it look after a couple hours?  What is your time estimation window set at?

[organofcorti]

How many getworks end up producing a share, on average? Is there an average?

[stoppots]

At wat speed would you think or recommend miners be mining at for trying out the variable difficulty server. Is a higher possible variance the only negative drawback in higher difficulty shares, from a miners perspective? (A slow 1gh/s miner perspective)

[stoppots]

When did the change to no longer sticky the top 10 pools in the pools section occur?

Bitcoin Forum > Bitcoin > Mining > Pools

Seems like less than a week ago they were still stickied. Am I safe in assuming this is directly related to EMC's growing popularity and the soon to be added Variable Difficulty Shares(or similar shortened version) to EMC's subject title?

EDIT:
Decided to use the search feature and came across this thread.

Top 10 Pools stickied?
https://bitcointalk.org/index.php?topic=103313.0

and this

Mining Pools
https://bitcointalk.org/index.php?topic=104664.0

[Graet]

When did the change to no longer sticky the top 10 pools in the pools section occur?

Bitcoin Forum > Bitcoin > Mining > Pools

Seems like less than a week ago they were still stickied. Am I safe in assuming this is directly related to EMC's growing popularity and the soon to be added Variable Difficulty Shares(or similar shortened version) to EMC's subject title?

EDIT:
Decided to use the search feature and came across this thread.
https://bitcointalk.org/index.php?topic=103313.0

yeah more likely diblo-d3 cant be bothered anymore, no-one responded to that thread or 2 similar in meta forum :/

[stoppots]

Hi Graet, perhaps I am being overly observant watching out for conspiracies again.

[os2sam]

Hi Graet, perhaps I am being overly observant watching out for conspiracies again.

Just because you ain't paranoid doesn't mean they aren't out to get you!
Sam

[cyberlync]

Not sure if it's a problem, it looks strange though. Connected to the var. diff. server.
I get the "New block detected on network before longpoll" message like 10-20 times, every time it shows up. Anyone else? Using CGminer 2.6.6 (Higher versions make my gpu crash and can't recover).

[Inaba]

I think on average there should be 1 share for 1 getwork at 1diff, but I'm sure someone can correct me if I am wrong.

Stoppots: The variable difficulty is for every miner of every speed, that's the  beauty of it.  If you are less than about 500 MH/s, you'll get 1diff shares almost exclusively just like the current servers, but if you're > 500 MH/s, your difficulty will start going up according to how fast you are.

[ocminer]

I'm really sorry for interrupting or asking such a dumb question, but I seemed to have missed whats the pro with the higher difficulty ?

Will I solve blocks faster ? Will I get paid more ? Will I have less stales ?!

Sorry & Thanks !

[LazyOtto]

Less network and application overhead load for you.

A lot less network and application/execution overhead load for the pool.

[ocminer]

I've switched some of my workers to the variable diff server for testing

[os2sam]

I'm really sorry for interrupting or asking such a dumb question, but I seemed to have missed whats the pro with the higher difficulty ?

Will I solve blocks faster ? Will I get paid more ? Will I have less stales ?!

Sorry & Thanks !

I don't think it's a dumb question.

My understanding is that higher difficulty shares will take longer for a miner to mine.  So a Difficulty 4 share would take your rig a approx 4 times as long to achieve so the pool will then pay 4 times the amount than it would for a diff 1 share.

So no you won't solve blocks faster, you won't get paid more and I don't know about stales.

The reason for this is for pools to prepare for the high hash power mining hardware that are coming out soonish.  The higher difficulty shares require less bandwidth to and from the pool and less processing power at the pool end.

That's my understanding reading what different pool op's and Bitcoin experts have been posting here and there.  I may be a little off, I usually am .
Sam

[kano]

To be specific, it's the 2nd part of the 2 part pair of roll-n-time and >diff1 shares

Roll-n-time means you ask the pool for work and can make multiple work items with the one work item you get from the pool
Typically it seams around 10 or 20 work items per single work request for someone in the 1GH/s to 2GH/s range.
Thus you are asking the pool for 1/10th or even 1/20th of the information, thus your requests to the pool for work are also 1/10th or 1/20th of the bandwidth
... and the pool has to do 1/10th or 1/20th of the getwork work for you that it needs to do without roll-n-time.

However, you still find the same number of shares and thus have to send back a share for each one to the pool.

Higher difficulty shares means you are telling the pools you have done more work, so you get paid for more work, each time you send a share back to the pool.
An example would be 2 diff shares.
A 2 diff share means you will only find half as many as you would if you were looking for 1 diff shares.
Since you only find half as many, they get paid twice as much.
As a result, you are also only sending half as many shares to the pool - so again you are using half as much bandwidth sending shares to the pool and the pool only has to handle half as many shares processing ... from you.

However, there is of course also a down side to >1 diff shares, it increases your variance.
However, anyone mining with a high hash rate is already mining with a lower variance than people with a low hash rate - so it's not really a negative thing in that respect - as long as the difficulty shares you are getting are appropriate for your hash rate.

See the posts by the BTC Statistics Squad for details about what difficulty shares are appropriate for your hash rate
(of course I am referring to people like organofcorti and Meni )