cverity
Newbie
Offline
Activity: 31
Merit: 0
|
|
August 23, 2013, 08:59:33 PM |
|
Yeah, they're both in the same machine, the only difference apart from the actual cards is which PCIe slot they're in, obviously. I do occasionally get HW errors on the HD 5830 (about 1 every 2-3 days), but I don't see the connection.
Strange, it's hard to imagine that any minor difference between the cards design would affect its ability to get the share from the card to the pool that significantly. Out of curiosity, is the 5830 also serving your monitor?
|
|
|
|
mueslo
Member
Offline
Activity: 94
Merit: 10
|
|
August 23, 2013, 09:37:33 PM |
|
Strange, it's hard to imagine that any minor difference between the cards design would affect its ability to get the share from the card to the pool that significantly.
Out of curiosity, is the 5830 also serving your monitor?
Nope, that's the 7950's job.
|
|
|
|
h2odysee (OP)
|
|
August 23, 2013, 09:43:08 PM |
|
Although there is still one phenomenon I cannot explain. I have a HD 7950 and a HD 5830. The HD 7950 consistently gets 2-5% rejects, while the HD 5830 gets 6-12% rejects, even if I set clock and intensity on the HD 7950 such that the hashrates are equal, so it isn't dependent on hash rates.
I'm guessing that older cards have more latency. So to predict your rejection rate, you'd need to add up the ping to middlecoin.com, plus whatever latency is associated with your GPU.
|
|
|
|
h2odysee (OP)
|
|
August 23, 2013, 09:49:34 PM |
|
Thanks for doing that, liquidfire.
I should really learn numpy, and scipy. They look fun to use. I could make use of them, to make my selling bot better.
|
|
|
|
|
cverity
Newbie
Offline
Activity: 31
Merit: 0
|
|
August 23, 2013, 10:42:44 PM |
|
Although there is still one phenomenon I cannot explain. I have a HD 7950 and a HD 5830. The HD 7950 consistently gets 2-5% rejects, while the HD 5830 gets 6-12% rejects, even if I set clock and intensity on the HD 7950 such that the hashrates are equal, so it isn't dependent on hash rates.
I'm guessing that older cards have more latency. So to predict your rejection rate, you'd need to add up the ping to middlecoin.com, plus whatever latency is associated with your GPU. That does seem to be the most reasonable conclusion. It's hard to imagine that the latency between the GPU and CGMiner is measurable compared to internet latency, but I'm no hardware expert. [Edit] If that's true, then you'd think everyone running a 5830 would get similar results. Anyone else with an older card experiencing the same thing compared to a newer card?
|
|
|
|
FlungSpun
|
|
August 23, 2013, 11:15:16 PM |
|
could be due to a mixture of cards forcing inefficient driver performance? perhaps there is a software bottleneck I seem to remember some chatter about cards of different generations not playing well together but didn't pay much heed as I'm all 7950s & 70s and 1 6950 on its own with the bios flashed
If some mixtures don't work maybe some are a bit iffy ....
|
|
|
|
CoinBuzz
|
|
August 24, 2013, 07:38:58 AM |
|
why cant we test it in a real pool?
we can set high difficulty (512) for about a week, then switch to lower difficulty and compare the results with each other.
That would be a real examination about this topic.
|
|
|
|
chup
Sr. Member
Offline
Activity: 736
Merit: 262
Me, Myself & I
|
|
August 24, 2013, 05:21:51 PM |
|
Not possible running a trial. Middlecoin creator said that 512 difficulty is hardcoded. Is higher difficulty favourising faster miners over slower? Answer is on the middlecoin.com main page. Just calculate rejected over accepted MH/s as a percentage and it will come that percentage is raising going down the table...
|
|
|
|
mueslo
Member
Offline
Activity: 94
Merit: 10
|
|
August 24, 2013, 10:45:22 PM Last edit: August 24, 2013, 11:06:15 PM by mueslo |
|
Just calculate rejected over accepted MH/s as a percentage and it will come that percentage is raising going down the table... Here, I made a little something. This should finally clear up misconceptions. http://mueslo.de/host/middlecoin/miners.pngRegenerates every 5 minutes.
|
|
|
|
FlungSpun
|
|
August 25, 2013, 12:11:12 AM |
|
So there is a little variance wobble for the small hash rates, seems to be around the same average as the rest of the miners ... well what do you know eh
if you animate those pngs you can see the variance dance.
|
|
|
|
FlungSpun
|
|
August 25, 2013, 12:28:10 AM |
|
Answer is on the middlecoin.com main page. Just calculate rejected over accepted MH/s as a percentage and it will come that percentage is raising going down the table... No it doesn't , no it isn't. if you plot a moving average against that data there are some small hash rate miners around 200th on the list getting better rejection averages than in the top 50. its dynamic and mueslos graph shows what going on nicely.
|
|
|
|
chup
Sr. Member
Offline
Activity: 736
Merit: 262
Me, Myself & I
|
|
August 25, 2013, 07:18:09 PM |
|
On the png I can see miners (+s) forming nice curves that are exponentialy rising with lower hashrate. At least now 21:05 25-th of August.
|
|
|
|
chup
Sr. Member
Offline
Activity: 736
Merit: 262
Me, Myself & I
|
|
August 25, 2013, 07:37:12 PM |
|
if you plot a moving average against that data there are some small hash rate miners around 200th on the list getting better rejection averages than in the top 50. its dynamic and mueslos graph shows what going on nicely.
The lowest hash rate miners has some total irrelevant data shown that are distorted by quantisation. Please, to see how 512 and 16 difficulty is different for small miners, start a small cpu/cuda miner with 10KH/s, mine for a day at middlecoin with 512 difficulty, and the second day at pool2.us.multipool.us with 16 difficulty. First day you will see 99% lines with "stratum detected new block", 0,5-0,75% lines with "yay" and 0,25-0,5% with "booo". Second day will be different story. Please note that my post is not in a way of flaming or making someone's hard work look worse, but in a way of making things better, if possible, because I like the middlecoin idea.
|
|
|
|
mueslo
Member
Offline
Activity: 94
Merit: 10
|
|
August 25, 2013, 08:13:54 PM |
|
On the png I can see miners (+s) forming nice curves that are exponentialy rising with lower hashrate. At least now 21:05 25-th of August.
Yep, but it jumps around a lot there due to there being very few miners there. On http://mueslo.de/host/middlecoin/ you can view the last hour.
|
|
|
|
|