Should have been out way earlier. This forum has some nasty people, since moderation is absent. I personally had to mute most of the people here. I’m thankful for your contributions, it helps us remain sane and fact-based.
2 Likes
I couldn’t read all the garbage that was posted, but yes higher C levels will use more power. Farmers who want to keep a small footprint can use the lower C levels which are very efficient (you can gain 15% over uncompressed, at barely any extra power).
4 Likes
Not really, its about both.
C9 is only good for free elec , farm size is irrelevant.
C8 is for cheap elec and appropriate size farm.
C7 is moderate elec and apropriate farm size.
Max did not hide this info.
So a c7 test is a better test, if you have dearer elec.
1 Like
If you could find an hdd that had 15% higher power consumption but 30% more capacity, would you say that farmers using that drive are turning chia into PoW? Compare that to chia farmers that add 30% capacity by buying more of the standard drives. Plot compression lowers the watts per farmable TB. This is the metric that should be optimized.
1 Like
That is a good point. I think however that maybe saying lowering the Watt per K32 (equivalent) is more accurate?
Or at least more clear to my brain 
Real space vs effective space gets confusing fast.
So in my case +22.25% (bb c4) more K32 on the same hdd space (350 TiB)
so uncompressed = +/- 3500 plots
compressed to c4 = +/- 4280 plots (+780 K32’s)
HDD power is same in both cases above, compressed plots add about 10W to CPU overhead. (10W is roundup based on actual results)
Adding 780 normal plots by way of adding extra 18TB hdd’s would require 5 hdd’s, each consuming 5-6W, so 25-30W in total.
So adding hdd’s in my example would lead to 15-20W more power consumption for the same number of K32’s.
Whether or not that translates into better security (more netspace) for the blockchain at the same or lower power consumption is a different and perhaps more difficult discussion.
- On the one hand, the netspace (measured in K32 equivalent) increases in a more efficient way than adding more hdd’s.
- On the other hand getting 5!% of that netspace can be done by the same efficient means.
2 Likes
we are getting there. Just wait for the next upgrade. It will use your boiler and heat pump electronics for farming.
While farming is almost eco friendly. Plotting burns like hell. With average plot lifespan, lets say 4 years, it may be interesting to find out. I am betting Chia wouldn’t be so green.
Imagine how much energy kids burnt just by trying compressed plots.
I would say it does, since there’s more hardware doing more work, thus the blockchain is more secure. It only becomes a problem if there’s centralization, such as asics chip production (which won’t happen for the algorithm chia uses).
I’m really disappointed.
Keep spreading your good mood @aurelius , your good advices.
A troll, an army of troll should never bother you on the internet.
Turning everything “discord” might be evil at a similar level.
1 Like
Your difficulty level of 18 is wrong imho for 4355 plots although I am not sure about MMX difficulty compared to Gigahorse Chia plots. Gigahorse Chia plots would need at least difficulty of 100 for that amount. This might explain the work being done by GPU.
1 Like
According to MMX…highest impact is between 1-20 difficulty. >20 there is very little difference.
GPU still does proofs every 5-10s, it cannot be changed. Difficulty/wattage is allegedly linked to checking the full proofs.
This is a thing that is bugging me…
There is some advice somewhere that you should set your pool difficulty so that you get a proof every few minutes. I think 10minutes.
So why is there a farm size limit for each GPU?
Farm size determines the number of plots, and the number of plots determines your chance of passing the filter at every block. Higher farm size means higher compute cost to calculate proofs at every block, because “compression” simply means “calculate it on-the-fly”. If you don’t have enough compute, you’ll be too slow and the benefits of saving space will be lost.
This is amplified by partials when mining in pool.
This is a good explanation but if I set pool difficulty high so that I have to work out a proof only rarely…then the gpu will be idle?
There’s some bit that I don’t get!
Yes. You can read about partials and pooling, and it will click in your mind I’m sure
It’s not idle, you are adding 1-2% extra work ontop of what the gpu is already doing for farming.
you have misunderstood - it is FULL PROOF calculation. That is explanation on MMX Gigahorse site. GPU still does partials ~10s, and uses ~80Wh. With pool diff set to 18, I have reduced wattage by 20W. Unless you have free energy, there is no point in GPU farming.
What you see in 1st post is the reality of compression farming. Despite, there is whining about “wrong” math.
When you are farming solo, you are already doing a lookup every 9-10 seconds. So the GPU already has to do that work.
Pool partials just add a bit to that work, but not so much. (unless you set the difficulty too low and it’s trying to do a lookup every second.)
let me do some simple maths, i have 100TB farm, it uses 60watts, i use c8 with nvidia k2200, now it uses 80 watts. i now have effective space of 142TB, about extra $0.5/day from compression. extra 0.48 Kwh used by K2200, would need electric cost of $1/Kwh for it to be not worth doing.
4 Likes
How did you achieve 60W farming, if PC with such a card idles at 60W without HDDs?
do you use Gigahorse or Flexfarmer?
how much do you earn per month? “extra 0.5$” is a weird, incomparable number. wattage is weird - looks like some sensor reading, not wall reading.
nice, now use Pi with 100TB normal (or slightly compressed plots) - where you get about <40W, and your profit will skyrocket 