OG Farming and Win Rate Drastically Down - UPDATED!

Chia Farming & Harvesting
1

Update at the end of this post. Can’t add an updated post to the thread as it appears to have been locked.

Hello,

I touched on this in a thread a week or two ago - but I just wanted to air some thoughts as I’ve been out of Chia for a good while and I am still farming with OG (non-compressed) plots, solo. Not in a pool.

Let me start by stating that my farm is working properly, missing no (or very, very few) signage points (may be 2-3 a week), plots are passing filter and are eligible for farming, response times are typically within 1 second and always well within 5 seconds. I am not sure I have ever seen a 5 second response time in fact. And my farm has all been working since April 2021 and I have won lots of XCH since this time.

Until… Around April 2023, my win rate dropped off a cliff. Nothing in my system has changed.

What I think has changed (and I am not sure of timing as I didn’t keep up with Chia for about a year…) is the introduction of compressed plots - NoSSD/MadMax etc ?

Yes, I know that the effective netspace will have increased accordingly, but I think there is more at play here than just size of farmed plots and netspace;

It seems to me that GPU/CPU compressed plots are ‘out competing’ OG plots full stop ? That is, if what I am thinking is correct, that no matter how many valid plots I have an how much space I am farming, there is someone, somewhere who is simply returning proofs sooner than I ever could through the GPU/CPU mechanism and thus winning the blocks before I get a chance to.

Does this sound plausible ? I confess I do not know the deep technical in’s and out’s of how the whole process works, but it seems to me that my OG solo plots are as good as useless, and the only way I will win in any significant way again is to re-plot my entire farm to say C7’s using GPU’s to farm.

Anyone care to shed any light on my thoughts ?

Thanks.

UPDATE:
Well I found out what the ‘problem’ was, feel a bit foolish really, it’s a simple one… :face_with_hand_over_mouth:

I changed my payout address back in April and forgot all about it. Since I had not been back to my office for over 6 months until recently, I had kept monitoring the old address on XCHScan as I had always done. It had stopped increasing balance, as it would do… So it looks like my farm has been working all this time, just sending rewards to the other payout address. Still, there has been a reduction in win rate but I guess that is because of the increases in netspace over this time.

Must pay more attention! :laughing:

3 Likes
2

welcome to the club. thats exactly whats going on, the reason is simple incompetent dev team, extremely stupid corps and CNI in common are full of shitbags.

look at the price. chia is over, chia is no more.

2 Likes
3

My brother is still solo with a relatively small OG farm, and still wins occasionally, his average effort over time is about right, perhaps you’re just experiencing a period of bad luck coupled with the increasing netspace.

2 Likes
4

1000
10001093×104 27.2 KB

I keep waffling on this issue… I just don’t know. We had someone block a 1000%'er over at Flex… It still could just be terrible luck.

But some of you have been running farms for a couple of years now, and if you’re noticing a trend, there could be something to it.

2 Likes
5

Well, I am not sure if all proofs are unique. That is, can I have a ‘proof’ to a challenge as well as someone else (potentially hundreds or even thousands of other people) ?

If so, back in the old days when we were all on HDD’s with OG plots, it was still the one who responded with the proof first that won the block. Now, if I have a proof and someone else running compressed plots also has a proof, he may get there in a fraction of the time I get there, even though I am still within a few seconds.

Is this how it works ?

6

Not convinced. With 12PB, my win rate was a couple of blocks a day sometimes.

Now I’ve not had a block since May…!

3 Likes
8

Yes, it apparently is and I thought I had read it before - from the Chia docs…

image
image1550×360 35.2 KB

So it would seem the first to respond with a proof (and there are likely multiple proofs per challenge, the larger the netspace the larger to number of proofs… ?) wins the block. So speed absolutely matters. Compressed plots I would presume give farmers that edge…

11

Well, given that you have 12PB and have not won a block since May, clearly something is wrong.

Foxy pool has an OG pool, I’d suggest joining that pool, and see how it goes, pools are a good way of seeing how your setup is functioning.

I was on their OG pool until I replotted to NFT, now on their NFT pool. Just looked at their OG pool, which is 111PiB in size, and the current average effort is 84%, if there was an OG plot problem I would expect their average effort to be on the high side, and it’s clearly not.

Just looked at their two biggest farmers and the average effort between those two it’s 100%. Now presuming they are running OG plots, this suggests there isn’t a problem, but I don’t know if you can farm nft or compressed plots on the OG pool.

3 Likes
12

I think normal plots would have the advantage here, compressed plots need to be computed on the fly, and response time actually increases. Running compression actually increases the time taken to respond, it doesn’t reduce it.

Please take what Ivan has to say with a pinch of salt :wink:

3 Likes
13

I was thinking the elements computed on the fly would be somewhat faster than reading larger amounts of data from a larger plot.

Is there an appreciable/measured increase for say C7 plots in average response times over OG plots ?

14

I will look into all this some more. I am convinced there’s nothing wrong, given that I see everything still happening now as I always have done when the last block (and all previous blocks) were won. Maybe it is just incredibly bad luck!

16

If you haven’t seen a win since May with 12PB, then something is very much wrong. Even with the increase in netspace, as of now you should average almost 2 wins per day. If it is somehow bad luck, you’ve been struck by lightning thousands of times by now…

For 12PB right now, going just 30 days with zero wins is a 1 : 5.04x10^23 chance.

1 Like
17

The actual proofs found on disk for challenges are relatively small, so the differences in the amount of data read in compressed plots vs C0 plots is relatively insignificant and both will have virtually identical disk lookup times.

The difference is that for C0 plots, thats it, thats all the latency in the local system. For compressed plots, they’re only ‘compressed’ because plotting was never completely finished. So after reading the data off the disk, compressed plots still need to effectively finish plotting to get the final proof. This additional latency can be significant and if it is over the 9ish seconds between signage points you can fall behind and lose out on wins.

For ~2PB of C0 plots my latency averaged under 1 second. For the same ~2PB of C7 BB plots (with 2 3060Tis), I had latency averaging about 5-6s I believe. Size of the farm, compression level, and the GPU used for harvesting are the main factors here.

18

I’ve been thinking about posting something similar for a few months now. I’m OG @1.5PiB, and after a year of consistent wins, my farm also “dropped off a cliff” after April. At the moment, my last win was 9/24, with similar time-between-wins going back to May. All systems normal excepting occasional updates.

I’m observing much fewer plots passing the filter on my four harvesters. My proof metrics look bad (>30%), and my missing signage points are oddly creeping up.

Given the price and management changes, I’m nervous, but continuing.

1 Like
19

No, as we have multiple wins per block
Provide adequate proof on time, you get a reward normally, as it always was.

I say normally, as there are exceptions, you havent posted the follow up to the selection of proofs provided, which fully explains the process.

1 Like
20

I really can’t see it being bad luck as others have said, you should be wining on a daily basis, twice a day, if you’d last won on June 1, and haven’t won since, you’d be around 26000% effort. Something somewhere is wrong, but I’ve no idea what.

I’d suggest asking on Discord. Given the advances with compression, it would also be worth replotting, but it may not solve the problem.

21

@DanRelfe In mid August, I started hitting higher multiples of gaps between wins – and it got worse in September. So you are not alone.

However, my farm is smaller than yours, and I am still winning blocks – it just takes longer.

For your farm’s size, you should be winning every day, or every other day. Certainly every few days.
Something had to have changed or broke, for you to have not won since May.

Just taking a guess…
– Could your ISP be blocking crypt-o related traffic?
– Can being synced be considered confirmation that your replies to block challenges are not being tossed by your ISP?
– Can a firewall setting be blocking replies to block challenges?

In May, did you install a new version of Chia?

Before your wins got reduced to zero, did your wins slow down?
In other words, did you go from 1 per day to 1 per every-other-day, or 1 per every-three-days? Was it a slow-down that eventually got to zero? Or did you go from your normal time-to-win average - straight to zero.

22

Anecdotal so take with however much salt that you want:

I have had times where the logs showed that I was submitting proofs and signage points but the pool never received them (thus in a pool and not solo). I tried debugging and never found out what was wrong, deleted the the config.yaml and reconfigured it and suddenly everything was working again. I’ve had this a couple of times and I don’t know why. It’s my first port of call now when I have a problem that I can’t get my head around. This has occured with NFT and Madmax plots on two different nodes (non related plot keys) and not at the same time.

Might be worth trying temporarily joining a pool to see what is being received.

23

Here is a recent block with multiple rewards

24

Agreed, this is a degree of bad luck that has effectively reached 0% chance of happening.

Still think joining the OG pool on foxy is one of the best diagnostics. Just go to their discord and hit up Felix, on how to join. he’s a pretty nice guy who also knows a lot about the chia chain.

1 Like