[Archive] BFL trolling museum

[streblo]

We explicitly want to avoid a difficulty spike, which is why we are wanting to release as many units we can at once.

Can you explain what you mean by 'a difficulty spike'? It seems releasing as many units at once is the epitome of a difficulty spike, insofar as the network hashrate could jump by ca. two orders of magnitude overnight with the difficulty (hopefully) quickly following. Now, as a pool operator you certainty know better than me especially with EMC's recent 700GH/s increase, so let me know your thoughts on this: I imagine this abrupt increase in hashrate could accidentally DDoS pools out of existence, with both bandwidth and CPU bottlenecks. Recall, most people have a few pools as backup for their miners. Hence once one pool fails, the hashrate will jump on another pool, causing it to fail, etc. In this way, failure among pools due to very high load is correlated. With tens of TH/s coming online overnight and with pools at capacity, competing blockchains and poor network communication (remember, relaying blocks and serving getwork requests use the same connection!) may lead to very bad things. Basically, we'd have a situation where the difficulty may be an order of magnitude (or worse) below where it should be.

I think a slow, as-organic-as-possible rise in difficulty would be best for pools, miners, and the network. In other words, ship the units as they're made (according to the 1/3rd schedule).  It will allow time for pools to increase the number of server instances and change share difficulty to, say, 10 or 100. It would allow miners to access reliable pools. And it would be best for the network so we don't have a difficulty which is far-from-equilibrium possibly causing all sorts of weird problems (exacerbated by the pools being overloaded).

As far as demo units go, anything below 1% of the current network hashrate (~170GH/s) is really negligible in terms of their effect on difficulty. In my opinion, yes, send a jalapeno to every developer, I'm sure Gavin, et al., would appreciate it!

[kano]

Who are the people getting them? :-)

I would hope that it would be DrHaribo (bitminter developer), ckolivas (cgminer developer) and luke-jr (bfgminer developer) so that they can tighten up their software to work with ASIC.

Except that wasn't the answer given in this thread

https://bitcointalk.org/index.php?topic=87934.msg1109152#msg1109152

I think he's saying that you're wrong about them needing someone to do the free software development.

Correct.  We do have a little thing that's been in development called EasyMiner.  It's languished on the back burner for a bit whilest we try to get equipment out the door, but it's not been entirely idle and we are going to be focusing more resources on it as we get into gear.  This is one of the reasons I have taken over the customer service and shipping processes, to free up some people who's resources can be better utilized elsewhere.

[cablepair]

in the spirit of doing something to help instead of just bitching about it:

I have created a thread dedicated to the topic of the BFL Shipping plan

https://bitcointalk.org/index.php?topic=102040.0

and I also created a poll which I encourage you all to vote in

maybe it will give BFL the nudge it needs to nail down a fair shipping plan

thanks!

[BFL_Josh]

I am working on coming up with a concrete reply and explanation of the shipping plan which I will post/distribute when I have it ready.

[cablepair]

thanks Josh! can you please post it to this thread:


https://bitcointalk.org/index.php?topic=102040

thanks man!

[SgtSpike]

And be sure to include where converted orders fall in the 1/3's!

[nedbert9]

FIFO order portion will be on the date the order was paid.  If two orders were paid on the same day, the lower order number gets priority.

If we send out demo units, it will only be a few Jalapeno's, which won't amount to more than a few tens of GH, not enough to affect the network hashrate in any appreciable way.  We explicitly want to avoid a difficulty spike, which is why we are wanting to release as many units we can at once, so no one has a marked advantage.  The demo unit business is not set in stone, I am just trying to get a feel for how the majority of the people want to handle verification that we have the hardware we claim to have.  I don't want to get into another situation where people are crying scam or what not.

Ok.  A few JP's are fine IMO, but I hope they'd get in to the hands of the miner dev's.

[The-Real-Link]

  Not sure how common this knowledge was but I was curious about warranties and the like and received the following information:

  BFL products have a warranty of 6 months.  If something dies, just like you would with a GPU, you send it in and they'll swap it for a new one.  The downtime would suck, yes, but at least you get the product fixed.
 
  Should something go south with Bitcoin period (and arguably if so, BFL would also tank as a company), they do *not* offer any returns for any ASIC product to my knowledge.

  Thus the contact person suggested for each purchaser to evaluate their investment into Bitcoin and BFL products accordingly.

[Gabi]

6 months?

Now i understand why they are NOT in Europe, here warranty by law is 2 years 

[AndrewBUD]

^ that is a good thing

[odee2004]

Anyone pre-ordered their Jalapeno product?

Also, does anyone know how much power will it comsume?

Thanks.

[freeAgent]

Anyone pre-ordered their Jalapeno product?

Also, does anyone know how much power will it comsume?

Thanks.

A lot of people have preordered the Jalapeno.  I've got one of them.  As far as power goes, nobody knows yet.  I'd imagine it will be a fairly low-power device.  I hear that it will not draw all its power over USB, though, so it'll be more than 2.5 or 5 watts.

[odee2004]

Anyone pre-ordered their Jalapeno product?

Also, does anyone know how much power will it comsume?

Thanks.

A lot of people have preordered the Jalapeno.  I've got one of them.  As far as power goes, nobody knows yet.  I'd imagine it will be a fairly low-power device.  I hear that it will not draw all its power over USB, though, so it'll be more than 2.5 or 5 watts.

Hmmm.. so no one yet has tested the Jalapeno.

It bothers me though that the Jalapeno with 3.5GH/s is much cheaper compare to the Single with only 835 MH/s.

[AndrewBUD]

Does it bother you that CPU's two years ago were not as fast as CPU's today and they cost more 2 years ago?

[odee2004]

Does it bother you that CPU's two years ago were not as fast as CPU's today and they cost more 2 years ago?

Ohhh.. so you mean that the "Single" was made two years ago?

[abracadabra]

Does it bother you that CPU's two years ago were not as fast as CPU's today and they cost more 2 years ago?

Ohhh.. so you mean that the "Single" was made two years ago?

You cannot relate "bitcoin time" to "real time".
Seeing as bitcoin is still relatively "new", yes, in bitcoin time, BFL singles were about 2 years ago.

[crazyates]

Does it bother you that CPU's two years ago were not as fast as CPU's today and they cost more 2 years ago?

Ohhh.. so you mean that the "Single" was made two years ago?

No, he means the current Single is an FPGA product that has been out for a while. The SC Jalapeno is an ASIC product they are still developing.

[kano]

Does it bother you that CPU's two years ago were not as fast as CPU's today and they cost more 2 years ago?

I'm actually surprised that it doesn't bother people that few will make any profit on their BFL FPGAs and then they must all double their money they give to BFL before they reach that profit mark and start again ... due to BFL themselves.

[imsaguy]

Does it bother you that CPU's two years ago were not as fast as CPU's today and they cost more 2 years ago?

I'm actually surprised that it doesn't bother people that few will make any profit on their BFL FPGAs and then they must all double their money they give to BFL before they reach that profit mark and start again ... due to BFL themselves.

What choice is there?  If a person doesn't do it, someone else will and then they are left behind.

[kano]

Does it bother you that CPU's two years ago were not as fast as CPU's today and they cost more 2 years ago?

I'm actually surprised that it doesn't bother people that few will make any profit on their BFL FPGAs and then they must all double their money they give to BFL before they reach that profit mark and start again ... due to BFL themselves.

What choice is there?  If a person doesn't do it, someone else will and then they are left behind.

There is no choice
Let's hope they can reach a break even on the money they spent on the FPGA and then the ASIC before something happens there also ....