Bitcoin Forum
June 16, 2024, 11:10:46 PM *
News: Latest Bitcoin Core release: 27.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: New Mining Companies  (Read 532 times)
brontosaurus (OP)
Sr. Member
****
Offline Offline

Activity: 441
Merit: 250



View Profile
June 02, 2013, 07:40:14 PM
Last edit: June 02, 2013, 09:14:50 PM by brontosaurus
 #1

Hi all,

Just joined after a look around. Read through some of the posts regarding new mining companies, and this one caught my interest- kncminer.

Had a look on their site, and nearly burst out laughing. Seem to think they can change instantly from 40nm to 28nm technology (in standard cell, no less) and get a prototype out by September;

CRAP

Have a 'special' algorithm that speeds up SHA-256 by 30%. Oh yeah? On a data dependency algorithm? Have they got a time machine?

CRAP

Are actively asking their users if they would like a Litecoin asic when they haven't even delivered a Bitcoin one. The two are not remotely similar. and a Litecoin asic is going to be HUGE due to the memory requirement.

CRAP

One crap is ok, everyone can be optimistic or makes mistakes. But three?

Oh, and their wattage figures are way, way out; for once, too high. For example, their 'Saturn' miner will consume a maximum of 500W. For a 175GH asic based design on 28nm? It's about an order of magnitude too high. And water cooling? Guess what -

ANOTHER CRAP

Perhaps it's all down to the 30% extra hash power from their special algorithm?


***** ADD ON INFO **********


According to KncMiner's site, their 'Mars' box costs $2795 and contains 48 Altera Cyclone IV FPGA's 115k LUT version. That equates to just under $60 each. Are they getting them off the back of a lorry?

Cheapest price for 60 units on Altera's own site is $315 EACH, so even if they're buying them in the 10's of thousands, they are NEVER going to get a price 5 times less than this! For them to make the machine and make a profit they would have to be buying them at less than $30.

And since they are so fond of 28nm technology, why not use the much faster (and cheaper at $209) Cyclone V?

Can anyone else see something wrong with this picture?

BTW, I reckon a properly routed Cyclone V could do 2 -3 GH with just one unit. And no water cooling.
rc0cl
Newbie
*
Offline Offline

Activity: 20
Merit: 0


View Profile
June 03, 2013, 05:00:42 PM
 #2


I understand your cynical take on the yet to be shipped miners.

For the Mars miner, they claimed a performance of 6000 Mh/s for $2795. That works out to be about 2.14 Mhs/s/$; great for FPGA based miners but rather uncompetitive with the ASIC miners (40-50 for larger ASIC rigs). The pricing seems to be set as a lead-in to their Jupiter ASIC miner. That 2795 for a Mars, comes with a voucher good for $2000 towards the purchase of that Jupiter miner.

You have to consider the entire purchase price for both miners at 6995 + 2795 - 2000 = $7790 for a hash rate of 356 Gh/s
MooC Tals
Hero Member
*****
Offline Offline

Activity: 644
Merit: 500


View Profile
June 03, 2013, 05:08:47 PM
 #3

Fuck it anything that will make BFL shit their pants is great big fucken CHECK MARK on my books. Time to break in the champagne.

I will wait and see how they do but I will wait and won't bash at least not yet.

Message to the OP: Good work and keep an eye on them I always enjoy someone like you to keep the business honest. Thank you
WillMilk4Coin
Member
**
Offline Offline

Activity: 83
Merit: 10



View Profile
June 03, 2013, 05:41:10 PM
 #4

....
Are actively asking their users if they would like a Litecoin asic when they haven't even delivered a Bitcoin one. The two are not remotely similar. and a Litecoin asic is going to be HUGE due to the memory requirement.
CRAP
...
Oh, and their wattage figures are way, way out; for once, too high. For example, their 'Saturn' miner will consume a maximum of 500W. For a 175GH asic based design on 28nm? It's about an order of magnitude too high. And water cooling? Guess what -
ANOTHER CRAP
.....
According to KncMiner's site, their 'Mars' box costs $2795 and contains 48 Altera Cyclone IV FPGA's 115k LUT version. That equates to just under $60 each. Are they getting them off the back of a lorry?

Just to comment on a couple of your points;

They didn't say they were making an ASIC for LTC. They would be making an FPGA based (a revision to their Mars product) LTC miner. https://www.kncminer.com/news/news-11

Their statement for power consumption fits in-line with the rest of their announcements as they happen. Originally, the Jupiter was 250Gh/s. Later, they updated and claimed they can reach 350. They've taken the approach of under-promising/over-deliver. That wattage is the maximum they can confirm at this time.

KnC is no longer offering the Mars product, nor do they seem intent on actually delivering any FPGA BTC miner to those that had done a pre-order registration for one. Just today they released a new newsletter with an explanation. https://www.kncminer.com/news/news-12

Essentially, we now have the standard pre-order and pay for an undeveloped ASIC mining device. They've also now opened the order books to pay for those confident enough to give them money.
brontosaurus (OP)
Sr. Member
****
Offline Offline

Activity: 441
Merit: 250



View Profile
June 03, 2013, 08:05:04 PM
 #5

The whole setup with Kncminer is all wrong. Isn't it a little odd that they pull the info from their website on the Mars project today of all days?

How could a bunch of professional engineers make such stupid mistakes as they did with the Cyclone IV 'specifications'?

And let me assure you, in this particular universe we inhabit, you cannot 'improve' the SHA256 algorithm.

If people want to put their money up for this, that's their choice. But first, why not ask kncminer exactly how they are funding this wonderful device?
Simply to get to prototype stage on a multi project wafer with a typical 10mm2 die on 28 nm will cost them about 250k$. That's not including the cost of renting the Cadence or Synopsis tools necessary to get from HDL code through simulation, Layout, DRC and tape out. Lets call it 300k$.

That gets them a handful of die. Of course, if they get enough 'pre orders' they could afford full mask set. Today that'll set you back about $2-3 million for 28nm. That's a shitload of pre orders, and what happens if they make a mistake?

Rather than the 'I want to believe' line, why not try to look at what's actually happening here and start asking some pertinent questions? If you need help in this respect, please ask.


J35st3r
Full Member
***
Offline Offline

Activity: 196
Merit: 100



View Profile
June 03, 2013, 09:02:27 PM
 #6

This has been discussed at huge length on the custom hardware subforum at https://bitcointalk.org/index.php?topic=170332.0
Serve out your newbie time and come and join the party over there.

1Jest66T6Jw1gSVpvYpYLXR6qgnch6QYU1 NumberOfTheBeast ... go on, give it a try Grin
brontosaurus (OP)
Sr. Member
****
Offline Offline

Activity: 441
Merit: 250



View Profile
June 03, 2013, 10:09:00 PM
 #7

Thanks for the invite.

I just hope to hell that people don't order this stuff until some very pertinent questions are answered.
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!