MysteryMiner (OP)
Legendary
Offline
Activity: 1512
Merit: 1049
Death to enemies!
|
![](https://bitcointalk.org/Themes/custom1/images/post/question.gif) |
October 05, 2012, 11:58:06 PM |
|
Simple question, probably not so simple answers and predictions. Let's imagine Butterfly Labs are not going to deliver Bitcoin ASIC at all. The people preordering have lost money or btc. And the block reward is going down to 25BTC/block.
Will the Bitcoin price rise, fall or stay the same?
|
bc1q59y5jp2rrwgxuekc8kjk6s8k2es73uawprre4j
|
|
|
optimator
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 06, 2012, 12:06:37 AM |
|
Essentially what you are asking is if the market has already taken into consideration the increase in difficulty due to the addition of ASIC hardware regarding the BTC to USD exchange rate (I'm assuming you are American....)
You might also want to consider if the market has already taken into consideration the halving of the mining reward. (That is, is the increase in BTC to USD related to ASIC hardware, reward halving, or something else?)
My opinion.. is it doesn't matter what BFL does. It won't affect the BTC to USD exchange rate. Why? Bitcoin proponents are optimists and a single company not delivering won't affect the exchange rate.
|
|
|
|
MysteryMiner (OP)
Legendary
Offline
Activity: 1512
Merit: 1049
Death to enemies!
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 06, 2012, 12:24:21 AM |
|
The increase in difficulty will only increase coin supply in short term until difficulty adjusts. The mass adoption of ASIC will only change who will get most of the newly mined coins. And what the new coin owners do with them. They keep them or sell them. If BFL does not deliver there will be no sudden shift in newly minted coin owners. I dont know how much of the mined coins are sold or hoarded. If most of new coins are sold then reward halving will make the price go upwards. The BFL will have no influence on this. It won't affect the BTC to USD exchange rate. Why? Bitcoin proponents are optimists and a single company not delivering won't affect the exchange rate. My guess are that majority of people buying/selling bitcoins are here either for speculation or using bitcoins as a medium of value transfer (SR). (I'm assuming you are American....) I'm not American at all but I use USD as a reference point. The Euro prices does not differ that much form USD prices.
|
bc1q59y5jp2rrwgxuekc8kjk6s8k2es73uawprre4j
|
|
|
odolvlobo
Legendary
Offline
Activity: 4424
Merit: 3325
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 06, 2012, 02:24:44 AM |
|
The difficulty does not directly affect the value of bitcoins, mostly because the same reward is generated regardless of the difficulty. However, if BFL doesn't come through (which seems unlikely) then there will be some loss in faith in the bitcoin economy, prompting some number of people to sell, and that can have an effect on bitcoin value.
|
Join an anti-signature campaign: Click ignore on the members of signature campaigns. PGP Fingerprint: 6B6BC26599EC24EF7E29A405EAF050539D0B2925 Signing address: 13GAVJo8YaAuenj6keiEykwxWUZ7jMoSLt
|
|
|
optimator
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 06, 2012, 03:51:08 AM |
|
The difficulty does not directly affect the value of bitcoins, mostly because the same reward is generated regardless of the difficulty. However, if BFL doesn't come through (which seems unlikely) then there will be some loss in faith in the bitcoin economy, prompting some number of people to sell, and that can have an effect on bitcoin value.
The increase in difficulty does not affect the supply of bitcoins over time but it may affect the cost to produce bitcoins. How much does it cost to produce a megahash? The cost tomorrow is unknown today. If the market believes the cost will increase the exchange rate should increase or the intrinsic value should increase. If the cost tomorrow can be approximated today then there would be no affect on the value of btc.
|
|
|
|
odolvlobo
Legendary
Offline
Activity: 4424
Merit: 3325
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 06, 2012, 06:49:40 AM |
|
The difficulty does not directly affect the value of bitcoins, mostly because the same reward is generated regardless of the difficulty. However, if BFL doesn't come through (which seems unlikely) then there will be some loss in faith in the bitcoin economy, prompting some number of people to sell, and that can have an effect on bitcoin value.
The increase in difficulty does not affect the supply of bitcoins over time but it may affect the cost to produce bitcoins. How much does it cost to produce a megahash? The cost tomorrow is unknown today. If the market believes the cost will increase the exchange rate should increase or the intrinsic value should increase. If the cost tomorrow can be approximated today then there would be no affect on the value of btc. The cost of mining a bitcoin is a little less than the value of the bitcoin, regardless of the exchange rate (except market fluctuations). There is an equilibrium. If it gets cheaper to mine a bitcoin, there will be more miners, raising the difficulty, and thus returning the cost of mining back to the equilibrium level. If it gets more expensive to mine a bitcoin, then miners will drop out because they can't make a profit. The difficulty will fall, and the cost or mining will drop back to the equilibrium level. When the ASICs come online, they will have incredible hash rates even though they cost no more to operate than current mining hardware. The result is that neither the cost nor the reward will change, even though the difficulty will go up by a factor of 20 or so. Either way, none of these scenarios affect the exchange rate for bitcoins.
|
Join an anti-signature campaign: Click ignore on the members of signature campaigns. PGP Fingerprint: 6B6BC26599EC24EF7E29A405EAF050539D0B2925 Signing address: 13GAVJo8YaAuenj6keiEykwxWUZ7jMoSLt
|
|
|
ElectricMucus
Legendary
Offline
Activity: 1666
Merit: 1057
Marketing manager - GO MP
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 06, 2012, 08:47:23 AM |
|
Any big scam has a bearish influence.
But it is possible it is overshadowed by bullshish influences which we have plenty atm.
|
|
|
|
MysteryMiner (OP)
Legendary
Offline
Activity: 1512
Merit: 1049
Death to enemies!
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 06, 2012, 01:07:05 PM |
|
The Bitcoin price is not tied to difficulty or cost of mining. For some people it always will be free to mine. Some who is mining more expensive than current bitcoin price might hold the coins in hope the price rises in future or sell them below mining costs to lose a little than lose everything.
I think the BFL will not deliver the ASIC at all. Making a working FPGA ir one thing, but ASIC with such specifications is not possible unless they use the latest 22nm fabrication process combined with some other unknown technology. There is no working samples, no update on fabrication progress, nothing. Just preorder and promises too good to be true.
|
bc1q59y5jp2rrwgxuekc8kjk6s8k2es73uawprre4j
|
|
|
ElectricMucus
Legendary
Offline
Activity: 1666
Merit: 1057
Marketing manager - GO MP
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 07, 2012, 04:19:26 PM |
|
I think the BFL will not deliver the ASIC at all. Making a working FPGA ir one thing, but ASIC with such specifications is not possible unless they use the latest 22nm fabrication process combined with some other unknown technology. There is no working samples, no update on fabrication progress, nothing. Just preorder and promises too good to be true.
I still think it's very probable they deliver a FPGA conversion product and greatly underdeliver on hashes/jule and barely meet hashes/usd. They have recuted enough suckers who will defend them and their 'investment' any way possible.
|
|
|
|
alan2here
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 07, 2012, 10:40:39 PM |
|
It's been ages since the product was designed hasn't it, sort of been in the "shipping" stage for months but nobody has any whatsoever.
Doesn't Moores Law, or the relevent part of the law of accelerating returns now make it impossible that the product can be competative, even if they had made some breakthroughs?
|
|
|
|
notme
Legendary
Offline
Activity: 1904
Merit: 1002
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 08, 2012, 04:58:02 PM |
|
It's been ages since the product was designed hasn't it, sort of been in the "shipping" stage for months but nobody has any whatsoever.
Doesn't Moores Law, or the relevent part of the law of accelerating returns now make it impossible that the product can be competative, even if they had made some breakthroughs?
No, design to hardware is always a long process. You don't want to make 10000 wafers with a bug.
|
|
|
|
trogdorjw73
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 08, 2012, 06:37:08 PM |
|
As a "hardware guru" of sorts, I have to agree with the seriously questionable nature of these ASICs. First, they're going to be on what process node? 22nm is definitely out as Intel is the only one doing 22nm, and they don't fab for other people (especially something as small as a Bitcoin ASIC). Realistically, 2xnm and even 3xnm are highly unlikely as they're the state-of-the-art and that usually means harder to work with. So let's go with something reasonable like 40nm from TSMC.
Second: Anyone remember all the issues with NVIDIA and AMD transitioning to 40nm back in the day? And those are big, experienced companies, not some fly-by-night group. NVIDIA was very late to 40nm, and their initial chips (e.g. GTX 480) were underwhelming. It wasn't until the GTX 580 that NVIDIA really got the hang of all the things to consider with TSMC 40nm. AMD meanwhile transitioned to 40nm with a smaller chip first, learned some lessons, and they were prepared for the idiosyncrasies when they created the Evergreen family of GPUs. BFL (and the other ASIC groups) have none of this experience, so they're basically flying blind right now and hoping for the best.
Butterfly Labs may have some awesome computer models of how their ASICs will perform on 40nm (or 32nm or 28nm or whatever), but when they finally tape out and get first silicon I can pretty much guarantee the things won't work. Then they have to debug (which requires some pretty serious hardware if you're encountering chip errors), come up with a revision that will fix the problems, resubmit, and maybe the second revision they can get hardware working. Then the clock speeds get dropped because the chips don't run as fast as they wanted without overheating or other errors, and you end up with maybe half of the promised performance if you're lucky.
Third: Even if everything works out well and BFL (or some other group) can deliver a working ASIC that will do 60 Gh/s at 60W, the amount of money required to design, test, validate, code, etc. such a device will be far higher than what we're seeing in terms of pricing. Several million dollars would be my minimum estimate, and if they sell to people at a 100% markup, they need to likely sell more than $10 million worth of ASICs. At a current pre-order price of $1300, do they really have 8000 orders lined up? From there, those people buying will need quite some time to recoup their investment. Even if pricing stays at $12/BTC, with the block halving they would need to mine every BTC for about eight months just to break even.
TL;DR: ASICs sound like the next big thing, but the claims of 60Gh/s at 60W for $1300 (or less!) are simply too good to be true. Bitcoinica/Zhoutong got out of BTC with how much money "stolen"? BS&T/Pirateat40 took how much BTC/money from investors with his ponzi? I could see BFL topping both of those by the time all is said and done. For ASICs to make sense, I think BTC has to have a much higher price and much higher usage in the real world -- and it would need to cut ties with the anarchistic groups that are currently in love with it and somehow get big investors and banks involved (which is something most of the current users would oppose). ASICs might come eventually, but I'll be shocked to see anyone doing 60Gh/s on a single chip in 2013, let alone in the next three months!
|
|
|
|
Sitarow
Legendary
Offline
Activity: 1792
Merit: 1047
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 08, 2012, 07:24:22 PM |
|
As a "hardware guru" of sorts, I have to agree with the seriously questionable nature of these ASICs. First, they're going to be on what process node? 22nm is definitely out as Intel is the only one doing 22nm, and they don't fab for other people (especially something as small as a Bitcoin ASIC). Realistically, 2xnm and even 3xnm are highly unlikely as they're the state-of-the-art and that usually means harder to work with. So let's go with something reasonable like 40nm from TSMC.
Second: Anyone remember all the issues with NVIDIA and AMD transitioning to 40nm back in the day? And those are big, experienced companies, not some fly-by-night group. NVIDIA was very late to 40nm, and their initial chips (e.g. GTX 480) were underwhelming. It wasn't until the GTX 580 that NVIDIA really got the hang of all the things to consider with TSMC 40nm. AMD meanwhile transitioned to 40nm with a smaller chip first, learned some lessons, and they were prepared for the idiosyncrasies when they created the Evergreen family of GPUs. BFL (and the other ASIC groups) have none of this experience, so they're basically flying blind right now and hoping for the best.
Butterfly Labs may have some awesome computer models of how their ASICs will perform on 40nm (or 32nm or 28nm or whatever), but when they finally tape out and get first silicon I can pretty much guarantee the things won't work. Then they have to debug (which requires some pretty serious hardware if you're encountering chip errors), come up with a revision that will fix the problems, resubmit, and maybe the second revision they can get hardware working. Then the clock speeds get dropped because the chips don't run as fast as they wanted without overheating or other errors, and you end up with maybe half of the promised performance if you're lucky.
Third: Even if everything works out well and BFL (or some other group) can deliver a working ASIC that will do 60 Gh/s at 60W, the amount of money required to design, test, validate, code, etc. such a device will be far higher than what we're seeing in terms of pricing. Several million dollars would be my minimum estimate, and if they sell to people at a 100% markup, they need to likely sell more than $10 million worth of ASICs. At a current pre-order price of $1300, do they really have 8000 orders lined up? From there, those people buying will need quite some time to recoup their investment. Even if pricing stays at $12/BTC, with the block halving they would need to mine every BTC for about eight months just to break even.
TL;DR: ASICs sound like the next big thing, but the claims of 60Gh/s at 60W for $1300 (or less!) are simply too good to be true. Bitcoinica/Zhoutong got out of BTC with how much money "stolen"? BS&T/Pirateat40 took how much BTC/money from investors with his ponzi? I could see BFL topping both of those by the time all is said and done. For ASICs to make sense, I think BTC has to have a much higher price and much higher usage in the real world -- and it would need to cut ties with the anarchistic groups that are currently in love with it and somehow get big investors and banks involved (which is something most of the current users would oppose). ASICs might come eventually, but I'll be shocked to see anyone doing 60Gh/s on a single chip in 2013, let alone in the next three months!
No one said they have ASIC's that do 60Gh/s on a single chip ![Cheesy](https://bitcointalk.org/Smileys/default/cheesy.gif)
|
|
|
|
tbcoin
Legendary
Offline
Activity: 1022
Merit: 1000
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 08, 2012, 07:29:59 PM |
|
As a "hardware guru" of sorts, I have to agree with the seriously questionable nature of these ASICs. First, they're going to be on what process node? 22nm is definitely out as Intel is the only one doing 22nm, and they don't fab for other people (especially something as small as a Bitcoin ASIC). Realistically, 2xnm and even 3xnm are highly unlikely as they're the state-of-the-art and that usually means harder to work with. So let's go with something reasonable like 40nm from TSMC.
Second: Anyone remember all the issues with NVIDIA and AMD transitioning to 40nm back in the day? And those are big, experienced companies, not some fly-by-night group. NVIDIA was very late to 40nm, and their initial chips (e.g. GTX 480) were underwhelming. It wasn't until the GTX 580 that NVIDIA really got the hang of all the things to consider with TSMC 40nm. AMD meanwhile transitioned to 40nm with a smaller chip first, learned some lessons, and they were prepared for the idiosyncrasies when they created the Evergreen family of GPUs. BFL (and the other ASIC groups) have none of this experience, so they're basically flying blind right now and hoping for the best.
Butterfly Labs may have some awesome computer models of how their ASICs will perform on 40nm (or 32nm or 28nm or whatever), but when they finally tape out and get first silicon I can pretty much guarantee the things won't work. Then they have to debug (which requires some pretty serious hardware if you're encountering chip errors), come up with a revision that will fix the problems, resubmit, and maybe the second revision they can get hardware working. Then the clock speeds get dropped because the chips don't run as fast as they wanted without overheating or other errors, and you end up with maybe half of the promised performance if you're lucky.
Third: Even if everything works out well and BFL (or some other group) can deliver a working ASIC that will do 60 Gh/s at 60W, the amount of money required to design, test, validate, code, etc. such a device will be far higher than what we're seeing in terms of pricing. Several million dollars would be my minimum estimate, and if they sell to people at a 100% markup, they need to likely sell more than $10 million worth of ASICs. At a current pre-order price of $1300, do they really have 8000 orders lined up? From there, those people buying will need quite some time to recoup their investment. Even if pricing stays at $12/BTC, with the block halving they would need to mine every BTC for about eight months just to break even.
TL;DR: ASICs sound like the next big thing, but the claims of 60Gh/s at 60W for $1300 (or less!) are simply too good to be true. Bitcoinica/Zhoutong got out of BTC with how much money "stolen"? BS&T/Pirateat40 took how much BTC/money from investors with his ponzi? I could see BFL topping both of those by the time all is said and done. For ASICs to make sense, I think BTC has to have a much higher price and much higher usage in the real world -- and it would need to cut ties with the anarchistic groups that are currently in love with it and somehow get big investors and banks involved (which is something most of the current users would oppose). ASICs might come eventually, but I'll be shocked to see anyone doing 60Gh/s on a single chip in 2013, let alone in the next three months!
No one said they have ASIC's that do 60Gh/s on a single chip ![Cheesy](https://bitcointalk.org/Smileys/default/cheesy.gif) Assumed that Single SC have 8 chips http://codinginmysleep.com/first-look-at-bfls-asic-hardware/
|
|
|
|
trogdorjw73
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 08, 2012, 07:39:53 PM |
|
As a "hardware guru" of sorts, I have to agree with the seriously questionable nature of these ASICs. First, they're going to be on what process node? 22nm is definitely out as Intel is the only one doing 22nm, and they don't fab for other people (especially something as small as a Bitcoin ASIC). Realistically, 2xnm and even 3xnm are highly unlikely as they're the state-of-the-art and that usually means harder to work with. So let's go with something reasonable like 40nm from TSMC.
Second: Anyone remember all the issues with NVIDIA and AMD transitioning to 40nm back in the day? And those are big, experienced companies, not some fly-by-night group. NVIDIA was very late to 40nm, and their initial chips (e.g. GTX 480) were underwhelming. It wasn't until the GTX 580 that NVIDIA really got the hang of all the things to consider with TSMC 40nm. AMD meanwhile transitioned to 40nm with a smaller chip first, learned some lessons, and they were prepared for the idiosyncrasies when they created the Evergreen family of GPUs. BFL (and the other ASIC groups) have none of this experience, so they're basically flying blind right now and hoping for the best.
Butterfly Labs may have some awesome computer models of how their ASICs will perform on 40nm (or 32nm or 28nm or whatever), but when they finally tape out and get first silicon I can pretty much guarantee the things won't work. Then they have to debug (which requires some pretty serious hardware if you're encountering chip errors), come up with a revision that will fix the problems, resubmit, and maybe the second revision they can get hardware working. Then the clock speeds get dropped because the chips don't run as fast as they wanted without overheating or other errors, and you end up with maybe half of the promised performance if you're lucky.
Third: Even if everything works out well and BFL (or some other group) can deliver a working ASIC that will do 60 Gh/s at 60W, the amount of money required to design, test, validate, code, etc. such a device will be far higher than what we're seeing in terms of pricing. Several million dollars would be my minimum estimate, and if they sell to people at a 100% markup, they need to likely sell more than $10 million worth of ASICs. At a current pre-order price of $1300, do they really have 8000 orders lined up? From there, those people buying will need quite some time to recoup their investment. Even if pricing stays at $12/BTC, with the block halving they would need to mine every BTC for about eight months just to break even.
TL;DR: ASICs sound like the next big thing, but the claims of 60Gh/s at 60W for $1300 (or less!) are simply too good to be true. Bitcoinica/Zhoutong got out of BTC with how much money "stolen"? BS&T/Pirateat40 took how much BTC/money from investors with his ponzi? I could see BFL topping both of those by the time all is said and done. For ASICs to make sense, I think BTC has to have a much higher price and much higher usage in the real world -- and it would need to cut ties with the anarchistic groups that are currently in love with it and somehow get big investors and banks involved (which is something most of the current users would oppose). ASICs might come eventually, but I'll be shocked to see anyone doing 60Gh/s on a single chip in 2013, let alone in the next three months!
No one said they have ASIC's that do 60Gh/s on a single chip ![Cheesy](https://bitcointalk.org/Smileys/default/cheesy.gif) Assumed that Single SC have 8 chips http://codinginmysleep.com/first-look-at-bfls-asic-hardware/Okay, so eight ASICs instead of one. That's fine, and it does work better with a 40nm Bulk SOI process, but now that link is showing a computer render of the board and chips. They need to get the board back, get the chips back, and start testing. From software concepts to working hardware is generally NOT a two month process. I'm still saying 60Gh/s at 60W in 2013 will leave me shocked. Even 30Gh/s at 60W would be damn impressive. If they do ship hardware in 2013 (which I'm still not convinced will happen), 15Gh/s at 120W is more in line with what I expect. And that's still basically an order of magnitude speedup over the current fastest GPU mining.
|
|
|
|
IMakeComps
Newbie
Offline
Activity: 36
Merit: 0
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 09, 2012, 05:55:24 AM |
|
Who said they only had two months to design this?
|
|
|
|
capn noe
Member
![*](https://bitcointalk.org/Themes/custom1/images/star.gif)
Offline
Activity: 98
Merit: 10
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 09, 2012, 06:38:57 AM |
|
Pretend for a second that BFL doesn't deliver.
How WOULD that affect the market? I'm curious, really. I assume a lot of pools are counting on them to repay loans for ASIC pre-orders?
|
|
|
|
trogdorjw73
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 09, 2012, 04:33:35 PM Last edit: October 09, 2012, 04:47:35 PM by trogdorjw73 |
|
Who said they only had two months to design this?
Their website says estimated ship date is November/December. Now perhaps my math is faulty here, but that seems like two months or less from now. </sarcasm> If the devices have taped out and they've received initial samples and are well into the debugging and validation process, I sure haven't seen any clear indication of this. There's a wafer shot, sure, but if that's the BFL-SC wafer and it's manufactured on a modern 300mm wafer, those chips are very big! There are around 300 (give or take -- I count about 270 that aren't clipped by the edge) usable chips on that wafer shot ( http://www.butterflylabs.com/wp-content/uploads/2012/05/BFL-SC.jpg), so they would be at least 210mm^2. That puts them at the same size as Intel's typical chips, but then Intel makes and sells millions of chips and we're talking an ASIC that's basically targeted at one thing: Bitcoin. And of course, they're not the only ones out there making these claims. I just find it highly suspicious that there are at least four companies all pursuing this same goal, with launch dates all relatively close together. And of course, why bother with a design that apparently uses more power and delivers lower performance if you're going to come to market four months late (e.g. Deepbit Reclaimer vs. BFL Little Single)? If they ship this year, I'll be happy to eat crow. Heck, if they ship next year and get >50Gh/s at <500W, I'll be surprised. Personally, I'm not going to give $600+ to some place for a pre-order of a completely unproven product.
|
|
|
|
SgtSpike
Legendary
Offline
Activity: 1400
Merit: 1005
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 09, 2012, 05:07:07 PM |
|
Personally, I'm not going to give $600+ to some place for a pre-order of a completely unproven product.
And that's well within your right. The people who have pre-ordered obviously don't mind sending money for an unproven product. So what's your point? Are you trying to warn people? If so, why? Are you just making these posts as a precursor to an "I told you so" post, when the time comes?
|
|
|
|
ElectricMucus
Legendary
Offline
Activity: 1666
Merit: 1057
Marketing manager - GO MP
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
October 09, 2012, 05:26:22 PM |
|
Are you just making these posts as a precursor to an "I told you so" post, when the time comes?
I hope he makes one. Those rare moments of joy are one of the few things left to strive for in the bitcoin forums. Nothing better than the laughs at the expense of those people who were so greedy that they ignored all warnings. I'm very proud of telling people exactly what to expect in the pirateat40 bitcoin savings and thrust thread. The unending stream of butthurt... priceless ![Grin](https://bitcointalk.org/Smileys/default/grin.gif)
|
|
|
|
|