Butterfly Labs is going to give lifetime warranty

[Jack1Rip1BurnIt]

The title says it all. Our favorite magazine talks about it here
http://bitcoinmagazine.net/butterfly-labs-offers-lifetime-warranty-asic-competition-surges-on/

[1715654587]

1715654587

[1715654587]

1715654587

[1715654587]

1715654587

[SkRRJyTC]

The title says it all. Our favorite magazine talks about it here
http://bitcoinmagazine.net/butterfly-labs-offers-lifetime-warranty-asic-competition-surges-on/

Why wouldnt they announce this sooner?  Why wait until now to tell customers they will be getting a warranty?  Did BFL doubt that they could offer this warranty at some point?

[Jack1Rip1BurnIt]

It's all about the hype. You can't spend it all at once. You have to give in doses to keep people medicated.

[ice_chill]

I will believe when I hear it from official BFL representative.

[d3c0n808]

Good to know they intend to stand by their product.  Hopefully this rumor turns out to be true. 

[reeses]

I do not think 'diaspora' means what you think it means.

[bobitza]

I will believe when I hear it from official BFL representative.

I heard about it from the Announcements area in the customers portal. I think that's where the magazine picked it from.

[KIDC]

I will believe when I hear it from official BFL representative.

Confirmed: https://forums.butterflylabs.com/content.php/121-ASIC-Products-will-carry-a-lifetime-warranty

[ice_chill]

I imagine if it fails in 5 years time, then they will more likely replace the unit of a newer model of similar speed than repair the old one ?

[superfastkyle]

A reputable company that offers a lifetime warranty, does it because the benefits outweigh the costs. It prevents any unaccounted for loss by making as many loopholes in warranty terms needed, ie not transferable, mandatory registration, warranty seals etc. That's just the way business works. If you have been mining for a while you are likely to outweigh cost/benefit all the time. Some people pick newer 7000 cards for warranty. Others will pay less for a 4 year old 5800 card. Reference designs often take a premium as they seem more reliable. BFL has no way of estimated said costs yet, no way of making any educated guesses the product doesn't exist. Another BFL lie

[BFL_Josh]

There is no loophole.  The warranty follows the board, so you can sell it and the warranty is still intact.  Thus you don't need to register it... because we aren't warrantying the fan (at least not past 6 months or a year), there aren't really any seals we are going to be placing on the unit, either. 

[superfastkyle]

My point was is no company can be expected to fulfill a warranty service they have not done a risk calculation on. The only computer hardware company I have ever heard of selling a lifetime warranty is xfx, and this year they decided it was not sustainable without having a non-transferable clause. You think you can make a device more reliably than xfx? Or maybe you don't and you think oh well asic devices only cost us a few dollars to make. I remember another famous company that went this route, Oakley with their sunglasses. They also have discontinued their lifetime warranty.

[markm]

It does provide a use for all the trade-ins, maybe.

If they are going to get stuck with a bunch of old versions as people upgrade, they could use the traded in ones as replacement units for people who ship in an old model on warranty.

-MarkM-

[YokoToriyama]

sounds to good to be true

[ice_chill]

All good memory makers give lifetime warranty.

[superfastkyle]

Maybe I'm missing something but these are my thoughts. I'm not a chip manufacturer or a electrical engineer, but I have built computer systems for almost 15 years. Your 3d model doesn't look right to me, it looks overly complicated. Here are my concerns, I would love you to prove me wrong.

It seems that each of the 8 asic chips has their own 3 or 4 phase vrm. Why would this be necessary? or even preferred? it seems like it would cut down on reliability and raise cost. If a 4 phase vrm can handle a 5870 chip at 150 watts why can't it handle all 8 asic chips at less than 100w

Why doesn't it look more like a video card anyways? Shouldn't the majority of the difference be less RAM, a usb port added and maybe a microprocessor to split the work among the 8 chips

what are the square blocks in the upper right? someone suggested capacitors. if so, why are there so many and such a big size. Its not like we are starting a big motor here.

[superfastkyle]

have you ever had memory fail? I thought I did once but it was incorrectly inserted, lol.

All good memory makers give lifetime warranty.

[Jack1Rip1BurnIt]

Wait, are there people mad now that they have found out if they buy an ASIC from BFL the product will be covered for life? I'm asking this question before I jump to conclusions. Oh btw Visiontek also has lifetime warranties... although I have never owned a product from that company. XFX is the shit! Just sayin'

[P_Shep]

Maybe I'm missing something but these are my thoughts. I'm not a chip manufacturer or a electrical engineer, but I have built computer systems for almost 15 years. Your 3d model doesn't look right to me, it looks overly complicated. Here are my concerns, I would love you to prove me wrong.

It seems that each of the 8 asic chips has their own 3 or 4 phase vrm. Why would this be necessary? or even preferred? it seems like it would cut down on reliability and raise cost. If a 4 phase vrm can handle a 5870 chip at 150 watts why can't it handle all 8 asic chips at less than 100w

Why doesn't it look more like a video card anyways? Shouldn't the majority of the difference be less RAM, a usb port added and maybe a microprocessor to split the work among the 8 chips

what are the square blocks in the upper right? someone suggested capacitors. if so, why are there so many and such a big size. Its not like we are starting a big motor here.

Please put the rendering down and step away.

[superfastkyle]

Please put the rendering down and step away.

I want someone to answer my questions seriously, I'm not afraid to learn. I don't have as big of a ego as some people on here. Someone who has experience or someone at BFL labs explain what we are looking at?

[bustaballs]

have you ever had memory fail? I thought I did once but it was incorrectly inserted, lol.

All good memory makers give lifetime warranty.

Memory has one of the highest failure rates of any computer component. If there's a hardware issue with a computer it almost always has to do with the ram or hard drive.

[Cranky4u]

As an Engineer, I ask myself what is definition of BFLs "life time warranty"? Is it;
1. The calculated operational lifespan of the circuit board including solder joints, fatigue rates, etc and comes out to 18 months?
2. Is it the industry standard of 25 years = life time?
3. Is it the mean lifetime remaining of the purchaser? i.e. Purchaser is caucasian male with life expectancy of 80 years and he buys it when he is 40 years old, therefore the lifetime warranty is 40 years?

What does the forum think?

More importantly, what does BFL think?

[P_Shep]

It's your accusative tone. You don't know what you're talking about, yet you're still postulating that it's all wrong.

It's just a space model, probably a preliminary one. So you can't regard what you see there as accurate.
What appear big capacitor shapes, may turn out to be smaller caps. Power supplies all need big caps.
Compare it to the pics of the FPGA single. The designs are similar.
Why so many VRMs? Are they even VRMs? Will there be that many in the final design? Perhaps many smaller power VRMs are cheaper than a couple of big ones. 60W is 50A @1.2v. That's pretty hefty.

[SLok]

As an Engineer, I ask myself what is definition of BFLs "life time warranty"? Is it;
1. The calculated operational lifespan of the circuit board including solder joints, fatigue rates, etc and comes out to 18 months?
2. Is it the industry standard of 25 years = life time?
3. Is it the mean lifetime remaining of the purchaser? i.e. Purchaser is caucasian male with life expectancy of 80 years and he buys it when he is 40 years old, therefore the lifetime warranty is 40 years?

What does the forum think?

More importantly, what does BFL think?

Lifetime warranty on a 3 year old OCZ dual-channel memory set was 30% purchase price refund when one of the sticks started to fail. So lifetime seems flexible. In the European Union there's a standard 2 year minimum on electric devices like laptops, tv's etc. Even when failure kicks in at 30 months, there would still be a chance of complete covering of repairs. It depends on the general life expectancy of the device. For a tv, one would expect 6-10 years, so repairs/replacements should be given/priced accordingly.
I would say 36 months for a static device like an SC would be appropriate?

[2112]

It seems that each of the 8 asic chips has their own 3 or 4 phase vrm. Why would this be necessary? or even preferred? it seems like it would cut down on reliability and raise cost. If a 4 phase vrm can handle a 5870 chip at 150 watts why can't it handle all 8 asic chips at less than 100w

Au contraire. Multiple smaller regulators increase reliability and increase overall device yield. The hashing chips are http://en.wikipedia.org/wiki/Embarrassingly_parallel .

If one (or more) of the many regulators fails the whole device continues to work at a reduced speed.

And vice versa: the device can be populated with partially defective hashing chips. If the testing discovers a fault in some section of some hashing chip that section could be simply powered down and the whole device will remain useable.

It is quite possible that in order to save money and time the individual hashing chips are only minimally tested before packaging and soldering into the board. The full tests are run only in the final shipping device.

Finally, if the regulators are software-controlled this configuration would allow individual under/over-volting of the individual sections within individual chips.

The comparison to the graphic card is simply invalid because the Bitcoin hasher uses no RAM whatsoever and does not require any high bandwidth bus interface like AGP/PCI/PCIe. It also has no high bandwidth output to the video monitor.

[crazyates]

have you ever had memory fail? I thought I did once but it was incorrectly inserted, lol.

All good memory makers give lifetime warranty.

Memory has one of the highest failure rates of any computer component. If there's a hardware issue with a computer it almost always has to do with the ram or hard drive.

Umm I don't think so. Working the past 2 1/2 years as I data recovery technician, I'll give you the dead hard drives. I see a lot of those, but that's kind of our focus. However, I've only seen a handful of cases where the memory was bad, and they were usually old sticks(10+ years), damaged through ESD, or both.

You wanna talk about failing components? Assuming we're not talking about DOA, I'd be willing to bet budging caps, dead hard drives, faulty PSUs, or OCZ SSDs all have higher failure rates than any RAM chip.

[Syke]

A lifetime warranty on a mining-only device is a nice feature, but likely the lifetime of such a product is about 2 years for most people. By then it's unlikely to be profitable to continue mining on it.

[bustaballs]

have you ever had memory fail? I thought I did once but it was incorrectly inserted, lol.

All good memory makers give lifetime warranty.

Memory has one of the highest failure rates of any computer component. If there's a hardware issue with a computer it almost always has to do with the ram or hard drive.

Umm I don't think so. Working the past 2 1/2 years as I data recovery technician, I'll give you the dead hard drives. I see a lot of those, but that's kind of our focus. However, I've only seen a handful of cases where the memory was bad, and they were usually old sticks(10+ years), damaged through ESD, or both.

You wanna talk about failing components? Assuming we're not talking about DOA, I'd be willing to bet budging caps, dead hard drives, faulty PSUs, or OCZ SSDs all have higher failure rates than any RAM chip.

I've worked for the past 7 years as a computer repair technician and 25%~ of hardware failures that I've worked on have been related to faulty memory. I've run into faulty memory on my own machines a couple of times as well. There's a damn good reason why memtest86 is so popular and why RAM diagnostic tools are often built into operating system boot discs and even the bios of many motherboards. RAM fails.. a lot.

You're right about the power supplies. That also have high failure rates.

[iongchun]

A lifetime warranty on a mining-only device is a nice feature, but likely the lifetime of such a product is about 2 years for most people. By then it's unlikely to be profitable to continue mining on it.

I absolutely agree with you, plus it brings good PR to BFL.
Like RAM modules or SD cards, customers lost interest to old products quite soon that life warranty is not a burden to vendors,
unless the failure rate is too high in the short term.

[MrTeal]

I do not think 'diaspora' means what you think it means.

That, and some of their fact checking could use some work.

Finally, DeepBit has released a line of ASICs of their own. DeepBit is well known for having been, at one point, the largest Bitcoin mining pool. DeepBit even exceeded 50% of the network’s total hash power for a few brief moments in 2011, although the mining pool has shrunk considerably in recent months, and is now not even close to first place[2].

[squeept]

have you ever had memory fail? I thought I did once but it was incorrectly inserted, lol.

All good memory makers give lifetime warranty.

Memory has one of the highest failure rates of any computer component. If there's a hardware issue with a computer it almost always has to do with the ram or hard drive.

Umm I don't think so. Working the past 2 1/2 years as I data recovery technician, I'll give you the dead hard drives. I see a lot of those, but that's kind of our focus. However, I've only seen a handful of cases where the memory was bad, and they were usually old sticks(10+ years), damaged through ESD, or both.

You wanna talk about failing components? Assuming we're not talking about DOA, I'd be willing to bet budging caps, dead hard drives, faulty PSUs, or OCZ SSDs all have higher failure rates than any RAM chip.

I've worked for the past 7 years as a computer repair technician and 25%~ of hardware failures that I've worked on have been related to faulty memory. I've run into faulty memory on my own machines a couple of times as well. There's a damn good reason why memtest86 is so popular and why RAM diagnostic tools are often built into operating system boot discs and even the bios of many motherboards. RAM fails.. a lot.

You're right about the power supplies. That also have high failure rates.

20+ years in electronics and computer related fields here: I've seen like 2 sticks of dead RAM ever. I'd say we're both outliers of the real data.

Heat and moving parts are the usual culprits. So I'd say hard drives and power supplies as well. I wouldn't say capacitors are generally likely to fail, but if you specify electrolytic caps, that's a whole different story.

Whenever I read a review on Newegg that says an item was DOA, I laugh because they almost certainly broke it themselves.

[mdude77]

sounds to good to be true

I had a lifetime warranty once on a stereo I bought a long time ago.  The stereo still works, but the company that sold it is long gone.

My point is, lifetime could mean lifetime of the company, not the device.

M

[mdude77]

I've worked for the past 7 years as a computer repair technician and 25%~ of hardware failures that I've worked on have been related to faulty memory. I've run into faulty memory on my own machines a couple of times as well. There's a damn good reason why memtest86 is so popular and why RAM diagnostic tools are often built into operating system boot discs and even the bios of many motherboards. RAM fails.. a lot.

You're right about the power supplies. That also have high failure rates.

20+ years in electronics and computer related fields here: I've seen like 2 sticks of dead RAM ever. I'd say we're both outliers of the real data.

Heat and moving parts are the usual culprits. So I'd say hard drives and power supplies as well. I wouldn't say capacitors are generally likely to fail, but if you specify electrolytic caps, that's a whole different story.

Whenever I read a review on Newegg that says an item was DOA, I laugh because they almost certainly broke it themselves.

I've been using computers for 20 years.  Maybe once did I have memory fail.  I'm pretty sure I had memory failure because of a lightning strike as well.

Hard drives, well, I've replaced a good number through warranty, and it wasn't "lifetime".  There's a reason HDD manufacturers give 3-5 years and memory manufacturers give lifetime.  As you said above, those things with moving parts go first.  Think fans, power supplies, HDDs. 

M

[Mobius]

As an Engineer, I ask myself what is definition of BFLs "life time warranty"? Is it;
1. The calculated operational lifespan of the circuit board including solder joints, fatigue rates, etc and comes out to 18 months?
2. Is it the industry standard of 25 years = life time?
3. Is it the mean lifetime remaining of the purchaser? i.e. Purchaser is caucasian male with life expectancy of 80 years and he buys it when he is 40 years old, therefore the lifetime warranty is 40 years?

What does the forum think?

More importantly, what does BFL think?

Like many statements from BFL - sketchy, suspicious and undefined.

[elux]

Okay, let's think about this for a minute...

What other highly complicated electronic device, marketed to the consumer gets sold with a free >5 year warranty?

This is BFL's first generation ASIC design. Not some small change to an existing, proven, reliable design.
It's an untested product. Not one device has been shipped to customer yet.

And they're suddently offering a unlimited "lifetime" warranty at no added cost.

They have no data on failure rates. What is the MTBF of the BFL SCs?
What is the rate of defective products from manufacturing? Is it 1/10 or 1/100000?
Who knows? Nobody knows. Because this data cannot exist yet.

The thing is, hardware design is hard, to the point that even Intel fucks up badly from time to time,
to the point that they have to issue total and hence extremely expensive recalls.

Any number of things can happen. Especially with a brand new design.
Especially when rolling out four products simultaneously. Especially for a 22 person company.

Bugs can be subtle, yet fatal. Components may not play along nicely.
Problems can take time to surface. Problems aren't always detected in testing.

Does BFL have the appropriate insurance in place in case something bad happens? (How much would that cost?)

At first glance, I fail to see see how this makes any economic sense.
(Except of course from a pre-order / marketing / sales perspective.)

Here's a final question, left simply as an exercise to the reader:
What would the world have to look like for this "risk free gift" to make sense?

[bitmar]

LOL    BFL is the first company in history which gives a lifetime warranty on product that does not exist and has not been tested. This is a very impressive....  

Compared to the BFL companies such as Intel, Nvidia, AMD looks pathetic

[firefop]

Maybe I'm missing something but these are my thoughts. I'm not a chip manufacturer or a electrical engineer, but I have built computer systems for almost 15 years. Your 3d model doesn't look right to me, it looks overly complicated. Here are my concerns, I would love you to prove me wrong.

It seems that each of the 8 asic chips has their own 3 or 4 phase vrm. Why would this be necessary? or even preferred? it seems like it would cut down on reliability and raise cost. If a 4 phase vrm can handle a 5870 chip at 150 watts why can't it handle all 8 asic chips at less than 100w

Why doesn't it look more like a video card anyways? Shouldn't the majority of the difference be less RAM, a usb port added and maybe a microprocessor to split the work among the 8 chips

what are the square blocks in the upper right? someone suggested capacitors. if so, why are there so many and such a big size. Its not like we are starting a big motor here.

On VRMs - this is the primary point of failure on *most* hardware, it might raise costs slightly to have multiples - but I think it's going to make for a better product in the long run - imagine 1 vrm fails, the others should still function only reducing the hashing by a percentage. Suppose one of 8 chips attached to a VRM fails, you'll probably end up frying the other chips.

It seems like they took the more expensive but more reliable route, started with a small single unit and added more of that module to the board.

[salfter]

All good memory makers give lifetime warranty.

have you ever had memory fail?

Yes...in fact, I just put in for an RMA with Crucial yesterday. Tried putting a stick I had laying around in my mining rig the other day. It locked up when I tried starting bitcoind, and kept either locking up or rebooting on fsck. Took it back out, fsck completed, and it's been OK since.

Before that, some memory in my work computer crapped out and needed replacement.

(Also, top-posting's bad, m'kay?)

[ice_chill]

sounds to good to be true

I had a lifetime warranty once on a stereo I bought a long time ago.  The stereo still works, but the company that sold it is long gone.

My point is, lifetime could mean lifetime of the company, not the device.

M

AMEN.

[defxor]

Electronics follow the bathtub failure model. Either they fail immediately, or after a very very long time. If you test the stuff before shipping you can be pretty much sure about the mean time before failure (mtbf).

Some electronics include moving parts. Those can fail pretty much at any time.

Some electronics are based on chemistry, like electrolyte caps. Those dry out in decades.

Some electronics have wear, think cells in an SSD with limited (although still in the thousands and tens of thousands) number of write cycles.

Some electronics are basically just fixed gates. Like ASICs. They don't wear, and the likelihood of one failing is incredibly small.

/defxor

Anecdote: At home I have an 1987 Atari ST, including a SM124 monochrome monitor and an RLL MFM 30GB hard disk in working condition. The picture quality is slightly unstable, most likely due to the caps in the PSU needing to be exchanged.

[superfastkyle]

Warranties/support can also put a company out of business. If anyone remembers Quantex/Cybermax computers in the early 2000's. They had good products at a great price, won lots of awards. Their supplier went out of business and then so did they. Leaving many people without support and with broken equipment

sounds to good to be true

I had a lifetime warranty once on a stereo I bought a long time ago.  The stereo still works, but the company that sold it is long gone.

My point is, lifetime could mean lifetime of the company, not the device.

M

[Syke]

Some electronics are basically just fixed gates. Like ASICs. They don't wear, and the likelihood of one failing is incredibly small.

/defxor

Anecdote: At home I have an 1987 Atari ST, including a SM124 monochrome monitor and an RLL MFM 30GB hard disk in working condition. The picture quality is slightly unstable, most likely due to the caps in the PSU needing to be exchanged.

Mining wreaks havoc on gates, whether it's a CPU, GPU, FPGA, or ASIC. Mining absolutely hammers it, 24/7/365. Mining will wear out any device far faster than normal usage.

[ice_chill]

Some electronics are basically just fixed gates. Like ASICs. They don't wear, and the likelihood of one failing is incredibly small.

/defxor

Anecdote: At home I have an 1987 Atari ST, including a SM124 monochrome monitor and an RLL MFM 30GB hard disk in working condition. The picture quality is slightly unstable, most likely due to the caps in the PSU needing to be exchanged.

Mining wreaks havoc on gates, whether it's a CPU, GPU, FPGA, or ASIC. Mining absolutely hammers it, 24/7/365. Mining will wear out any device far faster than normal usage.

Actually if you know how hardware functions, a 1GHz CPU will always switch the gates at that frequency when the PC is powered on, once the operating system is loaded it will send a continuous off command to stop the switching of the unused gates when there is nothing to process. So really it makes no difference whether the CPU is processing or is in pre-OS state and switching anyway.

Excess heat and current is what degrades the silicon.

[firefop]

Only heat and specifically, fluctuations in heat, given clean power.

Exactly - given proper care and feeding --- first gen asics should produce until they are ready to be replaced by 4th gen asics. No problem at all.

Especially if they went wish solidstate caps - as the render seems to indicate.

[YokoToriyama]

nice gotta love bFL

[firefop]

It's a good thing - exactly what BFL needed to keep the big money interested. This combined with getting a nice enclosure on the sc-rig... tips it away from other producers. At least for me.

[mrb]

I've worked for the past 7 years as a computer repair technician and 25%~ of hardware failures that I've worked on have been related to faulty memory. I've run into faulty memory on my own machines a couple of times as well. There's a damn good reason why memtest86 is so popular and why RAM diagnostic tools are often built into operating system boot discs and even the bios of many motherboards. RAM fails.. a lot.

20+ years in electronics and computer related fields here: I've seen like 2 sticks of dead RAM ever. I'd say we're both outliers of the real data.

I am not saying the following applies to you, but most people think RAM failures are rare because they are often unable to identify them. When they happen, they cause all sorts of random errors: BSODs, SIGSEGVs, file corruption, etc. The OS is not going to display an error message saying "RAM failure" (even if you use ECC RAM, ECC is not always able to detect multi-bit errors.) And even when you run memtest86, the errors it reports are not always caused by failing RAM; sometimes they are caused by an overheating CPU, faulty motherboard, bad PSU, etc.

It is only in rare cases that RAM failures are easy to identify (eg. when the computer won't even POST with bad RAM). These are the only cases that people will usually identify them as such.

[mrb]

You people need to realize that it is an easy commitment to make to offer "lifetime warranty" for Bitcoin mining ASIC hardware... The reason? In 3-4 years tops, nobody will be interested in running the Butterfly Labs SC ASICs anymore, let alone have them be repaired if they break.

Why is that? Because eitheir Bitcoin will have failed (unlikely IMHO), or Bitcoin will be even more popular, therefore the ASICs will almost surely have already been made obsolete by other next-generation ASICs, more profitable, more power-efficient, maybe 45nm or even 32nm.

[Jack1Rip1BurnIt]

You people need to realize that it is an easy commitment to make to offer "lifetime warranty" for Bitcoin mining ASIC hardware... The reason? In 3-4 years tops, nobody will be interested in running the Butterfly Labs SC ASICs anymore, let alone have them be repaired if they break.

Why is that? Because eitheir Bitcoin will have failed (unlikely IMHO), or Bitcoin will be even more popular, therefore the ASICs will almost surely have already been made obsolete by other next-generation ASICs, more profitable, more power-efficient, maybe 45nm or even 32nm.

+This

Awesome is what I say! If I buy their product I never have to worry about exactly what day I bought it and when the product will lose its warranty. Even if I only use the product for 2 or 3 years I still will not ever worry about it. (Peace of mind = Hysteria in the bitcointalk.org community)

[jojo69]

There is no loophole.  The warranty follows the board, so you can sell it and the warranty is still intact.  Thus you don't need to register it... because we aren't warrantying the fan (at least not past 6 months or a year), there aren't really any seals we are going to be placing on the unit, either. 

Josh,

Have you all discussed a policy regarding customer removal/modification/replacement of the heat sink?

[ice_chill]

Definitely not in 3 years, because ASIC can only be replaced by ASIC so we are likely to see similar speed increases as in GPU to GPU upgrade.

[mrb]

Definitely not in 3 years, because ASIC can only be replaced by ASIC so we are likely to see similar speed increases as in GPU to GPU upgrade.

...and the speed increase of GPUs over 4 years is sufficient to push 4-year-old GPUs out of the market. Therefore nobody will be mining with 4-year-old ASICs.

[Tinua]

What is meant by lifetime warranty?
They sold the SC for Bitcoin mining.
If a code change come in the future, this SC would not mining anymore and would so be a warranty claim? 

[bobitza]

If a code change come in the future, this SC would not mining anymore and would so be a warranty claim? 

I think you're mistaken the Warranty program with the Customer Upgrades program.

[ice_chill]

Definitely not in 3 years, because ASIC can only be replaced by ASIC so we are likely to see similar speed increases as in GPU to GPU upgrade.

...and the speed increase of GPUs over 4 years is sufficient to push 4-year-old GPUs out of the market. Therefore nobody will be mining with 4-year-old ASICs.

Well if we take the GPUs for their main purpose which is gaming and not mining, then a 4 years old GPU is HD4870, while the HD7970 is much better, the HD4870 is by no means obsolete.

A 4 year old CPU is Core 2 Quad 2.6Ghz, while not as good as the Core i7 3770k, it is by no means obsolete.

With ASIC it is different as you are not judging by frames per second, but rather by the money it makes, so they will run for even longer.

What is meant by lifetime warranty?
They sold the SC for Bitcoin mining.
If a code change come in the future, this SC would not mining anymore and would so be a warranty claim? 

Sure, if only we lived in LALA LAND

[Tinua]

What is meant by lifetime warranty?
They sold the SC for Bitcoin mining.
If a code change come in the future, this SC would not mining anymore and would so be a warranty claim? 

Sure, if only we lived in LALA LAND

was only meant in ironic way! 
But, if you think more deep....the device do not work anymore correctly and do what they have to do, bitcoin mining.......Summary...... 

[mrb]

...and the speed increase of GPUs over 4 years is sufficient to push 4-year-old GPUs out of the market. Therefore nobody will be mining with 4-year-old ASICs.

Well if we take the GPUs for their main purpose which is gaming and not mining, then a 4 years old GPU is HD4870, while the HD7970 is much better, the HD4870 is by no means obsolete.

My point is the 4870 is obsolete for mining:

Its revenue: 80 (Mhash/sec) * 3600 (sec/hour) * 730 (hour/month) / (2**32*3054e3 (difficulty factor)) * 50 (BTC/block) * 11.7 (USD/BTC) = $9.40 per month
Its power cost: 0.150 (kW) * 730 (hour/month) * 0.11 (worldwide average $/kWh) = $12.0 per month

So most of the world already loses money mining with a 4870. Even those with insanely low rates, say $0.04/kWh, will be unable to make any profit 2 months from now when the reward drops to 25 BTC/block.

Yes gaming hardware takes longer to become truly obsolete, because graphics quality is not that important to some gamers. But in the world of Bitcoin mining, being profitable or not determines if your hardware is obsolete or not.

[firefop]

...and the speed increase of GPUs over 4 years is sufficient to push 4-year-old GPUs out of the market. Therefore nobody will be mining with 4-year-old ASICs.

Well if we take the GPUs for their main purpose which is gaming and not mining, then a 4 years old GPU is HD4870, while the HD7970 is much better, the HD4870 is by no means obsolete.

My point is the 4870 is obsolete for mining:

Its revenue: 80 (Mhash/sec) * 3600 (sec/hour) * 730 (hour/month) / (2**32*3054e3 (difficulty factor)) * 50 (BTC/block) * 11.7 (USD/BTC) = $9.40 per month
Its power cost: 0.150 (kW) * 730 (hour/month) * 0.11 (worldwide average $/kWh) = $12.0 per month

So most of the world already loses money mining with a 4870. Even those with insanely low rates, say $0.04/kWh, will be unable to make any profit 2 months from now when the reward drops to 25 BTC/block.

Yes gaming hardware takes longer to become truly obsolete, because graphics quality is not that important to some gamers. But in the world of Bitcoin mining, being profitable or not determines if your hardware is obsolete or not.

Also - 4870 is obsolete for games now. I had a pair of 2gb vapor-x cooled version (only 4870 made with 2gb) - I replaced them with a single 5770 - which out performed them. so yes, in computer technology, 3rd generation is end of life. 4th is a doorstop.

[The-Real-Link]

It's a good comparative theory to equate GPUs but then again the purpose of the GPU for most people is driving multimedia and games, not mining BTC.  Thus, the standards radically jump year after year.

ASICs can certainly become smaller or more efficient but since that's ALL they do, as others said, I can't see a 4th Gen ASIC making the first gen worthless.  A year might go by for the second gen to come out but they could only increase performance by ~5%.  Now while that little jump might make sense to changeover by the 4th gen, sure, who wouldn't want to do that?  But it hardly makes the first gen completely obselete.

[mrb]

The-Real-Link: CPUs, GPUs, mining ASICs, etc, all follow Moore's Law. The number of transistors doubles every 18 months, hence doubling the performance for an embarrassingly parallel algorithm such as Bitcoin mining.

[Bogart]

These ASIC mining offerings are 1st-gen designs, rushed to market, because time is money in this game.

I'm sure there is lots of room for improvement in subsequent generations.  Process size is the most obvious area.  All of the makers who have stated their feature size are using 130nm or 110nm.

There is likely also room for improvement in the SHA256 IP cores themselves, and in their management by the supervisory mining logic.

It's not so much about the Gh/$ or Gh/chip as it is about Gh/J.

[Soros Shorts]

So do people think some day we can buy SHA-256 IP cores from anybody, kind of like the 8051?

[mrb]

It is already possible to buy SHA256 IP cores. But none of them on the market are optimized for mining, hence why Bitcoin mining vendors presumably developed their owns.

[scrybe]

have you ever had memory fail? I thought I did once but it was incorrectly inserted, lol.

All good memory makers give lifetime warranty.

Yes, I've lost about 1% of the RAM I buy for our data center, sometimes it just stops or starts corrupting data and has to be replaced. At home I've had a few sticks fail over the years, but the last one was 1st generation BGA PC150, so I've been luckier recently. 

FYI, if you have random crashes that don't trace back to drivers, try some different RAM, you might be surprised.

[Syke]

There is no loophole.

Of course there's a loophole. Here it is:

https://forums.butterflylabs.com/showwiki.php?title=FAQ:What+is+the+warranty+period+on+your+devices&highlight=lifetime

Lifetime is defined as the length of time we are manufacturing that product.

https://forums.butterflylabs.com/showwiki.php?title=FAQ:What+is+the+warranty+period+on+your+devices&do=comments

Once we stop producing a product, the warranty will no longer apply

World's shortest "lifetime warranty". You'll get one year out of it, tops.

[SgtSpike]

Figures... how come lifetime warranties never actually last a lifetime?  They should be called product life warranties.

[greyhawk]

Figures... how come lifetime warranties never actually last a lifetime?  They should be called product life warranties.

But then people would know what's up and no one would be fooled.

[crazyates]

Figures... how come lifetime warranties never actually last a lifetime?  They should be called product life warranties.

Lifetime of the product, not your lifetime.

[bcpokey]

Figures... how come lifetime warranties never actually last a lifetime?  They should be called product life warranties.

Lifetime of the product, not your lifetime.

Lifetime of the product line. Lifetime of the product would actually be an even smarter and more evil Warranty...

"Hey my product broke but it's under lifetime warranty so please replace it"

"Sorry Sir, lifetime warranty means the lifetime of the product, since your product is now technically dead, that means the warranty has expired"

Regardless we should stop using Lifetime warranties, as it is misleading, doubly so in an industry that will have ridiculously short product line lifetimes.

[Bogart]

"Sorry Sir, lifetime warranty means the lifetime of the product, since your product is now technically dead, that means the warranty has expired"

"aaaand it`s gone! this line is for people with the money only please step aside!"

[SgtSpike]

Figures... how come lifetime warranties never actually last a lifetime?  They should be called product life warranties.

Lifetime of the product, not your lifetime.

Bowtech.  Thus far, the ONLY company I know of with TRUE lifetime warranties.  Bought a bow in 2001 that the limbs broke?  Well here, we'll either fix it with limbs we still have in stock, or we'll give you a brand new bow of our latest and greatest!

That's how a lifetime warranty should be.

[crazyates]

Figures... how come lifetime warranties never actually last a lifetime?  They should be called product life warranties.

Lifetime of the product, not your lifetime.

Bowtech.  Thus far, the ONLY company I know of with TRUE lifetime warranties.  Bought a bow in 2001 that the limbs broke?  Well here, we'll either fix it with limbs we still have in stock, or we'll give you a brand new bow of our latest and greatest!

That's how a lifetime warranty should be.

Seems to me that all sorts of hunting gear is like that. My friend has got a pair of LL Bean hiking boots that he's in all the time. He brings them in when they're worn, ripped, anything, and they replace 'em. My dad's got a Buck knife with a true lifetime warranty - he's had it for 15 years that he keeps sending in to get sharpened, new hinge, new release, etc.

Idk why, but electronics just aren't like that.

[SgtSpike]

Figures... how come lifetime warranties never actually last a lifetime?  They should be called product life warranties.

Lifetime of the product, not your lifetime.

Bowtech.  Thus far, the ONLY company I know of with TRUE lifetime warranties.  Bought a bow in 2001 that the limbs broke?  Well here, we'll either fix it with limbs we still have in stock, or we'll give you a brand new bow of our latest and greatest!

That's how a lifetime warranty should be.

Seems to me that all sorts of hunting gear is like that. My friend has got a pair of LL Bean hiking boots that he's in all the time. He brings them in when they're worn, ripped, anything, and they replace 'em. My dad's got a Buck knife with a true lifetime warranty - he's had it for 15 years that he keeps sending in to get sharpened, new hinge, new release, etc.

Idk why, but electronics just aren't like that.

Well, probably because electronics are only truly useful for a short period of time, then people would purposefully break them to get a newer/better model.  But companies should not call it a lifetime warranty if it only covers the lifetime of the product, not the lifetime of the person.

[greyhawk]

Well, probably because electronics are only truly useful for a short period of time, then people would purposefully break them to get a newer/better model. 

People already routinely do that, where they "accidentally drop" a friends smartphone, then ask their insurance to pay for that friends new phone.

[SgtSpike]

Well, probably because electronics are only truly useful for a short period of time, then people would purposefully break them to get a newer/better model. 

People already routinely do that, where they "accidentally drop" a friends smartphone, then ask their insurance to pay for that friends new phone.

Exactly.  The abuse would be far greater if lifetime warranties were truly for the lifetime of the person.  People would stash away old, unused electronics in their closets, then find them years later and think, "hmmm, if I just laid a screwdriver across this circuit board while the device was turned on, I bet it would stop working and I could send it in to get the latest model!"

There are ways to somewhat prevent this, with water-detecting stickers on the insides of electronics and tamper-proof warranty stickers on the casings, but I'm sure I could kill just about any electronic device without raising any of those flags fairly easily.

So, I can understand why electronics companies do not offer true lifetime warranties, but STOP CALLING IT A LIFETIME WARRANTY!

[crazyates]

Well, probably because electronics are only truly useful for a short period of time, then people would purposefully break them to get a newer/better model. 

People already routinely do that, where they "accidentally drop" a friends smartphone, then ask their insurance to pay for that friends new phone.

Do you know how many phones I've returned just because I borked it up while messing around with the software? One time a busybox update got screwed up, and I just returned it for a new one rather than messing around with it. I love Radio Shack warranties.

[Phinnaeus Gage]

Well, probably because electronics are only truly useful for a short period of time, then people would purposefully break them to get a newer/better model. 

People already routinely do that, where they "accidentally drop" a friends smartphone, then ask their insurance to pay for that friends new phone.

Do you know how many phones I've returned just because I borked it up while messing around with the software? One time a busybox update got screwed up, and I just returned it for a new one rather than messing around with it. I love Radio Shack warranties.

Craftsman tools also come to mind.

[ldrgn]

And vice versa: the device can be populated with partially defective hashing chips. If the testing discovers a fault in some section of some hashing chip that section could be simply powered down and the whole device will remain useable.

What you're describing is certainly possible but I don't think it is practical for BFL and I'm surprised nobody else in this thread has called you out on it yet.  They're making small batches of a niche product - there is no big volume here.  The "switch a core off if it didn't work" model is only really practical if you're making chips by the boatload.  BFL needs a low chip failure rate to survive.

[2112]

What you're describing is certainly possible but I don't think it is practical for BFL and I'm surprised nobody else in this thread has called you out on it yet.  They're making small batches of a niche product - there is no big volume here.  The "switch a core off if it didn't work" model is only really practical if you're making chips by the boatload.  BFL needs a low chip failure rate to survive.

Please remember, that the mining software doesn't access the hashing chip directly, but through the the secret firmware. There will be no need for expensive lasering or fusing on the die. Just the firmware has to have a bitmap of the verified working hashers. So the whole situation is very different from the things like CPUs with whole sections of them disabled at the hardware level, because BFL can disable the defective cores at the level equivalent to BIOS.

I'm also assuming that they didn't do fully-unrolled hashers, but much smaller, iterative ones. Therefore I assumed that the chip will have at multiple tens or even hundreds of hashing cores.

On the FPGAs the hashers were unrolled primarily because the deficiency in the place-and-route software heurisics: for the rolled designs the P&R failed to converge or produced spectacularly bad layouts. ASIC designers use way more advanced layout-optimization software.

[Bogart]

Remember, the cores don't need to operate perfectly.

The occasional false positive or false negative will not really impact things much when it comes to mining.

[ab8989]

Switching cores off is a useful technique to be used on chips that have huge die sizes ( 200mm2 and up ) where defect probabilities go up and for each defect you either have to throw a very expensive chip to bin or do something else that also might cost big money. The disabling a core is also an expensive solution because it needs a lot of engineering and testing.

I believe this is not useful for BFL because the die sizes are so small making each chip so cheap to produce that is much more economical to just throw the faulty ones to bin rather than trying to do something clever with them. The small die size also means that defect probabilities are much smaller than on bigger dies as the defect probability is some constant multiplied with the die size.

Pulling some numbers from my ass: Not being able to switch cores off could cost them 50 chips each worth $3. How much engineering would you like to do to save $150? The situation is totally different if we are talking about saving 10000 defective chips each worth $200.

[2112]

Switching cores off is a useful technique to be used on chips that have huge die sizes ( 200mm2 and up ) where defect probabilities go up and for each defect you either have to throw a very expensive chip to bin or do something else that also might cost big money. The disabling a core is also an expensive solution because it needs a lot of engineering and testing.

I believe this is not useful for BFL because the die sizes are so small making each chip so cheap to produce that is much more economical to just throw the faulty ones to bin rather than trying to do something clever with them. The small die size also means that defect probabilities are much smaller than on bigger dies as the defect probability is some constant multiplied with the die size.

Pulling some numbers from my ass: Not being able to switch cores off could cost them 50 chips each worth $3. How much engineering would you like to do to save $150? The situation is totally different if we are talking about saving 10000 defective chips each worth $200.

The engineering required is near zero. Basically a gated clock buffer instead of plain clock buffer. And a single register wide enough to store those "clock enable" bits for each hashing core.

The overall design is extremely repetitive. The first constraint is thermal. The second one is power distribution rail bounce caused by the huge number of simultaneous switches. Gating the clock would be a standard hardware debugging technique for such a project. I could say that not including clock gating would be a design mistake. The primary objective is to facilitate debugging. Defect tolerance is an additional benefit obtained for free.

Again, the chip is so repetitive and so self-testing, that standard debugging aids (like JTAG) are nearly worthless. The chip is almost an analog or mixed-signal power chip: the primary constraints are thermal and parasitic impedances.

Edit: Furthermore, I think none of the Bitcoin ASIC manufacturers can afford to invest time and money into a proper chip testing. I'm thinking that all packaged chips will be soldered into the mining boards and the final testing will be in-situ. I don't even think that an investment into the proper test equipment would be worthwhile from the engineering point of view. All in all, those chips are just lottery ticket printing machines, it doesn't make sense to test if some rare tickets are missing or mangled. Each winning ticket is worth something for just a couple of minutes maximum.

[ab8989]

I believe you are making the common mistake of only counting interesting engineering and I agree, that is almost nonexistent in here. However in a proper ASIC engineering there is huge amount of non-interesting engineering especially when testing is involved. I also agree that BFL is probably skimping on the testing capabilities and that is another reason why you do not want to take on tasks that rely heavily on the testing to get them to work.

Also when talking about the technique of switching off cores I was not talking about BFL, but mainly thinking about applications where this kind of technique could theoretically make some sense meaning designs 1000 times bigger and more complicated and in there adding stuff in there is also 1000 times more complicated than in simpler designs but again in such small designs this makes absolutely no sense from economical point of view. You simply do not lift a finger trying to chase a savings of grand total of $150. There are other things 100 times more important to worry about. If we would be employed by BFL, even this discussion we were having in here would have cost the company more than $150 and we haven't even started actual work.

I believe normal ASIC manufacturing process involves testing the chips either before or after packaging and shipping only known working chips to customers. Many customers do not do inhouse testing the ASIC chips separately and neither tests the 100 other kinds of non-ASIC chips that go on to the PCB, but just solders them all on there and after that performs the testing for the full PCB. It is also possible that the defect on the ASIC is such that is totally unfunctional and goes up puff in smoke and you really do not want to trash many fully populated PCBs by having such chips soldered on so you need to separate somehow the faulty chips you want to use from ones you do not want to touch.