I do not want to talk about the obvious and fundamental. You just lost all credibility. This conversation does not make sense. Not at this level.
It is obvious that you can do ASIC 100x faster than FPGA and can do FPGA 100x faster than ASIC. Not the point. The conversation is about the real possibility of the implementation of ASIC several times faster than the current leading FPGA chips (like Artix-7, Kintex-7,Virtex-7,Spartan-6), having a small budget. Do you know a company that did it? I not found any example.
Yes! The table you quoted was showing multiple ASIC implementations from multiple companies that significantly outperformed FPGAs. You rejected that argument, thinking one cannot replicate a small ~300MHz logic unit (20k gates) many times across the die area, but I pointed out to you this is possible by giving the example of a CPU with 10+ million gates running at 500MHz. Then you buried your head into the sand saying "Do not compare CPU with ASIC". A CPU is an ASIC. Maybe a better example to convince you would be to talk a GPU, where shaders occupy most of the die area and are precisely that: a small logic unit replicated many times across the die area. Perhaps another way to show it to you is as is. In your table: - The FPGA in [16] has a die size of 317 mm^2. Most of the space is occupied by slices, but let's be really conservative and say only half the die area is occupied by slices, or 150 mm^2. So they did 1Gbps by utilizing about 30 mm^2 of the die area (150 mm^2 * 2120 slices / 10752 total slices on the chip.)
- On the other hand the ASIC in [21] which I managed to track down was a reference to http://www.cast-inc.com/ip-cores/encryption/sha-256/cast_sha256.pdf is a SHA-256 core doing 2Gbps by utilizing only 0.25 mm^2 (PDF says exactly 250040 um^2).
So, obviously the chip doing twice the work in 1/120th the die space provides a building block that can be utilized to make a full-size chip that significantly outperforms FPGAs. Don't say again that "FPGAs progressed faster than ASICs", this is not true I have already pointed out that they both top out at 28nm today, so the comparison of FPGAs vs. ASICs made at the time your table was composed is still valid today. Also I already showed here: https://bitcointalk.org/index.php?topic=95762.0 that manufacturing at 65nm costs only $500k which is clearly within reach of BFL's financial resources. |
|
|
|
FFS: http://en.wikipedia.org/wiki/Field-programmable_gate_array"Historically, FPGAs have been slower, less energy efficient and generally achieved less functionality than their fixed ASIC counterparts. A study has shown that designs implemented on FPGAs need on average 40 times as much area, draw 12 times as much dynamic power, and are three times slower than the corresponding ASIC implementations.[22]" 40 times less area * 3 times faster = 120x at the same process node It's really simple guys, every logic block and interconnect needs a bunch of transistors to control them, but on ASIC you just lay the circuit down as you want it to run. Because of this you have far more transistors actually performing the work rather than controlling internal functions. So even though the underlying features are the same at a given process node, FPGA suffers a disadvantage in area and speed, but makes up for it with flexibility. ASIC is locked in function, but very, very fast at execution. OMG !  These data are historical, now the difference is less because the FPGA is growing faster than ASIC. FPGA technology is more advanced and more popular than ASIC. These are not my words, but experts in the field. No, FPGAs are not more advanced. The best technology for both FPGAs and ASICs is 28nm today. I would love to hear a quote from your "experts". I could help you clarify how you misunderstand their words. You do not proved anything except that ASIC is more energy efficient. I know it. Conversation is whether ASIC can be "several times faster" than a similar FPGA.
You are one centimeter away from understanding my point. Let me try it this way. If you know that ASICs are more power efficient, that means they do more work per Joule (or per Watt if that makes it easier for you to understand). Therefore, when comparing a 10 Watt ASIC chip to a 10 Watt FPGA chip, you can guess the ASIC chip will accomplish more work, right? So if an ASIC is 100x more power efficient than an FPGA, the 10 Watt ASIC chip will perform 100x more work than the 10 Watt FPGA chip. In other words it will be 100x faster. That's it. |
|
|
|
You bring that table to attempt to prove your point, then find out it proves you wrong, so you retract it saying it is old and does not apply anymore?  The observation I made from this table is still valid to this day. I was comparing one of its 180nm ASIC with a 120nm Virtex 2 FPGA (180/120 = 1.5x feature scale difference). Today, you would be comparing a 65nm ASIC (presumed process node for BFL ASIC) against the 45nm FPGAs that all other Bitcoin mining vendor use (Spartan6 LX150), that's a 65/45 = 1.4x feature scale difference. So in both cases the power efficiency of FPGAs over the ASICs (that I am comparing them with) is the same, because power efficiency is directly inversely proportional to the square of the feature size. One more time: please put your money where you mouth is if you are so convinced of yourself: http://betsofbitco.in/item?id=665 |
|
|
|
Nobody produces ASICs to achieve several times better performance than FPGAs, there are other advantages.
Gibberish. I wrote - several times faster. ASIC may indeed be faster than FPGAs, but this is not so big difference. This is the difference of approximately to 100%, no 1000% and more. BFL offers us about 500% faster ASIC than existing FPGA. I found an old comparison table:  and newer: http://www.heliontech.com/fast_hash.htmWhat you need to understand is that this table does not show how many hashing cores can be placed on a chip. The XC2V2000 for example only has 10 thousand slices. The table shows a SHA-256 core takes ~1000 slices and does ~1Gbps [16], so by putting 10 cores per FPGA the whole chip would do only ~10Gbps. On the other hand, a 180nm ASIC of small size (50mm^2) has about 10 million gates. The table shows a SHA-256 core takes ~20 thousand gates and does ~2Gbps [21], so by putting 500 cores per ASIC the whole chip would do ~1000 Gbps. Tada! There is your 100x difference. Roughly what BFL is promising (Single 832 Mhash/s vs. Single SC 60 Ghash/s.) it is easy to say in theory, try to do this in practice  Actually, yes, it is easy to do and has been done. To continue my example, a 180nm ASIC with 10+ million gates at 350MHz+ was done 13 years ago by Intel with the Pentium III 500E: 180nm, 28 million gates, 500MHz, at a tiny 13 Watt(!) http://en.wikipedia.org/wiki/List_of_Intel_Pentium_III_microprocessors#.22Coppermine.22_.28180_nm.29I personally estimate BFL developed at 65nm, which is only $500k in NRE costs: https://bitcointalk.org/index.php?topic=95762.0 By all estimates, they have at least $1.5M+ of preorders (85+ Thash/s), so yeah they can afford 65nm. If you don't believe that BFL can do at least as good as that, then please put your money where your mouth is by betting on http://betsofbitco.in/item?id=665 (I did with 50 BTC). You could double your money if you are right. |
|
|
|
Nobody produces ASICs to achieve several times better performance than FPGAs, there are other advantages.
Gibberish. I wrote - several times faster. ASIC may indeed be faster than FPGAs, but this is not so big difference. This is the difference of approximately to 100%, no 1000% and more. BFL offers us about 500% faster ASIC than existing FPGA. I found an old comparison table: and newer: http://www.heliontech.com/fast_hash.htmWhat you need to understand is that this table does not show how many hashing cores can be placed on a chip. The XC2V2000 for example only has 10 thousand slices. The table shows a SHA-256 core takes ~1000 slices and does ~1Gbps [16], so by putting 10 cores per FPGA the whole chip would do only ~10Gbps. On the other hand, a 180nm ASIC of small size (50mm^2) has about 10 million gates. The table shows a SHA-256 core takes ~20 thousand gates and does ~2Gbps [21], so by putting 500 cores per ASIC the whole chip would do ~1000 Gbps. Tada! There is your 100x difference. Roughly what BFL is promising (Single 832 Mhash/s vs. Single SC 60 Ghash/s.) |
|
|
|
|
This was to be expected. BFL selected the cheapest foundry to make their ASICs. So their design was stolen by rogue employees, or even the unscrupulous company itself, and clone chips are now going to be manufactured, and sold on the grey market.
</joke-but-plausible-scenario>
|
|
|
|
|
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.
|
|
|
|
|
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.
|
|
|
|
...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 monthIts power cost: 0.150 (kW) * 730 (hour/month) * 0.11 (worldwide average $/kWh) = $12.0 per monthSo 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. |
|
|
|
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. |
|
|
|
There's simply not much else to buy with Bitcoins
*cough* There are hundreds of merchants accepting Bitcoin. Hundreds are listed on the wiki. There are more than a thousand merchants alone using Bitpay. There is so much to buy in bitcoins! |
|
|
|
|
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.
|
|
|
|
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. |
|
|
|
*more accusations* There's a full list of users with deposits on page 1 - there are a lot of fairly big users in there. Can't you just settle for the fact that you were wrong about this one for once? May I remind you that pirateat40 also had stellar reputation over multiple months before his theft. He also had fairly big depositors. He also had supporters, people just like you, ardently arguing that pirateat40 always paid, and that, therefore, there is no way he was going to turn out to be a fraudster. The rest is history. My guess is, based on your broken logic, you would have supported pirateat40 as well at the time, am I right? Think about that very deeply. And try to understand why your logic is broken. Bottom line, there is no free money. 1.5% weekly == 117% yearly. There is almost no way Vescudero is legit, especially when his operations match almost all the Ponzi red flags established by the US SEC, which I keep quoting again and again: http://www.sec.gov/answers/ponzi.htmI was right about pirateat40. I will be right about Vescudero again. Gullible investors will continue to lose their coins  |
|
|
|
Oh yeah, friedcat is even more shady than BFL and all other mining related biz together.  An GLBSE asset which lets you rent self-made asic mining power. That is so shady I don't even start trying to educate the suckers falling for it. Did I mention he is in China?  Well, since you refuse to bet in http://betsofbitco.in/item?id=665 (apparently poor confidence in your own claims) I will just say this: I will come back to this thread to tell you "I told you so a company as small as BFL can make ASICs that efficient (>350 Mhash/Joule)". See you in the near future PS: Korbman: when I started mining, it was with 4x5970 and the difficulty was 12251. I was solving a block every 7 hours. Yes it was that easy... |
|
|
|
my co-workers produced an ASIC for under a couple hundred thousand dollars. It took a couple years but that was essentially because he had never done it before. It seems everytime people mention ASICs they takes someone's WAG and multiply is a couple times.
And by the way, that price included a few thousand chips off the line.
Unless you provide some proof of your claim it stands alone. As a claim nothing more... Once more, certainly possible with FPGA conversion chips. ElectricMucus, you need to chill out and accept that producing ASICs can be "a few hundred thousand dollars". Here is a claim from another ASIC manufacturer (friedcat for his Bitfountain company's asicminer project): only ~$150k for 130nm and ~$500k for 65nm. |
|
|
|
|
Are you kidding? dank is obviously the world's best guitar player. He will become famous, living in a mansion, selling marijuana pipes and shit.
|
|
|
|
It is not my intention to taunt when I tell people to put their money where their mouth is by betting 'disagree' on http://betsofbitco.in/item?id=665 I tell them that to hopefully stop (or at least reduce) the length of the debates on the forum. No, your intention is this: Here is the breakdown of the distribution:
45% goes to the bet winners proportional to their bets.
45% goes to the bet winners proportional to their weighted bets.
5% goes to the user who submitted the bet.
5% goes to the site. I have mined 25+ thousand BTC the last 2 years. I couldn't care less of the comparatively tiny financial profits of this bet. |
|
|
|
Thanks Tom for answering my question! I love your openness, unlike BFL. Have you considered getting venture capital? Even if you don't really want that, I am sure some VCs would be interested in you (generally they love founders who don't seem to need the capital.) Talk to http://ycombinator.com/ if only to see what they say. |
|
|
|
|
Just for reference, I ordered the first day the bASIC was announced, and am order #338. So the first bASIC may be in the late 200s or early 300s. It would be nice if Tom could tell us exactly...
|
|
|
|
|
Show Posts
