DeathAndTaxes
Donator
Legendary
Offline
Activity: 1218
Merit: 1079
Gerald Davis
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 15, 2011, 11:16:25 PM |
|
Does that make sense to you?
Do 32 GPU have 54x the performance of 1 GPU?
No idea how that applies to hashing, but I wouldnt rule out superlinear scaling just because its superlinear. Of course you know how it applies to hashing. The hashing problem is very easy. Very simple. On average GPU it takes less than a millionth of a second to solve. Then the GPU just does it again and again and again (millions of times per second i.e. MH/s). Each problem is completely independent of the prior problems. There is no speed up from having GPUs "work together" and the problem requires negligible amount of RAM and bandwidth. The problem scales perfectly linearly to ALUs and clock speed. We have seen this on CPU, GPU, and FPGAs. The reason is that while hashing may seen hard it is an illusion. Even the weakest CPU miner can complete a hash in a nearly insignificant amount of time. What makes hashing "difficult" is that a valid hash is so rare that is takes on average a 4 quadrillion attempts to find a valid hash. It isn't doing something hard. It is doing something very easy, 4 quadrillion times. You can't speed that up by having boards work together. Each "problem" (hash attempt) is trivially easy. It can't be made easier.
|
|
|
|
P4man
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 15, 2011, 11:18:25 PM |
|
So you are saying a 2 chip asymmetrical solution, like ASIC + FPGA is nonsensical? Have you read the PDF I linked above?
|
|
|
|
DeathAndTaxes
Donator
Legendary
Offline
Activity: 1218
Merit: 1079
Gerald Davis
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 15, 2011, 11:25:04 PM |
|
So you are saying a 2 chip asymmetrical solution, like ASIC + FPGA is nonsensical? Have you read the PDF I linked above?
I think you misunderstood the article. They aren't talking about increasing performance by using multiple chips but increasing the performance internal to a single chip. If BF has a breakthrough in performance it wouldn't only occur in the 64 FGPA model but also in the 2 FPGA model as well. The paper talks about using FPGA w/ onboard PowerPC microprocessor to aid the unrolled hashing engine. While it is an interest concept the performance levels they got (1.4Gb/s) are not better than performance obtained by "garage FPGA programmers" and those FPGA w/ onboard PowerPC microprocessors are very expensive $1000+ each.
|
|
|
|
P4man
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 15, 2011, 11:34:51 PM |
|
I didnt misunderstand the article, for the most part, I totally did not understand it. Either my IQ is lower than my shoesize, or achieving high performance in hashing isnt quite as straightforward as you think it is. Ive read left and right that creating an asic front end to the fpga (for loop unrolling?) would make a lot of sense, and if thats the case here, its not hard to see where superlinear scaling could come from. A much faster asic.
Anway, I dont know nearly enough about this to speculate, but I know enough to know I know nothing. And to some extent, that probably goes for the majority of posters here.
|
|
|
|
DeathAndTaxes
Donator
Legendary
Offline
Activity: 1218
Merit: 1079
Gerald Davis
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 15, 2011, 11:44:47 PM |
|
I didnt misunderstand the article, for the most part, I totally did not understand it. Either my IQ is lower than my shoesize, or achieving high performance in hashing isnt quite as straightforward as you think it is. Ive read left and right that creating an asic front end to the fpga (for loop unrolling?) would make a lot of sense, and if thats the case here, its not hard to see where superlinear scaling could come from. A much faster asic.
That doesn't make any sense. A single hash is incredibly fast. Literally a millionths of a second or less to complete. If you tried to have inter-chip communication between ASIC and FPGA that communication bus would be on the gigabit per second scale with nano-second latency. While you may be able to build something like that given how trivial a single hash is to accomplish it serves no purpose. The article you linked to had nothing to do with multi-chip designs. I haven't seen any paper anywhere which involves multiple chip designs where each chip isn't independent of the other chips. There is plenty of room even in the Spartan LX150 to completely unroll the loop INSIDE the FPGA. There is no longer any looping invovled. The SHA loop has been transformed into a flat process. Every clock cycle the FPGA completes all 80 loops (with a little cheating) involving both the inner and outer hash for Bitcoin. There isn't anything else to unroll. Nonce goes in, hash comes out on each tick of the clock. Tick - hash Tick - hash Tick - hash If a valid hash is found return it, otherwise keep going. 4 billion ticks later - nonce range is complete - request new data from controller.
|
|
|
|
bulanula
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 12:13:52 AM |
|
Ok so here is the email I received regarding info on the 54 GH/s box: "The BitForce Rig Box product will cost about 80% of the per mh/s cost established with the pricing of the BitForce SHA256 Single. The standard post pre-order price of the Single is $599 ( thought it was $500 LOL ) for 1,050 mh/s (1.2 gh/s), or roughly 60 cents per mh/s. At 80%, the Rig Box comes in at roughly 50 cents per mh/s. Although the technology is the same, clustering allows for additional efficiency opportunities which can't be realized in the stand alone single.TOTAL BS Rig Boxes are about $25,000 The scam is not big enough for you BFL ? each with 54 gh/s performance. They don't come in smaller flavors. Well obviously you do not want to scam for less - efficiency is king, even in scamming " Take it as you will. Don't know about you but this email above 99% convinces me this is a total scam.
|
|
|
|
fred0
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 12:16:32 AM |
|
Ok so here is the email I received regarding info on the 54 GH/s box: "The BitForce Rig Box product will cost about 80% of the per mh/s cost established with the pricing of the BitForce SHA256 Single. The standard post pre-order price of the Single is $599 ( thought it was $500 LOL ) for 1,050 mh/s (1.2 gh/s), or roughly 60 cents per mh/s. At 80%, the Rig Box comes in at roughly 50 cents per mh/s. Although the technology is the same, clustering allows for additional efficiency opportunities which can't be realized in the stand alone single.TOTAL BS Rig Boxes are about $25,000 The scam is not big enough for you BFL ? each with 54 gh/s performance. They don't come in smaller flavors. Well obviously you do not want to scam for less - efficiency is king, even in scamming " Take it as you will. Don't know about you but this email above 99% convinces me this is a total scam. So, no pre-order for you then? ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) I mean, you admit a 1% chance it is NOT a scam.
|
|
|
|
grubles
Full Member
![*](https://bitcointalk.org/Themes/custom1/images/star.gif) ![*](https://bitcointalk.org/Themes/custom1/images/star.gif)
Offline
Activity: 180
Merit: 100
mistaken for gribble since 2011
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 12:18:32 AM |
|
Ok so here is the email I received regarding info on the 54 GH/s box: "The BitForce Rig Box product will cost about 80% of the per mh/s cost established with the pricing of the BitForce SHA256 Single. The standard post pre-order price of the Single is $599 ( thought it was $500 LOL ) for 1,050 mh/s (1.2 gh/s), or roughly 60 cents per mh/s. At 80%, the Rig Box comes in at roughly 50 cents per mh/s. Although the technology is the same, clustering allows for additional efficiency opportunities which can't be realized in the stand alone single.TOTAL BS Rig Boxes are about $25,000 The scam is not big enough for you BFL ? each with 54 gh/s performance. They don't come in smaller flavors. Well obviously you do not want to scam for less - efficiency is king, even in scamming " Take it as you will. Don't know about you but this email above 99% convinces me this is a total scam. Surely you can explain why you think this instead of simply calling them scammers. And you guys wonder why BFL doesn't want to participate in this thread...
|
BM-NBi5PcH8BBqDVp7WSJe2pNgqEVvTjd6T
|
|
|
DeathAndTaxes
Donator
Legendary
Offline
Activity: 1218
Merit: 1079
Gerald Davis
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 12:21:05 AM |
|
The standard post pre-order price of the Single is $599 for 1,050 mh/s (1.2 gh/s) Totally missed this the first time. Under what math system is 1050 MH = 1.2 GH?
|
|
|
|
legolouman
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 12:25:13 AM |
|
The standard post pre-order price of the Single is $599 for 1,050 mh/s (1.2 gh/s) Totally missed this the first time. Under what math system is 1050 MH = 1.2 GH? Mine, I like that idea, less is more. Hey can anyone donate a BTC? By which I mean 1.142857 BTC. Thanks!
|
If you love me, you'd give me a Satoshi! BTC - 1MSzGKh5znbrcEF2qTrtrWBm4ydH5eT49f LTC - LYeJrmYQQvt6gRQxrDz66XTwtkdodx9udz
|
|
|
bulanula
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 12:25:24 AM |
|
The standard post pre-order price of the Single is $599 for 1,050 mh/s (1.2 gh/s) Totally missed this the first time. Under what math system is 1050 MH = 1.2 GH? Guys, I think we have cracked it !!! BFL Labs come from another planet. They have managed to break the laws of physics and achieve 100% efficiency and use a completely different system of computation far superior to ours which yields greater Mhash/s than we ever dreamed. That explains ALL the problems with the different company names and all that. Bring on the $$$, bring on the pre-orders, bring on the fools : I am sold ![Tongue](https://bitcointalk.org/Smileys/default/tongue.gif) ! Hope they can keep the assembly line hot, because I want to buy 599 units of the magical device. It really is damn clear : until they let Inaba in ( under NDA or something ), nobody should pre-order unless they want to lose money to some scammers. Might as well donate that money to me instead for charity
|
|
|
|
legolouman
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 12:53:25 AM |
|
This is semi related to the thread and is awesome vintage pr0n none the less... http://www.copacobana.org/gallery.htmlI once tried to do what BFL is claiming to do now. Spent a lot of time looking into the maths. If BFL has what they say they have, they honestly should be selling it to the Government for millions... EXACTLY, if they are managing to do what they claim, why are they aiming it at only the Bitcoin Community? An FPGA array of this power has more uses than bitcoin.
|
If you love me, you'd give me a Satoshi! BTC - 1MSzGKh5znbrcEF2qTrtrWBm4ydH5eT49f LTC - LYeJrmYQQvt6gRQxrDz66XTwtkdodx9udz
|
|
|
RandyFolds
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 12:58:20 AM |
|
Ok so here is the email I received regarding info on the 54 GH/s box: "The BitForce Rig Box product will cost about 80% of the per mh/s cost established with the pricing of the BitForce SHA256 Single. The standard post pre-order price of the Single is $599 ( thought it was $500 LOL ) for 1,050 mh/s (1.2 gh/s), or roughly 60 cents per mh/s. At 80%, the Rig Box comes in at roughly 50 cents per mh/s. Although the technology is the same, clustering allows for additional efficiency opportunities which can't be realized in the stand alone single.TOTAL BS Rig Boxes are about $25,000 The scam is not big enough for you BFL ? each with 54 gh/s performance. They don't come in smaller flavors. Well obviously you do not want to scam for less - efficiency is king, even in scamming " Take it as you will. Don't know about you but this email above 99% convinces me this is a total scam. Surely you can explain why you think this instead of simply calling them scammers. And you guys wonder why BFL doesn't want to participate in this thread... See pages one through thirty-two.
|
|
|
|
2112
Legendary
Offline
Activity: 2128
Merit: 1073
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 01:26:59 AM |
|
FPGAs are a poor choice for sCrypt mining for the same reason that GPUs are: they require too much on-chip cache.
This isn't that obvious to me. It seems like ArtForz choose suspiciously small value for the scratchpad size when parametrizing scrypt() for TeneBrix: 131583. Converting that to bits gives me slightly over a megabit. This is well within the BlockRAM resources for the Xilinx chips that are currently popular among miners. It isn't out of the realm of possibilities why a custom microcoded implementation of ArtForz's scrypt() would be necessarily slower that on the common CPUs. The requiring mixing array should fit with the FPGA "cache". Here "cache" means it would not require going off the FPGA to an external RAM. Besides, ArtForz is known to be crafty that way and has a very specific sense of humour.
|
|
|
|
rjk
Sr. Member
![*](https://bitcointalk.org/Themes/custom1/images/star.gif) ![*](https://bitcointalk.org/Themes/custom1/images/star.gif) ![*](https://bitcointalk.org/Themes/custom1/images/star.gif)
Offline
Activity: 448
Merit: 250
1ngldh
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 02:03:33 AM |
|
EXACTLY, if they are managing to do what they claim, why are they aiming it at only the Bitcoin Community? An FPGA array of this power has more uses than bitcoin.
Re-read what BFL themselves posted. They stated that Bitcoin is NOT their primary objective. Why would we go to the trouble for such a small market? We didn't. Our product design is influenced by other factors beyond hash mining.
Regards, BFL
|
|
|
|
RandyFolds
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 02:07:47 AM |
|
Someone should apply to their job postings. I would do it, but I can't even bullshit tech crap, let alone sound like I really know it.
|
|
|
|
plastic.elastic
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 02:34:28 AM |
|
Anyone thinks is not a scam, put money where your mouth is... go pre-order the board.
I cant wait to see your faces in a month from now
|
Tips gladly accepted: 1LPaxHPvpzN3FbaGBaZShov3EFafxJDG42
|
|
|
plastic.elastic
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 02:41:25 AM |
|
Send them a resume, give them personal information? No one can meet them in real life or confirm where they are located. I thought about trying this but I doubt it would work. I will pay the guy $200 (in bitcoin) just to have coffee with Inaba and have him take Inaba into the "Lab". Assuming Inaba can take photos of both the guy and the lab. This offer still stands. Anyone thinks is not a scam, put money where your mouth is... go pre-order the board.
I cant wait to see your faces in a month from now
If you really have faith pay in bitcoin:) If you're dumb enough to be scammed, you deserve it. I cant take someone, whos confused with something obvious like Thermalright and Thermaltake,.... seriously.
|
Tips gladly accepted: 1LPaxHPvpzN3FbaGBaZShov3EFafxJDG42
|
|
|
legolouman
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 02:51:09 AM |
|
Send them a resume, give them personal information? No one can meet them in real life or confirm where they are located. I thought about trying this but I doubt it would work. I will pay the guy $200 (in bitcoin) just to have coffee with Inaba and have him take Inaba into the "Lab". Assuming Inaba can take photos of both the guy and the lab. This offer still stands. Anyone thinks is not a scam, put money where your mouth is... go pre-order the board.
I cant wait to see your faces in a month from now
If you really have faith pay in bitcoin:) If you're dumb enough to be scammed, you deserve it. I cant take someone, whos confused with something obvious like Thermalright and Thermaltake,.... seriously. Have you heard about the new Thermal-left computers? Really rare, only 5000BTC! Okay, I see what you're getting at.
|
If you love me, you'd give me a Satoshi! BTC - 1MSzGKh5znbrcEF2qTrtrWBm4ydH5eT49f LTC - LYeJrmYQQvt6gRQxrDz66XTwtkdodx9udz
|
|
|
fred0
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
November 16, 2011, 03:38:41 AM |
|
Send them a resume, give them personal information? No one can meet them in real life or confirm where they are located. I thought about trying this but I doubt it would work. I will pay the guy $200 (in bitcoin) just to have coffee with Inaba and have him take Inaba into the "Lab". Assuming Inaba can take photos of both the guy and the lab. This offer still stands. Anyone thinks is not a scam, put money where your mouth is... go pre-order the board.
I cant wait to see your faces in a month from now
If you really have faith pay in bitcoin:) If you're dumb enough to be scammed, you deserve it. I cant take someone, whos confused with something obvious like Thermalright and Thermaltake,.... seriously. There's only one way to definitively settle this. Psychic Hotline!
|
|
|
|
|