RHA
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 19, 2013, 11:24:46 PM |
|
The translation: Awesome results - just as he intended.
Great respect.
|
|
|
|
2112
Legendary
Offline
Activity: 2128
Merit: 1073
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 20, 2013, 05:05:57 AM |
|
Can someone translate what bitfury said in his long post?
OK, I'll try... Bitfury did just a short burst tests (lasting at most few seconds) of his chip at various voltages and various speeds. He still need to test it under a continuous load and at various temperatures. Now I switch to explaining what he didn't do. He didn't use any "standard cell", "library" or other purely digital logic EDA tools, which were thus far used by all other competitors. They used unrolled pipeline designs and their simulations were very inaccurate-purely digital. Due to the very approximate simulation and high toggle rate of the flip-flops in SHA-256 the real power used by competitor chips was significantly higher than the simulated power usage. Now I switch back to what he did do. After completing the initial digital logic design he had switched to a full-custom design process: he no longer designed at the gate or flip-flop level but at the individual transistor level. He no longer dealt with zeros and ones but with volts and amperes. He had used a very precise analog models of the transistors from the foundry and used analog circuit simulators: SPICE and/or BSIM. This was possible without multimillion budget because he didn't unroll the hashing kernels but kept them rolled and replicated them several hundred times over the area of the chip. The additional benefit was that his power simulations were very accurate: the simulation error margins were lower than the wafer manufacture process error margins. One of the goals he had choosen for his chip was to have them work with rather wide ranges of the supply voltages, in particular with uncommonly low voltages. To achieve that he had paid extreme attention to minimizing the switching noise both inside the chip and on the chip pins. In particular despite of the relatively low clock speed (100MHz or low hundreds MHz) the I/O design methodology was the one used for GHz signals (or high hundreds of MHz). Because of the above goals his chip isn't really plug-and-play for designers used to deal with typical digital logic design workflows and tools. To get all thats possible out of his chip requires familiarity with analog and mixed signal design. In particular it requires familiarity with transmission line modeling to design the best PCB with a large quantity of his chips. His goal is to publish http://en.wikipedia.org/wiki/Scattering_parameters required to accurately model the I/O terminals of his chip. He probably already has a good idea of what those S-parameters are from the high accuracy simulations using SPICE/BSIM. But possibly he either can't (because of NDA with the foundry) or doesn't want (for competitive advantage reasons) release his internal models and their parameters. Additionally, the S-parameters are affected by the quality of packaging of the chip and there is no real substitute for making the actual measurements, for which he plans to use devices made by http://www.signalhound.com/ . Obviously, I'm not bitfury, and the above is just my educated guess.
|
|
|
|
intron
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: 427
Merit: 251
- electronics design|embedded software|verilog -
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 20, 2013, 08:14:12 AM |
|
Hi bitfury,
Saw in your code you are clocking the SPI chain at 200 kHz:
speed = 200000;
if (ioctl(fd, SPI_IOC_WR_MAX_SPEED_HZ, &speed) < 0) { perror("Unable to set WR_MAX_SPEED_HZ"); close(fd); return -1; } if (ioctl(fd, SPI_IOC_RD_MAX_SPEED_HZ, &speed) < 0) { perror("Unable to set RD_MAX_SPEED_HZ"); close(fd); return -1; }
Why it is so slow? Just te be sure during development or are there other reasons? It's not a problem, just curious.
intron
|
|
|
|
|
coldguy
Member
![*](https://bitcointalk.org/Themes/custom1/images/star.gif)
Offline
Activity: 69
Merit: 10
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 20, 2013, 09:22:30 AM |
|
An RF digital SHA256 chip?
|
|
|
|
Flashman
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 20, 2013, 04:03:22 PM |
|
Very nice, however, I'm also thinking BFLs chips were looking this good this early, they didn't know they had power issues until they tried to run them at a decent frequency. The question is, where is your "ski jump" where the power/frequency goes exponential? BFL found their's was not as far to the right of the graph as they hoped. If the "ski jump" is to the right of your intended frequency spec, all well and good, great job. If it's to the left.... well, there's always 14nm..... ![Cheesy](https://bitcointalk.org/Smileys/default/cheesy.gif)
|
TL;DR See Spot run. Run Spot run. .... .... Freelance interweb comedian, for teh lulz >>> 1MqAAR4XkJWfDt367hVTv5SstPZ54Fwse6
Bitcoin Custodian: Keeping BTC away from weak heads since Feb '13, adopter of homeless bitcoins.
|
|
|
Mabsark
Legendary
Offline
Activity: 826
Merit: 1004
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 20, 2013, 08:47:15 PM |
|
Congratulations on a an excellent chip. Now, the only relevant question is when will I be able to purchase one?
These chips are completely irrelevant if restricted to 100TH and Metabank (who wont ship outside of Russia). Either sell us the chips or sell us the devices that use these chips, but stop waving them in our faces and telling us we can't have them.
|
|
|
|
meowmeowbrowncow
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 21, 2013, 01:58:24 AM |
|
It would make logical sense, if it also made technical sense, that the power circuitry allowed for a 'high speed/turbo' mode that would allow to seize on the opportunity of relatively low difficulty at the present moment.
And, given bitfury's forward thinking electrical usage design, the ability for manual user clocking / undervolting to the 6/2 ratio would be equally as beneficial once Bitcoin reaches near Petahash capacity.
I'd be somewhat disappointed if bitfury did not provide circuitry that allowed for both a mutually exclusive hi speed-hi consumption setting and then the 6/2 setting.
My 2 cents. If the bitfury units possessed both capabilities I'd be more likely to purchase.
|
"Bitcoin has been an amazing ride, but the most fascinating part to me is the seemingly universal tendency of libertarians to immediately become authoritarians the very moment they are given any measure of power to silence the dissent of others." - The Bible
|
|
|
Eric Muyser
Full Member
![*](https://bitcointalk.org/Themes/custom1/images/star.gif) ![*](https://bitcointalk.org/Themes/custom1/images/star.gif)
Offline
Activity: 224
Merit: 100
You can't kill math.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 21, 2013, 02:08:30 AM |
|
Can't write comments on bitbet.us site... It would be great if they're copy this as well in comments... Sent payouts: http://bitbet.us/bet/450/bitfurys-asic-will-work-with-power-1/DATE (2013-06-19 02:08:28) FOR BLOCKCHAIN LOOKUPS: BET OUTGOING WALLET PAID AMOUNT (WIN) 0.1 1CvadbWnVotoUhrrepxEGqESakmuqHtwnE 0.2 1.0 1NxPnXyN4XmxjysUxLXVXgNGFhYCEGFASV 1.9 10.0 19d6exSEyAPDXLLe3W3ucexpVXUABFeFsy 18.6 1.0 1HP9pTD9ueGYnkEA7Wj7D4SBCesTxXVsNb 1.9 0.044 1GnK5jZxzhFnKCu2yu89XP2JKvBME1em6A 0.1 5.0 16dQP7Nr1irruUTAbiiwYgNB7jQ8NpEVQ3 9.3 1.0 1DEVgggHwKngXf6YSn8ZNBvESTzCDiQWE2 1.9 0.002 1MoaWWXfSFWv6wYpKJEQiRjkmn8gHKBieK 0.1 0.052 19XHbghMRS9HhvWMoKhW8zKFYBSLVz69gB 0.1 4.0 1FzPgqnBrkBZanZwHaAmnd5wRxeDarwr9b 7.4 0.5 1CUanrri8gqXnU8e6bP2Tq8j2DSeBt5Tym 0.9 27.0 1MXf4cFD7EETFQEqD7QdLHbJeGVZfLnbaV 50.2 1.0 1DUv9KXp5CVaEdjvwg5GRUEZ5AhPVy8mnV 1.9 1.5 17xHzsvibCp9SKHParBaHVNZ5yykcjiRTk 2.8 0.05 13ZCqhVuvGL5RqF8WhUdiSFKDScZLZ7Jcz 0.1 0.01 1JCKtVk5W2LVsJiSHGXAcHeRMbZW6RXDAf 0.1 6 13X6T4iGWNHwPstvda7J1XDHbdiugio3j9 11.2 Wow... Bitfury! I don't know you from Adam, but I must say I am impressed. It seems integrity is in short supply in this forum, but you appear to have a surplus. Godspeed good sir. Amazing Bitfury! Congratulations and thanks for holding your word! Much respect.
|
@EricMuyser | EricMuyser.com | OTC - "Defeat is a state of mind; no one is ever defeated until defeat has been accepted as a reality" - Bruce Lee
|
|
|
Adeel06
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 21, 2013, 03:36:05 AM |
|
Hi, I didn't read through this all but I would be willing to test some chips. Don't need 30 (unless you guys want to part with 30? ![Cheesy](https://bitcointalk.org/Smileys/default/cheesy.gif) ). 16 would be more than enough to fill a PCB board that would probably run at 4.8gh/s x 16 = ~77gh/s. This of course is an underestimation, I am sure we can get levels over 5 gh/s. I do have a question for the op though, would you mind if I implemented self-made water cooling blocks on each chip so I could tinker with voltage and temp and try to squeeze the max out of the chips or do you want a "normal" set up to compare to the systems youre selling? By the way, for my legitimacy, I have owned online businesses since '03. My most recent project, www.limitlessbids.com , has excellent reviews on the various rating sites. I havent been working on the site though in the past 3 months because of school and bitcoin but it should suffice as proof that I'm not some scammer. - Adeel Adam Anwar
|
|
|
|
Adeel06
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 21, 2013, 04:19:25 AM |
|
Guess since it works they won't be sending any chips. Damn ;(. Wanted to see what would happen past 1.1V using a copper based water cooling block on every chip. The expense may outweigh the benefit, but since the chips were going to be free, I was willing to try it and see just how much we can get out of these. Blahhhhhhh ![Sad](https://bitcointalk.org/Smileys/default/sad.gif) . Anyone want to send me 16 chips written for SHA256 arch to overclock the shhh out of? ![Cheesy](https://bitcointalk.org/Smileys/default/cheesy.gif)
|
|
|
|
phillipsjk
Legendary
Offline
Activity: 1008
Merit: 1001
Let the chips fall where they may.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 21, 2013, 04:27:40 AM |
|
Adeel06, I logged in just to add you to my ignore list.
The first post explained what testing was required, and even why. I'll admit the title was a little misleading: you have to work for your "FREE MONEY".
|
James' OpenPGP public key fingerprint: EB14 9E5B F80C 1F2D 3EBE 0A2F B3DE 81FF 7B9D 5160
|
|
|
Adeel06
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 21, 2013, 05:20:14 AM |
|
Uh, I'm pretty sure he said he'd prefer air cooling, not that it is a necessity. Thanks sir.
|
|
|
|
nightyj
Member
![*](https://bitcointalk.org/Themes/custom1/images/star.gif)
Offline
Activity: 89
Merit: 10
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 21, 2013, 08:39:19 AM |
|
Guess since it works they won't be sending any chips.
bitfury is a serious man, he said that he will send chips to some developers fast and then he will contact the others that want to participate in the test and will send chips to them by slower mail I am shure that he will send chip to us, me and some of the forum members have invested in expensive power supply boards, variablle oscillators and other parts needed for the test .
|
|
|
|
|
muyuu
Donator
Legendary
Offline
Activity: 980
Merit: 1000
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 21, 2013, 09:59:51 AM |
|
Well done, chaps.
|
GPG ID: 7294199D - OTC ID: muyuu (470F97EB7294199D) forum tea fund BTC 1Epv7KHbNjYzqYVhTCgXWYhGSkv7BuKGEU DOGE DF1eTJ2vsxjHpmmbKu9jpqsrg5uyQLWksM CAP F1MzvmmHwP2UhFq82NQT7qDU9NQ8oQbtkQ
|
|
|
tytus
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 21, 2013, 10:04:47 AM |
|
We will start sending chips beginning of next week. Probably Monday. I will try to give more detailed info about this tomorrow.
|
|
|
|
nightyj
Member
![*](https://bitcointalk.org/Themes/custom1/images/star.gif)
Offline
Activity: 89
Merit: 10
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 21, 2013, 10:14:31 AM |
|
We will start sending chips beginning of next week. Probably Monday. I will try to give more detailed info about this tomorrow.
Great news ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif)
|
|
|
|
eldentyrell
Donator
Legendary
Offline
Activity: 980
Merit: 1004
felonious vagrancy, personified
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 21, 2013, 09:39:59 PM |
|
First I would like to say that chip is working. Complete confirmation using test-vectors.
Congratulations. I've added your chip to the η-factor rankings, which it currently leads by a factor of 5x (!). (Edit: bitfury's chip is 55nm not 65nm)
|
The printing press heralded the end of the Dark Ages and made the Enlightenment possible, but it took another three centuries before any country managed to put freedom of the press beyond the reach of legislators. So it may take a while before cryptocurrencies are free of the AML-NSA-KYC surveillance plague.
|
|
|
greaterninja
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
June 21, 2013, 09:57:29 PM |
|
Congratulations Bitfury! I should be #6 on the waiting list ; please send samples to me so I can tinker with them ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) I think I said this before, but I work for the #2 supplier of chip packaging solutions in the world. I have a Russian translator, and 1-2 Electrical Engineers on hand to help find improvements. I can probably find resources to help get this thing going faster.
|
|
|
|
|