Author Topic: [ANN] Sfards upload the latest programs of SF3301 [Updated 13/05/15]  (Read 3907 times)

Offline Sfards

  • Newbie
  • *
  • Posts: 9
Update 2/11/2015

SF100, the first 28nm Dual-Mode Miner is accepting pre-order now

Sfards, the only mining company has received venture capital investment, now announces that its Sha-256 and Script dual mining chip, using FD-SOI technology, is successfully taped-out and has been massively produced in the second quarter of 2015. It’s the only 28nm multi-algorithm chip, supports both Bitcoin and Litecoin (or other crypto currency using script).

Our SF100 miner (with 24 SF3301 chips) has already accomplished scale deployment and sales to big miners and investors. It achieves 1.7THS BTC hashing power (with 720w power consumption), 54MHS LTC (with 250w, power consumption), in total 960w power consumption at wall (based on test with 90% conversion rate power supply unit).

The Script hashing power of SF3301 is one-of-the-kind in global market. Its litecoin return is 3-5 times of its bitcoin return. Using SF100, Investors can have return on two coins at the same time, the feature is very beneficial, especially when litecoin hashing power increasing rate is very low and stable, and next halving is going to be 4 years later, which means by investing SF100, miners and have shorter ROI and longer shelf time.

Sfards has noticed that recent market of crypto currency has been much more dynamic, therefore, we have placed more chips order from foundry.
Sfards now accepts pre-order of chips and miners from global investors, delivery time is January, 2016.
We also have limited amount of chips and miners in stock, please contact us for details, thanks.

For more details, please refer to:
Website: www.sfards.com
Introduction of chip: http://www.sfards.com/pro/chip
Introduction of miner: http://www.sfards.com/pro/miner
Open sources: https://github.com/sfards
Contact:
Email: [email protected]


Please note, Sfards Technology has not authorized any reseller yet, the companies claiming they have chips or miners for sale might be committing fraud. We suggest you only deal with companies or individuals you have complete trust, if you have doubts, please directly contact us to inquire their soundness.












Update 15/7/2015
Please Be Careful!!!!
We only have one official SFARDS Skype account sfardsservice. If there is any other similar account such as sfardsservice [ASIC market] contacting you. Don’t listen to them and please double check the account you have talk with and make sure you add the right one :)




Update 23/6/2015

SF100-the first 28nm Dual-Mode Miner is getting into mass production

With the experience of developing dual-mode chip and miner manufacturing  succeeded from Gridseed, Sfards Technology has successfully developed and produced the first FD-SOI 28nm SHA-256 (Bitcoin) and Scrypt ( Litecoin ) dual-mode mining chip SF3301, in massive scale. After more than one month evaluating, prototyping, and extreme-condition pressure testing, the first 28nm dual-mode miner, SF100, is about to be produced massively.



SF100 dual-mode miner has 24 SF3301 chips, with the running modes as following:

  • Dual mining mode, Bitcoin hashing power 1700 GHS + Litecoin hashing power 47 MHS, the power consumption at wall of whole miner is 960W.
  • Bitcoin single mining mode: Hashing power 1700 GHS, power consumption 770W.
  • Litecoin single mining mode: Hashing power 47 MHS, power consumption 180W.
   
With 0.45W/GHS + 3.8W/MHS, the power efficiency of SF100 is by far the best in the market. It is also possible to overclock to certain extent (Sha-256: 2T, Scrypt: 55M).

The unique ability of dual-mode mining makes SF100 the best choice of mining community. It offers more choices to dig potentials not only of Bitcoin, but also of Litecoin and other altcoins. For fiat currency investors, dual-mode is good for them because SF100 maximizes their profits; for Bitcoin“fundamentalists”, it’s also preferable since the altcoin revenue helps them to cover the power cost of mining Bitcoin.

Based on recent Bitcoin and Litecoin price, a dual-mode SF100’s revenue equals to a 2.8-3.2T pure Bitcoin miner, which makes it 0.3-0.34W/GHS power efficiency. (Take June 19th price for instance, one SF100 generates 4.2 USD from Bitcoin and 3.8 USD from Litecoin, in total 8 USD per day).

Sfards Technology will be selling SF100 to public on July; the first batch quantity is limited. The SF3301 chip will be available for sale on August.

If you are interested at our products, please send email to [email protected]





Update 13/5/2015


Sfards upload the latest programs of SF3301 chips. Welcome to download the resources on GitHub.

More details: https://github.com/sfards/ASIC-SF3301





Update 11/05/15:

Welcome to join our latest contest, 1BTC TROPHY WAITING FOR YOU!     https://bitcointalk.org/index.php?topic=1057425.0




Update 04/05/15:

SFARDS’ SF3301 sample chip is available on May 4th

SFARDS is now selling a few SF3301 development boards and sample chips to developers who wish to customize and build their own hardware.


Sales Information:

Sample chip:                      200 USD per pack, include two chips (exclude shipping)
Development board set:      400 USD (exclude shipping)
Shipping info:                     Within 48 hours after full payment (Payment by Bitcoin or USD)

For more questions such as place order or large scale orders, please email [email protected] or add our skype:sfardsservice.

Features:

Sample chip:
More details about SF3301 chips, please kindly find the datasheet.
https://github.com/sfards

Development board set (includes one dev board, two power supply boards, radiator and fan):
Main Board
SF3301 : 1 chip
Heatsink : heatsink & 12mmx12mm fan
FBB Voltage : +/- 0~1.0V
Communication Interface : RS232 / TTL

Power Supply Board:
Output Voltage Range : 0.5~1.6V
Voltage Step:6.25mV
Max Current : 120A


Update 30/4/15:

SFARDS is now opening up the ASIC SF3301's datasheet for developers and anyone interested in the technology. One of our main goals is to allow numerous developers to test our new technology in a variety of environments, allowing for a truer test of the SF3301's boundaries and possibly finding an even more optimal running condition that we have yet discovered.

In addition to this we have made a number of sample chips available, more details including pricing to be released on 4th May.

Datasheet download website as follows:
https://github.com/sfards


Update 14/4/15:

Thank you all for waiting so patiently, today we'd like to announce some long awaited specifications for the SF3301!



The SF3301 fully utilizes the advantages of the FD-SOI technology. With increased forward body bias the chip is operational at lower voltages while maintaining a higher frequency. The chip boasts impressive power efficiency while affording high hash power, allowing for much lower wastage per hash. The ASIC’s lowest working voltage is 0.45V, with lowest BTC efficiency at 0.19J/GH, and lowest LTC efficiency at 1.75J/MH. In current market conditions compared to other mining technology, dual-algorithm capabilities can provide a clear advantage.

After strenuous testing of the SF3301, we reached hashrates of 152GH/s for SHA-256 and 3.17MH/s for Scrypt.

This month we will also be releasing a number of documents for the SF3301 as open source including the ASIC Development Board designs.

Next month we will release some design specifications for our debut miner.

Additionally in the near future SFARDS will be selling ASIC development boards and sample chips, making the SF3301 accessible to developers who wish to customize and build their own hardware. SFARDS welcomes any and all developers and communities to build upon our technology and contribute to the mining ecosystem.

Please contact [email protected] for more information.

Update 10/3/15:

After much hard work we’re happy to announce a brief progress update regarding our debut technology.

Firstly, tape out of our 28nm SF3301 semiconductor is currently underway. We are expecting to release our first miner in April 2015.

The SF3301 is the first of a new generation of dual-algorithm mining chips, the world’s first 28nm with the capability to mine SHA-256 and Scrypt simultaneously. The advantages of this ability translate to an extended period of effectiveness for the miner and increased profitability.

By using FDSOI technology not currently present in any other chip on the market, power efficiency when mining with SHA-256 is predicted to reach below the 0.3 J/GH range, and 2.0 J/MH when mining with Scrypt.

The SF3301 is extremely versatile, allowing you to mine multiple currencies with one device. This means price fluctuations can be harnessed and being forced to turn off your miner is far less likely.

A condition can arise where altcoins are utilized to cover electricity costs while bitcoin is continuously and simultaneously being mined.

This ability is something never seen before with 28nm technology.

Stay tuned for more updates soon!





We are SFARDS.

We are a new player in the mining industry but with us we bring some hefty experience in a shiny new wrapper.

Sfards is the result of the recent merger between Gridseed and WiiBox. We are funded by Li Xiao Lai (one of China's most influential bitcoin figures and an avid angel-investor) and have garnered substantial backing from a tier one VC firm.

Through the joining of these two companies we inherited the talent that conceived of the world’s first dual-miner machine. We plan to launch miners that have reached the next stage of ASIC technology in terms of both power and efficiency.

We are also expanding the capabilities of the WiiBox: an open source mining hardware control system.

Sfards will be involved in a few different areas of bitcoin mining. What's in store:

- We are currently working on semiconductor research, and plan to bring to the market some new ASIC technology.
- We plan to use this technology to build mining hardware.
- An innovative hash rate rental platform.

Please check out our website: www.sfards.com
or shoot us an e-mail: [email protected]

Social Media Profiles:
Facebook - www.facebook.com/sfardstech
Twitter - www.twitter.com/sfardstech
Weibo - www.weibo.com/sfards
WeChat - SFARDS
LinkedIn - www.linkedin.com/company/sfards-technology-ltd
« Last Edit: November 02, 2015, 11:27:26 AM by Sfards »

Offline Sy

  • Legendary Member
  • ******
  • Posts: 5178
Re: [ANN] Sfards: SF3301 Specifications Revealed [Updated 14/4/15]
« Reply #1 on: April 16, 2015, 06:10:23 AM »
Any first hand feedback?

Offline Searing

  • Full Member
  • ***
  • Posts: 312
  • Clueless
Re: [ANN] Sfards: SF3301 Specifications Revealed [Updated 14/4/15]
« Reply #2 on: April 16, 2015, 06:26:32 AM »


look at the thread on bitcointalk...they claim they are gonna ship in april ...but nada.....does not look good for this getting off the ground with the current BTC price imho

not a slam..just saying

Offline wiggl3r

  • Sr. Member
  • ****
  • Posts: 1217
Re: [ANN] Sfards: SF3301 Specifications Revealed [Updated 14/4/15]
« Reply #3 on: April 16, 2015, 01:39:12 PM »
So IF true.

220 watts per 1155 GH/s compared to Antminer S5 which is 590.

450 watts per 256 MH/s compared to the Alchminer at 2200.

Seems WAY to good to be true coming from a 28 nm chip.


Offline Sukarti

  • Hardcore Member
  • *****
  • Posts: 3418
Re: [ANN] Sfards: SF3301 Specifications Revealed [Updated 14/4/15]
« Reply #4 on: April 16, 2015, 02:11:02 PM »
So IF true.

220 watts per 1155 GH/s compared to Antminer S5 which is 590.

450 watts per 256 MH/s compared to the Alchminer at 2200.

Seems WAY to good to be true coming from a 28 nm chip.

These aren't legit chips yet, still in design/simulation phase. Long way to go before they're actually hashing.

I agree the numbers are awfully optimistic.

Guess we'll see.

Offline J4bberwock

  • Jr. Member
  • **
  • Posts: 27
Re: [ANN] Sfards: SF3301 Specifications Revealed [Updated 14/4/15]
« Reply #5 on: April 16, 2015, 02:32:15 PM »
So IF true.

220 watts per 1155 GH/s compared to Antminer S5 which is 590.

450 watts per 256 MH/s compared to the Alchminer at 2200.

Seems WAY to good to be true coming from a 28 nm chip.

They are the best figure, and J/GH isn't the same as w/GH at the wall.
You can usually expect 15-25% more than the chip efficiency because of the losses from voltage conversion, in the power supply, and on board to go from 12v to 0.5-1.0v

Those values are looking realistic to me, I'm currently running my good old G-Blade at 11-12w/MH efficiency at the wall on LTC mode, passive cooling.
If it was worth it, I could even go under 10w/MH at the wall by designing a new board for the GC3355 chips.

Offline Sukarti

  • Hardcore Member
  • *****
  • Posts: 3418
Re: [ANN] Sfards: SF3301 Specifications Revealed [Updated 14/4/15]
« Reply #6 on: April 16, 2015, 02:42:48 PM »
So IF true.

220 watts per 1155 GH/s compared to Antminer S5 which is 590.

450 watts per 256 MH/s compared to the Alchminer at 2200.

Seems WAY to good to be true coming from a 28 nm chip.

They are the best figure, and J/GH isn't the same as w/GH at the wall.
You can usually expect 15-25% more than the chip efficiency because of the losses from voltage conversion, in the power supply, and on board to go from 12v to 0.5-1.0v

Those values are looking realistic to me, I'm currently running my good old G-Blade at 11-12w/MH efficiency at the wall on LTC mode, passive cooling.
If it was worth it, I could even go under 10w/MH at the wall by designing a new board for the GC3355 chips.

15-25% is expected. The issue is most existing Scrypt ASIC manufacturers came in at 300% power usage.

That's the disparity we are all concerned about.

I hope Sfards can produce miners without a single pre-order given their financing.

Guess we'll see in the next 6 months or so.

Offline J4bberwock

  • Jr. Member
  • **
  • Posts: 27
Re: [ANN] Sfards: SF3301 Specifications Revealed [Updated 14/4/15]
« Reply #7 on: April 16, 2015, 03:16:50 PM »
So IF true.

220 watts per 1155 GH/s compared to Antminer S5 which is 590.

450 watts per 256 MH/s compared to the Alchminer at 2200.

Seems WAY to good to be true coming from a 28 nm chip.

They are the best figure, and J/GH isn't the same as w/GH at the wall.
You can usually expect 15-25% more than the chip efficiency because of the losses from voltage conversion, in the power supply, and on board to go from 12v to 0.5-1.0v

Those values are looking realistic to me, I'm currently running my good old G-Blade at 11-12w/MH efficiency at the wall on LTC mode, passive cooling.
If it was worth it, I could even go under 10w/MH at the wall by designing a new board for the GC3355 chips.

15-25% is expected. The issue is most existing Scrypt ASIC manufacturers came in at 300% power usage.

That's the disparity we are all concerned about.

I hope Sfards can produce miners without a single pre-order given their financing.

Guess we'll see in the next 6 months or so.

300% over the "best efficiency" is unfortunately easy to understand, it's a tradeoff between number of chips and hashing power.
End user wants a lot of hashing power, manufacturer wants to reduce the costs, and the asic chips are usually 50-70% of the total cost of a miner.
So, as a manufacturer, you have the choice of selling the same hashing power with worst efficiency for let's say 1000$, medium efficiency would cost you 1500$, and best efficiency will cost 2000$.
Depending on how healthy the market is, and what the other manufacturers are selling, you'll usually see the "medium efficiency" option sold.

The G-Blade was capable of over 7MH/s, but I was using close to 400w, definitely the worst efficiency we can get from those chips.
Stock value was 5.2MH, 100w, that's the medium efficiency option
I'm currently running 3.4MH @ 40w, that's close to the best results you can get with it.

The question is, 1 year ago, if you had the option to buy:
5.2MH @100w
3.4MH @40w
which one would have been your choice?

Seeing the interest I catched when I posted how to overclock, and how many (poorly) overclocked G-blades were sold by a reseller, I have no doubt that the 5.2MH (or even 6MH light overclock) is the answer.

I believe that, if the chips efficiency is confirmed, SFARDS will sell something close to 100GH-2.1MH per chip, with an efficiency close to 0.4w/MH in BTC mode, and 3.5w/MH in LTC mode.
We'll still have options to overclock/underclock.

Offline Sfards

  • Newbie
  • *
  • Posts: 9
Re: [ANN] Sfards: SF3301 Datasheet Open Source [Updated 30/4/15]
« Reply #8 on: April 30, 2015, 10:13:43 AM »
SFARDS is now opening up the ASIC SF3301's datasheet for developers and anyone interested in the technology. One of our main goals is to allow numerous developers to test our new technology in a variety of environments, allowing for a truer test of the SF3301's boundaries and possibly finding an even more optimal running condition that we have yet discovered.

In addition to this we have made a number of sample chips available, more details including pricing to be released on 4th May.

Datasheet download website as follows:
https://github.com/sfards

Offline Sfards

  • Newbie
  • *
  • Posts: 9
Re: [ANN] Sfards: SF3301 Datasheet Open Source [Updated 30/4/15]
« Reply #9 on: April 30, 2015, 10:14:55 AM »
SFARDS is now opening up the ASIC SF3301's datasheet for developers and anyone interested in the technology. One of our main goals is to allow numerous developers to test our new technology in a variety of environments, allowing for a truer test of the SF3301's boundaries and possibly finding an even more optimal running condition that we have yet discovered.

In addition to this we have made a number of sample chips available, more details including pricing to be released on 4th May.

Datasheet download website as follows:
https://github.com/sfards

Since the Chinese Labour Day is coming, we will out of office until May 4th, so more details and replies will be followed from then on. PS: That's vacation just for marketing colleagues, as for our engineer, they still hard working day and night for your guys  ;).

Offline J4bberwock

  • Jr. Member
  • **
  • Posts: 27
Re: [ANN] Sfards: SF3301 Specifications Revealed [Updated 14/4/15]
« Reply #10 on: April 30, 2015, 06:42:36 PM »
So IF true.

220 watts per 1155 GH/s compared to Antminer S5 which is 590.

450 watts per 256 MH/s compared to the Alchminer at 2200.

Seems WAY to good to be true coming from a 28 nm chip.

They are the best figure, and J/GH isn't the same as w/GH at the wall.
You can usually expect 15-25% more than the chip efficiency because of the losses from voltage conversion, in the power supply, and on board to go from 12v to 0.5-1.0v

Those values are looking realistic to me, I'm currently running my good old G-Blade at 11-12w/MH efficiency at the wall on LTC mode, passive cooling.
If it was worth it, I could even go under 10w/MH at the wall by designing a new board for the GC3355 chips.

15-25% is expected. The issue is most existing Scrypt ASIC manufacturers came in at 300% power usage.

That's the disparity we are all concerned about.

I hope Sfards can produce miners without a single pre-order given their financing.

Guess we'll see in the next 6 months or so.

300% over the "best efficiency" is unfortunately easy to understand, it's a tradeoff between number of chips and hashing power.
End user wants a lot of hashing power, manufacturer wants to reduce the costs, and the asic chips are usually 50-70% of the total cost of a miner.
So, as a manufacturer, you have the choice of selling the same hashing power with worst efficiency for let's say 1000$, medium efficiency would cost you 1500$, and best efficiency will cost 2000$.
Depending on how healthy the market is, and what the other manufacturers are selling, you'll usually see the "medium efficiency" option sold.

The G-Blade was capable of over 7MH/s, but I was using close to 400w, definitely the worst efficiency we can get from those chips.
Stock value was 5.2MH, 100w, that's the medium efficiency option
I'm currently running 3.4MH @ 40w, that's close to the best results you can get with it.

The question is, 1 year ago, if you had the option to buy:
5.2MH @100w
3.4MH @40w
which one would have been your choice?

Seeing the interest I catched when I posted how to overclock, and how many (poorly) overclocked G-blades were sold by a reseller, I have no doubt that the 5.2MH (or even 6MH light overclock) is the answer.

I believe that, if the chips efficiency is confirmed, SFARDS will sell something close to 100GH-2.1MH per chip, with an efficiency close to 0.4w/MH in BTC mode, and 3.5w/MH in LTC mode.
We'll still have options to overclock/underclock.

It seems I was guessing quite close to the actual specs:
Voltage(V)  Frequency(MHz)  FBB(V)  Power(W)  GHash  W/GHash/s
0.8                  870                ±0.6       77.68     139.20     0.56
0.6                  450                ±1.1       23.28      72.00      0.32

Voltage(V)  Frequency(MHz)  Power(W)  MHash  W/MHash/s
0.8                  450                 2.18         1.21       1.79
0.9                  700                 4.34         1.89       2.30

The LTC mode efficiency is even better than I expected.

It's now time for some work on an homemade board with 2 different power circuits to allow voltage/efficiency adjustment for each mode if you want to run both at the same time.

I did some google-ing on the body bias adjust thing, and it seems a really nice trick to allow better efficiency.

How is it used in the chip? triggered by software or automatic?
I can't see anything in the BTC registers regarding the FBB setting.

Offline jadefalke

  • Sr. Member
  • ****
  • Posts: 566
  • it's only me Honey, don't worry
Re: [ANN] Sfards: SF3301 Datasheet Open Source [Updated 30/4/15]
« Reply #11 on: April 30, 2015, 07:04:44 PM »
that looks really promissing, i hope the yield rate of this chips is good an we will see them in the wild soon.

Offline Sfards

  • Newbie
  • *
  • Posts: 9
Re: [ANN] Sfards: SF3301 sample chip is available [Updated 04/05/15]
« Reply #12 on: May 04, 2015, 10:02:13 AM »
Update 04/05/15:

SFARDS’ SF3301 sample chip is available on May 4th

SFARDS is now selling a few SF3301 development boards and sample chips to developers who wish to customize and build their own hardware.


Sales Information:

Sample chip:                      200 USD per pack, include two chips (exclude shipping)
Development board set:      400 USD (exclude shipping)
Shipping info:                     Within 48 hours after full payment (Payment by Bitcoin or USD)

For more questions such as place order or large scale orders, please email [email protected] or add our skype:sfardsservice.

Features:

Sample chip:
More details about SF3301 chips, please kindly find the datasheet.
https://github.com/sfards

Development board set (includes one dev board, two power supply boards, radiator and fan):
Main Board
SF3301 : 1 chip
Heatsink : heatsink & 12mmx12mm fan
FBB Voltage : +/- 0~1.0V
Communication Interface : RS232 / TTL

Power Supply Board:
Output Voltage Range : 0.5~1.6V
Voltage Step:6.25mV
Max Current : 120A

Offline Marvell9

  • Elite Member
  • ****
  • Posts: 1363
Re: [ANN] Sfards: SF3301 sample chip is available [Updated 04/05/15]
« Reply #13 on: May 04, 2015, 03:55:31 PM »
Thats a very expensive development board, but i suspect pricing is tailored towards deep pockets hardware developers  but you have to realize that hobyist developers could also develop hardware for the community with your chips and price the dev kits appropriately.

$200 per chip is scary and $400 for a board that can only be used to test 1 chip ?!

Offline J4bberwock

  • Jr. Member
  • **
  • Posts: 27
Re: [ANN] Sfards: SF3301 sample chip is available [Updated 04/05/15]
« Reply #14 on: May 06, 2015, 07:14:04 AM »
Thats a very expensive development board, but i suspect pricing is tailored towards deep pockets hardware developers  but you have to realize that hobyist developers could also develop hardware for the community with your chips and price the dev kits appropriately.

$200 per chip is scary and $400 for a board that can only be used to test 1 chip ?!

200$ for 2 chips (not really better), but I already have a 2 chips board design almost finished, and I'll take 4 chips to make my tests.