Bitcoin Forum
June 16, 2024, 06:30:18 PM *
News: Latest Bitcoin Core release: 27.0 [Torrent]
 
   Home   Help Search Login Register More  
Warning: One or more bitcointalk.org users have reported that they strongly believe that the creator of this topic is a scammer. (Login to see the detailed trust ratings.) While the bitcointalk.org administration does not verify such claims, you should proceed with extreme caution.
Pages: « 1 2 3 4 [5] 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 ... 58 »
  Print  
Author Topic: Dwarf FPGA – the anti-ASIC  (Read 43405 times)
shaninium
Sr. Member
****
Offline Offline

Activity: 489
Merit: 253



View Profile
May 21, 2018, 08:47:19 PM
 #81

Come on guys... give the OP the benefit of doubt until proven otherwise.  If the original quoted pricing holds, what's $350 in the big scheme of things?  I wasted more than that by dropping a GPU that no longer works... LOL!

When they do go on sale, I'd bet someone here is willing to gamble $350 on a few of them!

That's a bad plan but you're welcome to go forwards with it and you can also let us know how it worked out for you  Grin

If things can be easily tweaked to avoid possible scams why take the risk?

This isn't Vegas and the roulette table, it's science  Cheesy

Thats how scams work, the ones that dont see 350 as a big loss if they are scammed are the ones the scammer depends on. It would be nieve to think bitmain wont enter this FPGA mining scene after the x3 debacle. They already use them for other things.

vlad230
Sr. Member
****
Offline Offline

Activity: 616
Merit: 279



View Profile
May 22, 2018, 07:38:07 AM
 #82

Thats how scams work, the ones that dont see 350 as a big loss if they are scammed are the ones the scammer depends on. It would be nieve to think bitmain wont enter this FPGA mining scene after the x3 debacle. They already use them for other things.
Yes, agreed. I wonder how many scams of 300 euros one can be part of until they realize they're throwing away money  Cheesy Now, I'm not saying this is a scam, I'll give OP the benefit of the doubt.

Bitmain will definitely enter the FPGA scene if they haven't already in secret. But since FPGAs are more comparable to GPUs (when it comes to hash rates) and not as powerful as ASICs, this could be a turn for the better.
smoolae
Full Member
***
Offline Offline

Activity: 602
Merit: 106



View Profile
May 22, 2018, 09:27:27 AM
 #83

Still, if any solid proof comes out, stating and proving that this little thing indeed exists and hashes as good as told, I will definitely buy it. It just seems like the perfect way to mine without using mad amount of electricity.

Really hope this turns out as a real deal Smiley

ZaebaloRegitsa4
Jr. Member
*
Offline Offline

Activity: 64
Merit: 1


View Profile
May 22, 2018, 09:10:06 PM
 #84

But like other users said, it is too good to be true.
I do not think so. Look at this article: https://www.iacr.org/archive/ches2005/031.pdf
The hash function has a 524288 cycles, if I understood everything correctly.
From the article:
"Two new FPGA designs for the Advanced Encryption Standard
(AES) are presented. The first is believed to be the fastest, achieving 25 Gbps
throughput using a Xilinx Spartan-III (XC3S2000) device. The second is
believed to be the smallest and fits into a Xilinx Spartan-II (XC2S15) device,
only requiring two block memories and 124 slices to achieve a throughput of
2.2 Mbps. "

It means, XC3S2000-5 for $78 only can generate 1490,12 h/s; https://www.digikey.com/product-detail/en/xilinx-inc/XC3S2000-5FGG456C/XC3S2000-5FGG456C-ND/1951750
+ SRAM memory like https://www2.mouser.com/ProductDetail/GSI-Technology/GS8160Z36DGT-150?qs=sGAEpiMZZMt9mBA6nIyysJQc0%252bdJyBlRwbATLGJ3Hts%3d, $14 for 2,25 megabyte.
About a hundred bucks for one and a half kh. And it's using the components from the online store. If they use an external AES coprocessor, this can work faster.

DwarfMiner, tell us more about the architecture of the system  Cheesy
smoolae
Full Member
***
Offline Offline

Activity: 602
Merit: 106



View Profile
May 22, 2018, 09:13:10 PM
 #85

But like other users said, it is too good to be true.
I do not think so. Look at this article: https://www.iacr.org/archive/ches2005/031.pdf
The hash function has a 524288 cycles, if I understood everything correctly.
From the article:
"Two new FPGA designs for the Advanced Encryption Standard
(AES) are presented. The first is believed to be the fastest, achieving 25 Gbps
throughput using a Xilinx Spartan-III (XC3S2000) device. The second is
believed to be the smallest and fits into a Xilinx Spartan-II (XC2S15) device,
only requiring two block memories and 124 slices to achieve a throughput of
2.2 Mbps. "

It means, XC3S2000-5 for $78 only can generate 1490,12 h/s; https://www.digikey.com/product-detail/en/xilinx-inc/XC3S2000-5FGG456C/XC3S2000-5FGG456C-ND/1951750
+ SRAM memory like https://www2.mouser.com/ProductDetail/GSI-Technology/GS8160Z36DGT-150?qs=sGAEpiMZZMt9mBA6nIyysJQc0%252bdJyBlRwbATLGJ3Hts%3d, $14 for 2,25 megabyte.
About a hundred bucks for one and a half kh. And it's using the components from the online store. If they use an external AES coprocessor, this can work faster.

DwarfMiner, tell us more about the architecture of the system  Cheesy

Sounds great Smiley

Kahooli
Newbie
*
Offline Offline

Activity: 58
Merit: 0


View Profile
May 22, 2018, 09:43:51 PM
 #86

But like other users said, it is too good to be true.
I do not think so. Look at this article: https://www.iacr.org/archive/ches2005/031.pdf
The hash function has a 524288 cycles, if I understood everything correctly.
From the article:
"Two new FPGA designs for the Advanced Encryption Standard
(AES) are presented. The first is believed to be the fastest, achieving 25 Gbps
throughput using a Xilinx Spartan-III (XC3S2000) device. The second is
believed to be the smallest and fits into a Xilinx Spartan-II (XC2S15) device,
only requiring two block memories and 124 slices to achieve a throughput of
2.2 Mbps. "

It means, XC3S2000-5 for $78 only can generate 1490,12 h/s; https://www.digikey.com/product-detail/en/xilinx-inc/XC3S2000-5FGG456C/XC3S2000-5FGG456C-ND/1951750
+ SRAM memory like https://www2.mouser.com/ProductDetail/GSI-Technology/GS8160Z36DGT-150?qs=sGAEpiMZZMt9mBA6nIyysJQc0%252bdJyBlRwbATLGJ3Hts%3d, $14 for 2,25 megabyte.
About a hundred bucks for one and a half kh. And it's using the components from the online store. If they use an external AES coprocessor, this can work faster.

DwarfMiner, tell us more about the architecture of the system  Cheesy

I think you're completely ignoring memory latency for implementing the hash function.
GPUHoarder
Member
**
Offline Offline

Activity: 154
Merit: 37


View Profile
May 23, 2018, 01:05:23 AM
 #87

But like other users said, it is too good to be true.
I do not think so. Look at this article: https://www.iacr.org/archive/ches2005/031.pdf
The hash function has a 524288 cycles, if I understood everything correctly.
From the article:
"Two new FPGA designs for the Advanced Encryption Standard
(AES) are presented. The first is believed to be the fastest, achieving 25 Gbps
throughput using a Xilinx Spartan-III (XC3S2000) device. The second is
believed to be the smallest and fits into a Xilinx Spartan-II (XC2S15) device,
only requiring two block memories and 124 slices to achieve a throughput of
2.2 Mbps. "

It means, XC3S2000-5 for $78 only can generate 1490,12 h/s; https://www.digikey.com/product-detail/en/xilinx-inc/XC3S2000-5FGG456C/XC3S2000-5FGG456C-ND/1951750
+ SRAM memory like https://www2.mouser.com/ProductDetail/GSI-Technology/GS8160Z36DGT-150?qs=sGAEpiMZZMt9mBA6nIyysJQc0%252bdJyBlRwbATLGJ3Hts%3d, $14 for 2,25 megabyte.
About a hundred bucks for one and a half kh. And it's using the components from the online store. If they use an external AES coprocessor, this can work faster.

DwarfMiner, tell us more about the architecture of the system  Cheesy

Your sram supports a tiny bandwidth of 36bit*260 in read only, In read/write turnaround it only supports 133Mhz. That’s only 600MB/s. Since it takes 32MB/s for each hash you’re only getting 20H/s or worse.

AES is not the bottleneck.
squallw
Full Member
***
Offline Offline

Activity: 504
Merit: 122


View Profile
May 23, 2018, 06:00:10 PM
 #88

But like other users said, it is too good to be true.
I do not think so. Look at this article: https://www.iacr.org/archive/ches2005/031.pdf
The hash function has a 524288 cycles, if I understood everything correctly.
From the article:
"Two new FPGA designs for the Advanced Encryption Standard
(AES) are presented. The first is believed to be the fastest, achieving 25 Gbps
throughput using a Xilinx Spartan-III (XC3S2000) device. The second is
believed to be the smallest and fits into a Xilinx Spartan-II (XC2S15) device,
only requiring two block memories and 124 slices to achieve a throughput of
2.2 Mbps. "

It means, XC3S2000-5 for $78 only can generate 1490,12 h/s; https://www.digikey.com/product-detail/en/xilinx-inc/XC3S2000-5FGG456C/XC3S2000-5FGG456C-ND/1951750
+ SRAM memory like https://www2.mouser.com/ProductDetail/GSI-Technology/GS8160Z36DGT-150?qs=sGAEpiMZZMt9mBA6nIyysJQc0%252bdJyBlRwbATLGJ3Hts%3d, $14 for 2,25 megabyte.
About a hundred bucks for one and a half kh. And it's using the components from the online store. If they use an external AES coprocessor, this can work faster.

DwarfMiner, tell us more about the architecture of the system  Cheesy

I dont think that things work like you said.
Even if $100 = 500h/s for 7 kh/s will cost $700 and OP wanna sell for $350.

But it is very simple, like i said before, OP just need to give more information and prove the truth.
mettalmag
Legendary
*
Offline Offline

Activity: 1084
Merit: 1003


≡v≡


View Profile
May 23, 2018, 07:31:28 PM
 #89

interested
would like to see pics or videos here and not in zip files
also if there's some kind of waiting list would like to be there for MOQ
cheers
Goool
Jr. Member
*
Offline Offline

Activity: 182
Merit: 2


View Profile
May 23, 2018, 10:23:20 PM
 #90

We are a group of developers from different countries. After working over the last six months we are ready to present the result - an ecological "green" miner.
Usually, ASICs are designed for “industrial mining” and they are expensive. In addition, these devices threaten decentralization. Changing the mining algorithm involves the risk to not recoup investment.
GPU mining is more flexible but produces a lot of heat and noise. This accelerates global warming and increases the owner's costs. This forms a feud with gamers!  Grin

We have developed an intermediate product. This small device is able to replace one rig and produce 7 kh /  10 W on the v7 and 3.5 kh / 8W on the heavy.  https://drive.google.com/open?id=1h_6uMPuAfIud_22hsuTlX7IHwLfanmIW
Dwarf supports decentralization because personal mining will now become profitable in the countries with high electricity prices. The device is of low cost and does not threaten large losses if the sudden change in the algorithm happens. This is available to anyone who does not want to pay thousands dollars for ASIC.

Dwarf can be updated. If the next fork will be similar to the previous one changing the firmware will allow the device to function further. This can be done remotely using a unique device encrypted key. We plan to sell updates for a small price.

Now we are negotiating with the factory to produce 5000 units. We expect about 300 euro for each and can tell exact price within 10 days. The sale will be launched soon.
If you have questions about a major purchase, please write to the official mailbox (this is a unique project mailbox!) deleted@gmail.com .   dwarfminersales@hotmail.com

I regret to inform you f-n Google has blocked our mail.  Angry Too many letters came and this was regarded as commercial activity.
Now we do not have access to letters, they are all lost. Please forward your questions to our new mail dwarfminersales@hotmail.com
Now all your emails will automatically be forwarded to duplicate mailboxes and the situation should not be repeated.
We're sorry for what happened...


Why do we sell this? Because this board is not designed for industrial mining. It's not convenient way to use a thousand devices for one person. It is designed as an anti-ASIC for personal use in the living area.

Newbie with activity 1 selling ... ----> SCAM
ZaebaloRegitsa4
Jr. Member
*
Offline Offline

Activity: 64
Merit: 1


View Profile
May 27, 2018, 08:00:57 PM
 #91

I think you're completely ignoring memory latency for implementing the hash function.
I gave the example of the simplest and cheapest memory. You can find faster memory, distribute it in parallel. I used memory inside the FPGA and achieved 3.5 kh. But it was more expensive, of course.
DwarfMiner (OP)
Jr. Member
*
Offline Offline

Activity: 42
Merit: 1


View Profile
May 27, 2018, 10:16:03 PM
 #92

Hi all!

--This is very efficient. almost as efficient as the ASIC.
ASIC uses an array of chips for operation. You can see it on the baikal boards. We use one chip. It consumes as much as in ASIC.

--The circuit board doesn't look that exotic.  Doesn't need MIL-spec anything. 
This is a eight-layer board. All the magic is inside. Most chips have BGA packs. This can be mounted automatically only, with X-ray control.

--If they give it a good review, you'll be flooded with orders.  Your bankers will be impressed.
We are flooded with orders now. Five thousand is too small batch to wait for a review and wasting time up to fork. We made a discount now to cover the risks. The next batch will be bigger and more expensive.

-- Please put me on the waiting list  Grin thx
We have free pre-order. But the batch has not yet sold out Wink

--Your board looks strikingly similar to the first x11 fpga's, except they needed a fan to cool.
in a warm climate it need a fan too

--de hecho, por las fotos, parece que somos "vecinos"
algunas personas - si Wink

--Well, guys, if you are in the Euro zone, that change a lot.
Developers are located around the world but the leader is in Madrid.

-- I will fly in anywhere in Europe to see samples
there are only prototypes and they are all in China now. The whole batch will be made and sent at one time.

--DwarfMiner, tell us more about the architecture of the system  Cheesy
We use a hybrid chip containing FPGA, memory and AES coprocessor. This allows to change the firmware, in contrast to ASIC, in which the entire chip is made on a single crystal.
This means all typical elements are fast, but the algorithm section is made flexible.

--Even if $100 = 500h/s for 7 kh/s will cost $700 and OP wanna sell for $350.
iPhone about three times cheaper than its component parts cost in stores. Because the factory buys millions chips.
We have released a quarter of a million chips. It's much cheaper than buying one IC.

--would like to see pics or videos here and not in zip files
tomorrow

--Newbie with activity 1 selling ... ----> SCAM
Activity: 98 this is much better.
DwarfMiner (OP)
Jr. Member
*
Offline Offline

Activity: 42
Merit: 1


View Profile
May 28, 2018, 09:22:29 AM
Last edit: May 28, 2018, 09:59:29 AM by DwarfMiner
 #93

Hello everybody!

We are happy to inform you we will start selling!

We hope to answer faster in the sales period Smiley

A lot of time was spent to the organization of production. It was necessary to arrange the delivery our chips to China, discuss the terms of assembly rules, find a replacement for European components in Asia. Instead of one week, we spent four. But now the whole chain is built!

Yesterday we have begun manufacture of five thousand pieces. It will be ready on June 18. After that we will begin to send orders in the order of payments (first paid - first received). To cover your risks we offer a discount during this period.

Work demonstration you can see on our youtube channel: https://www.youtube.com/channel/UCz6buvT87nX-NuEkfcrNQGg
It does not have a fee, but updates will be charged. We expect $20-50 for each new algo. At now it is tested with monero, graft, haven, sumo and loki. The same coins should work too.

Our online store is available here: https://dwarfminer.webasyst.cloud/shop/  You can find there description and characteristics of the good. You must add to cart the number of devices and delivery to your country. The cost of delivery includes no more than four units (a parcel weight limit is 500 grams). If you need more, please contact us: dwarfminersales@hotmail.com

All orders (except monero payment) are made through the manager to eliminate confusion. The exchange rates will be fixed at the order's time.
skablast
Member
**
Offline Offline

Activity: 476
Merit: 19


View Profile
May 28, 2018, 10:35:53 AM
 #94

its not clear when will be shipped, well, to be precise:
its not clear who(company in china ?/europe?)  is going  to invoice, when payment will be requested compared to when the product will be shipped
smoolae
Full Member
***
Offline Offline

Activity: 602
Merit: 106



View Profile
May 28, 2018, 11:39:26 AM
 #95

Really want to buy a couple of those boards and you have also included the option to pay with PayPal. This just seems like a way to go with mining Smiley

melpheos
Jr. Member
*
Offline Offline

Activity: 557
Merit: 5


View Profile
May 28, 2018, 11:42:27 AM
 #96

Taking a leap of faith and put an order for 4...
I'll let you know in June if i got scammed :p
chris-swiss
Jr. Member
*
Offline Offline

Activity: 182
Merit: 4


View Profile
May 28, 2018, 12:39:02 PM
 #97

Hello ,

have make an order and make payement to PayPal, but you system have a bug, if you can be control its 4 FPGA for swiss country , and i don't have receveid email for valid my account .
thanks for your time
yrk1957
Member
**
Offline Offline

Activity: 529
Merit: 29


View Profile
May 28, 2018, 02:11:28 PM
 #98

Hello everybody!

We are happy to inform you we will start selling!

We hope to answer faster in the sales period Smiley

A lot of time was spent to the organization of production. It was necessary to arrange the delivery our chips to China, discuss the terms of assembly rules, find a replacement for European components in Asia. Instead of one week, we spent four. But now the whole chain is built!

Yesterday we have begun manufacture of five thousand pieces. It will be ready on June 18. After that we will begin to send orders in the order of payments (first paid - first received). To cover your risks we offer a discount during this period.

Work demonstration you can see on our youtube channel: https://www.youtube.com/channel/UCz6buvT87nX-NuEkfcrNQGg
It does not have a fee, but updates will be charged. We expect $20-50 for each new algo. At now it is tested with monero, graft, haven, sumo and loki. The same coins should work too.

Our online store is available here: https://dwarfminer.webasyst.cloud/shop/  You can find there description and characteristics of the good. You must add to cart the number of devices and delivery to your country. The cost of delivery includes no more than four units (a parcel weight limit is 500 grams). If you need more, please contact us: dwarfminersales@hotmail.com

All orders (except monero payment) are made through the manager to eliminate confusion. The exchange rates will be fixed at the order's time.

I looked up the address in the video at xmrpool and it shows 0 hahses submitted or 0 payments made!

455yPxusDp2EPBTWhLqmPS7ywgWvC8hJ7cgFL2ggpWRR6brgi9QS2yVZeBYHuWPpr9HfDkCVqnE5T11 LnTFgAD5DYN3CdD

Maybe the address I got is wrong or XMRPool does not shows older stats or ...
XrayHunter7
Newbie
*
Offline Offline

Activity: 67
Merit: 0


View Profile
May 28, 2018, 02:14:08 PM
 #99

DwarfMiner can you take off heatsink ? So We can see chip under heatsink.
XrayHunter7
Newbie
*
Offline Offline

Activity: 67
Merit: 0


View Profile
May 28, 2018, 03:18:29 PM
 #100

Hello everybody!

We are happy to inform you we will start selling!

We hope to answer faster in the sales period Smiley

A lot of time was spent to the organization of production. It was necessary to arrange the delivery our chips to China, discuss the terms of assembly rules, find a replacement for European components in Asia. Instead of one week, we spent four. But now the whole chain is built!

Yesterday we have begun manufacture of five thousand pieces. It will be ready on June 18. After that we will begin to send orders in the order of payments (first paid - first received). To cover your risks we offer a discount during this period.

Work demonstration you can see on our youtube channel: https://www.youtube.com/channel/UCz6buvT87nX-NuEkfcrNQGg
It does not have a fee, but updates will be charged. We expect $20-50 for each new algo. At now it is tested with monero, graft, haven, sumo and loki. The same coins should work too.

Our online store is available here: https://dwarfminer.webasyst.cloud/shop/  You can find there description and characteristics of the good. You must add to cart the number of devices and delivery to your country. The cost of delivery includes no more than four units (a parcel weight limit is 500 grams). If you need more, please contact us: dwarfminersales@hotmail.com

All orders (except monero payment) are made through the manager to eliminate confusion. The exchange rates will be fixed at the order's time.

I looked up the address in the video at xmrpool and it shows 0 hahses submitted or 0 payments made!

455yPxusDp2EPBTWhLqmPS7ywgWvC8hJ7cgFL2ggpWRR6brgi9QS2yVZeBYHuWPpr9HfDkCVqnE5T11 LnTFgAD5DYN3CdD

Maybe the address I got is wrong or XMRPool does not shows older stats or ...


Your adress is wrong. Its 455yPxusDp2EPBTWhLqmPS7ywgWvCY8hJ7cgFL2ggpWRR6brgi9QS2yVZeBYHuWPpr9HfDkCVqnE5T1 1LnTFgAD5DYN3CdD
It has 673 valid shares on xmrpool
Pages: « 1 2 3 4 [5] 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 ... 58 »
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!