All algos in X15,include the combination of 15 algos, ASIC is ready,they can change the firmware to update ，no need to re-tape
lyra2re2 ASIC is coming soon.
XZC will change algo soon.
LUX will change algo soon。
you will loss the battle field.
the most important thing is
There is no evidence to prove the FPGA hashrate and power consumption.
Anyone can purchase an FPGA and run the available open source RTL that exists now to see the general power consumption and performance. The issue is those without those skill set are waiting for someone else to show them. There are plenty of people on this forum that have experience and can confirm the hashrates and power are reasonable. Personally I’m not going to post detailed performance specs until the final piece of hardware I would sell someone is in my hand. Doing anything else is just grandstanding or bad business. If they can change the firmware for algorithms it isn’t an asic... I don’t doubt developments are coming though. The world doesn’t stand still. Just a little bit of picture or video of mining software and pool stats to prove it .but no a $10 power meter to measure the AC line. also no Anyone doubts his performance,is said to have no professional skills. You can make a lot of money with secret weapons，take a large amount of sales, the proceeds may not be high, and is likely some DEV change algo. I don't believe it. also I think it's a scam You can make a lot more money keeping things private, especially since whitefire990 isn’t selling anything. Also people developing this hardware don’t put a $10 meter on the AC line, they use PMBus and other professional tools to get exact power consumption of core, memory, etc. in the design. |
|
|
|
You’ve made two very aggressive anti-FPGA posts. Can you share your credentials and experience leading you to make such statements, or demonstrate why these things are not possible? You’ve made at least 4 claims that I personally know to be untrue.
Not agressive, pessimistic. FPGA cards like that exists from 2014 https://www.alpha-data.com/dcp/products.php?product=adm-pcie-ku3Many people(in Russia) pm'ed me and ask about this. They read this topic and think that it will be a GPU replacement. made at least 4 claims that I personally know to be untrue.
lyra2z will not fit fully pipelined (corrected). All other algos in my post _can_ be fitted and managed to run on fpga, but they will be to slow to ROI $4000 ($5000 soon) board. Anyone that thinks that FPGAs are going to destroy GPUs 10x is wrong, but being 10-200% more power and cost efficient total ROI - yes, that’s going to happen. I had a good conversation today I wish I could share, but sufficient to say that there are forces actively working to make sure the above remains true, and FPGAs have plenty of margin to play with to make that happen and continue to happen - GPUs have no room to move. There are also some major developments over the last 4 years. The 16nm vs 20nn fabric is a big changer, and if you look at the old spartix 6 night and day. SERDES speed is light years different now, to the point that the amount of logic that can fit on a single chip is absolutely not an issue. For every chip to chip pair you can handle 50+ internal bus bits at full fabric speed. Hard resources and URAM and even BRAM have exploded. FPGAs have gotten better, cheaper, and faster much more than GPUs have. The last open piece is high bandwidth access to big memory. It does exist now, but you have to architect it into your system. Within two years it will be there and there will no longer be a difference for that part. I also don’t think FPGAs are a centralization risk like ASICs, and as long as they remain closely competitive third hardware option I don’t think there will be a reason for teams to explicitly resist them. I haven’t gained a full sense of where the community stands in that however. Edit: are you trying to fit the whole chain in static logic fully pipelined? Lots of efficiency trade offs you make to do that, and it is not strictly necessary. Chip to chip FPGAs or look at partial reconfiguration options. Gains are possible. |
|
|
|
Hi,
Really interesting thread! I am beginning to mine using FGPA boards. I have been GPU and asic mining for the past year or so and have been interested in the FGPA side of things since hearing of it recently. I have purchased a Terasic DE0-Nano-SoC Development and Education Board to begin my learning. Can you say if any of your algo's will be compatible with the board mentioned? 1GB DDR3 ram and ARM dual processor.
Best,
Karl.
The DE0-Nano-SoC is a great board to get into learning this stuff. I have this and a few of the Max10 50kLE boards that I occasionally use for initial RTL hardware evaluation because their just so easy to work with. Vivado and Xilinx’s tools are steeper learning curve, but definitely doable. With that said, the Cyclone V 5CSEMA4U23C6N on that board is about 1/5th the logic of what I’m proposing and while it has high-speed DMA to the ARM cores, it doesn’t have any high speed IO such as PCIe to other peripherals. That makes it tough to use outside of single small algorithms (I.e. Keccak). The $199 version of the M.2 has ~100 Logic Elements and the $329 version has more than 200, capable of running at higher speed. The top package also has 1GB DDR3. Peak power consumption is around 15W from on-board M.2 Here’s an update summary for those catching up on this thread: 1. A few of us have been working on algorithms on FPGAs, and they’re profitable. Whitefire990 intents to release his with miner/dev fee and others of us have decided to share with the community in various forms as well. 2. On the high end the VCU1525, based on the VU9P is a very good candidate for this. It is currently a Xilinx development board on promotion for $3995, but only very small batches are being produced and the price is set to go up. Some algorithms require connecting multiple with high speed links to achieve the best performance. 3. Senseless (and possibly self, if I can be helpful) have been organizing essentially a group buy - but with FPGAs it’s less of a group buy and more of a group build. That is expected to ramp up the VCU1525 style (some power/cooling improvements) availblility in a similar price envelope to Xilinx dev version. Working on production in US and Europe so it can be available everywhere. 4. I’ve also decided to reveal one of the smaller FPGA options in the $200-350 price envelope that I had developed for internal use and deploy, to provide an entry level option. It has a slightly different set of capabilities from the big VCU1525, but for many things it does scale. For both hardware offerings from the community (and possibly others) orders are expected to start in June. These boards are not so much for Ethash (though they can be used to assist/accelerate GPUs in it) , or Equihash. They excel , can improve total system performance on, or are at an advantage on Keccak (and most SHA3 candidate) , Phi variants, NIST5, Timetravel10, Lyra/LyraRev2, etc. Someone else can chime in if I missed anything. Damn. Watching your stuff with interest. Got any specs on the smaller FPGA option? Which specs are you looking for? Device specs or specific algo performance? |
|
|
|
All algos in X15,include the combination of 15 algos, ASIC is ready,they can change the firmware to update ，no need to re-tape
lyra2re2 ASIC is coming soon.
XZC will change algo soon.
LUX will change algo soon。
you will loss the battle field.
the most important thing is
There is no evidence to prove the FPGA hashrate and power consumption.
Anyone can purchase an FPGA and run the available open source RTL that exists now to see the general power consumption and performance. The issue is those without those skill set are waiting for someone else to show them. There are plenty of people on this forum that have experience and can confirm the hashrates and power are reasonable. Personally I’m not going to post detailed performance specs until the final piece of hardware I would sell someone is in my hand. Doing anything else is just grandstanding or bad business. If they can change the firmware for algorithms it isn’t an asic... I don’t doubt developments are coming though. The world doesn’t stand still. |
|
|
|
What happens if coins move to ethash-like algorithms where the efficiency gains would be alot smaller?
That’s the nature of crypto, you have to decide what risks and opportunities are right for you. Currently what I’m proposing plays well in the acceleration role with GPUs for those algorithms. In 9-12 months both FPGAs and GPUs will be shipping with similar memory bandwidth and then FPGAs (imho) will be winning on power. Next gen Radeon Vega is rumored to have up to 1.28 Terabyte/s mem bandwith. IS there any FPGA (or announced) that would pull the same kind of perf ? The ultraram on the FPGA collectively have a total memory bandwidth of 5.1TB/s. The HBM2 on the next generation FPGA will have the same bandwidth as the vegas with HBM2 and those chips will still have ultraram on top of that. LOL VCU1525 has only 33 MB of uram - it is nothing. This price overvalued FPGA's can mine only few algos. If the anti-asic and anti-fpga policy will be continued, this FPGA will mine nothing. Main problem is small amount of high speed RAM and small amount of LUT. This is kicking out a lot of actual algos like equihash, ethash, CN-v7, X17, X16r, Lyra2Z, leaving fpgas with primitive low-profit algos like keccak, tribus, skunkhash, lyra2v2, etc. 33MB is a lot for on chip memory. VCU1525 also have up to 64GB of DDR4. DDR4 is slow. You cannot use it for ethash or CN-7 You’ve made two very aggressive anti-FPGA posts. Can you share your credentials and experience leading you to make such statements, or demonstrate why these things are not possible? You’ve made at least 4 claims that I personally know to be untrue. |
|
|
|
What happens if coins move to ethash-like algorithms where the efficiency gains would be alot smaller?
That’s the nature of crypto, you have to decide what risks and opportunities are right for you. Currently what I’m proposing plays well in the acceleration role with GPUs for those algorithms. In 9-12 months both FPGAs and GPUs will be shipping with similar memory bandwidth and then FPGAs (imho) will be winning on power. |
|
|
|
I can purchase AVNET AES-KU040-DB-G for $1250 and Nexys video Artix 7 for $472 from a retail shop in my city. I know the prices are high compared to US but I won't get big boards any time soon. So are these board any good ? I can purchase them tomorrow.
I can purchase AVNET AES-KU040-DB-G for $1250 and Nexys video Artix 7 for $472 from a retail shop in my city. I know the prices are high compared to US but I won't get big boards any time soon. So are these board any good ? I can purchase them tomorrow.
The Nexys board is lower speed grade of the $329 board I’m doing, with slower ram and no PCIe for interior with system/GPUs. Of the two listed, the XCKU040 is more capable in a stand alone fashion if you’re trying to be involved early. The downside of purchasing a random board is people won’t be making you miners for them like they can with GPUs - development work elaborate tuning etc. has to be done for every board individually. How can there be an online download and play solution that OP is proposing if you are saying that each board has to be tuned individually. Do you think that it's worth it to stock the XCKU040 (at the price listed) as the shopkeeper has only 1 left in stock. And will you ship your board to Asian countries too ? And do you think it's worth it for me who knows very very basic Java, C, python to spend my time learning to program FPGAs for mining ? (I am quick learner) Each board design has to be tuned individually, unless two boards are specifically designed to be compatible. If you are interested in learning and can afford the cost of entry and want it now, personally I would buy that board. There are some open Keccak designs that you could reasonably get setup to make a little bit and amortize the cost of getting the board over time. The skills you will learn are more valuable than the board. |
|
|
|
I can purchase AVNET AES-KU040-DB-G for $1250 and Nexys video Artix 7 for $472 from a retail shop in my city. I know the prices are high compared to US but I won't get big boards any time soon. So are these board any good ? I can purchase them tomorrow.
I can purchase AVNET AES-KU040-DB-G for $1250 and Nexys video Artix 7 for $472 from a retail shop in my city. I know the prices are high compared to US but I won't get big boards any time soon. So are these board any good ? I can purchase them tomorrow.
The Nexys board is lower speed grade of the $329 board I’m doing, with slower ram and no PCIe for interior with system/GPUs. Of the two listed, the XCKU040 is more capable in a stand alone fashion if you’re trying to be involved early. The downside of purchasing a random board is people won’t be making you miners for them like they can with GPUs - development work elaborate tuning etc. has to be done for every board individually. |
|
|
|
Hi,
Really interesting thread! I am beginning to mine using FGPA boards. I have been GPU and asic mining for the past year or so and have been interested in the FGPA side of things since hearing of it recently. I have purchased a Terasic DE0-Nano-SoC Development and Education Board to begin my learning. Can you say if any of your algo's will be compatible with the board mentioned? 1GB DDR3 ram and ARM dual processor.
Best,
Karl.
The DE0-Nano-SoC is a great board to get into learning this stuff. I have this and a few of the Max10 50kLE boards that I occasionally use for initial RTL hardware evaluation because their just so easy to work with. Vivado and Xilinx’s tools are steeper learning curve, but definitely doable. With that said, the Cyclone V 5CSEMA4U23C6N on that board is about 1/5th the logic of what I’m proposing and while it has high-speed DMA to the ARM cores, it doesn’t have any high speed IO such as PCIe to other peripherals. That makes it tough to use outside of single small algorithms (I.e. Keccak). The $199 version of the M.2 has ~100 Logic Elements and the $329 version has more than 200, capable of running at higher speed. The top package also has 1GB DDR3. Peak power consumption is around 15W from on-board M.2 Here’s an update summary for those catching up on this thread: 1. A few of us have been working on algorithms on FPGAs, and they’re profitable. Whitefire990 intents to release his with miner/dev fee and others of us have decided to share with the community in various forms as well. 2. On the high end the VCU1525, based on the VU9P is a very good candidate for this. It is currently a Xilinx development board on promotion for $3995, but only very small batches are being produced and the price is set to go up. Some algorithms require connecting multiple with high speed links to achieve the best performance. 3. Senseless (and possibly self, if I can be helpful) have been organizing essentially a group buy - but with FPGAs it’s less of a group buy and more of a group build. That is expected to ramp up the VCU1525 style (some power/cooling improvements) availblility in a similar price envelope to Xilinx dev version. Working on production in US and Europe so it can be available everywhere. 4. I’ve also decided to reveal one of the smaller FPGA options in the $200-350 price envelope that I had developed for internal use and deploy, to provide an entry level option. It has a slightly different set of capabilities from the big VCU1525, but for many things it does scale. For both hardware offerings from the community (and possibly others) orders are expected to start in June. These boards are not so much for Ethash (though they can be used to assist/accelerate GPUs in it) , or Equihash. They excel , can improve total system performance on, or are at an advantage on Keccak (and most SHA3 candidate) , Phi variants, NIST5, Timetravel10, Lyra/LyraRev2, etc. Someone else can chime in if I missed anything. |
|
|
|
How much would that plug & play FPGA cost ? Excluding shipping and VAT / custom duties.
Listed a couple posts back (USD at least) I’ve got three versions but seems most interest is in the highest speed. |
|
|
|
So you are saying these can be used along side GPU to accelerate. What algo’s can this work along with? Like Ethash? X16R?
How many can be used with 1 rig? Do you run 1 fgpa to 1 gpu? Or 1 gpu to a rig?
I designed for Ethash originally, but things are changing there. many more algorithms are possible and I’m beginning to work on that now. Let’s continue to underpromise and overdeliver. You can run essentially as many as you have slots or PCIe lanes for, but unlike normal GPUs on risers this actually needs the PCIe bandwidth to be beneficial. You can use it in the multiple M.2 slots on a lot of motherboards that support it, like the popular Prime Z170/270/370 variants). You can also just get a M.2 to PCIe adapter ($10-15 on Amazon). I’m testing using 4 in 16x bifurcated slots in host cards like the Hyper x16, but not all motherboards support that. If your GPUs are on 1x risers and you’re not using it standalone you’ll need ~4 GPUs per M.2 for maximum benefit regardless of algorithm - because you’ll be streaming data to/from the FPGA and you want to match its 4 PCIe lanes to your other devices. if you have one GPU in a normal 8 or 16x slot you may be able to fully utilize the accelerator paired with that GPU. The board is capable of reconfiguring itself, so for example it can configure itself for the current timetravel10 permutation and while it is not big enough to hold all 10 algorithms at once, it can accelerate the first 2-5 of them before offloading the rest to the GPU to finish. In this use case it can provide as much as 33MH of lift. Most chain-hashing algorithms can see similar benefits. |
|
|
|
If you’re looking in that range, I might as well start taking pre-orders for my M.2 accelerator since it’s ready. It has a 200k LE 7 series chip and 1GB of DDR3 + 4x PCIe (2.0 atm). $329 for the very fast version, $199 for the lower end (100k LE/512MB DDR3). Uses a M.2 M-Key / Nvme slot, or a $10 PCIe adapter.
How can I order this board? Can you share more infos? (Algos, hashrates etc...) I don’t want to be these guys that sell something on a render or photoshopped picture of a prototype, so I need two more weeks to get first production batch in, heatsinks on, etc. but if anyone wasn’t to reserve PM me contact info and I will share the info. Developed these for our own use, but (dig around on the reddit threads and you’ll see me talking about it pre forms development a few months ago). We originally developed these for ourselves but I think they can help the general community so I will sell near cost. I’m going to refrain from posting final hashrates until they are done on a final production model with the last power/cooling/speedgrade changes we’ve made. I'm interested in your "M.2 accelerator" especially given your affordable price range. Might be a good way to get feet wet with FPGAs. I sent you a PM regarding this and i understand you won't be posting figures or other details but the community may also benefit from the following general questions: 1. Is this plug-and-play hardware that already includes software which can easily be configured? 2. What about software updates, can we be assured that you will provide maintain these and not leave us with paperweights? 3. Does it come with warranty? Still swamped with final mass production preparations (I had to take an emergency detour to secure the necessary memory chips in quantity - DDR lead times have been 6 months for a while now. I took steps to mitigate that months ago but still had to pull some strings.) 1. It is plug in play, in that you can either use the (closed source - sorry :/) bitstreams I’ll provide (at the moment Ethash acceleration, Keccak, and CrpytoNight7 acceleration) + miner, or because of the selected chip you can use Vivado Webpack for free to build open RTL and use ccminer or others. - note I use the term acceleration because this is really built to work in hybrid with existing GPUs and system RAM. It can run stand alone as well, but best to get best returns. 2. You won’t be limited to my bitstreams, others can develop for it, and the design was made so it could be used for lots of general purpose acceleration. As far as I am aware there is no lower cost board available with these resources. 3. There will be some warranty, I need to determine what terms. It’s prerty hard to intentionally brick a GPU, but it is pretty easy to (un)intentionally brick an FPGA if you load your own software. |
|
|
|
I see, Thanks. So as a comparison, FPGA/SoC in Antminers is like a CPU in a traditional PC and the hash-boards are like GPUs (although with fixed logic)? There is an ARM CPU chip. The Zynq 7000 is an SoC with ARM+FPGA. i am not sure what the FPGA plays the role there but the ARM should be the host processor. Yep, nothing (currently) intense is happening on those 7000s. Most just simple system mana game to. But... expect massively bigger FPGAs in Bitmain tech soon... Using c/c++ to program an FPGA high performance app is a bit like writing a high performance app in Visual Basic “. Or maybe Python. But it can help learn. |
|
|
|
If you’re looking in that range, I might as well start taking pre-orders for my M.2 accelerator since it’s ready. It has a 200k LE 7 series chip and 1GB of DDR3 + 4x PCIe (2.0 atm). $329 for the very fast version, $199 for the lower end (100k LE/512MB DDR3). Uses a M.2 M-Key / Nvme slot, or a $10 PCIe adapter.
which 7 series chip ? sounds nice. There are only a handful that fit on an M.2 2280, and the most logic is in the Artix. Kintex 160 is next best. I’ve got a bunch of interested PMs I’ll be responding to tomorrow as I have some obligations tonight. The main reason I don’t want to list exact hash performance is the effect of cooling. Just like an nVME what I can get on a bench with great thermals might not match performance on the mother boards, which led to a heatsink modification I need to very final numbers with. Essentially you’ll be looking at 10% of a VU9P for <10% of the price |
|
|
|
PCIe 4x? does it need the 4x? ie, will running it on a single PCIe lane hurt performance?
i have an onda mining board with 8 full size 16x slots but i believe they run in 1x mode, so any PCIe ->m2 adapter card would run at 1x.
It costs next to nothing, basically only traces, all the logic is synthesized in the FPGA. Gives the board other uses besides mining. The reason why old ZTEX boards have much better resale value than the old mining-only boards from ngzhang. His Lancelot & Icarus boards were really crippled by not having any way to get signals to the IO pads of the FPGA. They had only 2 (out of 8 available) signals connected between USB USART chip, greatly reducing available bandwidth and protocols. Similar weak connectivity was between the two FPGA chips on the board. It makes the whole board unusable for anything else, not even toy projects could be reasonably built on them. This is accurate - plus one of the uses of this board is ETH acceleration, where it provides a few MH lift to GPUs isn’t the system that are on 1x connections. |
|
|
|
....+$2000 for a waterblock. ...
One waterblock for $ 2000 for one VCU1525 or suitable for rig up to 8 pcs? Photo? I'm saying that he's paying $2000 more getting the bittware board just to get a waterblock. [/quote I’m not sure why we are bothering with water blocks and extreme cooling. We can get two smaller chips with fast interconnect easily cooled for the same or less price... |
|
|
|
If you’re looking in that range, I might as well start taking pre-orders for my M.2 accelerator since it’s ready. It has a 200k LE 7 series chip and 1GB of DDR3 + 4x PCIe (2.0 atm). $329 for the very fast version, $199 for the lower end (100k LE/512MB DDR3). Uses a M.2 M-Key / Nvme slot, or a $10 PCIe adapter.
How can I order this board? Can you share more infos? (Algos, hashrates etc...) I don’t want to be these guys that sell something on a render or photoshopped picture of a prototype, so I need two more weeks to get first production batch in, heatsinks on, etc. but if anyone wasn’t to reserve PM me contact info and I will share the info. Developed these for our own use, but (dig around on the reddit threads and you’ll see me talking about it pre forms development a few months ago). We originally developed these for ourselves but I think they can help the general community so I will sell near cost. I’m going to refrain from posting final hashrates until they are done on a final production model with the last power/cooling/speedgrade changes we’ve made. |
|
|
|
I've been looking into a few FPGA board options and stumbled across a much cheaper board similar to the Xilinx KC705.
The FPGA is a XC7K325T with the following specs:
Logic Cells - 326,080
DSP Slices - 840
Memory - 16,020 kb
You should be able to find more details on Xilinx's website
The board also has 128 MB of flash and 8 GB of DDR3, if that's helpful.
Would this board be capable of profitable mining? I know that SHA-224 and Blake2b would fit on, Keccak should too, but have no idea what the hashrate and profitability would be around. What other algorithms can fit on to it? Hashrate estimates?
I’m not sure if you realize, but there is a drastic difference between this and the VCU1525. 1. Almost 8 times less logic cells. 2. 20x less on-chip memory 3. 7 series fabric (28vs16nm) much much lower clock speeds (~250 vs 600 MHz max) 4. The flash is for configuration 5. DDR3 is bandwidth IS nice. Over all Keccak you could maybe do 6-7 pipelines @ 200Mhz. - 1200-1400 MH. How cheap is cheap?. Overall the XC7K325T based board is about 16x less power than a VCU1525. If you’re looking in that range, I might as well start taking pre-orders for my M.2 accelerator since it’s ready. It has a 200k LE 7 series chip and 1GB of DDR3 + 4x PCIe (2.0 atm). $329 for the very fast version, $199 for the lower end (100k LE/512MB DDR3). Uses a M.2 M-Key / Nvme slot, or a $10 PCIe adapter. |
|
|
|
So what kind of lead time are we talking about before shipping? It’s going to be pretty hard to cover the $25-$30k investment for TOO long before I would need to start turning them on! We talking 1-3 months or longer?
And what about the bitstreams? Is Whitefire the only source at the moment? Does anyone know where we could get bitstreams from other sources? Of course for a fee!
I’m willing to provide bitstreams as well, or in collaboration. |
|
|
|
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  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. |
|
|
|
|
Show Posts
