Pages:
Author

Topic: ALLMINE INC - FPGA Cryptominer - page 53. (Read 51510 times)

jr. member
Activity: 33
Merit: 1
May 10, 2018, 04:40:08 PM
#36
There is a lot of talk on two threads here about these fpga cards for mining but very little information and zero evidence that any of it even works.  Sure, I'll spend $35k on a rig with a $450 per day return.  Prove that it works.
hero member
Activity: 1118
Merit: 541
May 10, 2018, 02:49:04 PM
#35
Did you read this:

Sorry, I've been meaning to make a post related to this.  We have been working with the OP to determine the optimal configuration and board for this application, as we have a few to choose from.  The XUPP3R may not be the optimal one, we have other boards, including one with larger core power supplies and even the VU13P instead of the VU9P. The OP is working to determine the combination of FPGA size, clocks, logic, memory and power use that maximizes ROI.  We do not want to sell you boards that may not be the best fit, that is in no one's best interest.

So you have not missed out, you perhaps have saved yourself from buying a non-optimal board (we actually have not yet taken any orders or shipped any boards to miners so no one is in that camp).  You can still contact us and we will get back to you as soon as we have enough details to let you place an order.

So VU13P might be better? I think you are going too fast. I am interested in 2 but things need to settle down.

Edit: And OP(whitefire99) is still working specs.

I've already been at this for over a year. I've been mining on these boards since June 2017 and working on them since Jan 2017. The XCVU9P is where you want to be at. The XCVU13P isn't produce in the same quantity and it will not be as cost effective. The fact they don't know which device to get and are just figuring this out now shows how late to the game they are.

Bittware is also not going to give you a good deal. For each board they sell they're going to gross about 2/3.

Maybe they mean the VU31P and not the 13P?  The 31P has 4GB of HBM2 memory, which would be better for equihash than the VU9P.


Those aren't in mass production yet.

sr. member
Activity: 588
Merit: 251
May 10, 2018, 02:31:23 PM
#34
Did you read this:

Sorry, I've been meaning to make a post related to this.  We have been working with the OP to determine the optimal configuration and board for this application, as we have a few to choose from.  The XUPP3R may not be the optimal one, we have other boards, including one with larger core power supplies and even the VU13P instead of the VU9P. The OP is working to determine the combination of FPGA size, clocks, logic, memory and power use that maximizes ROI.  We do not want to sell you boards that may not be the best fit, that is in no one's best interest.

So you have not missed out, you perhaps have saved yourself from buying a non-optimal board (we actually have not yet taken any orders or shipped any boards to miners so no one is in that camp).  You can still contact us and we will get back to you as soon as we have enough details to let you place an order.

So VU13P might be better? I think you are going too fast. I am interested in 2 but things need to settle down.

Edit: And OP(whitefire99) is still working specs.

I've already been at this for over a year. I've been mining on these boards since June 2017 and working on them since Jan 2017. The XCVU9P is where you want to be at. The XCVU13P isn't produce in the same quantity and it will not be as cost effective. The fact they don't know which device to get and are just figuring this out now shows how late to the game they are.

Bittware is also not going to give you a good deal. For each board they sell they're going to gross about 2/3.

Maybe they mean the VU31P and not the 13P?  The 31P has 4GB of HBM2 memory, which would be better for equihash than the VU9P.
newbie
Activity: 16
Merit: 0
May 10, 2018, 01:16:12 PM
#33
Im interested 3 or 4 units from mex
hero member
Activity: 1118
Merit: 541
May 10, 2018, 01:03:58 PM
#32
Im kean.

Just a noob question if you don't mind: Will your design require a different bitstream from @whitefire990 than the one he will be releasing for the
DK-U1-VCU1525-A-G by the end of this months?

I'am asking as I have a quote for 2 DK-U1-VCU1525-A-G in South Africa but If I need to wait a few weeks to get the more mining dedicated version from you I might be willing.

No, if he releases bitstreams for the VCU1525, it would work on the boards I'd make. And, the board I'd make would be equivalent to the product id you mentioned. The only difference is it will be cheaper once the $4,000 avnet promo ends. Avnet and Digi will both increase their price to $5200 for small MOQ. I would still be offering at $4k or less (hopefully, if we can get 10k orders).

jr. member
Activity: 252
Merit: 8
May 10, 2018, 06:43:55 AM
#31
no interest in selling your miner or dev services?

also is that normal to have the intro pricing then jack it up, whitefire also says its $4k now but 5k + next month

It's the first time I've ever seen a company do that with FPGAs, that's for sure. Only avnet is doing it, digikey wants $5200/board for low QTY. My pricing just depends on chip order quantity, the more ordered, the lower the price is going to get.

I don't trust % fees for dev services because they're so easy to avoid paying.


Well, it's going be inevitable if whitefire does release his with dev fee. Why not tap into the market and collect residual for yourself regardless? If you have something that is already done, and then whitefire releases his, people are going use whitefire no matter what, since his is the one that is offered to the public.

But if you offered it ahead of the game and/or is better, then at least you get to collect some residual regardless if people want to bypass the dev fee or not.

Not sure how much work that would be needed to do this, but the way I see it, for example;

Lets say 100 people use your service, 50 people bypass the dev fee if that's your concern.
I would rather have 50% of some free residual, then 0% of nothing.

That's just one perspective.
+1
sr. member
Activity: 512
Merit: 260
May 10, 2018, 03:21:08 AM
#30
Im kean.

Just a noob question if you don't mind: Will your design require a different bitstream from @whitefire990 than the one he will be releasing for the
DK-U1-VCU1525-A-G by the end of this months?

I'am asking as I have a quote for 2 DK-U1-VCU1525-A-G in South Africa but If I need to wait a few weeks to get the more mining dedicated version from you I might be willing.
jr. member
Activity: 37
Merit: 6
May 10, 2018, 03:03:56 AM
#29
I'm in! 10units/month
newbie
Activity: 3
Merit: 0
May 10, 2018, 02:56:33 AM
#28
Definitely interested if the order is possible. Would sell off some rigs and willing to purchase around 25-30 Units minimum depending on the price possible. Only problem is would need some programs to be released with reasonable dev fees, or reasonable upfront costs. I dont have problems with Dev Fees, as majority of the good miners all have them, if they are reasonable i will support, or willing to pay monthly/yearly/one time for the miner.

Also i understand where you are coming from with the backout from backers. I had investors lined up in 2017 to build a large mining farm with solar panels, and had them backout during the crash. I still know plenty of investors and could potentially line up more units after the initial order. Have to watch how this whole FPGA thing goes as it seems alot are talking about it now, before anything can be done, and would need multiple rigs myself to prove.


i'm in too and i can buy at least 50 units and me too in a farm with investors. please get in touch
member
Activity: 273
Merit: 12
May 10, 2018, 12:51:24 AM
#27
Definitely interested if the order is possible. Would sell off some rigs and willing to purchase around 25-30 Units minimum depending on the price possible. Only problem is would need some programs to be released with reasonable dev fees, or reasonable upfront costs. I dont have problems with Dev Fees, as majority of the good miners all have them, if they are reasonable i will support, or willing to pay monthly/yearly/one time for the miner.

Also i understand where you are coming from with the backout from backers. I had investors lined up in 2017 to build a large mining farm with solar panels, and had them backout during the crash. I still know plenty of investors and could potentially line up more units after the initial order. Have to watch how this whole FPGA thing goes as it seems alot are talking about it now, before anything can be done, and would need multiple rigs myself to prove.
sr. member
Activity: 544
Merit: 250
May 10, 2018, 12:50:34 AM
#26
no interest in selling your miner or dev services?

also is that normal to have the intro pricing then jack it up, whitefire also says its $4k now but 5k + next month

It's the first time I've ever seen a company do that with FPGAs, that's for sure. Only avnet is doing it, digikey wants $5200/board for low QTY. My pricing just depends on chip order quantity, the more ordered, the lower the price is going to get.

I don't trust % fees for dev services because they're so easy to avoid paying.


Well, it's going be inevitable if whitefire does release his with dev fee. Why not tap into the market and collect residual for yourself regardless? If you have something that is already done, and then whitefire releases his, people are going use whitefire no matter what, since his is the one that is offered to the public.

But if you offered it ahead of the game and/or is better, then at least you get to collect some residual regardless if people want to bypass the dev fee or not.

Not sure how much work that would be needed to do this, but the way I see it, for example;

Lets say 100 people use your service, 50 people bypass the dev fee if that's your concern.
I would rather have 50% of some free residual, then 0% of nothing.

That's just one perspective.
full member
Activity: 1179
Merit: 131
May 09, 2018, 11:11:38 PM
#25
This new FPGA trend sounds interesting and earnings do sound good to everybody but don't you think making so many devices available could cripple coins?
Every coin developer is striving to be ASIC resistant but when using FPGAs they won't be able to counteract them since you can change the algo as fast as they fork.

I mean difficulty would increase but you still would have a lot of coins flooding the market and driving them on a bear trend to very low values thus ruining your ROI too.

I hate to break it to you but in 6 months Nvidia is going to be making a lot more devices available than the few thousand OP is talking about.  You can either evolve with the technology or get left behind
hero member
Activity: 1118
Merit: 541
May 09, 2018, 10:55:05 PM
#24
no interest in selling your miner or dev services?

also is that normal to have the intro pricing then jack it up, whitefire also says its $4k now but 5k + next month

It's the first time I've ever seen a company do that with FPGAs, that's for sure. Only avnet is doing it, digikey wants $5200/board for low QTY. My pricing just depends on chip order quantity, the more ordered, the lower the price is going to get.

I don't trust % fees for dev services because they're so easy to avoid paying.


Really? This is standard operating procedure for Xilinx. Dev boards are usually sold cheaper than even just the chip cost for normal quantities. They want to make it easy for people to build things around their expensive chips, but still make all the margins on chips. I too have been talking to Xilinx and trying to get them to see there is a big market here, big enough to make up for their margins on some of these chips.

There is more FPGA hardware coming pretty soon (over the next 2-3 months)  from a few sources (disclosure - myself included). Generally speaking most are going to be in the same ballpark cost/performance, but there may be outliers or easier entry level costs.

The problem with FPGAs is it is very hard to protect the bitstream if you aren’t selling the hardware. You can sell hardware that will work with your encrypted bitstreams and also support unencrypted bitstreams (anyone can make new miners). You can also lock hardware to only your encrypted bitstreams. But if you don’t sell the hardware (and burn in the encryption keys) anyone can copy it. So that leaves virtuous open source, or people making profit on hardware alone. The tools and time to develop these things are very expensive, so you likely won’t find a lot of people giving it away for free. Development time is also much longer. If I make a change that actually needs tested on real hardware of VCU1525 size, expect hours for a very fast computer  to synthesize and route it before I can test it. A lot can be done in simulation, but the last mile takes a lot of time.

It is possible to use normal methods to try and protect the “command and control” normal software that drives the FPGA, but someone could reverse engineer that - just like OpenCL kernels.

If you think the FPGA is magic and the OpenCL compiler will give you 10-50x performance over your GPU, you will be sorely disappointed. Those of us seeing these gains are (afaik) writing very low level RTL to do so. There’s a big difference between OpenCL and RTL for FPGAs vs OpenCL vs ISA Assembky for GPUs.


Ya, I've never seen any of these companies offer a promo price for a new fpga dev board. Normally 'promo price' to these companies would mean 10x the mass production price.

I'm considering the solution of using a shell (like aws) pass some wires through and allow any fpga dev to build and compile against our shell. We'll then distribute the firmware and collect the developer fee on behalf of the dev. There are a lot of FPGA devs in this community, a lot more than have been posting in the threads. I'm guessing they just haven't seen it yet.

It seems like there's not a lot of enthusiasm that there's only 1 dev providing firmware to the community. It would be nice to provide a platform that would allow any dev to create firmware and profit from the work.
member
Activity: 154
Merit: 37
May 09, 2018, 10:45:39 PM
#23
no interest in selling your miner or dev services?

also is that normal to have the intro pricing then jack it up, whitefire also says its $4k now but 5k + next month

It's the first time I've ever seen a company do that with FPGAs, that's for sure. Only avnet is doing it, digikey wants $5200/board for low QTY. My pricing just depends on chip order quantity, the more ordered, the lower the price is going to get.

I don't trust % fees for dev services because they're so easy to avoid paying.


Really? This is standard operating procedure for Xilinx. Dev boards are usually sold cheaper than even just the chip cost for normal quantities. They want to make it easy for people to build things around their expensive chips, but still make all the margins on chips. I too have been talking to Xilinx and trying to get them to see there is a big market here, big enough to make up for their margins on some of these chips.

There is more FPGA hardware coming pretty soon (over the next 2-3 months)  from a few sources (disclosure - myself included). Generally speaking most are going to be in the same ballpark cost/performance, but there may be outliers or easier entry level costs.

The problem with FPGAs is it is very hard to protect the bitstream if you aren’t selling the hardware. You can sell hardware that will work with your encrypted bitstreams and also support unencrypted bitstreams (anyone can make new miners). You can also lock hardware to only your encrypted bitstreams. But if you don’t sell the hardware (and burn in the encryption keys) anyone can copy it. So that leaves virtuous open source, or people making profit on hardware alone. The tools and time to develop these things are very expensive, so you likely won’t find a lot of people giving it away for free. Development time is also much longer. If I make a change that actually needs tested on real hardware of VCU1525 size, expect hours for a very fast computer  to synthesize and route it before I can test it. A lot can be done in simulation, but the last mile takes a lot of time.

It is possible to use normal methods to try and protect the “command and control” normal software that drives the FPGA, but someone could reverse engineer that - just like OpenCL kernels.

If you think the FPGA is magic and the OpenCL compiler will give you 10-50x performance over your GPU, you will be sorely disappointed. Those of us seeing these gains are (afaik) writing very low level RTL to do so. There’s a big difference between OpenCL and RTL for FPGAs vs OpenCL vs ISA Assembky for GPUs.



full member
Activity: 325
Merit: 110
May 09, 2018, 06:52:19 PM
#22

Any interest?

Payment methods would be Crypto (obviously) and Credit Card / Paypal (identity verification would be required and shipping to your registered / verified address)



Shut up and take my money
hero member
Activity: 1118
Merit: 541
May 09, 2018, 06:49:26 PM
#21
no interest in selling your miner or dev services?

also is that normal to have the intro pricing then jack it up, whitefire also says its $4k now but 5k + next month

It's the first time I've ever seen a company do that with FPGAs, that's for sure. Only avnet is doing it, digikey wants $5200/board for low QTY. My pricing just depends on chip order quantity, the more ordered, the lower the price is going to get.

I don't trust % fees for dev services because they're so easy to avoid paying.
legendary
Activity: 1610
Merit: 1008
Forget-about-it
May 09, 2018, 06:40:35 PM
#20
no interest in selling your miner or dev services?

also is that normal to have the intro pricing then jack it up, whitefire also says its $4k now but 5k + next month
hero member
Activity: 1118
Merit: 541
May 09, 2018, 06:26:27 PM
#19
how would we load the various bitstreams onto these things? we would need special software?

No. The bitstreams can be loaded via USB.

Whitefire990 mentions he uses USB for programming and communication in his thread. His software would be the one to program the board.




newbie
Activity: 78
Merit: 0
May 09, 2018, 06:03:18 PM
#18
The VCU1525 schematics and bom are available for anyone to produce. I have already sat down with Xilinx (in Shenzhen) and negotiated chip pricing. If there is enough interest, I could facilitate a chip purchase, put the chips on boards and deliver them at a regular price of $4,000-$4,500. To hit a $4,000-$4,500 and stay at $4,000-$4,500 there would need to be at least 1,000 boards sold. If order volume went up to or over 10,000 units pricing could be dropped to $3,000-$3,500 per unit.

You would be able to use the device with the firmwares produced by whitefire990

Any interest?

Payment methods would be Crypto (obviously) and Credit Card / Paypal (identity verification would be required and shipping to your registered / verified address)


So this would require his software? what if I want to develop my own? would I have a paper weight?

If you are good enough to develop your own, you know the question to this.

legendary
Activity: 4354
Merit: 3614
what is this "brake pedal" you speak of?
May 09, 2018, 05:55:45 PM
#17
how would we load the various bitstreams onto these things? we would need special software?
Pages:
Jump to: