Topic: BlockBurner LLC - Crucible FPGA Scrypt Miner - Announcement Aug-19 - page 13. (Read 42381 times)

Good stuff!

Will pre order a few LTC hashers for sure, summer is coming. Heat is a issue

Announcement:

A dev team is officially being formed

Thank you all for your support! There will be more updates soon as our team comes together to start mapping out Stage 1, followed by round 1 pre-orders (or you may simply donate) when it comes time.

Operatr

sweet..

when you get to testing and software dev stage i can help more, until then ill be watching this closely..
gl and thx

At this point interest in this seem high enough, I am putting together a volunteer dev team to work with me. If you are an interested FPGA/Microcomputing engineer, software engineer, or have otherwise relevant skill in bringing such a product to market, please PM me.

Project Overview

Design Goals:

Modular Scrypt FPGA system
USB Connectivity
Stand alone/Rack convertible casing for scalability
Associated software package

I have had a few PMs and have seen questions regarding pre-orders for this project:


On Pre-Orders

Any pre-order campaign will be associated with the current stage of development. Unlike other producers there will be no pre-orders until a certain capital requirement is met meeting the estimated costs associated with that stage. At this stage it would be in generating a working prototype device. I am taking a community approach for complete transparency, every transaction would be made public knowledge as I think if you are willing to take a chance on us, you should know exactly what your money is funding and see it develop before your eyes.

This approach minimizes risk and gives a linear progression of development that is seen by the whole community.

I don't believe it is fair to hold pre-orders in a way that in a way fakes it as if it is a real product sold online, knowing full well it does not exist. I think this practice itself is fraudulent in nature itself.

Prototype Stage

Proto-adopters would be taking the bulk of the risk, as such we would work out some other kind of benefit to funding assistance at this stage. I am open to ideas on what you would like to see if you opted to be a proto-adopter.

A known price point will be known before any pre-order campaing begins with a known cap to hit, all pre-order capital going into third-party escrow until the needed amount is reached. Otherwise it would be returned to you. This could be receiving a prototype device to help with testing or some kind of future revenue sharing.

Production Stage Once a working prototype is created, we will then move on to casing actual production costs, and much like the Proto stage, will have a certain goal needed before any capital is invested.

To do this will require a crowd-sourced effort, which would be conducted through various forums as well as things like Kickstarter campaigns and the like.

They are versatile though some are more optimized in architecture for certain things. At the moment I am looking into FPGAs with Scrypt in mind over a more generic type such as the Spartan 6.

I think you are misinformed about what is and is not possible.

You can construct whatever width you like by putting multiple units in parallel.  This is commonly done, and is a general feature of FPGA's not unique to Xilinx.

The vendors put them into small blocks like that to improve the granularity / flexibility for the designer.  As a result, you effectively lose capacity (bits) when your chosen configuration doesn't map efficiently to the underlying memory organization.

Artix-7 is even better, but limiting the discussion to Spartan 6 which many people have already bought, here is some documentation:

See page two of this:
(a) http://www.xilinx.com/support/documentation/ip_documentation/blk_mem_gen_ds512.pdf

See page nine of this:
(b) http://www.xilinx.com/support/documentation/user_guides/ug383.pdf

See page two of this:
(c) http://www.xilinx.com/support/documentation/data_sheets/ds160.pdf


To get a x1024 memory using (a), you can see from (b) that one possibility might be (32) instances of (x32) width.
As far as the capability of the LX150 part commonly used on existing bitcoin mining boards, you will see in (c) that this devices has a total of (268) such blocks.
So accommodating the 128KB scratchpad in SCRYPT could be done with (64) blocks configured for (x32) width and (32) units in parallel.   The LX150 could possibly hold (4) such memories, but I think you run out of gates for SCRYPT arithmetic well before that.

The total RAM per block is 18KB. Each block has a 72-bit width. I don't really know where you're pulling your numbers from. Even if you calculate in parallel, 128/18 = 8 block RAM units required, with 72-bit widths each --> not 1024 bit width either.

My point is that off-chip memory for SCRYPT is entirely unnecessary.  It really doesn't need much and it can fit entirely onchip.  The latency of going off-chip would kill performance.

Something like this might be neat to attempt.
http://www.wpi.edu/Pubs/E-project/Available/E-project-031212-183607/unrestricted/FPGA_Design_for_DDR3_Memory.pdf


Please let me know if you need me to test it out...  

Yes, that's what I said.  (4.9 million == 4800 K)    (the exact number is 4939776 bits).



I'm not following your arithmetic.  Are you citing some document somewhere for either of those numbers?  If so, can you paste a link? 

From my previous post, an FPGA memory with 1024-bit width at (lets downgrade it to a more modest 200MHz) is 200 billion bits per second or 25GB / s.
This would be per-memory-instance (or, per hypothetical scrypt-core).

Total block ram on the whole chip for a spartan6 lx 150 (most expensive chip) is 4824 Kb.  http://www.xilinx.com/products/silicon-devices/fpga/spartan-6/lx.htm

Memory bandwidth for the block RAM is about 30-60 gb/s (your numbers above) while GPU internal bus is usually around 250 gb/s on higher end cards.

I'd be interested in helping out on the development side of things, shoot me a PM if you would like to discuss more. I work on FPGAs for a living, so I have the proper skillset.

I thought a FPGA was reprogrammable? Why can't someone just buy a blank FPGA with a decent amount of onboard memory and program the chip? Am I misunderstanding how to set up a FPGA?

Well, I'm not as familiar with GPU, but I doubt it is 10 times faster.  And I believe you have been misinformed regarding the capacity as well.

The Spartan-6 LX 150 used on many of the boards already built has 4.9 million bits of memory.  The memory in -3 speed grade part can run at up to 320MHz
Newer but similar priced Artix-7 have 13.4 million bits, with up to 509MHz in -3 grade parts.

As it relates to scrypt and it's 128KB scratchpad, the core loop accesses memory sequentially in 1024-bit widths.  Within an FPGA, you can have access to all 1024 bits in a single clock.  While you may not be able to achieve that performance point due to other issues,  1024 bits @ 320/500MHz is nothing to sneeze at.

If I was being picky, id say best thing for an fpga board, is being modular...

If you can design the backplane to be expandable, and sell fpga expansion boards as it were (thinking modminerquad style) then new miners could get in with a little investment and toy with fpga too

I am looking at the linear path Bitcoin took. GPUs are power intensive and generate a lot of heat, at some point these two factors will outweigh any benefit to mining LTC as it scales up. GPUs represent lowest hardware cost/highest operating cost generally, where FPGA generally costs a little more up front, but you save in the longer run with less energy needed to power it and superior heat efficiency negating the need for A/C units or other cooling, increasing your keep long term.

Current FPGAs used for Litecoin are geared toward Bitcoin more, I am looking at one that is optimized for Scrypt for superior hashpower.

2. I would definately buy one if it's really worth it. It should be cheaper than GPUs, quickier and you should consider the difficulty rise.

We will see- but I doubt it with be neither cheaper nor quicker. FPGA's were not cheaper then GFX card or Faster. Only use electricity, smaller and less heat.

Indeed. Dissipating the heat from 3 GPU rigs is hard enough for me, especially now that it's getting hotter outside. Opening a window merely makes things worse.

Good point. A lot less heat too.

I'd say that the logistics benefits alone make it worth it. It's very easy to run 20 FPGA's in an average home. It's not as easy running 20 GPU rigs inside a home, at least not inside mine.
I will buy Litecoin FPGA's as long as the price is reasonable.