FPGA minig software

Is there any mining software for FPGA devices?
I have cairnsmore, icarus and some other stuff.

There's currently no Zcash mining software for FPGA devices that I know of. The boards that you have are not going to be useful for mining Zcash: they lack memory. While FPGAs do have some memory inside them, it's way too little for efficient Zcash mining. If anyone implements Zcash mining on FPGAs later, this would be on (currently too expensive) FPGA boards with fast memory buses and some memory chips, or on (future) huge FPGAs (with sufficient SRAM in them), or maybe on (currently non-existent? but technically possible) multi-die FPGA+DRAM packages/MCMs some years later.

3 Likes

There are hobby projects around soldering ram modules onto Cairnsmore (Spartan 6) boards. That said, no idea how they are working out so far.

You seem acquainted with the market. It difficult for someone outside to dig through this plethora of board variations and product offers.

After xenoncat's assembler implementation for x86_64 arch, what is your take on how much work needs to be done to have a functioning fpga implementation?

What current board (if you can provide brand/model/url) would be the best choice ?

What solardiz said is correct.

That said if you have those FPGA's and just want to put them back to work, you can always mine VCash which uses Blake256 vanilla.

1 Like

Sorry to sound negative, but I don't expect any FPGA board currently on the market to be cost-effective for Zcash mining (except if you already have a suitable board anyway or/and in electricity costs over multiple years), and there's no point in asking your questions before there's an implementation (unless you were to make one yourself). It is possible for cost-effective boards to appear later if the economy of scale would work that way (although suitable many-core CPU-like chips or ASICs might become cost-effective sooner, or they might not), but right now FPGA boards with fast memory target some other niche markets (not cryptocurrency mining) where production volumes are low and multiple thousands of dollars per board is peanuts.

FWIW, you may be able to resell ZTEX 1.15y compatible FPGA boards for password hash cracking. We've just added support for them for cracking descrypt hashes to John the Ripper: http://www.openwall.com/lists/john-users/2016/11/06/1

Currently, this only works with ZTEX original (from Germany) and its direct US clones. Most of your boards will probably require some extra code changes to make them work as well. Maybe someone will add the support for them to JtR as well, in exchange for a few free boards. I may ask around. I guess you have lots?

1 Like

Amazon's "F1" instances might be a game-changer for Zcash mining on FPGAs - not requiring investment into the hardware, yet providing not-too-slow memory. There's no pricing information yet, as far as I can tell, so no way to tell if it'd be profitable (I am skeptical, but we just don't know yet). Total memory bandwidth of their 8 FPGA instances should be on par with 2 or so high-end GPU's, but maybe - just maybe - it could be used more efficiently there, and the BRAMs in the FPGAs would help, too.

3 Likes

With the speedups in GPU mining I no longer believe that FPGA mining is a viable path.

That's pretty cool- have they offered an ETA. ? I'm sure prices will be dramatic. May be cheaper long run to invest in some pcie fpga with fast dual 8 GB ddr. They're out there but spendy. Or one could do a home project but Amazon could help speed up the testing process.

Regarding ETA: "we are launching in developer preview form today in the US East (Northern Virginia) Region (we will expand to multiple regions when the instances become generally available in early 2017)" from https://aws.amazon.com/blogs/aws/developer-preview-ec2-instances-f1-with-programmable-hardware/

I tried tweeting at Jeff Barr with questions on whether the preview is free or paid, what pricing to expect, and whether or not the Vivado license lease is included in the free(?) preview and the pricing, but he didn't reply.

No point in buying boards with merely "fast dual 8 GB ddr". That's probably not fast enough. In fact, no point in buying any FPGA boards for Zcash mining at all yet, and likely later as well (not expected to be cost-effective), but I expect that a closer (non-)match would be boards with HMC or/and with a lot more than "dual" memory channels, and soon with HMC2. Here are some examples (which I expect to cost a fortune):



http://picocomputing.com/wp-content/uploads/2015/05/Product-Brief-SB-8501.pdf
http://www.hitechglobal.com/Accessories/HybridMemoryCube-HMC.htm

Many of these are in fact in PCIe card form factor, but that's not essential. The task has to fit in each board's own memory anyway, not requiring much PCIe bandwidth (so it could be a slower bus as well).

2 Likes

Those card examples look like fun!! Price point- I agree. But is very exciting for Amazon. I'm going to go take a look. Good points. I can't argue. Thank you :slight_smile:

What do you think would be a good setup for a hobbiest to test mine zcash? Some of the older gen chips seem descent. Add a mem controller memory power logic circuits...

If you have the skills, you'd plan for a specific design before you choose and acquire the hardware - and then you would know what hardware you want. If you don't have the skills, then the question does not apply until someone else designs and implements Zcash mining on FPGAs and shares their design. If you want to get acquainted with FPGAs, then Zcash is a poor choice for that, at least at this time (until someone else possibly paves the way).

1 Like

Great points solardiz. Thanks for the links and keep us posted :slight_smile:

They implemented it! https://twitter.com/ZcashMiner/status/815512912546541568

While this is possible, I see no reason to believe this specific claim.

My expectation is that even if Zcash does go through FPGA mining for a portion of its hashrate, that will be relatively brief and small scale (perhaps even less than we've seen for Litecoin, for which FPGA miners did appear briefly). A jump from GPUs to ASICs feels more likely, and ASICs will be the longer-term and larger-scale solution, although at this time the prospect of a possible Equihash parameters change discourages at least some kinds of possible ASIC designs. We'll see. And we shouldn't believe unsubstantiated claims like the above, despite of the technical possibility.

No they didn't. Just continuous tweets from a 100% scam 'service'. We have FPGA, yeah that's why you mine 5 Zcash a week right? jokers.

1 Like