Topic: Goliath Miner - page 2. (Read 10813 times)

There is a easier way to keep FPGA boards in game, just write sha1 bruteforcer for them.
It will keep them in a play for few more years!

We don't think the memory solution for CM1 will be the most optimal Litecoin solution possible with FPGAs and interconnect bandwidth may be, or not, somewhat of a limitation. There is a local FPGA with the memory that may help in this solution. Whilst we think CM1s will remain profitable (more earning than electric cost) in Bitcoin mining for some time to come yet but we would like to see the many CM1 owners with a way forward with their boards into Litecoin.

We don't have them yet in final form but all the elements for the accessing the memory have been done before by the team before so I would anticipate we could get that knocked together into a workable form some time in September.

Someone made the comment that GPUs are ASICs. If you are working on that basis FPGAs are also ASICs. However it is worth saying that FPGAs can often compete with, and often beat, GPUs albeit usually with a different structural approach. Outside maybe a few people already working on FPGA Litecoin solutions I doubt anyone actually has the knowledge to say which will be better - GPU or FPGA at the moment. Our point here is that there is a awful lot of CM1s out there with potentially nothing to do in maybe 6-12 months time when we foresee Bitcoin being no longer profitable and provided CM1s operate in profit no reason not to do Litecoin at whatever level a CM1 can achieve.

I will also say that anyone having a serious attempt to do a Litecoin solution should talk to us partly so we know what is happening but it will also allow us to offer some limited help maybe technically or even some form of financial reward for a solution. On the latter this might be royalty (say based on memory module sales), or a straight payment in some form, or even a bounty. We have not discussed this or set anything up yet so I can't say more at this point about what we might do.

Thank you hf_developer, this was very helpful. I was not aware of this effort and look forward to reading and understanding the code better.

The design only used the on-chip FPGA RAM of an LX150, which is fairly limited. With many FPGAs you can have multiple 64-bit external memory ports that all run at full speed, for example 4 ports * 64bits/port * 200MHz optimized design yields 6.4 GBytes/sec of memory bandwidth.

Even a basic LX150 includes integrated Memory Controller blocks for DDR1-DRR3 memories at up to 12.8 Gb/s peak bandwidth (from spec sheet). This chip may or may not be optimal for scrypt, other chips offer higher max bandwidth. I think the main point is memory bandwidth shouldn't be a bottle neck if done right.

Yes, there was a significant increase of hash rate for FPGAs in SHA256, but this is an effect of the possibility to fully unroll the SHA256 core. It then goes one hash per clock. MHz = MHash/s

This is algorithmically impossible for scrypt as this algorithm was specially designed to be resistant against that. In most linear algorithms you have two possibilities: Get a speed up with the need of more ressources, or save ressources but achieve a lower computation rate. If one side decreases, the other one increases and vice versa. Not so for scrypt. Both sides increase nearly equally.

Here you can see a scrypt demonstration on FPGA with hashrates ~ 2kh/s. (You need to be very experienced to make it twice that speed!):

https://github.com/kramble/FPGA-Litecoin-Miner

OK, if people that have already looked at this think the above is best you can get with FPGAs for scrypt, then there is no need to pursue it. Do you have any links pointing to the above results / findings for FPGA development?

However I find this result surprising and am not sure that I fully believe it. For SHA256 bitcion hashing, FPGAs were competitive with with 1600 core GPUs on a Watt/MHash ratio and close enough on a $/MHash ratio. This is why many miners converted from 1600 core GPUs to FPGA clusters in 2012. Remember the initial FPGA designs only offered <5Mhash per FPGA, but motivated people later brought that up to 200MHash per FPGA.

Fully agree that the # of memory ports that can be independently accessed in parallel is most important for an FPGA implementation. I suspect the most important aspect to a scrypt FPGA implementation is to balance the # of computational cores in the FPGA to the memory access attributes, not sure how that was done to yield the above results. GPUs offer extremely high memory bandwidth to service the large # of independent cores, which yes might give them the edge...

There are Litecoin ASICs already available on the market, no need to develop with FPGAs. They're called GPUs. Contains up to 1600 processors each, every processor has its own memory. Hashrates up to 800kh/s, you will never get a performance like that with FPGAs. Scrypt is not a question of how much memory you have, but more on how much memory portions you can access independently in parallel. Already available scrypt implementations on FPGAs have shown that hash rates of 4 kh/s per core can be achieved, 6-8 fitting into one FPGA, consuming 30W. Even if you increase throughput by factor 5 (....this will not happen), it can't compete GPUs.

I _might_ be interested in one such CM1 board with memory add-ons in order to try and hack a solution together in my spare time (which is very little these days between work and babies, and it is doubtful I would have enough time). Couple questions:

1) Have you done any sizing to estimate/optimize both the amount of memory and # of memory ports that would be optimal for scrypt and CM1? Otherwise the board configuration is a stab in the dark.

2) Would such a board come with the necessary IP blocks needed to interface with the add-on memory ports during Verilog development? This would make development easier and is usually provided with development boards.

3) If someone did successfully develop a scrypt miner, would you offer any business terms to make it available for others?

Part of our strategy here is to be able to adapt with the fast moving market and even visually CM4 looks very much like CM3. No idea yet if CM5-10 will follow the same chip patterns but there will be similarities in their structures seen in those boards. With the controller module as a plug in can get carried over to the next design and in extreme cases you might scrap a CM3 main board and replace with say a CM5 main board reusing the Controller and saving some cost in the upgrade. We also have plans for the Controller and the second release with probably be very different from the first. At the moment this is about keep timescales and what we can actually deliver.

The 400-500 GH/s chips still have to be proven as working and actually practical to use. Currently there are more questions than answers on these but of course we will look at them when it is relevant to do so.

On Litecoin we do have a memory add-on for CM1 going into manufacture. I don't want anyone to think we have Litecoin mining now running on CM1. We don't and won't for at least some time to come give our current workload. This new module will simply be available anyone that wants to try and implement it on a CM1. On the big boards we will come back to Litecoin at some point but no promises when that might be.

Yohan, I agree with the comments above. It would be great to have a scrypt hashing board based on FPGAs, and I think you guys would do well here.

I see no reason why an FPGA scrypt solution couldn't be developed. I used to design FPGA applications on existing boards (~10 years ago) and a board with an FPGA and multiple memory ports to cheap DRAM dimms could be developed that would be very competitive to GPU mining.

The advantage of this is Enterpoint could offer stable supply/pricing based on commodity FPGAs, DRAM and your boards, not this insanity with unreliable ASIC vendors

The issue with bitcoin mining is the next gen ASICs make anything other than them non competitive. You have multiple vendors coming out with ~500GHash/sec chips at very attractive power profiles. You can argue if they will meet the announced time frames, but regardless that is the level of efficiency that will be reached at some point.

This is why when you first announced Goliath I asked what the hashing engine would be based on. With the new ASIC hashing engines there is little room to succeed with BTC mining unless you acquire one of these engines (from very unreliable suppliers.)

Epic.

Just something of a little update. Please don't email/PM asking for more details of specification and pricing as we are on shutdown until Sepember (apart from a small number of people working on CM3/CM4 designs). We also won't say any more until announced in this thread. There is also not a definate time frame for availability as this will depend in part on chip availability that is beyond our control.

There won't be pre-orders for these systems now. They will simply be offered for sale when they are more or less ready to ship.

Cairnsmore3 - We have this design now integrated with of revision1 Goliath Controller and basically debugging the design. We expect to complete that in the next couple of weeks. We will build boards with the Avalon chips we already have following that. Given recent announcements by Avalon we hope to migrate this design to their gen2 chips but we don't have enough technical data yet to say that definately will be a smooth transition.

On the subject of chips we will be offering to buy chips via known supply chains e.g. Sebastian Ju and Zefir at prices to be determined by our view of market conditions. Don't contact us, or the guys, on that directly until we make something more of a formal announcement. We will set it up with the relevant people and the offer will come from them only if they wish to participate in this.

Cairnsmore4 - As some you have already guessed we have changed the CM4 concept to being based on Bitfury chips. We plan to have first hardware on the bench next week but there will be a lot of work still to do on this. We think the design will offer between 360 and 800 GH/s per board but until we do the work that isn't definate as a spec. Time frame for availability is very much down to the next deliveries of Bitfury chips.

Outside of these products we may have something for the small setup as well but more on that later. We will start a seperate thread for that when ready.

We will continue to work with ASIC suppliers as new chips appear to provide successor products in this range.

Possibly we will but it is kind of academic until we have a product ready to launch. For now we will quitely continue our developments of Cairnsmore3 and the new concept of Cairnsmore4. We now won't take advance orders on these until they are actually working and proven. When we are there we will then look at the market and decide if there is viable pricing structure for us to bother with selling units or simply fed our own mining requirements.

Are you going to be making price changes to reflect market value and also other company's tech or are you going to be sticking at 24k

Please say.. Bitfury's 5GHash chips. Nice evolution from the Avalon CM3.  ;-)

Our aim will be to continue with bigger rig concepts. Work on CM3 continues albeit at a slower pace and we are also looking at an alternative concept for the CM4 slot which might be quite interesting but needs much less work to complete. What we don't want to put down at the moment is anything like a fixed timescale until we are more or less ready with whatever of the concepts makes the grade.

So what does this mean then for your cm3 design are you still going to continue to work the project for the 300 to 500GH rigs ? or is this also being stopped too?

+1 The litecoin aspect is really interesting and could be a gamechanger in that currency. I imagine that would bring more success than an bitcoin ASIC would - seeing as theres a lot of run on the marked for those units lately. But noone seem to have been going forward with a litecoin setup - not in public that is.

The first one was already crushed, bASIC. Others have stopped in their tracks before even starting.


I guess this has already happened, we might be a step further already.

However with Mtgox getting ready for LTC, a small scale or larger scale LTC mining solution would be probably the most interesting at the momoment. From what I have looked at technically, it seems that even with larger memory and a good solving strategy litecoin will be far more resistant to FPGAs and ASICs than SHA. This is a good thing insofar as the explosion in hash rate will be more reasonable and less exponential than what happens to BTC right now.

If we were reasonably ready on a Litecoin miner IP design we would probably have carried on. CM4 still makes sense for that market but to be honest being short on people we had to let something go in the short term. Hopefully we can come back to this when we are less stretched. For the shorter term we will continue with the plan to manufacture a memory add-on for CM1. That doesn't mean we will have time to do much on Litecoin IP in Q3 but it will hopefully enable anyone that does want to have a go at a Litecoin design. Depending on what else is happening we will try and look at Litecoin proper in Q4.

As to other technologies we are open to developing other members of this family based on whatever technology is available and appears to be viable as a project. Once we have the Controller module sorted out with Cairnsmore3 most of that work will be reusable into any new board designs. There are going to be several months of instability in the ASIC marketplace and I would not be surprised if one, or more, of the companies offering solutions go out of business or withdraw from producing Bitcoin kit. During this period our main focus will be on our mainstream products and Bitcoin/Litecoin products will simply carry on in the background whilst the market sorts itself out.

That's a real shame, but seems to make sense in the current crazy ASIC environment. There's no sense in being late to the party if some of the other ASIC offers materialise at the end of the summer. Would be great to see bitfury ASICs on a CMx board at some point. Also, your litecoin mining hint for CM4 was cruelly inticing . I'd love to see a scrypt miner from you guys if you can do it!!