LTC FPGA discussion!

[shibaji]

The title of this post is going to be very ironic in coming months.

I predict most the benefits of fgpas will simply be lower power consumption per coin.

Your prediction is incorrect. Until the market for LTC matures enough so that someone takes a shot at ASIC, FPGA will rule.

[1715262376]

1715262376

[1715262376]

1715262376

[ElectricMucus]

I am curious what are you up to? Make a memory controller into the FPGA and pack a bunch of DRAM chips onto the board?

Wouldn't you like to know!

Of course I would, but since you aren't really forthcoming: I smell BS.

[jasinlee]

There are several methods to provide the memory, why would I disclose it publicly so others could use that as a starting point in their own design? That would be completely idiotic. I will point out basic stuff, but not design details or methods.

[jasinlee]

The title of this post is going to be very ironic in coming months.

I predict most the benefits of fgpas will simply be lower power consumption per coin.

It would be ironic, if I made the thread....

[bushstar]

I will be waiting patiently. Good luck to you.

[ElectricMucus]

There are several methods to provide the memory, why would I disclose it publicly so others could use that as a starting point in their own design? That would be completely idiotic. I will point out basic stuff, but not design details or methods.

From what I can tell there are 3 potential methods:

On FPGA blockram only, highest bandwidth but FPGAs with a good amount are freaking expensive.
External SRAM, easy implementation but not as cost effective
External DRAM, needs to dedicate FPGA resources for refreshing circuitry but memory is a lot cheaper and there might be spare logic blocks anyway (so that what I did bet on)

I don't really think there are other options. And from what I know the choice is obvious.

[jasinlee]

Then why ask ? Our method will be a bit unique so I dont want to be putting that out there. I am sure if you run a business you can respect that.

[ElectricMucus]

Then why ask ? Our method will be a bit unique so I dont want to be putting that out there. I am sure if you run a business you can respect that.

I don't I am merely curious, but I might consider buying one if I know what I would get. 

[tgsrge]

To anyone passing by:
nothing to see here, move along.

In this regard, Litecoin specialized/custom hardware will not be Bitcoin: part deux.

scrypt is made -specifically- to make custom hardware (such as fpgas, and others) attacks much more costly/less beneficial than otherwise it would be without it. Unless they have discovered a vulnerability/flaw/subtle property no one else knows about, the MOST they are going to get is power savings.

Dont expect anywhere near the same kind of advantage that asic/fpga has over other means of mining as in bitcoin. fpgas and the other forms of custom hardware wont have the same volume savings as gpus, the same amount of engineering, nor nearly the same amount of r&d $ to back it up.

again, all they will get is possibly some power savings, and even this being due to the fact they dont have to deal with the power overhead of the rest of the components in a pc gives other miners.

until (and if) this comes out and we see what it's capable of (or they post actual proof of the benefits of their custom hardware) i'm gonna stand by this, and others have no reasons to do otherwise.

[jasinlee]

When we have a complete product, I will publish info, I am not going to take preorders or screw with peoples heads with conjecture about our final speeds/design/kwh/delivery dates/color/size/weight/cup size. I will happily chit chat about it if it does not touch on something that could cause problems later.

[jasinlee]

To anyone passing by:
nothing to see here, move along.

In this regard, Litecoin specialized/custom hardware will not be Bitcoin: part deux.

scrypt is made -specifically- to make custom hardware (such as fpgas, and others) attacks much more costly/less beneficial than otherwise it would be without it. Unless they have discovered a vulnerability/flaw/subtle property no one else knows about, the MOST they are going to get is power savings.

Dont expect anywhere near the same kind of advantage that asic/fpga has over other means of mining as in bitcoin. fpgas and the other forms of custom hardware wont have the same volume savings as gpus, the same amount of engineering, nor nearly the same amount of r&d $ to back it up.

again, all they will get is possibly some power savings, and even this being due to the fact they dont have to deal with the power overhead of the rest of the components in a pc gives other miners.

until (and if) this comes out and we see what it's capable of (or they post actual proof of the benefits of their custom hardware) i'm gonna stand by this, and others have no reasons to do otherwise.

All internet engineers are welcome to post their opinions in Simran's thread. Have fun.

[mr_random]

To anyone passing by:
nothing to see here, move along.

In this regard, Litecoin specialized/custom hardware will not be Bitcoin: part deux.

scrypt is made -specifically- to make custom hardware (such as fpgas, and others) attacks much more costly/less beneficial than otherwise it would be without it. Unless they have discovered a vulnerability/flaw/subtle property no one else knows about, the MOST they are going to get is power savings.

Dont expect anywhere near the same kind of advantage that asic/fpga has over other means of mining as in bitcoin. fpgas and the other forms of custom hardware wont have the same volume savings as gpus, the same amount of engineering, nor nearly the same amount of r&d $ to back it up.

again, all they will get is possibly some power savings, and even this being due to the fact they dont have to deal with the power overhead of the rest of the components in a pc gives other miners.

until (and if) this comes out and we see what it's capable of (or they post actual proof of the benefits of their custom hardware) i'm gonna stand by this, and others have no reasons to do otherwise.

Yup. This.

If research computer scientists haven't found an effective way to parallel compute Scrypt I doubt a random dude off an internet forum will figure it out all by himself and his soldering iron.

Litecoin FGPAs will bring an advantage of course, but it will be nowhere near as dramatic as bitcoins ASICs.

[jasinlee]

Of course it wouldn't be, Asics are not FPGAs. And as you probably did not notice I have noted before, I am not the vhdl developer...the random dude to which you refer actually is an altera engineer.

[tgsrge]

All internet engineers are welcome to post their opinions in Simran's thread. Have fun.

No amount of "engineering" (read: buying crap, outdated, regurgitated, often counterfeited packages and pre-made ic from (more ofthen than not) dubious sources) is capable of making magic flying pink elephants appear in the sky.

[ElectricMucus]

When we have a complete product, I will publish info, I am not going to take preorders or screw with peoples heads with conjecture about our final speeds/design/kwh/delivery dates/color/size/weight/cup size. I will happily chit chat about it if it does not touch on something that could cause problems later.

That makes sense.
It is just - from what I can tell that this would go beyond the effort of making a FPGA bitcoin miner and I doubt that anybody would be able to steal the concept based on a few specs. The development board alone needed to prototype the concept would cost a relative fortune alone.
Lets just say I am surprised anybody goes through the hassle at this stage of Litecoin, or you could be just on time for the big rush.

Either way good luck, you'll need it.

[jasinlee]

Thanks, we have seen lots of indicators and heard of projects that we feel will cause ltc to be the leading coin in the next few months so, we like to think we are right on time.

[SyRenity]

As this thread turns, it quite reminds me of the initial Avalon postings. The entrepreneurs were met with ridicule, then with suspicion, then with accusations, and look like what it looks now in batch 3 pre-order.

I for one learnt the lesson of never discounting anything too early (until proven otherwise), and think you guys are definitely on time to market. I suggest everyone to wait until making their judgment and will be happy to support you as an early adopter.

[Wekkel]

Thanks, we have seen lots of indicators and heard of projects that we feel will cause ltc to be the leading coin in the next few months so, we like to think we are right on time.

Good to hear. LTC seems to have some good times ahead.

[SyRenity]

Speaking on subject, I read somewhere scrypt is too complicated to be done in ASIC, therefore FPGA is the only way to go. Is it true?

Because otherwise (and forgive someone as ignorant in HW design as me), why not skip straight to ASIC?

[tgsrge]

Speaking on subject, I read somewhere scrypt is too complicated to be done in ASIC, therefore FPGA is the only way to go. Is it true?

Because otherwise (and forgive someone as ignorant in HW design as me), why not skip straight to ASIC?

asics are more complex. besides, you never release the absolute best of what you will have to offer as your first iteration. You always need to leave something out for the next iteration so people have a reason to buy your product.