ASIC-Based Folding

[VectorCalculus]

I was wondering about possible future implementation of a folding client for ASIC-machines, as they are much faster at processing the math problems used in cryptocurrency mining than traditional GPU/CPU based solutions. A lot of the calculations encompassed therein are based on floating-point math, just like FAH. Because of all of this, I was curious as to whether or not such machines would provide a noticeable increment to the performance of the FAH network were they to be incorporated in the future. So, the question is, would they? And also, would the usage of such nascent technology be considered?

[toTOW]

The maths behind cryptocurrencies is pretty simple, repetitive and predictable : it's always the same algorithms, with no variability in the amount of data processed.

In FAH, the algorithms are much more complex, and inputs are variable (different size of proteins, different kind of atoms or forces, ...), so building ans ASIC would cost a lot of money and time, and its lifetime would be rather limited as new techniques often appear in molecular design ...

And don't forget that the market for Folding ASIC would be rather limited, so the cost would split across limited market, that would make it hard to get competitive solutions.

[ChristianVirtual]

Still there is https://en.m.wikipedia.org/wiki/Anton_(computer); based on custom ASICs. but agree: that's expensive and beyond our scope (how many PPD that monster would bring?)

[foldy]

How about a FPGA? Not so fast like ASIC but can load any hardware code.
It would load a fah core and project or unit which was compiled for fpga.
When fah core or project or unit changes then the FPGA is reprogrammed again.
Requirement for success: Execution time on FPGA including loading time must be faster than current GPUs and use less power.

[toTOW]

Still there is https://en.m.wikipedia.org/wiki/Anton_(computer); based on custom ASICs. but agree: that's expensive and beyond our scope (how many PPD that monster would bring?)

This has already be mentioned in another topic : viewtopic.php?p=274342#p274342

[toTOW]

Still there is https://en.m.wikipedia.org/wiki/Anton_(computer); based on custom ASICs. but agree: that's expensive and beyond our scope (how many PPD that monster would bring?)

This has already be mentioned in another topic : viewtopic.php?p=274342#p274342

foldy> not a bad idea ... but try to get a board for general public, it will be hard (usually sold by manufacturer to industrial customers) and expensive (since you won't buy many chips at once)

It's so easier to stick to GPUs

[7im]

Also:Re-purposing Bitcoin ASIC miners

[tnaa97]

Question from a newbie non programmer.

Could F@H create a SHA 256 Coin for some of the diseases? if so we could "Recycle" some of our older ASICs like Antminer S3?

Apologies if there is really no way.

Thank you in advance for your thoughtful time.

[Joe_H]

Question from a newbie non programmer.

Could F@H create a SHA 256 Coin for some of the diseases? if so we could "Recycle" some of our older ASICs like Antminer S3?

Apologies if there is really no way.

Thank you in advance for your thoughtful time.

There really is no way the Pande Group and the F@H consortium could get into creating or working on alt-currency. It would probably be considered misuse of grants and other funds provided to them for research. As it is they have limited funds that they can use from those grants on programming.

Unfortunately, as mentioned here and elsewhere, the type of computations made by ASIC's are not suitable for folding itself.