Theoretical question regarding Gridseed GC3355 chip...

[skyhawk]

So apparently the general consensus is that the SHA-256 hashing function on the Gridseed miners isn't worth bothering with, and is more of a hindrance than anything else. All the interest in these units is directed at the fact that they're the only Scrypt ASIC in town.

If the team behind the GC3355 was to strip out all the SHA-256 circuitry, could they make a similar chip with superior Scrypt performance, or is the Scrypt performance mostly limited by the very expensive (space-wise) banks of RAM required to perform it? If the RAM is the limiting factor, then is it safe to consider the SHA-256 hashing to be basically just "bolted on" functionality that would otherwise be wasted space on the finished product?

I ask because the developers can't possibly be unaware of the lack of SHA-256 demand for their product. What are the odds of them releasing a superior Scrypt-only chip in the near future?

[CaptEmulation]

They way I heard, I really have no idea if this is correct-- The SHA256 function is on the chip because scrypt coins still use some SHA256 in their workflow-- thus the SHA256 hardware is necessary.  The dual mining is a bonus from that requirement.

[nybbler905]

They way I heard, I really have no idea if this is correct-- The SHA256 function is on the chip because scrypt coins still use some SHA256 in their workflow-- thus the SHA256 hardware is necessary.  The dual mining is a bonus from that requirement.

yep, the way I read script mining was that it was X iterations of SHA256 for each script operation where X is what the script part deals with along with using the results per SHA256 in to the next SHA256 of the single Scrypt hash.  whole lotta SHA256 hashing per 1 round of scrypt hash.  If I found out what X was, i'd prolly already have worked out IF i can move my Erupters to a dedicated PI based scrypt miner