3 ms·
It's interesting that you didn't consider internal SRAM; 2 MB of SRAM is less than 10 mm2 in a 10 nm process so an entire CryptoNight core might be ~12 mm2. Thi
by wmf 9y ago
It's interesting that you didn't consider internal SRAM; 2 MB of SRAM is less than 10 mm2 in a 10 nm process so an entire CryptoNight core might be ~12 mm2. This would be roughly twice the area efficiency of x86.
- dragontamer 9y agoQDR-IV is SRAM. I think buying QDR-IV and externally interfacing would be easier. But I haven't thought too much about this aspect of the problem. In effect, QDR-IV already exists and is already mass produced. So if you wanted SRAM, there it is. Already to order from numerous suppliers. I don't see much reason why you'd want to design it into an ASIC when its a ready off-the-shelf part. Overall, I think the optimal design would use RLDRAM3 anyway. RLDRAM3 is cheaper and uses less Watts than QDR-IV. The tRC latency issue on RLDRAM3 doesn't seem to affect the expected workload of a memory-heavy Cryptocoin, so QDR-IV (aka SRAM) wouldn't have much of an advantage over RLDRAM3 anyway.