3 ms·
David Patterson (who else) experimented with the idea almost 20 years ago. Idea was to put CPU into SRAM, remove caches and connect to the rest of the world wit
by nabla9 9y ago
David Patterson (who else) experimented with the idea almost 20 years ago. Idea was to put CPU into SRAM, remove caches and connect to the rest of the world with fast serial interfaces.
http://iram.cs.berkeley.edu/ http://iram.cs.berkeley.edu/
I'm guessing that mixing CMOS and SRAM into same process is hard to do in large scale.
- jacquesm 9y agoSRAM can be CMOS.
- kurthr 9y agoWith water stacking, its not unreasonable to have high interconnect density for two different (highly optimized eg DRAM and FinFET) processes with high yield. I expect it to happen sooner than later. Latency to extremely large caches could be significantly improved. I expect it to happen. https://semiengineering.com/whats-next-in-scaling-and-stacking/ https://semiengineering.com/whats-next-in-scaling-and-stacki...