In a new paper, Vitalik Buterin, co-founder of Ethereum, examines one of the most overlooked limitations in computing: memory access time, and how to change the way cryptographic systems are developed by better understanding it. Although subjects appear to be very technical, they are directly related to the potential efficiency of blockchain, encrypted proofs, and even AI models.
Memory access issues
Buterin argues that it is wrong to assume that reading and writing memory takes a fixed amount of time, as is often done in computer science. Rather, he presents a model that proposes that memory size cue brutes determine memory access times. Simply put, retrieving data from memory gradually slows down as data is acquired from memory, as it requires moving larger physical distances.
He provides evidence from the real world to support this, showing how long it takes to access data, from CPU caches to RAM, along with memory size, is surprisingly consistent with his theoretical model. This realization goes beyond mere academic capabilities — it fundamentally changes the way we think about algorithm optimization, especially in the field of encryption.
Blockchain Memory Management Solution
Vitalik Buterin illustrates this point using an example that includes elliptic curve encryption, a fundamental part of blockchain security. To accelerate the process, developers often pre-calculate large tables of numbers. However, as it is caused by increased memory accesses, if the table is too large and fits in cache memory, the speed increase can be lost. The smaller cache fit table proved to be faster than the larger ones stored in RAM in one of his tests.
The conclusion is simple, but profound. Encryption efficiency includes more than faster processors. It also involves more intelligent memory management. Understanding the principles he laid out could dictate future hardware optimizations for blockchain and zero-knowledge systems as the industry moves to specialized hardware like ASICS and GPUs.
