3 ms·
44mb of cache really highlights the bottleneck of how raw compute is cheap, but the real limit is memory latency.
by waazy 1mo ago
44mb of cache really highlights the bottleneck of how raw compute is cheap, but the real limit is memory latency.