【CPUの闇？】コア数を減らしても、案外性能落ちない説（ゆっくり解説）

The number of cores in CPUs has increased significantly in recent years. In the early 2010s, 4 cores were the norm even for high-end models, but now it's not uncommon to see 10 or more cores. But do we really need that many cores? In particular, in games, performance per core is more important than core count, so some people think that 10 or 16 cores are even more unnecessary. In this video, we change the number of cores in the BIOS of the 16-core 7950X and examine in detail how reducing the number of cores and threads affects performance. ■Contents 0:00 Introduction 0:32 Explanation of Parallelization (Amdahl's Law) 5:36 Test Details 6:12 Test Results 13:50 Measurement Results, Including Temperature and Power Consumption 17:03 Summary 21:01 Conclusion 22:44 About Cinebench 23:22 Chit-chat ■Notes Benchmark values ​​are the average of three runs (one run for temperature and voltage, and five runs for the detailed Cinebench R23 analysis). While increasing the number of samples would improve accuracy, time constraints limited us to three to five runs, so please understand that there may be some discrepancies. CPU package power consumption is measured using a sensor. Please note that if you attempt to obtain more realistic values ​​using a 12VEPS terminal, the results will be approximately 10-20% higher due to factors such as power transmission loss. ■Recommended    • 最近のハイエンドCPU、TDP半分でも性能大して変わらない説（ゆっくり解説）      • 農地を建物に変えてしまおう！？ 舞蔵国日記 #6（ゆっくり実況）   Past Testing    • CPUのコア数が増えると性能も向上するのか？（ゆっくり解説）   #CustomPC #CPU #Ryzen #Slow