社区长文:整条链相当于多少赫兹的电脑Long-read: how many hertz is the whole chain?
BruceBlue 转发推荐 Joooooo「链上虚拟硬件」系列第二篇。文章以约 50.7 万个 NAND 的「吟游诗人」电路为参照,按整条 BSC 的预算推算:官方逐门解释下每秒约走 0.14 拍,编译后可到十几拍;256 路可并行处理多组独立输入,但同一任务中后一步依赖前一步结果时仍需顺序执行。作者还修正了上篇与原生合约的成本对比,指出原生代码优化后原有优势明显缩小,并保留更正与反驳的过程。BruceBlue shared and recommended the second post in Joooooo's "on-chain virtual hardware" series. Using the roughly 507,000-NAND "Bard" circuit as a reference and extrapolating to all of BSC, the post estimates about 0.14 peta-ops per second under the official gate-by-gate explanation, rising to the teens of peta after compilation; 256 lanes can process many independent input sets in parallel, but steps within one task that depend on previous results still run sequentially. The author also revised the cost comparison with native contracts from the first post, noting that optimization of native code sharply narrows the earlier advantage, and kept the correction and rebuttal process on the site.
