Picture a cutting a small mark into a stone wall. The spot is chosen because it will not move, and its exact height is written down. From then on, anyone who needs to know how high or how far something is comes back to that mark and measures from it. Two things make the mark useful, and they are worth separating: it is fixed, and it is agreed on. Every later use of benchmark keeps both. The mark itself disappears — it becomes a sales figure, an exam result, a test program — but the job never changes. Somebody puts a point on the wall, and after that, everything else is described by where it stands next to it.
The supplied Wiktionary entry dates the word to about 1842 and analyses it as bench + mark: a mark cut into stone by land surveyors so that a bench — a bracket in their jargon — could be mounted to hold measuring equipment. The figurative standard-for-judging sense appears about 1884.
所附的維基詞典把這個字定在大約一八四二年，拆成 bench ＋ mark：測量員在石頭上刻一道記號，用來架設 bench（他們行話裡的一種托架），好把儀器固定上去。「拿來評斷的標準」這個引申用法大約在一八八四年出現。這條線索正好說明今天的用法為什麼都圍著「一個不會動的起算點」打轉。
Wiktionary entry supplied in this evidence packet (source_ref kaikki:3f6f69a190036ae4:1006550966): "From bench + mark. First use appears c. 1842. Originally a mark cut into a stone by land surveyors to secure a bench ... The figurative sense first appears c. 1884."
Four branches
刻在石頭上的那個點
這一家就是那道記號本身，還沒有被借去講別的東西。後面每一家都是把這個點換成別的東西。
noun · C1
A mark cut into stone or metal at a place whose exact height and position are already known, so that surveyors can return to it and measure from it. It is shown on maps as a fixed reference point.
水準點；測量基準標記
被當成標準的那個數字或成績
記號離開了石頭，變成一個數字。固定和公認這兩個條件沒有變：有人決定「就用這個」，之後別的東西都用離它多遠來講。
noun · B2
Something fixed that you use as a standard, so that other things can be judged or measured against it. Prices, exam results, salaries and company figures are all compared with a benchmark to show whether they are high, low or normal.
基準；標竿
拿自己去對著那個點量
從「那個點」變成「拿東西去那個點量」的動作。記號還在，只是這次動的是被量的人。
verb · B2
To judge how well a company is doing by measuring it against another company of the same kind that is known to do the job well. The usual pattern is benchmark A against B.
以…為基準衡量；對標
在電腦上，那個點變成同一套測驗
這一家把固定點放在「測驗」上：所有機器跑一模一樣的工作，剩下的差別就只有機器本身，成績才並排得起來。跟公司對公司完全無關。
noun · C1
In computing, a test program that is run on a machine or on software to find out how fast it does a set job, so that different machines can be compared on exactly the same work.
效能測試（程式）
以下是外部詞典記錄的用法，原文照列，我們還沒有寫成教學內容。上面的說明才是我們自己寫的；有些用法可能上面已經講過了。
benchmarknoun/ˈbɛn(t)ʃmɑːk/
A standard by which something is evaluated or measured.
A surveyor's mark made on some stationary object and shown on a map; used as a reference point.
computingengineeringmathematicsA computer program that is executed to assess the performance of the runtime environment.
benchmarkverb/ˈbɛn(t)ʃmɑːk/
To measure the performance or quality of (an item) relative to another similar item in an impartial scientific manner.
To measure the performance or quality of (an item) relative to another similar item in an impartial scientific manner.
資料來源：英文維基詞典 English Wiktionary（CC BY-SA 4.0）。 原文照列，未經改寫。