自從電郵的出現，我還以為「開會」是一個可以淘汰的工作了，原來不是。文字始終還是不能取代聲音，語法仍是不能代替情緒，而清楚的一段電郵，依然及不上見面一句「好的」。
有人說印刷是學術發展第一大發明，因為訊息
悶蛋
有一種人叫做悶蛋，你我都一定認識一兩個。說悶蛋像杯白開水，是老掉牙的比喻，而且不很貼切，因為我們總要喝白開水，卻不一定要跟悶蛋打交道。悶蛋更像的是杯極淡的茶，徒有茶色，卻無茶味；我們或一不留神，或礙於
我之為我，無可奈何
問朋友對我的網誌有甚麼意見，他很直率，說我寫的東西 "so polished, so cultured"，而且有刻意製造形象之嫌。
第一點我絕無異議。本人天生凡事都喜 polish 一番，寫文章如是，下廚如是，打掃家居如是，連戴著的眼鏡也要經常除下來抹，以
第一點我絕無異議。本人天生凡事都喜 polish 一番，寫文章如是，下廚如是，打掃家居如是，連戴著的眼鏡也要經常除下來抹，以
Things that limit the performance of hardware acceleration
Suppose that you have an infinitely fast hardware accelerator, one thatcan compute something of interest in no time at all. What externalissues limit the total performance advantage that you can get by puttingthis hardware accelerator in a system?
I can think of the following limiters:
- main memory speed limits , the latency andbandwidth limits of system RAM. This limits how fast you caninteract with system memory.
- the speed limits of the underlying hardware that you're talking to,if you are. For example, hardware RAID
當明天有約會時
自四、五年前長時間用了不適合的護髮用品，再加上當空姐後經常紮髻，我的頭髮由掃把般厚變得異常稀薄。
為了拯救垂死的頭髮，上班時任由baby hair彈出來也不肯多用定型spray，過千元的洗髮水+護髮素，付錢時面不改容。
兩年前
Merry Christmas!
Merry Christmas! May you have a blessed day with your family, or with the Holy Family in your heart!
從吃人肉說起
Linux's non-strict overcommit is the right default
I've written before about Linux's overcommit settings and the general background of virtual memorylimits . Several years of experience ofrunning various general use systems with strict overcommit and payingattention to the numbers have convinced me of something: non-strictovercommit is the right default on Linux , and probably on any modernUnix in general.
I have come to this view through experimental evidence, namely thatall of our user login machines generally run with significant amountsof committed address space, yet they have lots of free memory