電腦效能應用與安全研究室 Performance, Applications and Security Lab


我們的研究範圍很廣,從嵌入式系統、手機應用、一直到雲端計算、商務應用、資訊安全都有做。
我們的合作研究夥伴很多,包括聯發科、IBM、中研院、資策會,還有和台大、清大、交大的教授合組研發團隊
,包括高階應用處理器架構研究、虛擬化技術、異質計算、系統軟體等重要技術的研究與創新,我們很關切台灣人才與產業的未來。

2011年1月31日 星期一

醒思:業界與學界


剛好是今天起床前在思考的問題,所以名之為『醒思』:為什麼現在資工畢業的學生還要蜂擁前往明明知道會被狂操三五年的硬體廠( e.g. MTK)?大家不知道burnout(過勞)有多可憐嗎?最近遇到好幾個案例,發覺不少基層的工程師心態和想法怪怪的,不通人情,偏執,情緒起伏很大,很難溝通,動不動就要脅對方要把事情鬧大。

根據政府單位的『過勞自我預防手冊』(http://www.iosh.gov.tw/upload/netbook/preoverwork/preovw.htm
「過勞」(burn-out)指的是一種身心耗弱狀態,乃是長期處在高度心理壓力之下的壓力反應。過勞的症狀與成因複雜,多數研究認為過勞是長期暴露於工作壓力下的結果。過勞與工作壓力是重要的公共衛生議題,所謂「過勞」、「過勞猝死」、及找不出特殊病因的「慢性疲勞症」,在高度工業化國家已是極受重視的職業健康問題。

當然,在一個不是『高度工業化國家』,過勞或許就被看成是個人問題或是個人選擇,沒有受到機關和政府的重視。著名的鴻海跳樓事件,一開始被淡化,要累積到十跳之後,才吸引輿論噠伐,而後中國官方介入,而後鴻海以提高工資和若干措施回應之。

回到台灣,在時間就是金錢的硬體廠作事很可憐,一方面要跟時間賽跑,一方面不知道未來在哪裡。以IC設計產業為例,比別人快一個月把新的IC做出來,就比別人多賺這一個月的錢。當然,數量龐大的話,這個月賺進來的錢就很多了,這是MTK等IC設計公司能夠日進斗金的關鍵,主要靠的是人才,不是資本。時勢造英雄,成名廠商吸引人才前來,按表操課,客戶要什麼就在最短時間設計出來,管他客戶是Apple還是山寨,不必多想些什麼,也不必培養真正創新研發人才。

我們電機和資工界許多的第一流人才,就是到這類地方,貢獻他們的黃金歲月,為他們自己累積財富,為公司的管理階層帶來收益。當然,這些錢為國家帶進稅收,也會促進經濟活動,利益全民。這些都很好,所以國家就讓這些無名英雄在經濟的戰場上,拼命在最前線作戰,台灣的經濟,這幾十年來都靠這些默默犧牲的無名英雄。

軍人保家衛國,公教人員為大眾服務(注:大學教授不算在內,沒有十八趴,沒有什麼保障),就算過了黃金歲月,國家還是會善待軍公教人員。但是業界這些burnout邊緣的無名英雄的宿命呢?運氣好,有人賺了一票,夠一輩子花用,但是大多還是繼續過著burnout邊緣的日子,一直到身體、精神、家庭出狀況為止。運氣不好,賺的不夠多,但是腦袋和身體已經burnout的中年人,廠商不想要,走投無路者大有人在。大多數運氣中等的,就在那邊一直苦撐著,所謂科技新貴,幾年賣命下來,能夠買房子,卻難得有福消受。有趣的是,為了繼續過這種『好日子』,每天還是必須汲汲營營的追趕日新月異的新科技,即便是周遭擁有一流的腦袋的朋友,到了四十歲以後,還是吃不消。

更可悲的是,明天在哪裡?坦白說,我對研究新科技極有興趣,和對於投資哪些新科技公司會賺錢這碼子事完全兩回事。我去Sun做事的時候,是Sun的顛峰時期,而後我在Sun的所見所聞告訴我,好的工程師、好的新科技、甚至收買所謂的好公司,不保證公司一定會賺錢。台灣科技業的明天在哪裡?真的很難講。

在過去,去創業或是去這些非常操的公司,無非是種賭博,或許還是贏面較大的賭博。現在呢?你想賭嗎?科技研發本來就是高風險的投資和賭博,只是用血汗去跟人家(新興的中國和印度)拼命,能夠撐到幾時?以業界大老闆和管理階層的思維,那裡便宜哪裡做,不會跟你講義氣,不會照顧你下半輩子,你還要賣命嗎?

本著教育者的良心來講這席話,讓學子們知道這些很多既得利益者不願告訴學生、一般大眾濛濛懂懂的現實,不是我們在大學應該做的嗎?讓學生們在學校學到真正有用的研發技能,將來有能力幫助業界突破現今科技界僵局,不是我們所謂研究型大學應該努力去做的嗎?還是任憑業界守舊的心態宰制我們第一流的學生,還把原罪歸咎於學界培養不出有用之才,笑我們為什麼不讓學生早點畢業到業界接受訓練?嚴重扭曲的大學精神,以及逐漸被鄙視的大學教育,難道在象牙塔的長官們還漠不關心?是渾渾噩噩,還是無動於衷?

該不該念博士?


每隔一陣子,就會有學生問到,該不該念博士? 不少正在念博班的學生,不瞭解自己為什麼要念博士,不知道自己適不適合念下去,也不知道念完之後要做什麼。問我值不值得念?老實說,這是個很複雜的問題,也是很切身的問題,答案因人而異,很難回答。

這篇2010年底經濟學人(The Economist)上的文章"The disposable academic - Why doing a PhD is often a waste of time",引發不少迴響和辯論,是個不錯的參考資料。在此為各位考慮念博士或是已經上了賊船的學子們作個整理:

我的個人意見:

就所謂的投資報酬率來說,第一,選擇學校很重要。第二,選擇領域很重要。第三,選擇專業很重要。念完博士,是不是有超乎經濟價值的東西呢?有的,但要你自己去找出來。

這篇文章有些以偏概全,真正關心這個議題的人,該去讀一讀這篇文章所引發的迴響。跟一般的討論版不同,在經濟學人(The Economist)這個版上,可以看到不少認真的討論以及討論者的背景資料。

For my Ph.D. students, please read the article and the comments -- either quit Ph.D. and find yourself a real job immediately, or stop complaining and focus on your research!

今天早上花了一個小時讀了最前面的五十篇comments,順便整理了一下其中比較有意思的。後面還有大約140篇,大致上意見是重複出現的,不然就是個人經驗。如果我是學生,我會全部看完,再想辦法得到個人的了悟。

以台灣的學術界來說,除非你天縱英明,否則念完博士後,找教職很困難。然而,台灣在電腦這個領域,由於電腦越來越複雜,以及近年來電腦產業研發門檻和需求大幅提昇,業界越來越受重視學有專長能夠獨當一面的博士級研究人員,這是目前和其他領域有很大不同的地方。很多公司都找我要研發人才,光是近半年來談過的就有IBM, HP, Qualcomm,Intel,趨勢,廣達,英業達,聯發科,晨星,資策會,工研院,缺人缺的很嚴重。

如果你想進一步詢問我的想法的話,請讀完這篇文章,還有大部分的comments,並且對於你個人特質和研究領域有足夠的覺知之後,我們再來聊一聊!

對於有趣的comments,我整理如下:

1. 選擇學校很重要,相關的comments:

(Rand0) "One Ivy-League president, asked recently about PhD oversupply, said that if the top universities cut back others will step in to offer them instead."

I'm confident that Harvard PhDs are in strong demand, as almost every university wants "PhD, Harvard University" to appear on their faculty roll as many times as possible.

Don't go to a school that would never hire its own graduates.

2. 選擇領域很重要,相關的comments:

(Rand0) Its the exact same for PhD programs. Don't get a PhD because you don't have any better prospects (and certainly don't pay for one if you don't), it won't help you in the long run.

(HuskyPhD) I got my Ph.D. a few months ago in a scientific field. I didn't want to stay in research, but I figured with my credentials, I could easily land a job. I applied for around 40 jobs, and I received exactly one offer.

3. 選擇專業很重要,相關的comments:

(quasimetric) Here's the solution: find an industry you want to work in, make contacts in that industry, ask them what questions they're thinking about. Make the answer your thesis.

(Bowl Weevils)Much of this problem could be solved if US universities actually hired faculty at a rate equal to the demand from students. American production of PhDs may have doubled since 1970, but there are also 100 million more Americans, many of whom are young and are increasingly told that they must get a quality university education to survive.

4. 念完博士,是不是有超乎經濟價值的東西呢?相關的comments:

(Albert Dutch) I happen to have a PhD; and a great career in industry. My experience could not be more different to the correspondent's: I have always felt privileged, and the PhD was the beginning of my beautiful journey. I have worked in areas which are exciting and beautiful. A good combination when added to a nice salary. It's true that some people get higher salaries, but very few of them love what they do as I do. This is the beauty of a PhD: you love what you do.

Very few people in life are as lucky as PhD students. The luxury to work on something they like, expanding the boundaries of knowledge and learning a set of skills which will make a difference in their careers. My advice for to-be-PhD students is to be aware of that, and enjoy the experience. If they do not, then quit.

(Carpjaws) I did a lab-based PhD (admittedly about 20 years ago now) and, overall, it was an enriching experience. I currently don't work in as a lab researcher, but I think some of the skills learned have been useful in my different lines of work since. Lab-based research can teach useful lessons in teamwork, diplomacy and camaraderie, more so perhaps than say a history PhD spent alone in an archive. In my day, it was also often over pretty quickly (I was 24 when I was awarded my PhD, hardly too old to go looking for changes in direction).

(pmkim) no sane person does a PhD for an increase in salary and the numbers toted by the author are well known. Perhaps a better indicator would be job satisfaction. University professors have a famously high job satisfaction rate (among the highest in the US) and higher echelon researchers at large companies (the sort that usually has PhDs) also do.

(David Karger) [Professor of CS at MIT] It makes no sense to talk about PhD's in terms of salary bumps. Getting a PhD, and even more so going into academia, is a lot like going into art: it involves a willingness to sacrifice earnings in search of other kinds of fulfillment. Anyone with PhD in the hard sciences is smart enough to make plenty as a quant on wall street; they've made a decision that certain kinds of job satisfaction matter more to them than money. By the same token, it's silly to whine about low pay rates for PhDs. To take a job you love and then expect to be paid really well for it is asking for a double portion.

(davidstuart) You don't have to be a genius to earn a PhD. One does have to be smart, and what is most important is the ability and willingness to learn independently, apply oneself over a long time to a difficult problem, overcome complex and frustratingly difficult hurdles, and to stick with the program and to love your work. If you want to hire a person to conduct a multi-year project of a difficult and highly technical subject, what better selection criteria could there be than someone who has earned a PhD? If you want to work in such an environment, how better to learn and demonstrate your qualifications?

5. 關於文章的瑕疵:

(Boguespierre) "Since then America’s annual output of PhDs has doubled, to 64,000."
Followed two paragraphs later by: "In a recent book, Andrew Hacker and Claudia Dreifus, an academic and a journalist, report that America produced more than 100,000 doctoral degrees between 2005 and 2009." I can see why you decided upon theoretical ecology rather than mathematics.

(pmkim) the author contends that universities are overproducing PhDs. The mere fact that unemployment among PhDs is by far lower than for bachelor or high school graduates flies blatantly in the face of this assertion. Last figure I recall is that PhD unemployment has been steady below 2% for many years (Science & Engineering PhDs). While it is true that only a small fraction stays in academia and eventually becomes a professor, the overwhelming majority then pursues a great career elsewhere.

(mbmusgrove) All PhDs aren't the same. Sounds obvious, but by the title of this article, it isn't to everyone. The difficulty of acquiring them as well as their demand are all over the spectrum, as is the pay, which is often averaged into 1 figure, as in this article, which is inaccurate at best and worthless and misleading at worst. A History or Econ PhD isn't going to make nearly what a marketing professor is, for example. Ironically, Econ depts are the ones churning out armies of PhDs and lowering the value.

6. 其他人的個人經驗:

(MacJr) I got my PhD recently from a top university and right after graduation got a position as an assistant professor (tenure-track). It felt great at that time. But now I am growing disillusioned with the academic career. Don´t get me wrong. I love what I do and my research is getting slowly published. But what disappoints me the most is the pay. You work 6 days a week, 10h/day. And you make relatively good money that affords you a middle class type of life. But the effort/benefit ratio seems way too high in this career. I can´t help but think that I am doing a terrible use of my intelligence and effort. The outside options (own consultancy company) just seem so much better. Not only a less socially isolated type of work, but also way more financially rewarding. As a signal, a PhD degree is still great. But career-wise can really be a completely waste of time.

(Count7) I would suggest a satifying alternative to the Ph.D. is to get a master's degree in several fields. Ph.D. programs are often so specialized that your knowledge base has limited application in the real world.

(mbmusgrove) I have an MBA and my wife is about to wrap up her PhD in marketing. She is in a University of 30,000 students and is THE ONLY STUDENT in her graduating class. Other marketing PhD classes here have between 1-3 students, and by the time they're ready to defend, that # will probably be knocked down to one or two (historically)as well, due to the difficulty of the program. She'll have her dissertation done in March, graduate in May after 4 years, and an MBA on top of that. I won't say the name of the school because that isn't the focus, which it would certainly become, but it isn't a top 10. She has already signed a contract for a tenure-track position at a major US University for next fall. Her income will be over three times the local per capita, and her schedule is 'flexible' to say the least. She had 14 interviews for a tenure-track marketing position this fall.

(rstanny) We were having exactly this discussion thirty years ago when I took my PHD. They said if we could just hold on till the 90s there would be lots of retirements among the Korean War vets. (Didn't happen--or, if it did, the effects weren't detectable.) My advice is don't do it unless you're both driven to do research or teaching and have a plan for finding happiness coping with the significant probability of becoming flotsam in the struggle for academic survival. If you do decide to try: (1) do not spend anything on your advanced degree except GI Bill money (or other surplus money you could never, ever, possibly need for food, shelter, or other important stuff), (2) do not marry before achieving tenure, except to a tolerant, easily self-employable professional (such as an MD), who will be willing to relocate, several times, to wherever it is you need to go during your periods of academic wandering, and (3) for goodness sake don't marry someone in your field, who will only be job-market competition, and who will probably be as smart as you, not to mention better looking.

(Pebbles8) I was putting in far too much work for the pay and abuse I was receiving and looking at jobs prospects in my small city (I wouldn't want to move since I am married and prize strong ties with my family) it was unlikely I would be able to get one making more than 32k with a PhD. I left with a masters, 4 publications and a more practical outlook on the world. I am now working in research and development at a small biotech company (~40 hours a week) for more than a post doc (at least 60 hours a week) would pay and getting much better job experience than if I had finished the final 18 months of my PhD (probably mainly kissing the behinds of my committee members.

Many of the comments say that the situation must not be that bad since intelligent students are still choosing that path. What they fail to mention is these are intelligent, idealistic 21-23 year olds who have never failed and who think they are they ones who will succeed.

(P.Allen) Unless you are really exceptional, your life as an academic will be hard work. You may not have all that much timetabled work, but you will get referee requests, requests to organise conferences, administrative duties and student problems on top of the timetabled teaching. You will not be judged on your performance in any of these (for promotion: but do them badly and your colleagues will hate you), and only to a fairly limited extent on your teaching performance. To get a promotion or another job, you must do research, and it has to be at least reasonably good and in some quantity. You will not be able to do this on 40 hours a week total. You will need to be thinking about your research at weekends, during your family holiday, over Christmas, and in the evenings as well as 9-5. What's more, you need to accept that maybe 95% of the time you spend thinking will simply consist of finding out that a certain attack does not work - that's valuable information but almost never publishable. You will not be paid well for this: you could almost certainly make twice as much money without working any harder in a finance job.

On the other hand, you are doing it because you enjoy it: it's rewarding to discover something new, it doesn't feel like a 'daily grind' to do research (though writing it up can, and especially rewriting it after the referee report comes back). You have a lot of freedom to choose your working hours, you can work with friends (usually), you can usually get paid two or three times a year to go to some interesting (though usually cheapish) place for a conference and combine it with a holiday.

This is how an academic also expects a PhD student to behave: but you can't really tell by interviewing if your prospective student will work hard enough. You might be lucky, and get hold of an obviously brilliant student who will do well even if they are lazy, but probably you won't. You have to gamble that you get a student who will actually help your research, who will be willing to put in the time making many failed attempts to solve their problem before they finally find a solution, instead of giving up and stopping working (that's common enough). If you get a student who isn't willing to put the time in - well, what can you do? You can be honest, and tell them to quit (but then they may well complain to your boss that you aren't trying). You can try to make them quit by dumping scut-work on them and making it obvious you don't rate them (some academics are not nice people). You can do all the hard work on their problem yourself, and tell them to do the final calculations and write-up (and they will never feel the reward that comes from actually solving something). Or you can let them run on until the funding runs out.

So - taking a PhD student is a gamble. It pays off more often than not, so academics will keep doing it. But if you start a PhD expecting it to be easy, or because you want to stay a student (which means: you don't want to work a 40 hour week, let alone a 60 hour week), or just because you didn't know what else to do - then you will probably be the academic's bad gamble, and there isn't much chance it will go well. At the end, what will your supervisor write for a reference? Either something bland, or something honest, and neither will go far to getting you a good job.

(J2Bryson) Good PhD students are essential to academic research, but bad ones take time and hold you back. Nevertheless they are seen as essential to the prestige & bottom line of our institutions and we are pressured to take them. Still, you are entering into a long-term, close, asymmetric relationship with someone whose career / life you will have profound impact on, so I always try to be honest about the outcomes and to be sure a candidate's motivations are sensible. Maybe consequently I wind up with a lot of mature students who know what the alternatives in life are and are really making an informed choice -- and also already have another career to go back to if they need to. I hate seeing universities recruiting their own best undergraduates to stay directly after graduation -- that is the worst thing for everyone. The students would learn more from experiencing another institution, and if the institution can't attract PhD students from other peer or superior institutions then they don't deserve to have them. One positive note on the economics -- more and more people in my field (AI) are writing grants for graduate research officers rather than PhD students. This allows candidates to be paid more than a PhD stipend and just pay their own fees (which in the UK are quite low for EU students) out of their salary. By the way, one totally incorrect part of the article -- PhD students do NOT bring in funding. WE have to fund THEM. That obligation is what keeps us up all hours writing grants.