The schedule of the project at work is really bad. My part has been slipped for almost 3 weeks already. I suppose to have 1.5 months to verify my block before I take off for vacation. However my work depends on the deliverable from other team members. Both the RTL design and the testbench components are late and cause a cascade effect to my schedule. Now I only have a little more than 2 weeks before my vacation and those parts are still not working. Today I got another release from Saskatoon, but it turn out the code is so buggy that it is unusable. It seems the first milestone of flushing out the datapath has to wait until I come back from Germany. The managers keep saying my block is not on the critical path so it can slip a bit. I hope the schedule will not get out of control and put me in the hot seat when it’s time to tape out.
Tag Archives: cubicle land
Myth of reuse
In the quarterly design analysis meeting, the manager of a cancelled project blamed the failure partly on the delay in verification. In that project, the verification schedule is underestimate at least by a factor of four. I think his comment is not totally fair to the verification team. One of the problem rooted from the myth of reuse. When there is some existing infra-structure available, the management just assume there is zero effort in reusing them in a new project. The reality is most of those so-call reused component are not structured nicely for reuse, and worse many of them lacks proper documentation due to schedule pressure. A true reuse component should have the quality of verification IP purchased from 3rd party vendors. Anything thing short of that quality should use a different term instead of reuse to stop confusion in project planning. Most of the time that falls into the port category, which require complete understand of the original code in order to use it with the new device. Sometimes that should be categorize as salvage, which is marginally better than writing fresh code. Mislabel salvage as reuse will only spell disaster later in the project
Layoff
There is another round of layoff today in the company. We had already lost track of the number of layoff, it maybe the fifth or the sixth round. It is not affecting the Headquarter in Burnaby at all. Over 60 engineers are let go in the Santa Clara design center, which is over 2/3 of the design team down there. It’s mainly due to the company decided to exit the custom MIPS core business, therefore we no longer need their expertise. We can foresee the closure of that division when we use the licensed core from MIPS instead the custom core in our own product. Upon hearing the news around lunch time, our usual group have a concern on the fate about Suzanne. Lucky that later in the day, she sent an email to all of us saying she is safe. From my experience and what I heard from other companies, it’s always more secure to work in the corporate headquarter or at least the regional main office. Tiny remote sites often become the first victim of downsizing.
Talk to the CEO
Today during lunch Bob Bailey, the CEO of PMC, sat with our table. I knew this is going to happen when we picked his usual table in the first place. This is the first time I had ever talked to him since I joined the company. He is pretty much up to my expectation, a typical CEO. Can’t really say I like him or dislike him, he just exists, so my concern is how to make him most beneficial to myself. Most of the conversation are weather talks, nothing much other than bragging long hours or questions about the prespective of our products. His response is pretty typical too. At last I put up my courage to bring up the idea of giving two monitors to every engineer to increase productivity. He seems quite interested in the idea and asked me to give him follow up information. So I wrote up a memo in the afternoon and send it off to him together with the reseach paper I had. I don’t know what this will lead to, probably he will just throw it away without looking. If he indeed like the idea, then finally I will have dual monitors after trying so many times. I have nothing to lose anyways.
legal language
Tonight our project team had a team building dinner with the leaders from Saskatoon. The food and beer are great, and most important the expense is covered by PMC. In the dinner, we had talked about how expensive it is to file a patent. It costs tens of thousands of dollar to hire a lawyer to word the patent using those legal terms. Then I came up with an idea, why don’t we have software automating this process? Legal language is a subset of normal language, and each word has a specific meaning, which is similiar to logic statements understand by computers. The goal is to create a software that will translate contracts and patents written in daily english into a legal document. It would be a great idea for inter-discipline research among the liguistic, CS, and law departments.