Originally posted by: Yoxxy
.
In fact it is the engineers that have doomed the company by going to a single die quadcore.
Seems to me if the Phenom was viable as a design (and the roadmap called for dual core Phenoms at one point) you could definitely go the Intel route and sandwich two dies in a package if 4 cores per die is the limiting factor. That doesn't seem like it'd call for a lengthy development process at all. Doubly so since I recall reading the K10 is just an optimized, tuned K8.
So if I was in charge and got reports that the flag ship product intended to draw high margins wasn't going to compete with Intel's mainstream part I'd be down there asking the engineers what they CAN do. I'd eat the crow, admit that hyping 4 cores on a single die was premature and isn't doable with current tech and ship a competitive part.
And if my production people can't deliver *anything* then it's my fault for not having good enough checks & balances and letting the company fall into such a state of disrepair. It's my fault if my marketing team isn't savvy enough to point out flaws in the product roadmap and unable to find good niches for our upcoming products. And it's my fault for not attracting and retaining visionary chief technology people educated and savvy enough to detect that the guys in the trenches are leading the company down the garden path.
Still blaming the leadership here, sorry. And worst case leadership should recognize they don't have the resources to compete and either scale down and scurry to a niche or pack up shop. Not collect fat pay checks while 'steady as she goes' takes the company into the tar pits. Ego has no place at the top of a publicly traded company.
45nm can not come fast enough.
Agreed.
As far as 'death by a thousands cuts' vs one 'final cut' -- the single deep cut is optimal, but it never happens that way. See my original post re: why nobody ever does it correctly. This 10% is just a start, expect another cut of roughly the same size over the next 2 quarters, and another 5-10% natural attrition rate over the next 6 months.
On-going cuts are absolutely terrible for morale. People get into a bunker mentality waiting for the next shoe to drop. They do everything in their power to make sure the guy in the next cube or office looks more ready for the chopping block rather than create value for the shareholders. Also unemployment insurance costs go through the roof, and the company gets a reputation on the street as a binger and purger. Top people at all levels stop wanting to come work for you unless you offer them an outrageous compensation package.
Gradual cuts over a long period are also associated with hiring freezes -- the labor department takes a pretty dim view of layoffs while hiring. Which means as people are lost due to natural attrition (see above re: low morale) they need to be replaced by contractors -- contractors don't show up as employees on the books, can't quit and can be found for a job slot very quickly. This is less efficient from a productivity standpoint because contractors have a ramp-up learning process and once they leave they take the knowlege with them. Plus it costs more money because you now have to cover a body shop's overhead and profit margin as well.
Ongoing small cuts are great from a theoretical standpoint if humans weren't involved. Unfortunately they are.