4/22    Windows device driver engineers needed.  -jeffwong
        /csua/pub/jobs/Transmeta
        \_ With this posting, transmeta is now officially in the
           "We've given up competing, we are now trying to sell out to
            microsoft/intel/amd" category.
            \_ How does having Windows drivers constitute selling out?
               If they expected all of their users to use Linux, they'd
               be out of business already.  Look how successful the Linux
               distro companies have been.
               \_ because one of the claims was that they could
                  just transparently replace (from a software point of view)
               \_ D00D R3D H4T RUL3Z! 7H3Y W111 D357R0Y M$ 4ND 7H13R
                  U53R5 4ND RUL3 7H3 W0R1D! 1N5T411 17 B3F0R3 17'5 T00
                  L473!
               \_ because one of the claims was that they could just
                  transparently replace (from a software point of view)
                  an intel chip for massive gains.
                  Seems it's not such a perfect replacement if they need
                  custom drivers, eh? Start looking for a new job, jeff.
                  I give you 1, maybe 2 good years there.
                  \_ Average length of employment of an engineer in the valley
                     these days is between 18 and 24 months.
                     \_ That average is skewed because of the high
                        concentration of dot-bombs in the valley. My point
                        being that transmeta looks to be another one.
                     \_ Really? I was at Cisco for 4 years and I was
                        still a "junior" in terms of years of service.
            \_ Perhaps because their "low-power" designs were not so
               spectacular after all. TM chips ran at lower power at
               the expense of performance and we all know that what
               matters isn't power but energy (E = P*t) consumed
               by a laptop. If a chip consumes energy at half the rate
               of another but runs at half the speed, which one would
               you buy?
               \_ D00D CRU053 15 7H3 B357 CH1P 3V3R M4D3! 11NU5 541D 50
                  0N 11NSUX.K3RN3L! CRU053 B3475 7H3 P4N75 0F 1NT31 4ND
                  4MD, M070R014 4ND D3C! 37337 R3D H4T GN00/11NSUX 15
                  W1CK3D F457 0N CRU053!
                \_ What percentage of your apps are CPU-speed limited?  -tom
                   \_ A process doesn't have to be CPU bound to benefit
                      from a faster processor tom.
                        \_ It has to be CPU bound to use the same amount of
                           power in the equation above.  I think for the
                           vast majority of real-world applications, a slower
                           chip is better in a laptop.  -tom
                           \_ Why? So you have IE running as an I/O bound
                              process. Now a user clicks on a button and
                              something happens. IE executes a short task.
                              On a TM it takes 2 seconds while on another
                              processor it takes 1 but for the most part
                              they consume the same amount of energy
                              because TM's power consumption is half of
                              the alternative. The TM loses in this case.
                              EDP (energy delay product) is a better metric
                              for power efficient devices like this because
                              it factors in performance and not just power
                              and energy.
                                \_ CPU is almost always starved for data
                                   these days.  The speed of Netscape/IE is
                                   virtually unaffected by CPU speed
                                   difference--they're always waiting for
                                   disk or net.  The difference between
                                   200 Mhz and 400 Mhz is enormously larger
                                   than the difference between 400 Mhz and
                                   800 Mhz.  Why do you think people stopped
                                   buying PC's?  -tom
                                   \_ At least on PIII there is a noticeable
                                      difference between 400, 550 and 600.
                                      Beyond 600 I tend to agree with you.
                                      An athlon 1.2 GHz is not that much
                                      faster than a 600 because you can't
                                      give it enough work to do in most
                                      cases or it blocks on device io.
                                      I believe that the future belongs
                                      not to single fast procs but to
                                      multiple cheap slow procs. MPP
                                      is the way to go.
               \_ It dynamically lowers its voltage and frequency according
                  to your workload.  Why would you want to be running at
                  the maximum clock rate when you're typing or reading a web
                  page?  That is a waste of energy. Besides, a PIII laptop
                  can't go very fast when it's batteries are dead.
                  \_ I'm not defending the Pentiums either. Personally, I
                     think Intel sells crap. But most processors today
                     (at least those concerned with power savings) employ
                     power savings features during idle periods like clock
                     gating or disabling input latches to unused units.
                     Power consumption during idle periods by the CPU is
                     rather inconsequential compared to power drawn by a
                     spinning disk, LCD monitor, DRAM refresh, and other
                     crap running. I doubt TM chips fair much better during
                     idle periods than other alternatives.
                     \_ I'm not defending the Pentiums either. Personally, I
                        think Intel sells crap. But most processors today
                        (at least those concerned with power savings) employ
                        power savings features during idle periods like clock
                        gating or disabling input latches to unused units.
                        Power coproperly phrased question.   here, I'll put
                        you out of your misery:
                        http://arstechnica.com/reviews/peripherals.html
                     \_ D00D U R 50 WR0NG! CRU503 R0X! R3P3N37 UR B45PH3M35
                        W4Y5!