max (spamformy@ss.net) on 2/8/09 wrote:
---------------------------
>>That means there's no room for growth for IPF, its all a slow
>>decline now, unless Sparc or POWER get canned first.
>
>I really do wonder why ppl think power is going to get canned in the server space?!
>last time I checked ibm was selling a shitload of them and is developing the architecture
>on several fronts: power7, cell and blue gene.

BlueGene was, IMHO, very promising, but it looks like it lost internal competition battle to inferior Cell tech.
Would be glad to be wrong about it.

>
>while the latter two might not matter outside of hpc, power mainframes and z are
>closely related chips, both very profitable and developed mostly on the same budget.

Don't get it. Power6 and z10 share some of low level blocks, but pretty much nothing at higher microarchitecture level. Also, I don't think they share much at compilers and OS level.

>the hpc and darpa wins for power further alleviate development cost.
>
>I haven't seen anything that would convince me that cell is going to stay. but
>then sony might still stick with the architecture despite recent rumors.

Except for excessive power budget Cell looks like a great floating point DSP also well suitable for medical imaging and silicon inspection. Not so great gaming or HPC engine so. Could IBM micro could moderate a clock rate and start selling the beast where it belongs? Or sell the whole thing to Freescale that seem to have better understanding of dedicated systems market.