[Info-vax] ITIC 2014 - 2015 Global Server Hardware, Server OS Reliability

[JF Mezei]

On 15-06-01 07:59, Bob Koehler wrote:

>    Take the cost of doing that, and spread it out over the number of
>    chips to be made, and no I don't think it would be competitive.

I realise that Alpha is commercially dead. My question pertained more to
whether the techniques used for Alpha's performance at the logic level
are still considered state of the art today.

If you look at CSI/Quickpath, it was started in 2004, and obviously
inspired from tech developped for EV7 and for Power. Has Quickpath
evolved far beyond Alpha in terms of logic, or is it still pretty much
still on par with it once you remove clock rates from the equation ?

Would the Alpha EV7 branch predictions pipelining and other logic based
performance optimizations still be considered state of the art (because
it hasn't changed much since then) or have other architectures leaped so
far ahead at that level that Alpha is geriatric by now ?

I didn't imply there was a business case to restart Alpha, just
wondering how much the chip tech has evolved from the last Alpha, once
you remove process shrinks and multiple cores from the equation.