The collected works of oreissig - Page 2

not literally

There are imho worse aspects to complain about Oracle Solaris than the price for their DVD kit.

The first thing that comes to my mind is network boot, which is really easy to set up in a Solaris-VM on x86.

Solaris might ask for root-pw in order to invoke a shell in single user mode. I'm not entirely sure about this, but mounting the fs from another instance definately works.

Here's a guide on how to set up Solaris as a install server (this also works for other versions than 8)

What's your exact problem? You don't like current OSes, okay I don't think there's much to debate here, that's your opinion.

But what about the CPU architecture? How often to you program assembler, so that you contact with the actual architecture? It was a lot of fun in the 90s, but actually it doesn't matter if it gets the job done, and one cannot make a case for Intel building bad CPUs, that's not the case. In the not-low-power segment, Intel does a pretty decent job and because it is so freaking hard to build a CPU, I don't expect this to change too soon. I don't think the old commercial RISCs (POWER, SPARC, let's also include Itanium) will ever scale down again, because there's nothing to get from there. The PC market is saturated, sales are actually going down. And the workstation market disappeared as PCs based on standard components outperformed specialized hardware, or rather that they were fast enough that no engineer or designer found it worthwile to spend 20k instead of 2k. I own a zoo of classic RISCs myself, but as much as I enjoy them, I also understand the reasons why they disappeared, won't appear again and why there's nothing to do about it.

Now interesting developments happen in the low-power segment. I expect a rage against tablets in 3...2...1....
But simply based on selling figures, mobile devices are the current wave of innovation. Frankly, I don't think that the currently available devices will be like what people will be using 20 years from now, they all need to find the right formfactor running the right software, and of course they all need hardware that's both energy efficient and powerful. Those attributes are primarily driven by how the chip is implemented and not by the instruction set. Actually the instruction set has become a minor detail in todays CPUs, which is why Intel was able to stay competitive with its CISCs against the RISCs, because they simply added a thin converter to drive the internal RISC while staying compatible. In times when CPUs are in their billions of transistors, such a conversion is not that complicated compared to the whole chip. Which is why basically all new designs are based on ARM, which gives the manifacturer an instant basis of available software for their SoC.


So what's my point? Yes there was heavy consolidation on the architecture front, but it was completely understandable why this happened and why it will probaly stay that way.
In case you are serious about experimenting with strange architectures, there are powerful FPGA boards available starting at something like 100$, where you can run any custom design (up to a certain complexity) or even create your own. There are even openly developed CPUs like the OpenRISC. This is an opportunity that didn't exist in the 90s, where building a chip meant investion millions to build an ASIC.

an R4k Indigo² with Pre-IMPACT graphics in a purple case?

I have an IMPACT-ready R4k in a teal case, so my guess is a skin swap.