From: Globemaker on

Open Source Crypto Processor Project Could be Developed by Volunteers

Rev. 0 Proposal for a Computerbunker

A gate array vendor will be chosen to provide one logic chip. It will
be used to implement the microprocessor with variable instruction sets
which can be defined according to a key. The 256 bit instruction width
provides uniqueness for each computerbunker owner. A variable compiler
will use the key to make executable binary files of instructions that
are not useable by any computers not using the owner‘s key. Also, the
owner‘s computebunker cannot execute instructions that are compiled
for standard CPUs. This provides a logical isolation of the
computerbunker from the corruptible internet. Engineers and scientists
will be invited to contribute simulations of the logic design, after
the FPGA field programmable gate array IC is selected. Your
recommendations concerning the chip choices are needed now.

One practical value of the resulting microprocessor is due to its
independence from government control. A blank FGPA silicon chip can be
burned with some ones and zeroes to achieve this.
From: Mok-Kong Shen on
Globemaker wrote:

> A gate array vendor will be chosen to provide one logic chip. It will
> be used to implement the microprocessor with variable instruction sets
> which can be defined according to a key.[snip]

What use can be made out of such a computer with a "novel" instruction
set? Could it render computing more efficient? If not, it is the
result of computing that counts, not how the computing is done. So,
in particular for crypto, I can't see what you intend to achieve.

M. K. Shen

From: Globemaker on
> What use can be made out of such a computer with a "novel" instruction
> set?

One application is for a business to buy a thousand computers for its
thousand employees. Then, all of the employees can run secret software
that nobody else can use and no virus can infect. This provides
"ownership" of the computing power, instead of using a shared
operating system licensed with a 50 page agreement from a monopoly
that is subject to government parasites. This new computer
architecture allows private programs to be executed while never
exposing the intellectual property to thieves and government scum.

>Could it render computing more efficient?

No, it is slower than a 2010 computer and faster than a 1990 computer.

>If not, it is the result of computing that counts, not how the computing is done.

Yes. The result may be a proprietary computer vision algorithm, for
example, that only your team shares.

>So, in particular for crypto, I can't see what you intend to achieve.

This will be revealed in more detail during the next four days.
Imagine Sony using a variable instruction set for playing Sony music
for paying customers. Non-paying hackers would be unable to execute
the programs without paying for the unique key for the unique
instructions that work on a single CPU. All other Sony CPUs will get
different instructions to execute the same algorithms.
From: Earl_Colby_Pottinger on
Or you could just use one of the ROM-able OSs out there with an
embedded Emulator.

The idea that you need a gate-array is just a way to make this system
more complicated, replacing the BIOS of a standard computer with said
ROM would be far cheaper. And note, I said ROM not EEPROM. And even
a standard programmable BIOS chip could securely used if the write
signal requires a removable jumper to be installed during programming,
then removed before the machine is given to it's user.
From: Mok-Kong Shen on
Globemaker wrote:
[snip]
> Yes. The result may be a proprietary computer vision algorithm, for
> example, that only your team shares.

That's in principle not very different from a proprietary encryption
algorithm that is available in binary only, isn't it?

M. K. Shen