I have a favourite quote for situations like this, where a person might seem to be over ambitious and is not quite convincingly confident:
Quote:
I'm reminded of Thomson's Rule for First-Time Telescope Makers: "It is faster to make a four-inch mirror then a six-inch mirror than to make a six-inch mirror" - not that I even own a telescope.
Which is to say, if you aim to make a super complex CPU, you'll have little difficulty starting with a simple 8 bit CPU, and it would be a good idea anyway. Every artist has a portfolio: start building yours.
On the same lines: if you aim to build a team around you, have some evidence that you can work with other people as part of a team. Be a collaborator on public projects. Write bug reports, and show how clear you can be, how helpful, how polite. Engage in online discussions, and show that you can resolve misunderstandings, can take on board new information, can explain your point.
Update on the project, explaining how we'd get round the patent issues:
It'll likely be done in a mixture of Cadence OrCAD, Icarus Verilog and/or Synopsys software, with additional libraries either freely downloaded or bought from Mentor Graphics, GlobalFoundries, TSMC, IBM and/or these companies. Any thoughts from those here?
This CPU, if the design works right, the streams of x86 CISC instructions can be ganged into massive 512-bit wide macro-operations, and it isn't horribly bugged like *cough* current Intel CPUs with their kernel memory sapping/SYSCALL privilege escalation flaw, should perform between 4-40x (!!!) better in singlethreading applications, albeit with a large die size (>600mm^2 for 8 cores) and perhaps low clock speed, on a modern 14nm process, and the first ones off the factory line will sell for over $500,000 a piece. To get around copyright and patent issues, we will strike contracts with AMD and VIA to license the x86 technologies.
This means Passmark singlethreaded scores of over 80,000, Cinebench scores of 300,000 or more and no more excuses for people claiming their CPU is bottlenecking their graphics card.
The CPU ID will most likely look like this:
CertifiedCBR or CertifiedCTN
801486-class
8 core(s), 16 thread(s)
Model 0 Family 0 Stepping C0
Extensions supported:
MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, VIA 64, VT-x, AES, AVX, AVX2, AVX-512, CVT16, FMA3, FMA4
Any thoughts on the overall idea? And yes, I will hopefully soon be looking at simulating a full 8086 system in Icarus Verilog or perhaps tweaking an existing design, from, say, OpenCores. At the moment I'm working as a tester on the 86Box PC emulator project, and have looked over large parts of its code including the CPU dynamic recompiler.
You can't ever "get around patent issues" with any of these companies.
They could sue you because they didn't like the color of your shoelaces, and bleed you for years over it.
Oh sure, you might "win" legally in the bitter end, but will be doing it from the soup kitchen line.
There is no room in the rich man's world for those of us trying to do the world a solid by delivering a better product.
