Changes

KS: Why did you never do them?

KS: Why did you never do them?

BM: Because, you know, think about it. If you add all these other things in it, what are you getting? A speed-up, right? I already was doing what I wanted to do, and my microprocessor was outperforming the 68000 and the 8086, but when you trip over that boundary, to like the 8086… Look at what’s in the 8086 compared to the 816, look at what’s in the 68000, compared to the 816, and you’ll see that what you’re getting is a whole lot more complication, and that’s why, when my nephew worked for me to get his electrical engineering degree from ASU, both my nephews Rod Bearish, who I was just talking to… he’s a software engineer, and Andrew Hall, worked with, for IBM… Intel, after he left my company, he worked with them in Chandler, and then he was moved to Portland Oregon for designing the Pentium Pro microprocessor… and that Pentium Pro ran at 200 MHz at the time. Andrew worked with about 500 other engineers. So when you trip over…. going from my 816, to something competitive, you need about 1,000 engineers. And so that’s why I never did it. Now did I specify it? Can you find a datasheet for the 65c832? You can. Do we have a product name to the 65c832pxb, for programmable accelerator board, we do, we have one of those. We sell those. So the thing is, is that when you go from the 816, you have to add about 500 engineers. And the reason why, is because you have to compete with Intel, AMD, and at the time IBM. MIPS, you have to go to the Sparc, which came out of Stanford I think, no I don’t know which one came out of Stanford. I think MIPS came out of Stanford, and Sparc came out of Berkeley or something like that, so had the RISC-processors, you had Cambridge behind the ARM processors. And when you do that you have just elevated yourself to a whole new ecosystem, and I was happy with what I did with the ‘02 and 816. So every time I look at a 32-bit processor, I end up with “why not recommend the ARM processor?” Those guys came to me in 1983, to design the ARM for them, and I said “no, you’re asking me to do something that’s not compatible with the ‘02 and the 816.” They wanted to me to stop basically the 816, because the British broadcasting Company, British Empires Acorn, is where ARM came from. It was called Acorn RISC-Machine. And it was Advanced RISC Machine, now it’s just ARM, because it was never really a RISC-machine, but it followed the RISC-concept. And so therefore ARM doesn’t want to be known as a RISC-processor, and it’s because they really aren’t. However they have a lot of the characteristics. So now let’s go back to something. If you go back to the addressable-register architecture, you understand, we’ve had floating-point libraries for 25 years, 30 years. A floating-point operation is possible because of the way the 6502 instruction set works. So you can do a double-precision floating point number in software. That’s what we had. That’s the floating-point library. So if you want to add in a floating-point processor, like they did in the 486, when they went from 386 to 486, that’s when I bought for $125,000, I switched over from my Talma GDS2 system, to a network of Micron 486 processors, because I knew once they put the floating-point processor in there, then the software would use that, so all of my design tools would run on that processor. I switched from a Data General Eclipse to a network of, I think it was 15 or so Micron computers, we had to wait for the 486 to come off the production facility, before we could take delivery on, I think they were about $7000 per node, per PC. And so when you look at that the addressable-register architecture, if you can do all of these complex operations, like for instance encryption, you can do that with my processor. So the only thing the only thing you’re doing when you’re getting a 2 GHz quad-core Intel i7, all you did was speed up your software. But I can do this problem set with my 816. So then you go oh yeah, but speed is a big deal. I go “yeah, speed is a big deal, but look what it takes to get the speed there for a quad-core i7,” or however many cores they have now, I don’t even know how many, but the point is, it’s going to take manufacturing organization like Intel has, and you’re going to have to drive that down to about 20 nm technology to make sense out of that, and then as an end result, I’d recommend you buy it! I don’t have to design it. I really designed the ones I wanted, back in the 80s, about 30 years ago.

BM: Because, you know, think about it. If you add all these other things in it, what are you getting? A speed-up, right? I already was doing what I wanted to do, and my microprocessor was outperforming the 68000 and the 8086, but when you trip over that boundary, to like the 8086… Look at what’s in the 8086 compared to the 816, look at what’s in the 68000, compared to the 816, and you’ll see that what you’re getting is a whole lot more complication, and that’s why, when my nephew worked for me to get his electrical engineering degree from ASU, both my nephews Rod Bearish, who I was just talking to… he’s a software engineer, and Andrew Hall, worked with, for IBM… Intel, after he left my company, he worked with them in Chandler, and then he was moved to Portland Oregon for designing the Pentium Pro microprocessor… and that Pentium Pro ran at 200 MHz at the time. Andrew worked with about 500 other engineers. So when you trip over…. going from my 816, to something competitive, you need about 1,000 engineers. And so that’s why I never did it. Now did I specify it? Can you find a datasheet for the 65c832? You can. Do we have a product name to the 65c832pxb, for programmable accelerator board, we do, we have one of those. We sell those. So the thing is, is that when you go from the 816, you have to add about 500 engineers. And the reason why, is because you have to compete with Intel, AMD, and at the time IBM. MIPS, you have to go to the Sparc, which came out of Stanford I think, no I don’t know which one came out of Stanford. I think MIPS came out of Stanford, and Sparc came out of Berkeley or something like that, so had the RISC-processors, you had Cambridge behind the ARM processors. And when you do that you have just elevated yourself to a whole new ecosystem, and I was happy with what I did with the ‘02 and 816. So every time I look at a 32-bit processor, I end up with “why not recommend the ARM processor?” Those guys came to me in 1983, to design the ARM for them, and I said “no, you’re asking me to do something that’s not compatible with the ‘02 and the 816.” They wanted to me to stop basically the 816, because the British broadcasting Company, British Empires Acorn, is where ARM came from. It was called Acorn RISC-Machine. And it was Advanced RISC Machine, now it’s just ARM, because it was never really a RISC-machine, but it followed the RISC-concept. And so therefore ARM doesn’t want to be known as a RISC-processor, and it’s because they really aren’t. However they have a lot of the characteristics. So now let’s go back to something. If you go back to the addressable-register architecture, you understand, we’ve had floating-point libraries for 25 years, 30 years. A floating-point operation is possible because of the way the 6502 instruction set works. So you can do a double-precision floating point number in software. That’s what we had. That’s the floating-point library. So if you want to add in a floating-point processor, like they did in the 486, when they went from 386 to 486, that’s when I bought for $125,000, I switched over from my Calma GDS2 system, to a network of Micron 486 processors, because I knew once they put the floating-point processor in there, then the software would use that, so all of my design tools would run on that processor. I switched from a Data General Eclipse to a network of, I think it was 15 or so Micron computers, we had to wait for the 486 to come off the production facility, before we could take delivery on, I think they were about $7000 per node, per PC. And so when you look at that the addressable-register architecture, if you can do all of these complex operations, like for instance encryption, you can do that with my processor. So the only thing the only thing you’re doing when you’re getting a 2 GHz quad-core Intel i7, all you did was speed up your software. But I can do this problem set with my 816. So then you go oh yeah, but speed is a big deal. I go “yeah, speed is a big deal, but look what it takes to get the speed there for a quad-core i7,” or however many cores they have now, I don’t even know how many, but the point is, it’s going to take manufacturing organization like Intel has, and you’re going to have to drive that down to about 20 nm technology to make sense out of that, and then as an end result, I’d recommend you buy it! I don’t have to design it. I really designed the ones I wanted, back in the 80s, about 30 years ago.

KS: So when you are designing that with your little team of four, six people, you said that you did simulation in software beforehand, but then when it came to actually designing the chip, are you actually sitting down at the drafting board with rubolyth and drawing things out... on plastic and paper?  

KS: So when you are designing that with your little team of four, six people, you said that you did simulation in software beforehand, but then when it came to actually designing the chip, are you actually sitting down at the drafting board with rubolyth and drawing things out... on plastic and paper?