For either educational purposes or SoC experiments it fires me up to think of a transistor level FPGA version of a '386, eventually.
This kind of thing is the eventual long term application of this guys project, aside from abstract art.
(I'm in "submitting too fast" punishment mode, so this edit is a reply to theatrus2:
Yes you are correct and inspired by the 6502 project, Peter Monta wrote a GPL utility that converts from transistor netlists to Verilog (or was it VHDL?). You can probably find it on SF or github. As far as I know its in a working state, not just vaporware or something.
So its also a cool project in that it inspires interesting software development.)
Intel used to release good-sized posters of those chips (I might have one) with enough detail to sorta make out transistors. I lamented later chips as they were so complex the details all blurred into a metallic grey, even at poster sizes.
Yeah the entire reward structure seem kind of nonsensical to me, basically a single person has to donate the entire cost behind an imaging, even if the result is a CC-licensed image; and then you have the upper-tier where someone can own the copyright over an image; ignoring the fact that if someone pays for that for some chip on the list and someone else pays for the CC-image of the same chip the image you own copyright to will be essentially the same as the one that is placed into the CC, even if he goes through the entire imaging process multiple times (which seems like a waste of time, but would technically skirt the licensing clash produced).
Would make a lot more sense, IMO, if it went with a more pooled approach where smaller donations got, as you mentioned, a poster, perhaps with tiers based on presentation (just a poster in a tube, framed poster for more money, etc), dropping things all the way down to say $1-5 to get just a high-res image suitable to use as a desktop background (I realize that anything released under the CC from this could be used this way anyway (and you can donate less than $75); but the point is making an option for people who would like to throw a few bucks his way because the project is cool, but aren't interested in dropping $75-400 on it, but would also like to feel like they are getting "something" out of it, even if just a digital image.
The longest a patent in the US can last is 20 years. According to Wikipedia the 386 was released in 1985, and '89 for the 486. Therefore any patents should have expired by now unless they applied for the patents years after release of the related product.
This is all about metal side imaging. The exciting part happens behind the metal, and requires you to image through a thinned wafer to see the actual transistors. IARPA has recently funded a good amount of research into this, so I am sure someone thinks this is an interesting security issue.
This title is way too vague. I know the domain clues you in a bit, but at first glance I was hoping for some nostalgic writing about the 386 and 486 architectures.
I think these chip layouts are already copyrighted. Sure photos have their own copyrights but you are probably infringing by producing and selling the "copyright" to these. IANAL.
Aside from that, if all you are providing is digital images of them, and the incremental cost of emailing out an image link to a funder is $0, why be stingy with what funders get?
To me this looks like nothing more than a "please fund my hobby" campaign, which I will not support.
John McMaster does some amazing stuff. Here's an example of how he took a section of a chip (Williams Special Chip 1, a blitter device used in a lot of video games) and reversed it back down to a pair of gates:
Side note: these two pages use very similar diagram colors. I wonder if there's a common source (textbook?) that both authors read that used those colors.
"For most people it started with the 80386 because it allowed running Windows and helped lead to the rise of the Internet."
whaaat.
A) In a vague, popular sense, the x86 series "started" with the 8088 and the IBM PC, obviously.
B) The 386 was contemporary with Windows 2.0 which was hardly popular. It was really used for Windows 3.0 which was a bit better. But the first majorly popular Windows release was 3.1, which was contemporary with the 486 (though it could run on 386s too).
C) The number of Windows 3.x users with direct Internet connections was awfully low. At that point everyone doing consumer-facing networking was on AOL and similar services. Associating "Windows" and the "rise of the Internet" gets you to at least Windows 95, which was a Pentium-era product.
I guess that intro statement doesn't really matter to the project, which does seem kind of cool, but it's also the first thing you read going to the page.
Inspire is not quite the word I would use, I think. The reason so much was done on the 386 is because Intel finally dropped that god-awful segmented addressing in favor of a flat memory model. In addition, protected mode became actually useful and was necessary for a Unix-like OS to be created (where you have multitasking and protected sections of RAM). Doom was also a killer-app for the 386.
There are Unix-like OS's on MMU-less architectures. Though of course there are limitations (like not supporting fork(), only vfork()). But you're right that it made it a lot more attractive to try to do a proper Unix.
There were even commercial Unix workstations based on MMU-less CPU's. An example is the Sun 1 workstation, that used an 68000 (Since the first version - the 68000 - is not fully restartable, various hacks were used on 68000 designs requiring an MMU; I'm not sure what the Sun-1 did, but one example that was used was running two of them in lockstep, and have the MMU halt the second one when the first one triggered a page fault, so that it'd be possible to inspect the CPU state before a bus-error would mess it up. From the 68010 onwards this was unnecessary)
> Associating "Windows" and the "rise of the Internet" gets you to at least Windows 95, which was a Pentium-era product.
Not quite: back in '94 the iconic Web client was a Windows 3.x PC running Trumpet Winsock http://thanksfortrumpetwinsock.com/ and Mosaic. Those machines could well have been Pentiums (or 486es) though.
I ran Winsock initially on a 386 with 3.1. At this time, the dial-up business was booming and I'm not talking AOL/Compuserve. There were entire websites dedicated to listing and sorting through the hundreds of local ISPs cropping up, who had the latest 28.8 modems, best advertised modem ratios, etc... These companies were all setup well before Windows 95 came around.
Well, the 386 had a pretty long tail. I'm pretty sure I remember 286 PCs still being sold in 1990 or so and 386 PCs being sold after win3.1 came out as well. And with how expensive desktop computers were in the late 80s, people kept them longer too I think.
But this is all anecdata, and I was pretty young. But the first machine I ran windows 3.1 on was definitely a 386 (and I ran 3.0 on a 286).
Some placeholder text got left in:
"Remember, keep it concise, yet personal. Ask yourself: if someone stopped reading here would they be ready to make a contribution?"
Is there a delay on Indiegogo or has a fundraiser that's interesting enough to hit HN front page for an hour only managed to attract a single $5 pledge in that time? (Which strikes me as odd given it's getting voted up.)
But I do wonder about the funding goal. It sounds like the equipment is already purchased. Asking other people to "recoup" his costs for equipment that he already owns and uses in his hobbies seems a bit unfair. Likewise, with vague funding goals like enabling research and funding equipment upgrades... and the fact that he is matching the price a commercial outfit would charge to do this... I just get the sense that we aren't being asked to fund a project as much as the public is being asked to become a paying client for a professional service.
Not that the service sounds bad - the results would be cool. I'd just like to know what the actual project costs will be, and have the goal match that. Anything above and beyond that goal, of course, he should feel free to use to recoup costs and upgrade.
41 comments
[ 371 ms ] story [ 5446 ms ] threadhttp://www.visual6502.org/wiki/index.php?title=6502_-_simula...
For either educational purposes or SoC experiments it fires me up to think of a transistor level FPGA version of a '386, eventually.
This kind of thing is the eventual long term application of this guys project, aside from abstract art.
(I'm in "submitting too fast" punishment mode, so this edit is a reply to theatrus2:
Yes you are correct and inspired by the 6502 project, Peter Monta wrote a GPL utility that converts from transistor netlists to Verilog (or was it VHDL?). You can probably find it on SF or github. As far as I know its in a working state, not just vaporware or something.
So its also a cool project in that it inspires interesting software development.)
Would make a lot more sense, IMO, if it went with a more pooled approach where smaller donations got, as you mentioned, a poster, perhaps with tiers based on presentation (just a poster in a tube, framed poster for more money, etc), dropping things all the way down to say $1-5 to get just a high-res image suitable to use as a desktop background (I realize that anything released under the CC from this could be used this way anyway (and you can donate less than $75); but the point is making an option for people who would like to throw a few bucks his way because the project is cool, but aren't interested in dropping $75-400 on it, but would also like to feel like they are getting "something" out of it, even if just a digital image.
btw does anyone knows if this is considered reverse engineering by law?
There is a sui generis right in IC mask designs, but it's limited to 10 years: http://en.wikipedia.org/wiki/Integrated_circuit_layout_desig...
Aside from that, if all you are providing is digital images of them, and the incremental cost of emailing out an image link to a funder is $0, why be stingy with what funders get?
To me this looks like nothing more than a "please fund my hobby" campaign, which I will not support.
The copyright of a photograph of the chip belongs with the person who took the photograph, as normal.
http://uvicrec.blogspot.com/2013/03/williams-special-chip-1-...
http://www.righto.com/2013/01/a-small-part-of-6502-chip-expl...
Side note: these two pages use very similar diagram colors. I wonder if there's a common source (textbook?) that both authors read that used those colors.
http://www.eet.bme.hu/~benedek/Vlsi_Design/Lectures/CMOS_Lay... has a color table that seems to match up.
whaaat.
A) In a vague, popular sense, the x86 series "started" with the 8088 and the IBM PC, obviously.
B) The 386 was contemporary with Windows 2.0 which was hardly popular. It was really used for Windows 3.0 which was a bit better. But the first majorly popular Windows release was 3.1, which was contemporary with the 486 (though it could run on 386s too).
C) The number of Windows 3.x users with direct Internet connections was awfully low. At that point everyone doing consumer-facing networking was on AOL and similar services. Associating "Windows" and the "rise of the Internet" gets you to at least Windows 95, which was a Pentium-era product.
I guess that intro statement doesn't really matter to the project, which does seem kind of cool, but it's also the first thing you read going to the page.
http://en.wikipedia.org/wiki/386BSD
There are Unix-like OS's on MMU-less architectures. Though of course there are limitations (like not supporting fork(), only vfork()). But you're right that it made it a lot more attractive to try to do a proper Unix.
There were even commercial Unix workstations based on MMU-less CPU's. An example is the Sun 1 workstation, that used an 68000 (Since the first version - the 68000 - is not fully restartable, various hacks were used on 68000 designs requiring an MMU; I'm not sure what the Sun-1 did, but one example that was used was running two of them in lockstep, and have the MMU halt the second one when the first one triggered a page fault, so that it'd be possible to inspect the CPU state before a bus-error would mess it up. From the 68010 onwards this was unnecessary)
Not quite: back in '94 the iconic Web client was a Windows 3.x PC running Trumpet Winsock http://thanksfortrumpetwinsock.com/ and Mosaic. Those machines could well have been Pentiums (or 486es) though.
But this is all anecdata, and I was pretty young. But the first machine I ran windows 3.1 on was definitely a 386 (and I ran 3.0 on a 286).
Edit: ah.
But I do wonder about the funding goal. It sounds like the equipment is already purchased. Asking other people to "recoup" his costs for equipment that he already owns and uses in his hobbies seems a bit unfair. Likewise, with vague funding goals like enabling research and funding equipment upgrades... and the fact that he is matching the price a commercial outfit would charge to do this... I just get the sense that we aren't being asked to fund a project as much as the public is being asked to become a paying client for a professional service.
Not that the service sounds bad - the results would be cool. I'd just like to know what the actual project costs will be, and have the goal match that. Anything above and beyond that goal, of course, he should feel free to use to recoup costs and upgrade.
"Remember, keep it concise, yet personal. Ask yourself: if someone stopped reading here would they be ready to make a contribution?"
Similarly, you could check out the Silicon Zoo (http://micro.magnet.fsu.edu/creatures/). Very cool stuff!
http://www.chip-architect.com/news/2012_04_19_Ivy_Bridges_GP...
http://chip-architect.com/news/2003_09_21_Detailed_Architect...
Of course the ones he's planning to photograph are older generation ones but worthy nonetheless.
Would be interesting to map structures on the metal mask to chip level structures but that's for later.