OpenSPARC and Power.org, Who has it Right? 125
Andy Updegrove writes "Last summer, IBM set up Power.org, to promote its PowerPC chip as what it called 'open hardware.' This year, Sun launched the OpenSPARC.net open source project around the source code for its Niagara microprocessor. But what does 'open' mean in the context of hardware? In the case of Power.org, Juan-Antonio Carballo said, 'It includes but is not limited to open source, where specifications or source code are freely available and can be modified by a community of users. It could also mean that the hardware details can be viewed, but not modified. And it does not necessarily mean that open hardware, or designs that contain it, are free of charge.' True to that statement, you have to pay to participate meaningfully in Power.org, as well as pay royalties to implement - it's built on a traditional RAND consortium model. To use the Sun code, though, its just download the code under an open source license, and you're good to go to use anything except the SPARC name. All of which leads to the questions: What does 'open' mean in hardware, and which approach will work?"
Re:Binary minds want to know. (Score:3, Interesting)
Right/Practical (Score:3, Interesting)
The more interesting question is "what use is an open core?"
Open source software has obvious utility in that it can be used by millions of people for a wide variety of jobs. All you need is a computer to get started.
Open hardware, on the other hand, is useful only for education or simulations unless you happen to have a fab plant.
If education and experimentation can be served by a "non-free" license then is there really any benefit to having a "free" license? I suspect by the time off the shelf technology is available to create CPUs based on current designs, they will be centuries obsolete. Even US copyrights and patents will have expired by then (unless they change the laws again) so it's a bit of a moot point.
Now I grant this might be a bit of a narrow viewpoint - for example some of the Lisp hardware designs would be very interesting to work with - but since the hardware costs of this sort of manufacture make the information needed to do it only one component of the (EXCEEDINGLY expensive) whole, I'm not sure the marginal benefit of having "free" cores will be very interesting, at least for something like a modern CPU.
Of course, there are non-economic considerations, but I don't really see overwhelming benefits for the "free as in freedom" model as opposed to the "free except for producing your commercial product based on them" model.
Re:Right/Practical (Score:3, Interesting)
Or MOSIS [mosis.org]....
A few thousand bucks and a working chip design will get you parts these days, in suprisingly modern fab processes (a few tens of thousand for 0.13u and 90nm).
e-iOpenBuzzword.com (Score:3, Interesting)
Then there was the i craze. iPod,iMac,ivillage.com, BMW's iDrive, ....
Maybe "open" is the new cool prefix to use. I'm sure anyday now someone will be selling OpenPods, sending openMail...
Does it matter? (Score:3, Interesting)
SPARC and PowerPC are pretty clearly niche and/or legacy architectures now. IBM has ceded the mainstream desktop to x86, and SPARC lost that battle a long time ago. The only question most people care about now is whether their x86 system is 32 or 64 bit, Intel or VIA or AMD.
Right now, most open source software tends to be tested and hacked on to at least make it run on PowerPC, for the benefit of Mac users. As the PowerPC Mac users switch to x86, who's going to care about PowerPC compatibility? I remember what it was like running Linux on PowerPC before OS X, and it wasn't pleasant--lots of stuff x86 Linux users took for granted didn't work, or you were stuck with old versions, because nobody had bothered to port, test or debug it.
SPARC and PowerPC will continue in the high-end server niche, but I think that niche is increasingly going to be squeezed by x86 too. Why deal with the possible risk of having your enterprise application break on PowerPC Linux, or being stuck with old versions of software, when you could run it on a big x86 Linux system and run the same binary 90%+ of the app's users are relying on every day? Sometimes there's safety in numbers.
PowerPC has the embedded space, of course, and maybe that'll be enough to sustain it as a target for general purpose code. I guess video game toolkits and related libraries will continue to be ported to PowerPC, at least.
But to go back to "openness"--in the embedded and video games space, who cares if the design is "open" or not? All the PowerPC video game consoles are locked down proprietary systems, as are various other embedded PowerPC systems like TiVo and car computers. And in the high end server space, I don't know that anyone cares there either--System i and System z seem to do OK without having open standard CPUs.
[Opinions mine, definitely not IBM's, obviously... and I may be completely wrong, perhaps openness is important in those niches?]
"Real design" (Score:3, Interesting)
I remember reading ~1992 that anyone doing real software development could afford to pay for a real supported compiler. They were downplaying the benefit of GCC and the other GNU tools. Well as hardware has gotten cheaper and faster and the Internet has expanded and more people have gotten tech-savvy, guess what? Lots of people are doing *REAL* software development with FLOSS software tools.
Back when you needed $10000 worth of hardware + OS licenses etc. to make software development feasible, paying $300 for a compiler was no biggie. But now you can get an awesome complete workstation for $600 and even the $100 Microsoft OS tax starts to seem like a pretty crappy deal.
I imagine that hardware design will increasingly go the same way. Obviously, there are a lot more hurdles to go before we'll be fabbing chips in our basements. But I work in electronics research in a physics department, and people are doing amazing stuff like printing integrated circuits with inkjet printers... commercial equipment to do that is now selling for $100k.
Hacker appeal? (Score:3, Interesting)
Sure, everybody just wants a nice cheap powerful x86 box these days, even Mac users. But there'd be a lot of hacker appeal to having the source for the processors available. Everybody would tinker with their processors and implement them on FPGA [wikipedia.org] (a cheap way of fabbing chips on a small scale, basically). Instead of people boasting about their tuned kernel, we'd be boasting about our tuned processors. "I got my OpenSPARC running on a Xilinx FPGA and I optimized out the floating point unit so I could add more cache."
And as we've seen with Linux, W3, Apache, etc., hacker projects can turn into BIG business down the road!!
Frankly, I'd switch over to OpenSPARC or OpenPOWER or OpenRISC if I could. I already can and do hack around with the source code to my software, and I'd love to be able to do the same with the hardware...
Both of them are right (Score:3, Interesting)
But it's no way as cool as the Solaris port to PowerPC.
Sun is involved in both, you see!
So who cares? They're both right.
Re:"show me the code" (Score:1, Interesting)
Uh, the license to OpenOffice.org is LGPL, something the slashdot crowd usually is okay with. Your statement could be interpreted as implying there is something wrong with CDDL; if you believe so, you should state it.
Regarding UltraSparc III and IV, those are older designs and aren't particularly interesting. In fact, they're probably less useful than the the CoolThreads design, because there are a bunch of other Sun-ish things it assumes exist (MP cores are far more about cache coherency and chipset then just straight-line execution). It takes resources to Open Source anything (think about the legal review alone), so it's not surprising that Sun would Open Source whatever it thinks will provide the most impact on the market.