Slashdot Mirror
Intel Opens Itanium Specs
Quite a number of people have recognized the power of open source development - Intel definitely has by opening the specs for Itanium. They've got major blueprints up on the Web. Good sign of the success of open development - but I wonder if AMD's resurgence has anything to do with this.
What else could they do? "Oh sure, we have this new kickass architecture and we would like get people to develop for it. Due to our paroid "don't tell them how it works"-policy we won't be releasing any specs though, so you'll just have to guess the right opcodes."
At least the CNet journalist didn't actually claim that Intel was "open sourcing" the Itanium itself - just that it was being friendly to open source development for the Itanium by releasing specs.
Pity I can't say the same for /. -- "Intel Open Sources Itanium". In the headline of all places!
Oddly enough, even the usually savvy The Register stuffed this one up too - "Open source CPUs..."
-Spiv.
You're confusing speed with MHz again. There's no reason for Motorola to crank the MHz much above 500 MHz at the moment, since at that level (combined with Altivec and the rest) it's already competing nicely with the 1000 MHz offerings of AMD and Intel. If they did, then they'd just be leaving themselves less room to expand in the future. IBM demonstrated some 1000 MHz PPC chips almost two years ago, so it's not an engineering matter that's keeping them back.
And if you want to discuss thermal breakdowns, then why bring up AMD? Their processors are a semi-order of magnitude hotter than Motorola's.
"If one is really a superior person, the fact is likely to leak out without too much assistance" -- John Andrew Holmes
Up until now, Intel was keeping even the databooks on these chips under lock and key. By giving these out now, they're letting anyone who wants to port an OS to their chips a fair chance.
Now, instead of just Windows NT, Solaris and Linux(the three OS's they were supporting) anyone can go do a IA-64 port.
They were being secretive before, claiming that the data contained in those
No, they're not "Open Sourcing" their chip, and I don't really see where *Intel* said that. But they are being Open Source Friendly. Don't flame someone for making a good effort, even if it's not as much as you want.
What about MIPS? Didn't that start out as an academic project, and wouldnt it therefore be publically accessibleß
The illegal we do immediately. The unconstitutional takes a little longer.
--Henry Kissinger
This might be a stupid question. But I'll ask anyway (sort of obvious sine your reading this).
Of what use is this information Intel is giving out?
Another issue that hamstrung the Motorola 68k family was the fact that Motorola didn't seem to do very well improving the speed of the CPU.
I do know that Intel (and more recently AMD) are able to crank up the CPU core speed pretty easily. This shows that Intel and AMD engineers had a pretty good idea how much faster the clock speed could go without causing thermal meltdown problems. After all, how come Motorola has not gotten the PowerPC G4 CPU past 500 MHz? Shouldn't they be able bump it up to 1,000 MHz or more?
Raymond in Mountain View, CA
In theory, at least. Just about any Open Source OS user has a compiler somewhere to patch up software, but most of us don't have bunny suits or a fab in their backyard...
--
--
fat lenny's gonna lick your brain today.
Motorola charged for their documentation, while Intel made it readily available for anyone who wanted.
Are you sure about this? I remember getting a free programmers guide for their 65HC11 microcontroller. Maybe they have different policies for their different products.
Scuttlemonkey is a troll
PDF documents really suck for online reading - pdf readers generally have crappy text search, if any at all - they're not internally hyperlinked, and you're restricted in your choice of readers.
The one good thing about pdf is it looks pretty when you print it out. But I for one never intend to print these out - if Intel's past policy is anything to go by, they will give you the official bound books for free.
Hasn't Intel heard of Docbook? Second question: has Intel good a clue who the target market for these docs are, and what their needs and desires are?
--
Life's a bitch but somebody's gotta do it.
Saying Intel is "open sourcing" Itantium because they are releasing architecture and programming info is like saying Black and Decker is "open sourcing" their toasters because they posted instructions on how to put the bread in.
At MOST this makes Intel "Open Source friendly", but I would argue that it just makes them pragmatic. How else am I going to create devices and compilers for a platform except if I have the specs? Duh.
--
Have Exchange users? Want to run Linux? Can't afford OpenMail?
Linux MAPI Server!
http://www.openone.com/software/MailOne/
(Exchange Migration HOWTO coming soon)
Wow.... I can't believe it! No more mindless bandwagon-jumping from Intel, but some real dedication to the Open Source cause!
I'll tell you what I'm going to do: I've got a few ideas on how to really improve the floating point speed on these guys, so I'm gonna put together a few patches and then whip out a few chips and test 'em and see if they work. I'll submit those to Intel, for next week's CVS snapshot, and then I've got a few tweaks I want to make to the speculative execution. After I submit those patches and get them tested for a while, I'll whip up another batch of chips and put them on my website free for download!
What? I need a multi-billion dollar fab for this?
What are you talking about?
Oh. I see.
Um, anyway... Back to the Linux kernel....
I'm gonna buy a bunch of resistors and transistors and solder me up an Itanium.
Do you think it will fit on a 6x6 breadboard?
George
How is Intel "losing"? I'm a Mac user myself, but Iknow that Intel is far from losing. Doing this can only help push Itanium into the mainstream server market. As far as AMD is concerned, I'm sure they make great chips, but when the day is over, they are still based on Intels design.
--
Don't lead me into temptation... I can find it myself.
That said, anyone planning on developing cutting-edge software should care. This information, which is normally protected by Non-disclosure agreements,is normally only available to approved software and hardware vendors/manufacturers. This really levels the playing field by giving every programmer equal access to this information.
kwsNI
Go look at the intel website link and it will take you about 5 seconds to see that this has nothing to do with open source. No designs have been published, no schematics, not even decent scale picture of the insides.
In fact what has happened is intel has published it's usually array of developer documentation online. This is the same information that has been freely available for all it's over processors. Yes intel has made these freely available online (a good thing), but even this is not new.
What we have here is a case of a journalist slapping the term 'open source' on a news item to get a bit of attention, either that or a journalist who never reads the material, some how 'technical specifications and programmer docs' has become 'blueprints', a bit a leap of imagination if ever I saw one.
This reminds me a lot of that altervista 'open sourceing' which just turned out to be the html code for a search box.
Hmm... Maybe I didn't look closely enough, but all I see are programming manuals and overviews of the chip's architecture. This is much less information than they released for past chips. I wouldn't trump it up as "opening their hardware" unless they did something totally uneard of like posting its complete ASIC design.
They've also always said they'd release the specs for IA-64, but they also said they intended to keep it under NDA until it was ready. That didn't stop them from giving access to VA etc. and funding Linux and GCC development for it, with the intent of releasing it once they'd be releasing specs.
It's nice to see their specs are available, but there's nothing more to it than business as usual.
Do I get a free chip if I submit an improvement?
No, you go down to your basement foundry, fire up all the GNU-brand fab equipment you downloaded from the FSF, and you pump out your own Itanium.
The whole idea is ridiculous! This is almost the same kind of idealism that lead the Chinese government during the 'Great Leap Forward' to convince people all across China to build backyard steel foundries. For the more dense idealists in the crowd itching to reply to this comment- no, it didn't work in China.
It's so ludicrous it can't help but make this website look like a bunch of damn fools for even putting up an article about it.
This used to be the case, but with modern CPU designs, it is very difficult to get the instructions in the right order to allow for out of order execution, keeping multiple pipelines full, etc.
You can do this by hand, but your code won't be very maintainable. In assembly, there is a tradeoff between readability/maintainability and speed of execution. Your better off writing in a high level language and let the compiler mangle your code for you.
Of course, there are some exceptions, but compilers are getting better, not worse.
Well, at least there is some serious blue color in
the page, so it's not entirely misleading article.
Phew, I almost lost my faith in the journalism.