Slashdot Mirror
Cornell University Sues Hewlett Packard
bmc writes: "Haven't seen this on any of the big news sites, but the local paper is reporting that Cornell is suing HP for patent infringement. The alleged infringement covers HP processors manufactured from 1995 to the present. How common is it for big universities to get involved in lawsuits like this?"
Here's Cornell's press release about it.
GROGGS: alive and well and living in
The patent number for this is: 4,807,115
Surf on over to US patent and Trademark Office and do a search with the patent number here:
Search uspto.gov by patent number
Or read it here if I don't bung up the the HTML.
Err he did. Perhaps you are thinking of compilers optimizing the order of instructions? That is not what this is about, this is about the actual hardware taking the instructions and reordering them on the fly. If my memory serves, which it usually doesn't, it requires something on the order of 250,000 transistors, so if you are asking why no one thought of this before 1989, it's because we didn't have 3 fucking billion or whatever transistors on each chip back then, so it would have increased the number of transistors by an order of magnitude. Of course, if you ask me, the compiler should do reordering, but what should I know, I got a big fat C in the *only* low level essentially EE (aka architecture) course I had to take - god bless the old-school theoretical computer-scientists-as-wanna-be-mathemeticians attitude at this school!)
-DH
ps - goddamn just about everything else at this school.
Here's the abstract of the patent:
"An instruction issuing mechanism for boosting throughput of processors with multiple functional units. A Dispatch Stack (DS) and a Precedence Count Memory (PCM) are employed which allow multiple instructions to be issued per machine cycle. Additionally, instructions do no have to be issued according to their order in the instruction stream, so that non-sequential instruction issuance occurs. In this system, multiple instruction issuance and non-sequential instruction issuance policies enhance the throughput of processors with multiple functional units."
If that description fits the processors you mention, then you have a point.
I survived the Dick Cheney Presidency 7 to 9 AM 7-21-07
Instruction issuing mechanism for processors with multiple functional units
Abstract: An instruction issuing mechanism for boosting throughput of processors with multiple functional units. A Dispatch Stack (DS) and a Precedence Count Memory (PCM) are employed which allow multiple instructions to be issued per machine cycle. Additionally, instructions do no have to be issued according to their order in the instruction stream, so that non-sequential instruction issuance occurs. In this system, multiple instruction issuance and non-sequential instruction issuance policies enhance the throughput of processors with multiple functional units.
The point it, it wasnt licenced. If I have a monopoly, you arent allowed to do it. Period. That is what a monopoly is. My monopoly can be invalidated in any number of ways. Lack of proper disclosure. Revelation of prior art. Those two come to immediately to mind. How long I wait to mention to you that I have a patent, is not one of the ways to invalidate my patent. Issued patents are all published. The assumption is, if you do something, you are supposed to make sure it ISN'T patented. If you do it for 10 years, without cheking if it is patneted I can come to you, in 10 years, and say "Hey, that is mine, you have been violating my monopoly, you owe me money." That is a right granted to me, by my monopoly.
Assuming that cornell has a legitamate patent, it doesnt matter why they didnt complain earlier. Further, it doesnt matter whether or not you HARM me by violating my monopoly. It is mine, granted by the govenrment, from the moment of patnent issue. If you are doing something, that I have a patent on, and you dont get a license, you are going to owe me money.
And the statemnt "Dude i dont know anything about friggin IP law " is rather the point of my post. You dont know anything about IP law. Why are making irrelevant proclamations, specifically
"what matters is how the technology was being licensed and why it took so long for Cornell to realise that HP was "stealing" the work done at Cornell",
it wasnt being licenced (read the article) and why it took cornell so long has nothing to do with cornells rights. You werernt being infomative, you were bieng mis-informend, and you have gotten modded up for it. I suppose I am railing as much againt the Moderators as your post. Your post was just not relevant. It says "The point is X" when X has next to nothing to do with the point. Now, if you want to say "Patent law should prescribe people from suing, if they do not aggresivley defend their patents, the way trademark law will invalidate a trademark not aggressivly defended" I guess that would sort of be relevant. But that isnt what you said. You said, "what matters is", and there you are just wrong. It has been that slashdot is a resource for people to learn about stuff. You want to just say something, ok. But I would hope that poeple who are mod-ing will pay attention to how they mod. I mean, seriously, you are at 3 now. When something is modded to 3 I would hope that it knows what it is talking about. And yes, it is slashdot, where people supposedly care about what their rights are and arent. So we should know about them. And not just say silly things. And hopefully not get arbitrarily modded up for saying silly things.
No matter how crappy you think Itanium is. HP and Intel codeveloped the IA-64. They actually dumped their own RISC processor in favourite of Itanium. Look what they are getting into now...
kawai
Cornell planned to sue Intel as well, but in the the affair was settled with a generous donation by Intel to the university and an Intel Fellowship for Thorng.
(Thorng was my advisor at Cornell).
http://www.news.cornell.edu/releases/Dec97/Torng.b s.html
b s.html
For the Lazy.. http://www.news.cornell.edu/releases/Dec97/Torng.
Err no he didn't. Tomasulo invented the way we do OOE many, many years before there were "3 billion fucking transistors" on a processor (which is still a slight exaggeration even by today's standards). The first machine to employ OOE was the IBM 360/91. Here are some pictures of the first (that I know of) 360/91 in operation in 1968 at NASA. Even though the machine was (by 1968's standards) blazingly fast, the 360/91 failed miserably and only a few were actually produced. The reason this machine failed was because it was incapable of handling interrupts properly. The 360/91 only supported imprecise interrupts, which meant that instructions causing the interrupt as well as subsequent instructions could continue to execute if they were already in the pipeline. This is generally not a Good Thing(tm).
I don't know where all of the Cornell students seem to get the impression that Torng invented OOE. Hopefully this is not from Torng himself, as anyone who studies modern processor design can tell you the first thing you learn about with OOE is the Tomaluso algorithm, which although it was invented over 30 years ago is still used in modern processors largely unchanged from the original design. This is not to say that Torng had no role. He did in fact substantially facilitate OOE, by devising a system to allow mutliple instructions to issue simultaneously (as his patent claims). Torng himself pays homage to Tomasulo in the patent, referencing his paper from 1967.
NOTE: this is a cut+paste from my reply to another post somewhere else. I just wanted to set the record straight. BTW it's typically abbreviated OOE (the "of" doesn't get a letter).
Err no he didn't. Tomasulo invented the way we do OOE many, many years before there were "3 billion fucking transistors" on a processor (which is still a slight exaggeration even by today's standards). The first machine to employ OOE was the IBM 360/91. Here are some pictures of the first (that I know of) 360/91 in operation in 1968 at NASA. Even though the machine was (by 1968's standards) blazingly fast, the 360/91 failed miserably and only a few were actually produced. The reason this machine failed was because it was incapable of handling interrupts properly. The 360/91 only supported imprecise interrupts, which meant that instructions causing the interrupt as well as subsequent instructions could continue to execute if they were already in the pipeline. This is generally not a Good Thing(tm).
I don't know where all of the Cornell students seem to get the impression that Torng invented OOE. Hopefully this is not from Torng himself, as anyone who studies modern processor design can tell you the first thing you learn about with OOE is the Tomaluso algorithm, which although it was invented over 30 years ago is still used in modern processors largely unchanged from the original design. This is not to say that Torng had no role. He did in fact substantially facilitate OOE, by devising a system to allow mutliple instructions to issue simultaneously (as his patent claims). Torng himself pays homage to Tomasulo in the patent, referencing his paper from 1967.