SGI v. AMD: Chief Judge Rader on Claim Construction

Silicon Graphics, Inc. (SGI) v. ATI Technologies, Inc. and Advanced Micro Devices (AMD) (Fed. Cir. 2010)

In his first opinion as Chief, Judge Rader provided the parties with a lesson in claim construction. This decision is a case-in-point for the high value of professional patent drafting because the decision turns on the esoterics of claim construction. This decision also highlights the primacy of claim construction in the infringement analysis. Namely, the question of whether ATI chips infringe was boiled-down into an interpretation of the claim language.

SGI’s patent is directed to graphics processing hardware that uses floating point calculations.

Meaning of “A”: The asserted patent claims include “a rasterization circuit . . . that rasterizes the primitive according to a rasterization process which operates on a floating point format.” Although the accused ATI chips do use floating point numbers, the ATI rasterization process also uses fixed-point numbers. The district court held that the ATI chips could not infringe because their rasterization process does not operate in floating point format “as a whole.”

On appeal, the Federal Circuit re-parsed the claim language and rejected the district court’s construction as erroneous. Rather, the fact that a circuit rasterizes using fixed point format does not preclude the circuit from also using “a rasterization process which operates on a floating point format” as required by the specification.

The use of the indefinite article “a” in the claim, when coupled with the list of processes provided in the specification, makes it clear that the claims’ references to “a rasterization process” means “one or more rasterization processes.”

The limitation “a rasterization process which operates on a floating point format” therefore means that “one or more of the rasterization processes (e.g., scan conversion, color, texture, fog, shading) operate on a floating point format.” This construction is also in line with the rest of the specification. Nowhere does the specification teach that all rasterization processes must operate on a floating point format.

Summary of the Invention: The claim also requires that the rasterization circuit perform “scan conversion.” The district court required that the scan conversion be accomplished entirely using floating point numbers. On appeal, the Federal Circuit agreed with that interpretation. In making its decision to limit the breadth of the term, the court looked to a statement from the Summary of the Invention that: “the scan conversion process is now handled entirely on a floating point basis” and another statement from the specification that “this rasterization process is performed exclusively in a floating point format.” In several places, the specification indicated that various operations could be performed using fixed point numbers. However, the court held those statements insufficient because they were never specifically directed to the scan conversion process. “Thus general language in the specification permitting some operations to be done in fixed point does not work to contradict the specific language that requires scan conversion in floating point.”

70 thoughts on “SGI v. AMD: Chief Judge Rader on Claim Construction

  1. 70

    For Juxt-a-post:
    Using both “at least one” and “a” within the same claim, or within the same claim set, might just raise some issues. I have not seen it addressed in case law, but it could send a district court into spasms. The correct answer should be that both mean the same thing, but that might be dispelled by the specification, arguments in the file history, claim differentiation, or other extraneous factors (the color of your tie when appear for summary judgment). I would avoid it myself, but I would not jump for joy if I found it in a patent I was trying to avoid.

  2. 69

    You could have a point ping. I’m thinking how the little Englanders constantly rubbish everything that is “European”. When they can point out where the Europeans are going wrong, and come up with a better solution, they get attention. But if all they do is assert that the UK patent system is already perfect, they might well amuse, but at the same time they lose the attention of their audience, both domestic and foreign.

  3. 68

    just one more argument against “harmonization”

    Harmonization is the siren song of those who envy the Gold standard of the US Patent.

    Be more like us, because we aint good enough to be like you.

  4. 67

    “you don’t understand the difference between “anticipate” and “read” ”

    And it appears that you think there is one…

    If you are European as I assume you are, your comment represents just one more argument against “harmonization”

  5. 66

    Regarding post by DC | Jun 16, 2010 at 01:22 PM:

    Query: If a claim is drafted using “at least one” to merely emphasize (reserve) plurality breadth for a particular element, but using “a” for other claim elements, could there be a concern (or a presumption) that the claim elements using only “a” might be construed as reading on only the singular, rather also the plural, of the element?

  6. 65

    Quoth RWA:

    “Luke, you have it exactly backwards. A prior art disclosure of 4 legs would anticipate a pending claim to 1 leg. An issued claim to 1 leg reads on any accused device with one or more legs”.

    RWA, I fear you don’t understand the difference between “anticipate” and “read”. You said “a chair with 4 legs reads on one with 1 leg”. No it doesn’t. It will anticipate it, to be sure, but that isn’t the statement you put forward; time to shift those goal posts back. As to whether an issued claim to 1 leg reads on a claim to a chair with 4 legs, that depends on the wording of the claim(s). Does the 1-legged claim use comprising/including (open, 4 legs will read), or having (could go either way; depends on the facts), or consisting of (closed, won’t read)? There are lots of nuances there.

  7. 64

    Poorly stated. This is not our grammar.

    ‘ The use of the indefinite article “a”
    in the claim, when coupled with the list of processes
    provided in the specification, makes it clear that the
    claims’ references to “a rasterization process” means “one
    or more rasterization processes.”’

    The use of the indefinite article “a” in the claim indicates *one of a number of processes*, broadly defined within the specification to be associated with “rasterization”. Instruction that, in the general case, the selective singular indefinite article “a” or “an” refers to more than a singular element is bad basis for law. A child might ask for “an apple” from a basket containing three Gravensteins, four Westfield Seek-No-Furthers, two Newton Pippins and five Opalescents and anticipate that she would receive one of each under this “law”. “A” troy ounce of gold would in the general case be one or more troy ounces, a fact that would make me rich since I could buy many at the price of one, while selling “a” troy ounce at a time to those unfamiliar with the “rules of Grammar”, preferably at the expense of a patent lawyer or vexatious and overreaching patentee. The claim at issue would encompass a parallel processing element capable of any number of concurrent rasterization processes, when the specification (though making not much else clear) discredits that interpretation of the claim.

    The precedents make clear that “an inner layer” can be comprised of a number of laminae, but only a single grouping of laminae is “an” inner layer. Rather than making the claim broader, the poor grammar and lack of specificity makes the claim indefinite.

    The Honorable Chief Judge Rader is “a” poor grammarian, one of a plurality of poor grammarians sitting on “a” Federal Appeals court.

  8. 62

    Here endeth the lesson.

    Um no. Nothin like simply being wrong and simply proclaiming victory – wait – I get it – you are a sockpuppet of 6!

    The point of the thread is that prior art to one or more ordinary legs, anticipates a claim to 1 leg.

    That be true – but that not be what you said.

    Sorry chuckles –

    Mere elimination of features is generally not patentable to begin with

    I lol – you, compounding your error with more fluff when you should simply admit that you were wrong. Eliminating features can be patentable if you maintain the performance (nay even improve) without those features. Take most mechanical devices – inventing something that can perform the same with half the parts (I not even going to 1/4 the parts as in 1 leg rather than four) sure as heck could easily be patentable. O course, “mere elimination in and of itself aint telling enough – but what you said: “As taught in Patent Law 101, a claim to a chair having one leg reads on a chair having four legs”. won’t cut it.

    You didn’t say “reads on a chair with one or more ordinary legs”. Ya want to amend your original statement, be my guest.

    But if ya wanna throw words like “meaningless” at me – go ahead – youse can join good ‘ol Ned on the eat-your-own-words bandwagon. You be nice and he may even share his tabasco sauce.

  9. 61

    “an application that eliminates features is not anticipated by prior art with the thus-eliminated features”

    Are you trying to hold yourself out as some kind of authority ping?

    Mere elimination of features is generally not patentable to begin with. But thanks for your meaningless hair splitting of a perfectly good example. The point of the thread is that prior art to one or more ordinary legs, anticipates a claim to 1 leg. There is no more to this example than that.

    Here endeth the lesson.

  10. 60

    That is, you have to get your improvement into the claims.

    That’s the point I was trying to make (i.e. “I claim a chair with only one leg” will not be anticipated by a prior art reference that teaches a chair with four legs “I claim a chair with four legs” – note that I did not say that the prior art reference claim is “a chair with a leg”). RWA’s statement is simply wrong. He may have meant to say something else, but will need an amendment to get there.

    farrago” ?

  11. 59

    Ping: A disclosure to a chair with 4 legs would necessarily anticipate a claim to a chair with one leg. Really. You might be able to add negative limitations, or positive limitations, to differentiate a barber’s one legged chair from an overstuffed Queen Ann four legged chair, but a claim merely for a chair with one leg is certainly anticipated by a prior disclosure of a four legged chair. That is, you have to get your improvement into the claims.
    But let’s get back to the case here: The opinion is a bit of a glancing blow to the actual issues. The district court did not deal with the “a” versus “a single” dichotomy. It grappled unsuccessfully with the “addition of features to an infringing device does not avoid infringement” rule. The “as a whole” language was a farrago of construction concepts, which could be more clearly explained as an incorrect rule that addition of other processes in the circuit with all the elements of the claim avoids infringement, or perhaps an incorrect construction of the claim to require one process to the exclusion of any other process. The Federal Circuit and Silicon Graphics, rather than restarting the analysis from a legitimate point instead tried to deal with the issues as framed by the District Court, which leads to this confusion about what the meaning of “a” is, which was not really at issue.

  12. 58

    I had this exact conversation with a client one time. He said, “so if I say I had two beers last night, it could mean I had five beers?” Yep, in a patent claim.

    That shouldn’t be a very long conversation. It’s not even a counter-intuitive concept in everyday life.

    “Hey, do you have $20?”
    “Yes, I have $50.”
    “Great. Does your chair have a leg?”

    Alternatively:

    “Hey, got a minute?”
    “Sure, I have all day.”
    “Great. Did you have two beers?”

  13. 57

    “As taught in Patent Law 101, a claim to a chair having one leg reads on a chair having four legs”.

    Really? Sounds rubbish to me.

    And yet, it is black letter law. “Comprising one” means one or more. If you want to claim exactly one of something you need to say “only one” or “consisting of one” or the like.

    I had this exact conversation with a client one time. He said, “so if I say I had two beers last night, it could mean I had five beers?” Yep, in a patent claim.

  14. 56

    A prior art disclosure of 4 legs would anticipate a pending claim to 1 leg.

    Oh really? So a disclosure that teaches that you need 4 legs anticipates an improvement that eliminates the need for 3 of the legs? – the chair example doesn’t lend itself well; but I hope you get the message that an application that eliminates features is not anticipated by prior art with the thus-eliminated features.

  15. 55

    “As taught in Patent Law 101, a claim to a chair having one leg reads on a chair having four legs”.

    Really? Sounds rubbish to me.

    A claim to a chair comprising (not having; having is not presumptively open) 4 legs reads on a claim a chair comprising *a* leg, yes.

    Luke, you have it exactly backwards. A prior art disclosure of 4 legs would anticipate a pending claim to 1 leg. An issued claim to 1 leg reads on any accused device with one or more legs.

  16. 54

    This morning, the Supreme Court published 5 opinions. Sorry folks — again no Bilski.

  17. 53

    at least one couple beers.

    Could that also be an odd number of beers?

    Better to ask (English-style) “What exactly was the writer of the claim using the language of the claim to mean”

    That would be purposive construction, yes?

    The patentee gets to decide exactly what the claims say. If he had a readily available alternative wording that would have better expressed his meaning, it’s his own problem that he didn’t opt for it. Let him put his intended meaning in writing as best he can. If it looks awkward in the claim, he can write it into the spec.

    Just so you know, I’m trying really hard to not imagine you describing this case in your classiest Alaskan accent as “an unfair attack there, again, based on the verbage that the patentee had used”. You betcha.

    But, again, this case is not about whether “a” means “at least one”.

  18. 46

    Luke, the UK Interpretation (of documents) Act that Parliament passed in 1880 something decrees that the singular shall include the plural, and the plural the singular, unless the context requires departure from that norm. So, “a” does mean “at least one” and translators ought to be able safely to use “a”. And then, if I claim a body having “a” head, I don’t see the problem about reciting in claim 2 that “the” head comprises “at least two” heads (think Gorgon). Of course, the enlightened drafter might use “head” (but not of course “head means”)in claim 1 and “head element” or “sub-head” or some such terminology in claim 2 but, hey, these days life’s too short for perfect drafting. Better to ask (English-style) “What exactly was the writer of the claim using the language of the claim to mean”

  19. 45

    “As taught in Patent Law 101, a claim to a chair having one leg reads on a chair having four legs”.

    Really? Sounds rubbish to me.

    A claim to a chair comprising (not having; having is not presumptively open) 4 legs reads on a claim a chair comprising *a* leg, yes.

    But if you claim *having* a specific numbers of legs, you mean that specific number, and only that specific number.

    The canon of “a” equaling “one of more” only applies to open transitional phrases. But I can’t think of a US cite at the moment. But I do recall a UKPO decision of Aug 15, 2006, where “formed of a” meant one only.

    Ah, I remember the US cite, it’s Abbot, CAFC 2003: in a Markush group “a” was held to mean only “one”, and not “one or more”.

  20. 44

    Regarding “a” being rewritten as “at least one”, this can be important when dealing with specifications originating from Japan. Japanese has no exact equivalents of definite or indefinite articles, or singular or plural, for that matter.

    So in Japanese, you can easily have a claim set saying something equivalent to: Claim 1. “An X, comprising a Y”. Claim 2. “The X of claim 1, wherein the Y is 2 Ys.”

    This not only sounds unnatural in English, it is asking for a rejection from the examiner for claim 2 who will likely object that the, singular, Y, can hardly be two Ys.

    But since in the original Japanese, there is no “a” as such in Claim 1, the better translation is Claim 1. “An X, comprising at least one Y”, which gives a perfectly natural Claim 2 of: “The X of claim 1, wherein said at least one Y is 2 Ys.” (This particular claim format being about the only time I prefer using said to the) But very few translators do this.

  21. 43

    You’d be surprised at how well you can construe claims after at least one bratwurst and at least one couple beers.

    Fixed, even as I suffer from at least one horrific hangover.

  22. 42

    With out reading the whole thing its seems to me that the important difference is between “on” and “in”.

    A rasterization process operates “on” floating point if the input data is in floating point and it operates “in” floating point if all of the internal intermediate calculations are done using only floating point.

  23. 40

    Paul, you might want to access the entire linked article. The linked document contains only a portion of part IV of a larger article on claim construction. Parts 1, 2 and 3 are not included, but discuss the law as it is, not as it should be.

  24. 37

    Nice reference to an article which tells us what the law SHOULD BE. But I have more than enough difficulty trying to keep up with what the law IS (if anyone knows).

  25. 36

    The linked article argues that claim construction is not directed to what one of ordinary skill in the art would understand the claim terms to mean, a question of fact if ever there was one, but rather what the claim terms mean to a patent attorney of ordinary skill practicing in that art.

    Expert opinion from skilled patent attorneys welcome.

    link to cyber.law.harvard.edu

    Still there are questions of fact about what claims mean, and rules of construction. While the question might ultimately be one for the court to decide, I do not see why a jury must be excluded from the process.

  26. 34

    I hear two beers is approximately the equivalent of being a District Court judge in Wisconsin.

    Six is equivalent to 6.

  27. 33

    the juries do a better job because they give a claim its clear meaning. Juries don’t give overly pedantic interpretations of what the meaning if “is” is.

  28. 32

    You’d be surprised at how well you can construe claims after a bratwurst and a couple beers.

    I hear two beers is approximately the equivalent of being a District Court judge in Wisconsin.

  29. 31

    I hope you’re not saying that juries did a better job of construing claims because of their greater technical or patent skills

    You’d be surprised at how well you can construe claims after a bratwurst and a couple beers.

  30. 30

    Once the wise courts took claim construction away from the collective wisdom of the jury and placed it with the singular wisdom of the Judge (more often than not having no particular technical or patent skills)

    I hope you’re not saying that juries did a better job of construing claims because of their greater technical or patent skills.

  31. 29

    Mr Morgan is correct. Once the wise courts took claim construction away from the collective wisdom of the jury and placed it with the singular wisdom of the Judge (more often than not having no particular technical or patent skills) the cost to the parties skyrocketed and arguments about the USPTO issuing bad patents became an infringer’s battle cry. I for one would like to see a return to the good old days before Markman.

  32. 28

    “But is the claim still valid?”

    You mean from an “all-computer-claims-are-obvious-gahrbahge” perspective?

  33. 25

    For Paul F. Morgan and Fischer Ames: Using ‘A’ or “At least one” does not matter: A, or at least one judge, can screw up the interpretation of both. The problem here is not the perfectly clear claim, but the muddled “as a whole” concept used by the judge. The “at least one” language does have the additional benefit of making a complicated claim inscrutable, but I avoid it when I can.
    By the way, who cares about this case? A new chief judge applied the most rudimentary claim construction rule in a most rudimentary manner, and got it right? Big Deal.

  34. 24

    there’s at least one person in Wisconsin who learned something from the case.

    The lessson is: don’t trust defendants when they try to narrow the claims in a manner which rebukes the canons.

  35. 23

    I disagree that this case involve the esoterics of claim construction. I see nothing new or surprising in this case, other than the district court making such an elementary mistake.

  36. 22

    Red M if the claim says “circuit that does A” and the accused circuit can do either A or B, then does it infringe? DCt said no, CAFC said yes.

    The District Court should be ashamed.

    Rader did the right thing. A great start. GO RANDY GO!!!!!!!!!!!!!

  37. 21

    Give me a break. That construction is the default, as every professional knows.

    I didn’t say it was a particularly contentious question. I said that was the question in this case.

    The result may have been obvious to you, but there’s at least one person in Wisconsin who learned something from the case.

  38. 20

    IANAE

    No, it’s about whether a thing that has properties other than the claimed property is covered by the claim, i.e. if the claim says “circuit that does A” and the accused circuit can do either A or B, then does it infringe? DCt said no, CAFC said yes.

  39. 19

    IANAE This case is about whether “a thing having a property” can permissibly mean “one thing having the claimed property, but also it has another thing lacking the claimed property…

    “Can permissibly”? Give me a break. That construction is the default, as every professional knows.

    This is a boring case. So now the claim is nice and broad because the Federal Circuit correctly did not bite on the defendant’s desperate attempt to narrow the claim. But is the claim still valid?

  40. 18

    Fisher Ames

    Not sure if you noticed but in my hypo, it is not about “a/the”, it is about the difference between “”a circuit that operates on a FPF” and “a circuit that operates on FPF”. Should that be construed the way the district court did, since it no longer has the “a” for meaning “one or more”?

  41. 17

    IANAE,

    You bring up a very good point which may have been too subtle for this Federal Circuit panel to grasp. What isn’t clear is whether the alleged infringing device had the “rasterization process” expressly recited in the Claim, as well at least one other “rasterization process” not expressly recited in the Claim but described in the patent spec. If the alleged infringing device doesn’t at least include the “rasterization process” expressly recited in the Claim, there can be no literal infringement (under the All Elements/Limitations Rule). It would then come down to DOE.

  42. 15

    “…ATI chips could not infringe because their rasterization process does not operate in floating point format “as a whole.” ”

    A good example of the kind of twisted logic that arises from sloppy application of claim construction principles (like saying that prior art “reads on” claims).

    Nowhere in patent law was it ever said that the accused device must be read upon “as a whole” by the claim.

    Talk about creative advocacy.

  43. 14

    The patent holder’s claim limitation reads “a rasterization circuit . . . that rasterizes the primitive according to a rasterization process which operates on a floating point format.”

    The master, million dollar defense is to add “as a whole” to the claim so that it has the effect of “a rasterization circuit . . . that rasterizes the primitive according to a rasterization process which operates on a floating point format as a whole.”

    The CAFC’s brilliant solution is to give primary weight to the patent holder’s claim language and specification and secondarily note something outside the patent holder’s claim language and specification, namely that “It is well settled that the term ‘a’ or ‘an’ ordinarily means ‘one or more.”

    Observations
    1. Great job CAFC! Perhaps we finally will reach the promise land where primary weight is given to the patent holder’s claim language and specification. Even if we’re unsure exactly what that looks like. Even if we don’t yet know precisely how to get there. I now feel more confident that we’ll get there.

    2. How much did the accused infringer spend on that defense? I probably would have tried to settle up front to save the accused infringer $$$ — and cause me to lose out in legal fees — which is why these things never come my way.

  44. 13

    Mr. Morgan,

    “a” should not require a “correct claim interpretation.” Certainly it is far more established than “at least one,” which would cost at least 10 million to interpret.

  45. 12

    Those criticising Mr. Ames’ advice to use “at least one” instead of “a” apparently think they would be doing their clients a favor by costing them “a” million [or more] dollars in extra legal fees going through and from a District Court to the CAFC, as here, just to finally get a correct claim interpretation for “a” that could have so easily be avoided ab initio?

  46. 11

    This case isn’t about whether “a thing” means “one thing” or “one or more things”. That’s well settled.

    This case is about whether “a thing having a property” can permissibly mean “one thing having the claimed property, but also it has another thing lacking the claimed property”, or whether it must mean “one or more things, all of which must have the claimed property”.

  47. 10

    Fisher Ames, what are you smoking? This sounds like the kind of reactionary thing that a crazed in-house counsel would demand in the wake of this.

    The more sensible lesson is simply that “a” still means “one or more” as it always has.

    As taught in Patent Law 101, a claim to a chair having one leg reads on a chair having four legs.

    A claim to a rasterization circuit that rasterizes according to a floating point format reads on an accused chip that operates using floating point numbers, even if it can use fixed point numbers.

    Don’t complicate life by demanding that every “a” be changed to “at least one of”

  48. 9

    “Shifting sands”? On claim construction? Oh dear, oh dear. I’m sorry to read that Fisher.

  49. 7

    If “a” means “one or more” (Hutz) why must I replace it Fisher with “at least one”? Is it something to do with the Rule that “or” is indefinite (so now “a” is indefinite too)? I’m hungry for more such education.

  50. 6

    replace every “a” in your draft claim with –at least one– or otherwise find a way to get rid of the indefinite article. also, not sure why the review court makes a distinction between “a rasterization process” versus the hypo –the rasterization process– as the latter would have lacked antecedent basis.

  51. 5

    Thanks Lionel. Haven’t read the decision yet. But your comment increases my suspicion that, often, what is good drafting for a PCT filing is bad for the USA and vice versa and there’s no way around that puzzle. Good news for folks like me. So, long may it continue.

    EG, thanks for the Link.

  52. 3

    I don’t see a link either and I should clarify that my opinion is based upon the summary provided (as it is 90% of the time).

  53. 2

    Would it have made a difference if you had left out the definite article altogether, as in “a circuit that operates on floating point format”?

    Also, I can’t find a link to the actual decision.

  54. 1

    This actually reinforces two aspects

    (1) “a” means “one or more” which is what the vast majority of us have always believed, although I have seen some comments to the contrary.

    (2) Avoid definitive statements or achievements of the invention statements that can be read as limitations by the court. Before Max chimes in, there are international considerations that need to be balanced if filing overseas.

Comments are closed.