[00:00:00] Speaker 04: The next target this morning is number 22, 1139, Intel Corporation Against Patent, XPP, Schweizer AG. [00:00:11] Speaker 04: Mr. O'Quinn. [00:00:13] Speaker 01: Thank you, Judge Newman, and may it please the court, John O'Quinn on behalf of Intel. [00:00:17] Speaker 01: I'm happy to answer questions on any of the issues regarding the 047 patent, but I'd like to focus on the substance of the board's claim construction and its approach as to two terms specifically. [00:00:28] Speaker 01: quote, according to a state and hysteresis characteristic, each of which affects a different set of claims. [00:00:36] Speaker 01: And Judge Prost, to the procedural questions that you have asked, if the court agrees with us on these claim constructions, given what is not disputed in the record, the court can outright reverse. [00:00:46] Speaker 01: And it would not need to remand or to reach the APA issues that are also presented in this appeal. [00:00:53] Speaker 00: What went down here? [00:00:55] Speaker 00: I mean, the main argument, well, one of the main arguments, and I'll ask your friends too, is that the board, which another board, working simultaneously with one overlapping board member between the board here and the board there, reached at least arguably [00:01:12] Speaker 00: very different, I won't say diametrically opposed results, but one could construe that, or at least something different, or at a minimum something that you think would have required some explanation if the board had one, as to why the cases were different, required a different construction. [00:01:29] Speaker 00: And none of that was done. [00:01:30] Speaker 00: Am I misstating what went down, and do you have any explanation for how that happened? [00:01:36] Speaker 01: I don't, Judge Prost. [00:01:38] Speaker 01: I mean, I agree with you that it does seem somewhat inexplicable. [00:01:41] Speaker 01: You do have, in the case of the 535 proceeding, which also involved a Nicole reference, and specifically found that Nicole teaches that processing element clock frequency may indeed be driven by actual load. [00:01:54] Speaker 01: That's page 67 of the 535 final written decision, which is attached to our opening brief. [00:01:59] Speaker 01: And I do think that is diametrically opposed to what the panel here [00:02:04] Speaker 01: And the authoring PTAB judge in this case was on the other panel. [00:02:09] Speaker 01: So I think it is inexplicable. [00:02:12] Speaker 00: But the arguments were pretty much the same. [00:02:15] Speaker 01: They were, Judge Prost. [00:02:16] Speaker 01: I mean, if you look at the briefing in that case, and we make reference to that in our briefing here on appeal, the same argument about whether or not you were dealing with [00:02:26] Speaker 01: an anticipated load versus an actual load and whether or not that made a difference, the exact same arguments were made in both cases. [00:02:36] Speaker 01: In that case, [00:02:37] Speaker 01: The board seemed to simply decide it on the facts agreeing with us. [00:02:41] Speaker 01: In this case, which was contemporaneous in time, these were decided within 30 days of each other, the board seemed to treat it as a claim construction issue and then construed the claims in a very limited and narrow way to literally read in the word existing. [00:02:56] Speaker 01: to focus and basically adopt the embodiment that you find at the bottom of column 5 of this patent, and to exclude the embodiments that are described in columns 3 and 4 without even acknowledging what it was doing. [00:03:09] Speaker 01: And to be clear, PAC doesn't dispute that embodiments in column 3, lines 1 through 15, and it's referenced again in column 4, 17 through 19, are read out under the board's construction here. [00:03:21] Speaker 01: So that inconsistency at a bare minimum. [00:03:23] Speaker 02: What's the status of that IPR? [00:03:26] Speaker 01: So the status of that IPR, Judge Hughes, is it's still pending before the board on our motion for rehearing or reconsideration on different sets of claims and different issues. [00:03:37] Speaker 01: So the issues on which we've identified the inconsistency here on appeal [00:03:41] Speaker 01: are final for purposes of the board. [00:03:44] Speaker 01: I suppose that they are theoretically appealable by pact one day whenever the board issues a decision on rehearing, but that's not part of what is currently pending before the board. [00:03:54] Speaker 01: But that is why you do have a present inconsistency that only came into existence when the board issued its second decision. [00:04:01] Speaker 00: Can I ask about the third issue though, because there's also an arguable inconsistency in terms of the board having found a motivation to combine these two references in the other proceedings. [00:04:11] Speaker 00: That's right. [00:04:12] Speaker 00: But not here. [00:04:13] Speaker 00: But is that a record problem? [00:04:16] Speaker 00: I mean, they said that there's substantial evidence to show that it would have been beneficial. [00:04:21] Speaker 00: And if that evidence was missing, and the question is, was it missing here? [00:04:25] Speaker 00: Was it different here? [00:04:26] Speaker 00: Was there a different record here in which it could justify reaching a different result? [00:04:30] Speaker 01: Yeah, so two points, Judge Prost. [00:04:36] Speaker 01: I mean, first of all, I don't think that there's anything that's missing from the record here that was in the record there in terms of the combination. [00:04:46] Speaker 01: Now, just to be clear, the combination that we're talking about goes to claims 10 and 11, and it goes to the temperature on a chip and whether or not you would be motivated to combine these. [00:04:56] Speaker 01: And I'm happy to rest on our briefs on claim 10 that also affects [00:05:00] Speaker 01: claim 11. [00:05:02] Speaker 01: But I don't think that there is a difference in the records that is relevant to this. [00:05:08] Speaker 01: I think that, again, you just have the board [00:05:10] Speaker 01: you know, saying one thing in one case, where in one case it specifically says, like, quote, as Petitioner points out, Nicole expressly contemplates that temperature tracking would be desirable in its power management system. [00:05:23] Speaker 01: You know, we credit the expert's logical, clear, reasoned testimony that a person of ordinary scale in the art would have been motivated to consult the Boccia reference and implement the low activity states in Nicole's system. [00:05:35] Speaker 01: That's page 24. [00:05:36] Speaker 01: of the 535 decision. [00:05:39] Speaker 01: And I don't see how that you reconcile that with the board here seeming to think that our expert didn't address the issue at all, because it was essentially the same issue. [00:05:49] Speaker 00: OK. [00:05:49] Speaker 00: So as you started to acknowledge that in each of these cases, we can go in different directions, even if we largely or partly agree with your argument. [00:05:58] Speaker 00: And in this case, on the two issues you want to focus on, the different patents, we could say there's an inconsistency between what the board said here and what the board said there. [00:06:08] Speaker 00: That, at a minimum, requires that the board explain how it could reach two different results. [00:06:14] Speaker 00: And if it can't reach two different results, well, [00:06:17] Speaker 00: something's got to give, right? [00:06:18] Speaker 00: Correct, Judge Prost. [00:06:20] Speaker 00: Or we could alternatively say, we've got the record in front of us. [00:06:24] Speaker 00: It's a claim construction issue. [00:06:25] Speaker 00: We can use their analysis in the other cases as authority or something to look at for the basis for reaching the claim construction. [00:06:34] Speaker 00: And just go ahead and decide it here. [00:06:35] Speaker 01: You absolutely can. [00:06:36] Speaker 01: And that is certainly true in terms of the issue of, [00:06:43] Speaker 01: according to a state. [00:06:45] Speaker 01: And I think you can just simply decide it straight up as a matter of claim construction. [00:06:49] Speaker 02: So if we agree with your argument that the board erred in the claim construction by narrowing it, does that render any dispute about whether there's evidence or not kind of null? [00:07:07] Speaker 01: You would moot the other issues that are presented if you agree with us on according to a state because because because they've added [00:07:16] Speaker 02: In your view, I think somewhat persuasively, an additional requirement that they isn't in the language of the claim and isn't in the claim construction, or maybe they didn't do a claim construction, but isn't in their application of the claim language in the other case. [00:07:33] Speaker 01: That's exactly right, Judge Hughes. [00:07:34] Speaker 02: You're asking that for us to reverse and find the broader claim construction, and then there's no dispute that these [00:07:41] Speaker 02: This is all disclosed. [00:07:42] Speaker 01: Yeah, that's absolutely right, Judge Hughes. [00:07:44] Speaker 01: And the reason that there's no dispute here, I mean, as the board said at Appendix 11, quote, PAC does not dispute that a processor load could be a state, as recited in Claims 2 through 4. [00:07:55] Speaker 01: So you have the board accepting that there is this causal relationship between load and the clock frequency. [00:08:05] Speaker 01: I mean, the board acknowledged this. [00:08:06] Speaker 01: For example, at Appendix 20, it said, [00:08:08] Speaker 01: that, quote, Nicole teaches the notion of adjusting operating frequency to load. [00:08:12] Speaker 01: The only thing that the board then did wrong after acknowledging this is saying, yes, but the load that is being used to adjust the clock frequency in Nicole, it doesn't count because it is an anticipated load. [00:08:25] Speaker 01: It is a future load as opposed to the load that is currently actually in the configuration registers [00:08:32] Speaker 01: being run at this moment in time. [00:08:34] Speaker 02: And that's wrong. [00:08:35] Speaker 02: Is the gist of this basically that in this case, the board is requiring some kind of individualized response to a load rather than a kind of pre-programmed response that still adjusts frequencies, but is kind of one size fits all or maybe two or three size fits all? [00:08:53] Speaker 01: So I would put it slightly differently, Judge Hughes. [00:08:56] Speaker 01: I think what the board is doing in this case is it's saying that the only load that matters [00:09:02] Speaker 01: is the load that is actually being currently executed in the processor at that moment in time. [00:09:09] Speaker 01: And it does this because it makes reference to the embodiment at the very bottom of column 5, where it talks about what is being currently run. [00:09:17] Speaker 01: And that ignores what's being explained at the top of column 3, lines 1 through 15, where the 047 specification very clearly explains that [00:09:28] Speaker 01: You can pre-select. [00:09:29] Speaker 01: That's the patent's words. [00:09:30] Speaker 01: It uses it over and over in column two and column three and column four. [00:09:34] Speaker 01: Pre-select clock frequency. [00:09:35] Speaker 01: How? [00:09:36] Speaker 01: By pre-selecting the configuration state. [00:09:39] Speaker 01: And we know the configuration state can be a load, or a load can be a configuration state. [00:09:43] Speaker 01: The only question is, can it be pre-selected? [00:09:47] Speaker 01: As opposed to, does it have to be the load that is actually in the registers the way that the board seemed to think? [00:09:53] Speaker 01: at appendix 15. [00:09:54] Speaker 01: And column 3 puts the lie to that. [00:09:57] Speaker 01: And if there were any doubt about it, if you look at column 4, line 17 to 19, it then says, quote, referring back to the things in column 3, clocking may be preselected together with the configuration. [00:10:10] Speaker 01: So you're preselecting configuration. [00:10:12] Speaker 01: And it says, quote, the state is then the configuration state and or is at least determined by it. [00:10:19] Speaker 01: So that state, [00:10:20] Speaker 01: is a state that has been pre-selected. [00:10:23] Speaker 01: And that puts the lie to the way that the board tried to limit this to things that are only currently being run in the registers as opposed to things that could happen moments before. [00:10:32] Speaker 01: Because as my friend on the other side said in the prior argument, [00:10:36] Speaker 01: I mean, we're in the computing world, and things in the computing world are happening very quickly, one on top of the other. [00:10:42] Speaker 01: And that is exactly what both the 047 patent are addressing and Nicole is addressing as well. [00:10:49] Speaker 01: And Nicole fully teaches [00:10:51] Speaker 01: according to a state, because it is adjusting clock frequency based on the workload. [00:10:57] Speaker 01: And the workload that is assigned to a particular processor, that means that particular processor is in a different state than one that doesn't have a workload assigned to it. [00:11:08] Speaker 01: Let me turn to hysteresis characteristic, if I may, and specifically [00:11:14] Speaker 01: The board erred in its understanding of hysteresis in the context of computer programming, reading the claims as excluding what the board referred to as deliberate predetermined choices and program the computer system. [00:11:26] Speaker 01: And it also conflated a hysteresis characteristic with hysteresis itself. [00:11:30] Speaker 01: And to be clear, the board agreed that Nicole in its figure three shows a non-linear time lag delay. [00:11:38] Speaker 01: You can see that in appendix 50. [00:11:40] Speaker 01: And that is a classic characteristic of hysteresis, having a non-linear time lag delay. [00:11:46] Speaker 01: That is literally the definition. [00:11:48] Speaker 01: Hysteresis is derived from the Greek meaning that it is lagging behind. [00:11:55] Speaker 02: And that is what the- Well, just because your time is short did jump ahead. [00:12:01] Speaker 02: My understanding is the board rejected that because Nicole was programmed to do that rather than [00:12:09] Speaker 02: kind of, again, I'm going to say it the wrong way, probably, but an individualized hysteresis response. [00:12:16] Speaker 02: But I assume you disagree with that in terms of the claim doesn't require it. [00:12:21] Speaker 02: But my understanding is that in a different IPR, they accepted that the hysteresis characteristic could be programmed. [00:12:31] Speaker 01: Yeah, that's absolutely right. [00:12:33] Speaker 01: I mean, in the 5-4-1 decision, which is final, it's not pending on rehearing, you had a different reference called Cling, where you did have hysteresis that was deliberately programmed into the system. [00:12:45] Speaker 02: I mean, this is not as clear as the other ones, where it's the exact same references. [00:12:51] Speaker 02: So is there anything in Cling that would teach more about this, as to why you would be [00:12:57] Speaker 02: you could have a programmed hysteresis, whereas Nicole does it? [00:13:02] Speaker 01: I don't think so, Judge Hughes. [00:13:03] Speaker 01: I mean, there is a difference with Kling. [00:13:04] Speaker 01: I mean, to be clear, what Kling is teaching is hysteresis using bands. [00:13:09] Speaker 01: So you can accomplish hysteresis in different ways. [00:13:12] Speaker 01: The end goal is to have a lag in the system so that you are not constantly fluctuating on off, on off, on off. [00:13:19] Speaker 01: And that's exactly what Nicole itself specifically teaches. [00:13:24] Speaker 01: you know if that is why Nicole builds in the step functions so that because Data and workload is constantly coming in instead of jumping from 70 to 140 and then back down to 70 you're taking steps so that if the workload changes instead of having jumped from 70 to 140 and [00:13:45] Speaker 01: You know, you only have to go from, say, 120 to 120. [00:13:47] Speaker 00: I find all this confusing, but I thought it was helpful. [00:13:50] Speaker 00: And I don't remember if either or both of you used the thermostat example. [00:13:54] Speaker 01: So to take the thermostat example, the way that thermostats work is they can work in a couple of ways. [00:14:00] Speaker 01: But one, in all cases, [00:14:02] Speaker 01: It is deliberate. [00:14:03] Speaker 01: It's predetermined. [00:14:05] Speaker 00: And so is it your view that, just to follow up with Judge Hughes' question, that is it clear that in the 541 proceeding, the thermostat would clearly fit within one of those? [00:14:17] Speaker 00: For sure. [00:14:17] Speaker 00: And it would not? [00:14:18] Speaker 00: And what the board said, it would not apply here? [00:14:21] Speaker 01: Yeah. [00:14:22] Speaker 01: I think the thermostat analogy indisputably applies in the 541 proceeding. [00:14:27] Speaker 01: The board seems to take issue with whether the thermostat analogy applies here. [00:14:31] Speaker 01: Let me explain very briefly why it does, because I think the board really fundamentally misunderstands what is going on here. [00:14:39] Speaker 01: You have an operating system that reacts to variations in the system load. [00:14:44] Speaker 01: Nicole tells you that at column five, lines 23 to 26. [00:14:48] Speaker 01: And so the variation in load is like the variation in temperature. [00:14:52] Speaker 01: And what will happen is that when there are changes in the load, [00:15:00] Speaker 01: the system is designed that you then change the clock frequency. [00:15:04] Speaker 01: But just like when there's a change in the temperature, you don't do it as soon as the temperature occurs. [00:15:09] Speaker 01: You build in a threshold. [00:15:11] Speaker 01: In this case, it also builds in a delay. [00:15:16] Speaker 01: And the way it builds it in is instead of immediately turning on or immediately turning off, just like with the thermostat, instead of immediately turning on or off, you wait till it hits a certain temperature band. [00:15:27] Speaker 01: Here, instead of immediately turning on or off, [00:15:29] Speaker 01: It does it in steps. [00:15:31] Speaker 01: It does it in steps. [00:15:32] Speaker 01: And the reason for that is so that if a change in the load occurs, so that it either needs to go back up or come back down, it doesn't have to be oscillating on, off, on, off, which is exactly what Nicole teaches that you want to avoid, because that prevents noise in the system and possible circuit failure. [00:15:53] Speaker 01: That's appendix 1556. [00:15:55] Speaker 01: Column 5, lines 30 to 33. [00:15:58] Speaker 01: I know that I'm past my time and past my rebuttal, but I'm happy to continue answering questions on this, because I do think that it's an important point where the board both erred in the way it construed hysteresis in expecting that you had to have hysteresis as opposed to something that just simply a characteristic of hysteresis. [00:16:16] Speaker 01: And I think it also misunderstood what it is that Nicole is actually doing. [00:16:20] Speaker 01: And I'm helping to try to answer further questions about that if it would be helpful. [00:16:24] Speaker 04: Anything else that does fine? [00:16:26] Speaker 04: Yeah. [00:16:27] Speaker 04: OK, thank you. [00:16:28] Speaker 04: We'll send you a rebuttal, Mr. Quinn. [00:16:29] Speaker 01: Thank you, Judge Newman. [00:16:31] Speaker 04: Mr. Weister? [00:16:32] Speaker 00: Thank you, Your Honor. [00:16:33] Speaker 00: May I please the court? [00:16:34] Speaker 00: Let me just say, you're pulling heavy duty today. [00:16:36] Speaker 00: I'm doing my best. [00:16:37] Speaker 00: Thanks. [00:16:37] Speaker 00: Thank you. [00:16:38] Speaker 00: It's a division of labor for you. [00:16:39] Speaker 00: Thank you. [00:16:40] Speaker 03: I'd like to start with the 535 decision, because it came up, and it's an important point. [00:16:45] Speaker 03: So first point, the 535 decision came out about a month before the decision in this board. [00:16:51] Speaker 03: Intel did not bother to bring it to the court's attention, to the board's attention. [00:16:55] Speaker 03: There was a mechanism for it to do so. [00:16:57] Speaker 03: And the board- I don't understand. [00:16:59] Speaker 00: I mean, well, or you could have brought it to their attention too, right? [00:17:02] Speaker 00: I mean, why would Intel's requiring obligation be any higher than yours? [00:17:07] Speaker 03: Certainly. [00:17:08] Speaker 03: We could have, but they are the ones who are relying on it for the first time in appeal. [00:17:12] Speaker 03: And my point, just briefly, because I want to address the merits of it, is that they could have given the board an opportunity to understand their argument and to address it. [00:17:21] Speaker 02: Is there no overlap between the boards on this? [00:17:23] Speaker 03: There was an overlap. [00:17:26] Speaker 02: I mean, I get it. [00:17:28] Speaker 02: Maybe it would have been better for them to bring it to their sentient. [00:17:30] Speaker 02: But how can the board, when they have an overlap of judges, [00:17:35] Speaker 02: be complying with the APA by issuing inconsistent decisions. [00:17:44] Speaker 03: Our adversary in his reply brief has a chart saying, well, there were two very similar arguments, and they were indeed similar. [00:17:51] Speaker 03: I acknowledge that. [00:17:53] Speaker 03: But the key difference, which the board did not have an opportunity to address, or perhaps could have done it to Esfanti, but the key difference, absolutely key, is the claim language. [00:18:02] Speaker 03: The claim language here is not just according to a state. [00:18:05] Speaker 03: It's adjustable at runtime according to a state. [00:18:09] Speaker 03: That claim language was indeed argued in this particular proceeding. [00:18:14] Speaker 03: It was not available by PAC. [00:18:15] Speaker 02: Where in the board's decision does it base this argument on at runtime? [00:18:20] Speaker 03: The board quotes the words at runtime when it's talking about, but your honor, respectfully, [00:18:25] Speaker 03: Because it's claim construction, it's something this court addresses de novo. [00:18:30] Speaker 03: It's not subject to the Chenery Act. [00:18:31] Speaker 02: I understand, but we're talking about the inconsistencies between the decision. [00:18:35] Speaker 02: Yes. [00:18:35] Speaker 02: And you're pointing out differences in claim language. [00:18:38] Speaker 02: I didn't see the board relying on that difference in claim language anywhere to explain why it was reaching a different conclusion. [00:18:46] Speaker 02: So at the minimum, this would have to go back for them to explain [00:18:51] Speaker 03: This. [00:18:52] Speaker 03: That is an option. [00:18:54] Speaker 03: But because this is de novo. [00:18:56] Speaker 02: But their argument, their claim construction is not based anywhere, as I can tell, on that at runtime language. [00:19:04] Speaker 02: It's based on their definition. [00:19:06] Speaker 02: And I mean, obviously, you've read this more closely than I have, I hope. [00:19:10] Speaker 02: But it's based entirely on their definition of a court date. [00:19:16] Speaker 03: I respect. [00:19:17] Speaker 00: Would you care to speculate how this happened? [00:19:20] Speaker 03: The two board decisions? [00:19:21] Speaker 00: Yeah. [00:19:22] Speaker 00: I mean, with an overlapping board. [00:19:24] Speaker 03: Well, I was. [00:19:26] Speaker 00: I mean, one of several things. [00:19:28] Speaker 00: I mean, either they didn't notice, which is hard to believe, or they thought the two were easily reconcilable, so they didn't have to say anything about it. [00:19:37] Speaker 00: But I think you're even kind of not trying to make that argument. [00:19:42] Speaker 03: I actually am trying to make the argument that they're reconcilable and that this court, because it's de novo in either of those proceedings, can do the reconciling. [00:19:51] Speaker 03: And to Judge Hughes' point, did the board focus on at runtime [00:19:55] Speaker 03: What the board did so, but not in those exact words, and I'd refer the court to appendix 13, because it's not just about these pre-selecting states that we heard Mr. O'Quinn talking about. [00:20:06] Speaker 03: What the board talks about on appendix 13, and this is the first full paragraph, is besides a configuration state, the 047 patent refers to other factors that the VPU, that's the processor, may use to set the clock frequency. [00:20:20] Speaker 03: And among those is temperature, drop in supply voltage, and condition of a power source. [00:20:25] Speaker 03: There is this aspect of pre-selecting, but the point of this patent and the reason why the claim language is so important and is different from the 535 decision is that you have to look at other factors that are also part of state. [00:20:38] Speaker 03: Those include temperature, for example. [00:20:41] Speaker 03: And when do you know those? [00:20:43] Speaker 03: You don't know what the temperature is to use to set the clock frequency until runtime. [00:20:48] Speaker 03: So Judge Hughes, did the board use at runtime? [00:20:52] Speaker 02: I understand what you're arguing about. [00:20:53] Speaker 02: Can we, instead of looking at 13, look at 12? [00:20:57] Speaker 02: Because this is what I remember from reading this, is that their sole focus is [00:21:05] Speaker 02: the language according to. [00:21:06] Speaker 02: There's no discussion in this of at runtime. [00:21:09] Speaker 02: They can quote it. [00:21:10] Speaker 02: But then in the middle paragraph here, we agree with Pac that Intel's analysis fails to consider the languages according to. [00:21:19] Speaker 02: Well, that language was in the other case, wasn't it? [00:21:22] Speaker 03: It certainly was. [00:21:23] Speaker 03: I don't disagree with that. [00:21:25] Speaker 02: And that's the language they're relying on to come up with a different construction. [00:21:30] Speaker 03: But I believe that the discussion at appendix 13 is relevant, because what states are they talking about when they're talking about according to a state? [00:21:39] Speaker 03: Certain of those states, which are expressed in this collective word state, include things like temperature and power consumption that are known only at runtime. [00:21:48] Speaker 03: So I agree they didn't use the words at runtime, but they referred to certain things that are only known at runtime. [00:21:53] Speaker 03: And so indirectly, they were relying on that. [00:21:56] Speaker 03: And even if they weren't, though, [00:21:58] Speaker 03: Again, I think it's important. [00:22:00] Speaker 03: Yes, this court's reviewing the board, but this is a claim construction issue. [00:22:05] Speaker 03: It's reviewed de novo, and this court can consider anything in the intrinsic evidence as well as the board's decision and come to its own interpretation. [00:22:12] Speaker 03: And I'd like to address, beyond just the fact that temperature is known only at runtime, why there are additional reasons in support of the board's construction here. [00:22:20] Speaker 03: Now, we talk about the sort of way the claim is phrased. [00:22:24] Speaker 03: To Judge Hughes' point, according to a state implies causation, right? [00:22:29] Speaker 03: First, you're going to determine what the state is, and then you're going to set. [00:22:32] Speaker 03: Something has to come first, and something has to come second. [00:22:34] Speaker 03: The thing that comes second is the clock frequency, based on that plain language. [00:22:38] Speaker 03: We also have this argument that my adversary makes about embodiments. [00:22:42] Speaker 03: And it says, well, there's an embodiment. [00:22:44] Speaker 03: I forget if it's column five or another column. [00:22:47] Speaker 03: There's an embodiment that talks just about preselection. [00:22:50] Speaker 03: PACS interpretation, according to Intel, does not deal with that embodiment, then PACS construction must be wrong. [00:22:57] Speaker 03: But that's wrong as a matter of this court's case law. [00:23:00] Speaker 03: They rely on the OD case. [00:23:01] Speaker 03: That's O-A-T-E-Y. [00:23:03] Speaker 03: But this court has subsequently explained distinguishing OD, and among other cases, the PSN case, 525, F3 at 1166, that all claims need not cover all embodiments. [00:23:14] Speaker 03: And so yes, there is an embodiment where you preselect the clock frequency and you kind of leave it alone and all that happens at runtime is you just execute. [00:23:22] Speaker 03: That does exist. [00:23:24] Speaker 02: Is there other claims in this patent that would have covered that? [00:23:28] Speaker 03: Embodiment there there are other well, so we're talking Specifically on these claims is two through four so claim one for example I believe would cover that situation two through four have the language adjustable at runtime according to a state of and then there's three different possibilities so [00:23:45] Speaker 03: Other claims would potentially cover that, but the point that I want to make is that other embodiments, and again, this is Appendix 13 of the board's decision referring to Appendix 104, include things like temperature and power consumption that are only known at run time. [00:24:00] Speaker 03: So you have that sort of notion from the other embodiments, and claims two through four need not encompass all of the embodiments in the patent. [00:24:09] Speaker 03: Finally, I'd like to reference the prosecution history, Appendix 1452. [00:24:13] Speaker 02: Well, I get that point, but just to follow up. [00:24:15] Speaker 02: And I guess if I look at claim one, I'm going to be convinced that it does cover this embodiment that they say is excluded. [00:24:22] Speaker 02: Because you're exactly right that every claim doesn't have to cover every embodiment. [00:24:27] Speaker 02: But we don't read embodiments out entirely unless it's very clear. [00:24:33] Speaker 02: And so you don't get a say. [00:24:37] Speaker 02: you know, this language that would read out an embodiment unless some other claim might cover it. [00:24:46] Speaker 03: Yes, but I do think it is covered by one, because one doesn't have according to a state. [00:24:51] Speaker 03: We'll see what your friend says. [00:24:53] Speaker 03: Prosecution history briefly, 1452 of the appendix. [00:24:56] Speaker 03: It talks about processing to do. [00:24:58] Speaker 03: And our adversary says things like, well, processing to do means future, so it means expected state. [00:25:03] Speaker 03: But what that prosecution history also says, and this is back to my temperature point, is if a processor is slower or faster for any reason, [00:25:13] Speaker 03: And that brings in, again, these things like temperature that are known only at runtime. [00:25:17] Speaker 03: That's what this patent is about. [00:25:18] Speaker 03: It's about taking into account all these factors that exist. [00:25:21] Speaker 02: I mean, you're doing a really good job of rewriting the board's decision to come up with more convincing explanations than what the board gave. [00:25:28] Speaker 02: I don't see any of this in the board's decision. [00:25:31] Speaker 03: So if I go back to appendix 13, the board's paragraph beginning, besides a configuration state, the 047 pattern refers to other factors. [00:25:40] Speaker 03: All of those other factors are ones that would known only at runtime. [00:25:42] Speaker 03: And I'll list them again. [00:25:43] Speaker 03: Condition of a power source, drop in supply voltage, responding to a temperature sensor signal, with citations to the specifications. [00:25:50] Speaker 03: So I do believe this is consistent with the board. [00:25:53] Speaker 03: I've talked about the 535 decision. [00:25:56] Speaker 03: Let me briefly talk about hysteresis. [00:25:58] Speaker 03: Number one. [00:26:00] Speaker 03: At the board, Intel didn't make this distinction between, well, it just has to some way resemble hysteresis. [00:26:05] Speaker 03: What they argued at appendix 209, which is from their petition, is that Nikol uses hysteresis. [00:26:13] Speaker 03: Not that it's in some way consistent with it. [00:26:15] Speaker 03: It uses hysteresis. [00:26:16] Speaker 03: And the board explained here why that's not the case. [00:26:20] Speaker 03: And I think, Judge Hughes, you summarized that, but I'll try again, which is basically that Nikol is not having, there's no response. [00:26:30] Speaker 03: You're pre-programming everything. [00:26:32] Speaker 03: You're not setting some parameters like a thermostat and letting it respond. [00:26:35] Speaker 03: You're saying exactly what is load going to be. [00:26:38] Speaker 03: It's going to go from 70 to 140, back to 100. [00:26:42] Speaker 03: It's one step, by the way, not multiple steps. [00:26:45] Speaker 03: You're not letting it fluctuate based on parameters. [00:26:48] Speaker 03: By contrast, in the other proceeding, the 541 decision, different reference, cling, that does look like a thermostat. [00:26:56] Speaker 03: It involves not temperature, but power consumption. [00:26:59] Speaker 03: And it's got these two thresholds. [00:27:00] Speaker 03: If you hit the high threshold, it's consuming too much power. [00:27:03] Speaker 03: They throttle it. [00:27:04] Speaker 03: They send a signal to slow it down. [00:27:06] Speaker 03: But they don't say exactly what the power consumption is. [00:27:09] Speaker 03: They just slow down the processing. [00:27:11] Speaker 03: And it falls naturally to the lower threshold. [00:27:14] Speaker 03: and then it can be increased again. [00:27:15] Speaker 03: That's like the thermostat. [00:27:17] Speaker 03: They did not rely on Kling in this petition. [00:27:19] Speaker 03: There's no dispute about that. [00:27:20] Speaker 03: They relied on Nicole. [00:27:21] Speaker 03: Also, by the way, Kling refers to hysteresis. [00:27:24] Speaker 03: And that's at page 10 of the appendix 3543. [00:27:29] Speaker 03: Uses the word history since Nicole did not. [00:27:31] Speaker 02: I understand what you're arguing again. [00:27:33] Speaker 02: But when I read the board's decision on this the basis it seemed to find Nicole insufficient is that Nicole was a programmed hysteresis requirement. [00:27:45] Speaker 02: But isn't it true that cling also is programmed? [00:27:49] Speaker 03: But what cling leaves not exactly set out in advance is power consumption by the processor. [00:27:55] Speaker 03: So Kling says, I'm going to slow you down. [00:27:58] Speaker 02: Again, the board, one, they didn't address any of this stuff. [00:28:01] Speaker 02: But two, this is not the distinction they relied on about this power stuff and stuff you're talking about. [00:28:07] Speaker 02: They just relied and said, [00:28:10] Speaker 02: Nicole is programmed, which doesn't comply with the Hershey's-Therese effect. [00:28:16] Speaker 03: Respectfully, I don't think the board in this case had to address Kling, which wasn't a reference. [00:28:21] Speaker 03: But in terms of what the board said about Nicole, the board said there's nothing that's reacting to anything. [00:28:29] Speaker 03: There's not power consumption. [00:28:30] Speaker 03: There's not temperature. [00:28:31] Speaker 03: It's not reacting to some thermostat programming. [00:28:34] Speaker 03: There's no reaction. [00:28:37] Speaker 03: It's not so much it's not programmed. [00:28:40] Speaker 02: Can we look at what you're saying the board said on this point? [00:28:43] Speaker 02: Because again, you're giving me a lot more information than I read from what the board said. [00:28:57] Speaker 04: Well, while you're looking, [00:28:57] Speaker 04: To add to Judge Hughes' comment, a lot more information than is in the claims. [00:29:03] Speaker 04: Was there any proposal as these issues and distinctions and refinements evolved in the debate before the board? [00:29:13] Speaker 04: Is there any attempt to put these possibly significant dispositive distinctions in the claims, either by amending the claims rather than just arguing [00:29:27] Speaker 04: saying well this is how you can read it and this is what we propose. [00:29:33] Speaker 03: So on the question of adjustable at runtime according to a state, there were no formal proposed claim constructions. [00:29:40] Speaker 03: But the claim construction dispute did arise and was addressed as such by the board. [00:29:46] Speaker 03: And at the hearing, the board asked about the word state, asked both parties what it meant, appendix 736, appendix 739. [00:29:53] Speaker 03: So there was an opportunity to address that question. [00:29:55] Speaker 02: Did they ask about at runtime? [00:29:57] Speaker 02: Because that seems to be where you're hanging your hat this morning, is at runtime. [00:30:03] Speaker 03: Well, they asked, I mean, I think, again, I guess the question is, were the words at runtime used in the board's question? [00:30:12] Speaker 03: I'm not sure I'd have to look back at the transcript, but there was an opportunity to address that claim language. [00:30:17] Speaker 03: And so it wasn't, to the APA question, this wasn't a situation where there was no opportunity to address the claim construction issue. [00:30:24] Speaker 03: On hysteresis, no claim construction issue, because the board used the very definition that was proposed by Intel for hysteresis. [00:30:33] Speaker 03: returning to Judge Hughes' question. [00:30:35] Speaker 02: The claim construction proposed by Intel didn't say anything about whether it could be programmed or not. [00:30:42] Speaker 02: That's the thing. [00:30:43] Speaker 02: When I looked at this board's decision, and yeah, maybe their expert made a more detailed argument than he should have or she should have, but the board's distinction wasn't based on that. [00:30:56] Speaker 02: It was based on the fact that Nicole was programmed [00:31:01] Speaker 02: I'm pretty sure they used the word program, didn't they? [00:31:03] Speaker 03: I think they used the word program, but if I could point to the language in the board's decision that I do think makes the point that I'm trying to make today. [00:31:15] Speaker 03: It's not whether it's- Where are we? [00:31:17] Speaker 03: We're on Appendix 50, judges. [00:31:20] Speaker 03: In the middle of the page, it's the second sentence, the first full paragraph. [00:31:25] Speaker 03: There's this language. [00:31:26] Speaker 03: We agree with PACT that this delay is not a non-linear response by the system, but a deliberate predetermined choice. [00:31:34] Speaker 03: So it doesn't use the word program in that sense. [00:31:39] Speaker 03: But the point we agree with is that part of hysteresis [00:31:42] Speaker 03: Any use of that word is that there's some reaction you're not predetermining all of and I think you could make a distinction between inputs and outputs Okay, so you're in Nicole. [00:31:54] Speaker 03: He's predetermining all of the outputs. [00:31:56] Speaker 03: Nothing is reacting to anything else and [00:31:58] Speaker 03: By contrast, in Kling and in the thermostat example, you're setting some programming, like when is the heater going to click on? [00:32:06] Speaker 03: When's it going to click off? [00:32:07] Speaker 03: Here are some thresholds that we're going to use. [00:32:09] Speaker 03: But you're not dictating in advance the exact temperature of the room. [00:32:13] Speaker 03: That is left sort of react to what you have programmed. [00:32:17] Speaker 03: That is programmed hysteresis. [00:32:18] Speaker 03: That would count as hysteresis. [00:32:20] Speaker 03: That's not what nickel does. [00:32:21] Speaker 03: It is what Kling does. [00:32:22] Speaker 03: And it is what our patent does. [00:32:25] Speaker 03: Finally, very briefly, on the temperature question, the board did say and had substantial evidence to say that the Intel expert had not, he had not explained why you would combine the temperature sensors in a system of Batia with [00:32:43] Speaker 03: plurality of regions on a single chip. [00:32:45] Speaker 03: And he did not rely just on the fact that neither reference put that all together, but he also said that the expert did not support why you would be motivated to combine. [00:32:52] Speaker 03: And in the TQ Delta case, this court has said that the board is a- In the other case where they did find the combination, was the expert testimony any different? [00:33:02] Speaker 03: The testimony was similar, but they cited more of it in that case. [00:33:08] Speaker 03: In our case, the board, I think it's appendix 2772 was the paragraph that was relied on here. [00:33:17] Speaker 02: If it was this insufficiency of evidence, I get it. [00:33:20] Speaker 02: If it's the same testimony or largely the same testimony, how can it support two different conclusions just because the board looks at some of it in some case and not some of it in the other case? [00:33:32] Speaker 03: I don't think it's a question of the board looking at some or others. [00:33:34] Speaker 03: I think we have two different records, and it's incumbent upon Intel as the petitioner to point to what evidence the board should consider. [00:33:41] Speaker 03: And it did not point to all of the same evidence in this proceeding. [00:33:45] Speaker 03: The court has no further questions. [00:33:47] Speaker 03: OK. [00:33:47] Speaker 04: Thank you. [00:33:51] Speaker 04: Mr. O'Quinn, you have your rebuttal. [00:33:57] Speaker 02: I'm most interested in you responding to him on the hysteresis stuff. [00:34:02] Speaker 02: You can respond on other stuff, too. [00:34:05] Speaker 01: I'm happy to start with that one very quick point that you had also asked about, Judge Hughes. [00:34:11] Speaker 01: Does claim one cover the state stuff? [00:34:13] Speaker 01: And my answer to that is if you look at claims two, three, and four and their analogs, those are the only claims in the patent that use the term state. [00:34:22] Speaker 01: Claim one doesn't use the term state. [00:34:23] Speaker 01: None of the other claims use the term state. [00:34:27] Speaker 01: column three is where it's describing what a state is. [00:34:30] Speaker 01: If I may, let me turn to two history systems, unless there are more questions on that. [00:34:34] Speaker 01: I mean, the board's decision does turn entirely on its idea that what you have is a deliberate predetermined choice. [00:34:42] Speaker 01: And again, just like the temperature in the room is not predetermined to know what the reaction of the thermostat is going to be, the load in the system is not predetermined. [00:34:55] Speaker 01: And the whole point [00:34:56] Speaker 01: of Nicole and frankly of the 047 itself are to react to variations to system load in real time. [00:35:05] Speaker 01: And so what he is essentially saying is that if instead of the temperature delay being built in, in a thermostat example, instead of it being built in by bands, like you want it at 70, so you set 69 and 71, instead you just built in a time delay. [00:35:19] Speaker 01: It's 70, and if the temperature goes up and stays up for more than two minutes, then I'm going to turn on the air conditioner. [00:35:25] Speaker 01: And if it comes down and stays down for more than two minutes, then I'm going to turn off the air conditioner. [00:35:30] Speaker 01: That is exactly what Nicole is disclosing at figure three, with the slight difference being [00:35:36] Speaker 01: instead of it delaying turning on, figure three shows you that it is proceeding in steps and the reason to do that is to reduce wear and tear on the system from constantly turning it off and on so that instead you're taking it stepwise so that if something countermanding in the load comes in just like if the temperature in the room changes then [00:36:01] Speaker 01: Instead of it not turning on or, you know, instead of continuing to ramp up to 140, it can ramp back down to 170 from somewhere in between and vice versa. [00:36:12] Speaker 01: And that is the classic understanding. [00:36:15] Speaker 01: This sort of lag is the classic understanding of what hysteresis is. [00:36:19] Speaker 01: And the board didn't seem to disagree with that. [00:36:23] Speaker 01: The board seemed to accept that you would otherwise have hysteresis here, except for the fact that it was deliberate and predetermined. [00:36:31] Speaker 01: And the board, of course, did not rely on that rationale in the 5-4-1 proceeding, where again, it is deliberate on what is going to happen when it reaches a particular threshold. [00:36:43] Speaker 01: Our points below, just to be clear, the arguments we made below didn't turn on whether it was deliberate or predetermined. [00:36:53] Speaker 01: And what the board here is doing is engaging in acclaimed instruction to read out things that otherwise are classic characteristics of hysteresis, with really no explanation for doing so that is at least consistent with what it did [00:37:10] Speaker 01: in the other proceeding. [00:37:12] Speaker 01: I'm happy to answer additional questions about this point. [00:37:15] Speaker 01: I think the court can just simply outright reverse by correcting the board from its [00:37:21] Speaker 01: limiting hysteresis characteristic to, in its words, non-deliberate or non-predetermined hysteresis characteristics, I think you could thought out reverse. [00:37:32] Speaker 01: But at a minimum, I think given the inconsistency, the court should at least remand. [00:37:37] Speaker 01: If there were any other questions about adjustable at runtime, they spent a lot of time on that. [00:37:41] Speaker 01: I realize I'm past my time. [00:37:43] Speaker 01: If I may, let me make one point briefly on that. [00:37:46] Speaker 01: And that is this. [00:37:47] Speaker 01: The term adjustable at runtime can't carry the weight that they're placing on it. [00:37:50] Speaker 01: That term exists. [00:37:51] Speaker 01: in all of the claims. [00:37:53] Speaker 01: It's in claim one. [00:37:54] Speaker 01: What it's talking about being adjusted at runtime is the clock frequency. [00:37:59] Speaker 01: It doesn't say adjustable according to a state at runtime. [00:38:04] Speaker 01: That's the wrong anesthetics, not the antecedent. [00:38:06] Speaker 01: It says the clock frequency is adjustable at runtime according to a state. [00:38:10] Speaker 01: And what they want to do is to limit it to a subset of states and to exclude other states. [00:38:16] Speaker 01: Our position doesn't exclude the states that are currently running. [00:38:20] Speaker 01: It just includes things that are in the queue that are being prepared to be run. [00:38:24] Speaker 01: Because again, this is all happening very fast in the computing environment. [00:38:28] Speaker 01: I'm happy to answer any additional questions. [00:38:30] Speaker 01: I thank the court for its indulgence. [00:38:34] Speaker 04: Any more questions? [00:38:34] Speaker 04: I think we have it. [00:38:35] Speaker 04: Our thanks to both counsel. [00:38:37] Speaker 04: The case is taken under submission. [00:38:40] Speaker 04: And that concludes this panel's arguments for today. [00:38:43] Speaker 01: Thank you, Judge Neiman.