[00:00:00] Speaker 05:
The next case is VLSI Technology versus Intel, 2023-12-66.

[00:00:07] Speaker 05:
This is Vaquili.

[00:00:11] Speaker 01:
Good morning, Your Honors.

[00:00:12] Speaker 01:
May it please the Court, Cameron Vaquili of IRL of Manila, on behalf of VLSI Technology, the appellant and owner of U.S.

[00:00:19] Speaker 01:
Patent 8020-014.

[00:00:22] Speaker 01:
The board's findings of obviousness for the challenged claims of the 014 patents suffered from several critical deficiencies whose undisputed character precludes meeting explicit limitations recited in the claim.

[00:00:34] Speaker 01:
In addition, I intend to discuss how the board relied on a disjointed reading of the prior reference, who, guided by impermissible hindsight, and failed to fulfill its burden to show either motivation or expectation of success.

[00:00:47] Speaker 01:
I think the clearest shortfall of the board's finding is its reliance on the so-called L2 access ratio from WHO for the claimed relationship between an estimated power gain and estimated power loss resulting from powering down.

[00:01:01] Speaker 01:
The key facts are undisputed.

[00:01:04] Speaker 01:
One, the board alleged L2 access to be the claimed estimated power loss resulting from powering down.

[00:01:10] Speaker 01:
Not L2 access plus some other thing, but L2 access.

[00:01:14] Speaker 01:
Two, the board alleged leak to be the claimed estimated power gain resulting from powering down.

[00:01:21] Speaker 01:
Three, L2 access is the energy per access to L2 memory.

[00:01:27] Speaker 01:
And four, leak is the energy per cycle.

[00:01:32] Speaker 01:
Now, if I could use an analogy, you can imagine that I sell hamburgers.

[00:01:36] Speaker 01:
I'm McDonald's or Burger King.

[00:01:38] Speaker 01:
And I want to know how much revenue am I going to lose if I shut down a location and how much cost am I going to gain?

[00:01:44] Speaker 01:
What the board is telling me is that all I need to do to figure out how much revenue I'm going to lose is estimate the price for hamburger.

[00:01:51] Speaker 01:
Well, that's clearly not enough.

[00:01:52] Speaker 01:
I need to know how many hamburgers I sell at the location.

[00:01:55] Speaker 01:
One hamburger, a thousand hamburgers a day, it's a big difference.

[00:01:58] Speaker 01:
And similarly, the board is telling me all I need to know is the cost per hour to operate the restaurant to figure out how much cost savings I'm going to have.

[00:02:08] Speaker 01:
Well, am I going to shut down for one hour, one day, five years?

[00:02:12] Speaker 01:
very different cost savings in those different scenarios.

[00:02:17] Speaker 01:
This is an exact analogy to the claim power gain and power loss resulting from powering down.

[00:02:23] Speaker 01:
This board, excuse me, this court actually acknowledged this reality in its prior May 26, 2021 decision in appeal 20-1744, where it stated, and I quote, in footnote one, the dynamic power cost of a cash miss varies with the number of operations required in response to the miss.

[00:02:43] Speaker 01:
The court said, more power will be consumed if data from the cash must be deleted.

[00:02:48] Speaker 01:
Less power will be used if the cash is not full.

[00:02:52] Speaker 01:
Well, L2 access doesn't tell you anything about how full or empty the cash is.

[00:02:56] Speaker 01:
It just says the energy per access.

[00:03:00] Speaker 01:
The ratio of the price of a hamburger to the cost per hour doesn't tell you how much revenue will be lost or cost gained if you shut down a restaurant.

[00:03:07] Speaker 01:
An exact analogy, the L2 access leak ratio doesn't tell you how much power is gained or power is lost, resulting from powering down a portion of the cash.

[00:03:16] Speaker 01:
It doesn't and it can't.

[00:03:20] Speaker 03:
It would appear that we decided that question against you in the first appeal.

[00:03:26] Speaker 01:
Well, I don't think that issue was

[00:03:29] Speaker 01:
addressed in the first appeal.

[00:03:30] Speaker 01:
The question there was more narrow.

[00:03:33] Speaker 01:
The question was the board's finding that L2 access leak was the basis of a determination of whether to power down.

[00:03:41] Speaker 03:
We said this satisfies the Clintonization of determining whether to power down.

[00:03:46] Speaker 01:
Exactly.

[00:03:47] Speaker 01:
Determining whether to power down, but not whether that was an estimated power gain or estimated power loss resulting from powering down.

[00:03:53] Speaker 01:
That's the question currently before the court.

[00:03:56] Speaker 01:
Now I agree, there's no dispute that this court has ruled that L2 access leak forms the basis of the determination of whether to power down.

[00:04:05] Speaker 01:
That's ironclad.

[00:04:06] Speaker 01:
But is L2 access leak a power resulting from powering down?

[00:04:10] Speaker 03:
It's part of the same claim limitation.

[00:04:11] Speaker 03:
I'm sorry?

[00:04:12] Speaker 03:
Part of the same claim limitation.

[00:04:14] Speaker 01:
Well, it's a long limitation.

[00:04:15] Speaker 01:
It has different parts.

[00:04:16] Speaker 01:
And there was a narrow issue that the board had decided.

[00:04:18] Speaker 03:
They had said that the L2 access lead ratio... When we said it satisfies the claim limitation, you'd say we only said it satisfied part of the claim limitation?

[00:04:27] Speaker 01:
Your Honor, I don't believe the court reached the full issue, because what had happened was the board had stated that L2 access leak was not used to determine whether to power down.

[00:04:37] Speaker 01:
That was the original final written decision.

[00:04:40] Speaker 01:
This court reversed on that point.

[00:04:41] Speaker 01:
That was the only thing before the court at the time.

[00:04:44] Speaker 01:
The question now is, does L2 access read on an estimated power loss resulting from powering down, and does the leak read on an estimated power gain?

[00:04:55] Speaker 01:
And the answer is no, it can't.

[00:04:57] Speaker 01:
it simply can't, and that was not before the court in the first appeal.

[00:05:00] Speaker 00:
Similarly... We did say previously that the L2 access represents the power cost of accessing the L2 cache when the L1 cache is powered down and a cache miss ensues.

[00:05:13] Speaker 00:
I mean, I recognize that that was, you know, in discussing the determining limitation, but how do you respond to that?

[00:05:21] Speaker 00:
Because that seems to be

[00:05:23] Speaker 00:
to be supported by substantial evidence.

[00:05:26] Speaker 00:
That seems to be correct to me, and it goes a little bit to the issue we're looking at here.

[00:05:31] Speaker 01:
Well, L2 access is certainly a factor that's related to the amount of power that might be consumed if you have a lot of what's called dirty bits.

[00:05:39] Speaker 01:
It's the energy that is required to access stable memory.

[00:05:45] Speaker 01:
It's relevant, but there is no information about how much dirty bits there are and who.

[00:05:51] Speaker 01:
And that issue was not before the court in the first appeal.

[00:05:54] Speaker 01:
It simply wasn't.

[00:05:55] Speaker 01:
In fact, Intel argued vociferously that that wasn't before the court at the time.

[00:05:58] Speaker 01:
And I don't believe the court ruled on that issue.

[00:06:02] Speaker 01:
Similarly with leak, leak is a separate factor where we don't know the amount of time that we're shutting down for it.

[00:06:08] Speaker 01:
It's just not there.

[00:06:08] Speaker 01:
It's not there in the reference.

[00:06:11] Speaker 01:
The board hasn't pointed to it.

[00:06:12] Speaker 01:
Intel hasn't pointed to it.

[00:06:14] Speaker 01:
So we believe that L2 access and leak cannot be the estimated power gain and power loss resulting from powering down, and that's an issue of first impression in this appeal.

[00:06:25] Speaker 03:
So just to understand correctly, what you're saying is when we said it satisfies the determining limitation,

[00:06:30] Speaker 03:
You're saying we only meant that it satisfied part of the determining limitation.

[00:06:36] Speaker 01:
Well, I believe that what the court... Is that correct?

[00:06:39] Speaker 03:
I mean, the determining limitation is determining whether the power to have at least a portion of the component responds to a relationship, et cetera, et cetera.

[00:06:48] Speaker 03:
We said it satisfies that limitation.

[00:06:50] Speaker 03:
You're arguing now it doesn't.

[00:06:52] Speaker 01:
Well, again, in my understanding, what the court was ruling on was whether L2 access leak was the basis of a determination of whether to power down.

[00:07:00] Speaker 01:
That's what the board had said it wasn't, and this court reversed and said that it was.

[00:07:05] Speaker 01:
But whether L2 access and leak are an estimated power gain and estimated power loss wasn't before the court in the prior appeal.

[00:07:12] Speaker 01:
This court itself, I believe, noted that.

[00:07:21] Speaker 01:
I can perhaps find the citation for that on my rebuttal time.

[00:07:26] Speaker 01:
If you have any further questions, or I may continue to the next point.

[00:07:31] Speaker 01:
So in addition, claims require an estimate of each of these values to be used in its determination.

[00:07:38] Speaker 01:
And again, no estimate of L2 access or leak is used in connection with

[00:07:45] Speaker 01:
determining whether to power down.

[00:07:46] Speaker 01:
Now L2 access, the L2 access leak ratio certainly is, and this is what the court found last time.

[00:07:54] Speaker 01:
But estimates that are calculated in who are not used in connection with that determination.

[00:08:00] Speaker 01:
We have two estimates that come up in a literature review.

[00:08:02] Speaker 01:
One is four nanojoules per access for L2 access.

[00:08:05] Speaker 01:
One is .45 nanojoules per cycle for leak.

[00:08:08] Speaker 01:
And they also calculate the ratio to get 8.9 for a typical L2 access leak ratio.

[00:08:12] Speaker 01:
Do they subsequently use four nanojoules or 0.45 or 8.9 in any calculation or any determination of decay intervals or whether to power?

[00:08:22] Speaker 01:
No.

[00:08:23] Speaker 01:
It's completely absent.

[00:08:25] Speaker 01:
What they do is they throw around different numbers.

[00:08:27] Speaker 01:
Five, 10, 20, 100.

[00:08:30] Speaker 01:
Numbers that they acknowledge are assumptions.

[00:08:33] Speaker 01:
In other words, made up numbers.

[00:08:36] Speaker 01:
Whose language?

[00:08:37] Speaker 01:
They say, we assume, quote, we assume that the L2 access leak ratio is equal to 10.

[00:08:42] Speaker 01:
Of course, the claim requires a determination on whether to power down based on two separate estimated quantities.

[00:08:49] Speaker 01:
Estimated power gain, estimated power loss.

[00:08:52] Speaker 01:
What we have are assumed values for one quantity, a ratio.

[00:08:56] Speaker 03:
What you lost the first time when we said it's immaterial to use experimental method relies in part on assumed values, right?

[00:09:06] Speaker 01:
Well, so again, Your Honor, I believe the

[00:09:10] Speaker 01:
Our understanding is that it was immaterial to the question of whether L2 access leak formed the basis of the determination of whether to power down, not on whether it qualified as using estimated values.

[00:09:27] Speaker 01:
Again, the narrow issue before the court at the time was the board's finding that L2 access leak was not the basis of the determination of whether to power down.

[00:09:35] Speaker 01:
The court reversed.

[00:09:37] Speaker 01:
But whether it qualified as a power gain or power loss resulting from powering down, and whether estimated values of it were used in connection with the determination, those questions were not before this court in the first appeal.

[00:09:58] Speaker 01:
And in any case, the who does not return to using its value of 8.9 to plug it back in or see what the optimal decay interval would be for that, because it's not concerned with using estimated values, it simply is not.

[00:10:12] Speaker 01:
Additionally, the board's obviousness determination rests on findings of motivation to combine and reasonable expectation of success that are both critically deficient.

[00:10:22] Speaker 01:
The board's finding on motivation rests on two mistaken premises.

[00:10:25] Speaker 01:
First, the board found that Takahashi teaches first evaluating a change of cash miss rate and only then powering down a cash way based on that evaluation, which they argue would render it compatible with who, which has a different way of determining whether to power down.

[00:10:42] Speaker 01:
And second, they backstop the argument by saying, well, you know, it doesn't matter what the order of operation is, because Takahashi's pre-process operation will fulfill its stated purpose anyway.

[00:10:54] Speaker 01:
These are both mistaken, and I'll take them in turn.

[00:10:56] Speaker 01:
The first point, the evidence overwhelmingly contradicts the board's finding.

[00:11:00] Speaker 01:
It shows that Takahashi's system first shuts down the cash way, and then measures the resulting change in cash miss rate to determine whether to power back up.

[00:11:10] Speaker 01:
Not whether to power down,

[00:11:12] Speaker 01:
Intel's position adopted by the board is that they say, quote, Takahashi's not based on counting actual cash misses.

[00:11:20] Speaker 01:
They say Takahashi expressly contemplates predicting a change in cash miss rate by hypothetically powering down.

[00:11:28] Speaker 01:
There's two problems with that.

[00:11:29] Speaker 01:
One, there's nothing in Takahashi about any prediction, calculation, hypothetical anything.

[00:11:35] Speaker 01:
It just doesn't exist.

[00:11:36] Speaker 01:
And number two, there's ample evidence that Takahashi does count the actual number of cashways that result from powering down.

[00:11:44] Speaker 01:
Takahashi refers to its cash miscounter.

[00:11:48] Speaker 01:
for measuring or counting and storing the number of cash miss operations.

[00:11:52] Speaker 01:
This court in its prior decision stated Takahashi uses the previous measurement of the cash miss rate to track whether the cash miss rate is improving or worsening, not to predict what might happen or hypothetically, but to see if it's getting better or worse.

[00:12:06] Speaker 01:
In its abstract, Takahashi says,

[00:12:09] Speaker 01:
It evaluates the cache miss rate, quote, before and after switching the number of ways.

[00:12:15] Speaker 01:
It shuts down the way, then it sees if it needs to power it back up.

[00:12:19] Speaker 01:
And most explicitly, it comes right out and says in transitioning from the pre-process mode to the two-way mode, quote, the number of activated ways is changed from one way to two ways.

[00:12:29] Speaker 01:
It's clear that it's already shut down the way.

[00:12:32] Speaker 01:
It does so automatically.

[00:12:34] Speaker 01:
and then it powers it back up if the cash misrate increases.

[00:12:37] Speaker 01:
It couldn't be any clearer on this point.

[00:12:39] Speaker 01:
This directly contradicts the board's erroneous finding on this point, which is not supported by substantial evidence, and highlights the incompatibility with who, which determines when to power down, not whether to power back up.

[00:12:51] Speaker 04:
You'll have to determine when to power down also.

[00:12:55] Speaker 04:
You're into your bottle time.

[00:12:57] Speaker 01:
In Takahashi, you mean?

[00:13:04] Speaker 04:
Oh, I gotcha.

[00:13:06] Speaker 01:
I'll rest on that point, Your Honor.

[00:13:08] Speaker 05:
Thank you.

[00:13:17] Speaker 02:
May it please the court, Dominic Massa on behalf of Intel Corporation.

[00:13:21] Speaker 02:
With me today is my colleague, David Yin, and from Intel, Ashok Pinto and Greg Kalbaum.

[00:13:29] Speaker 02:
Your honors, respectfully, as Judge Dyke pointed out, each of the arguments we heard today is precluded by your original consideration of this case several years ago.

[00:13:41] Speaker 02:
But moreover, the board's decision was supported by substantial evidence on each of these points.

[00:13:47] Speaker 02:
I read the court's prior decision, as Judge Dyke did, which is that the claim limitation determining whether to power down, including the phrase that's in that limitation resulting from powering down, this court found was met.

[00:14:07] Speaker 02:
And the court's decision was clear on that.

[00:14:11] Speaker 02:
But let's see what the board did and how

[00:14:15] Speaker 02:
Substantial evidence supports that.

[00:14:17] Speaker 02:
As this court noted and as the board found, whose basic premise is to measure the static power saved by turning off portions of the cash and then to compare it to the extra dynamic power dissipated by turning off cash lines.

[00:14:37] Speaker 02:
And that's at appendix 888.

[00:14:39] Speaker 02:
The issue that VLSI raises is that somehow power saved by turning off is not power resulting from powering down.

[00:14:49] Speaker 02:
But those are the same things.

[00:14:51] Speaker 02:
And this court has already found that, and the substantial evidence supports it.

[00:14:55] Speaker 02:
There's no doubt that WHO is trying to determine when to turn off a cash line.

[00:15:02] Speaker 02:
That's at 891.

[00:15:04] Speaker 02:
WHO also discloses at 898.

[00:15:08] Speaker 02:
an adaptive decay approach that chooses decay intervals at runtime.

[00:15:15] Speaker 02:
to match the behavior of individual cash funds.

[00:15:18] Speaker 02:
This is the operational system that VLSI says is lacking in who.

[00:15:24] Speaker 02:
But who describes a system that works at operation?

[00:15:28] Speaker 02:
It knows both the rate of energy being dissipated as well as the amount of who describes multiplying

[00:15:39] Speaker 02:
the rates, the L2 leak ratio, by the number of cash misses.

[00:15:47] Speaker 02:
That's in who and that's substantial evidence that the board could rely on.

[00:15:54] Speaker 00:
What about the argument that our prior opinion didn't address the requirement of an estimated power gain and an estimated power loss, and that there's somehow some difference by the emphasis on an estimated value?

[00:16:10] Speaker 02:
Again, Your Honor, I believe that the original panel did address that.

[00:16:16] Speaker 02:
And as Judge Dyke again pointed out, the panel found that it was immaterial.

[00:16:21] Speaker 02:
The same issue was raised over whether there were estimates or whether there were simulations.

[00:16:30] Speaker 02:
And the panel found that it's immaterial.

[00:16:34] Speaker 02:
And this is at 3.55.

[00:16:40] Speaker 02:
of the opinion.

[00:16:42] Speaker 02:
It is also immaterial that whose experimental method relied in part on assumed values of the L2 leak ratio.

[00:16:50] Speaker 02:
And I think the outside is reading a lot into the words estimate, whether who calls it an estimate in some places or a simulation in some places or assumed values.

[00:17:01] Speaker 02:
I don't think there's any specific

[00:17:04] Speaker 02:
construction here, there is none, and there's nothing to limit the term estimate in the 014 patent to exclude a simulation versus an assumption.

[00:17:12] Speaker 02:
I think those are all fairly estimates in the scope of the claim.

[00:17:22] Speaker 02:
The issue here is one that VLSI has raised late in the day, and I think contradicts the court's original opinion.

[00:17:30] Speaker 02:
But it's also clear that these estimates and simulations and assumptions in WHO are related.

[00:17:39] Speaker 02:
And there's substantial evidence in the record to show that.

[00:17:43] Speaker 02:
So this is WHO at 891 in the appendix.

[00:17:47] Speaker 02:
Section three is when this 8.9 number comes up, which is calculated, which VLSI says is a calculated value.

[00:17:57] Speaker 02:
And then section four is where VLSI says this is a totally separate teaching that is a simulation value.

[00:18:03] Speaker 02:
First of all, you read a priori reference for all it teaches and who teaches both of these calculated values and these simulated values.

[00:18:14] Speaker 02:
apart from that, who relates those back explicitly?

[00:18:19] Speaker 02:
Vilsa's argument in the paper is there's no discernible relationship between these two values.

[00:18:24] Speaker 02:
Who says it explicitly?

[00:18:26] Speaker 02:
It says as calculated in section three, and this is at 891,

[00:18:30] Speaker 02:
as calculated in section three, the dynamic energy required is roughly nine times.

[00:18:36] Speaker 02:
So you have this 8.9 calculation in section three.

[00:18:40] Speaker 02:
Section four, who says, refers back.

[00:18:43] Speaker 02:
As we stated in section three, it's roughly nine.

[00:18:47] Speaker 02:
So now we're at roughly nine.

[00:18:49] Speaker 02:
And then what who says at 8.97 is that it's describing a graph in figure eight.

[00:18:54] Speaker 02:
Again, DLSI says there's no discernible relationship here, but what WHO actually says is it relates it back explicitly.

[00:19:02] Speaker 02:
In this graph, and this is Figure 8, we assume that the L2 access-leak ratio is equal to 10, as discussed in Section 3.

[00:19:12] Speaker 02:
So what it's doing is it's taking this calculation of 8.9.

[00:19:15] Speaker 02:
It's saying that's roughly 9 when it's doing some math.

[00:19:19] Speaker 02:
And then later on when it's doing a graph, it says we're going to assume it's 10.

[00:19:26] Speaker 02:
These are not separate teachings.

[00:19:27] Speaker 02:
These are all within the scope of it.

[00:19:29] Speaker 02:
And not to get too deep into math, if you go back to the section 3 calculation where the 8.9 comes from, that's based on dividing the access to the weak.

[00:19:41] Speaker 02:
which was a range of 3 to 5 divided by 0.45.

[00:19:48] Speaker 02:
If you just do the math on that range, that's a range of 6.6 to 11.11.

[00:19:54] Speaker 02:
So it chose the middle of that, which was 8.9, discussed it, then it said it's roughly 9, and then chose a value of 10, which falls right within the range.

[00:20:03] Speaker 02:
So this is really a red herring argument, these first two arguments by BLSI, who discloses the entire

[00:20:10] Speaker 02:
The entirety of that claim limitation, including in a system which is operational and shuts down and knows exactly what's going on, it uses these estimates and simulations that are all explicitly related to each other.

[00:20:25] Speaker 02:
It's not picking and choosing from disparate teachings in who.

[00:20:29] Speaker 00:
Can I ask you about reasonable expectation of success?

[00:20:33] Speaker 00:
I understand that was not really brought up in the opening.

[00:20:36] Speaker 00:
One of the arguments that's made is that the board didn't provide a detailed explanation of why there was reasonable expectation of success.

[00:20:45] Speaker 00:
Rather, it was conclusory.

[00:20:47] Speaker 00:
What is your response to that?

[00:20:48] Speaker 00:
Why was that OK in this case?

[00:20:50] Speaker 02:
I believe, Your Honor, it's OK in this case as it is in other cases.

[00:20:56] Speaker 02:
And I believe Your Honor sat on a panel recently, Electa versus Zapp Surgical.

[00:21:03] Speaker 02:
This was a case that was decided after

[00:21:05] Speaker 02:
The briefing was closed here, but this is a case, 81, F4, 1368.

[00:21:10] Speaker 02:
In the elective ZAP, the court recognized that often there's intertwining between the motivation to combine analysis and the reasonable expectation of success.

[00:21:24] Speaker 02:
And that in the elective case, I think, was the extreme.

[00:21:27] Speaker 02:
There was no discussion.

[00:21:29] Speaker 02:
of expectation of success by the board.

[00:21:33] Speaker 02:
But this court found, in a precedential opinion, that even that it could be implicitly found.

[00:21:39] Speaker 02:
OK?

[00:21:40] Speaker 00:
We're not doing that.

[00:21:41] Speaker 00:
Most importantly, because I hear what you're saying, but that is, of course, a case-by-case determination.

[00:21:47] Speaker 00:
So here, where is it implicitly addressed?

[00:21:50] Speaker 02:
And I'd say, actually, Your Honor, I was going to say, that's the extreme case.

[00:21:53] Speaker 02:
And here, it was explicitly addressed.

[00:21:56] Speaker 02:
So the board.

[00:21:58] Speaker 02:
did discuss both in the context of motivation to combine and reasonable expectations of success.

[00:22:08] Speaker 02:
And it made explicit findings on both of those issues.

[00:22:19] Speaker 02:
I can get you a citation to it in one second.

[00:22:27] Speaker 02:
But while I'm looking for the citation, Your Honor, what we have here are two systems that are very, very closely related.

[00:22:35] Speaker 02:
We're in the world of computing systems with caches, with multi-level caches, with deciding when to power down a portion of the cache.

[00:22:44] Speaker 02:
And both Takahashi and Hu describe looking at various parameters available regarding the power loss and the power gain to then make that determination.

[00:22:55] Speaker 02:
and with thanks to Mr. Yin at appendix 36.

[00:23:00] Speaker 02:
So this is in the final decision.

[00:23:05] Speaker 02:
There would have been a reasonable reputation of success in doing so as both references teach using information related to cash misses to determine when to power down a portion of the cash and who additionally teaches turning off cash lines based on the comparison between estimated power loss

[00:23:24] Speaker 02:
and estimated power gain.

[00:23:26] Speaker 02:
So it's an explicit finding.

[00:23:27] Speaker 00:
One of the arguments made is that there's no reasonable expectation of success for the dependent claims.

[00:23:33] Speaker 00:
Is that necessary?

[00:23:34] Speaker 00:
That's really what I was referring to.

[00:23:39] Speaker 02:
So first of all, Your Honor, again, that argument applies only to the dependent claims who would have no import on independent claims 1 and 12.

[00:23:48] Speaker 02:
But besides that,

[00:23:49] Speaker 02:
Vealside made no differentiating arguments as far as motivation to combine or reasonable expectation of success with respect to the dependent claims.

[00:23:59] Speaker 02:
So they never argued before the board or really in this briefing that there's something particular about the additional limitations found in the dependent claims.

[00:24:09] Speaker 02:
So the board can take the analysis that it used for the independent claims and apply that to combining Takahashi and Hu for the dependent claims.

[00:24:21] Speaker 00:
So for the dependent claims, the only issue is whether that combination where there's a motivation to combine and there's reasonable expectation of success actually teaches the limitations of the dependent claim?

[00:24:33] Speaker 00:
Is that what you're saying?

[00:24:34] Speaker 02:
So what I'm saying here is, in particular with these dependent claims, the petition found that all those elements were in Takahashi.

[00:24:43] Speaker 02:
So you had the first couple elements in Takahashi.

[00:24:46] Speaker 02:
You had this determining limitation from who, and then the dependent claim elements from Takahashi.

[00:24:51] Speaker 02:
So when you're talking about the motivation combined, and you're talking about the reasonable expectation of success, you're talking about that element from the teaching of who can that fit in with Takahashi.

[00:25:02] Speaker 02:
That's resolved by the board with respect to claims one and claim 12.

[00:25:06] Speaker 02:
That's no longer an issue when you're talking about the dependent claims.

[00:25:09] Speaker 02:
What you have then is just the addition of a dependent claim element.

[00:25:13] Speaker 02:
That's just taking Takahashi qua Takahashi.

[00:25:16] Speaker 00:
And Takahashi as modified by Hugh, right?

[00:25:19] Speaker 02:
Takahashi as modified by Hugh, adding just another teaching from Takahashi into that mix.

[00:25:24] Speaker 02:
So you don't need to combine Takahashi and Hugh for the dependent claim elements.

[00:25:31] Speaker 02:
Thank you, Your Honors.

[00:25:34] Speaker 05:
And also, Mr. Flakeli has about two and a half minutes.

[00:25:40] Speaker 01:
Thank you, Your Honors.

[00:25:42] Speaker 01:
Just a couple of quick points.

[00:25:44] Speaker 01:
Firstly, I'd just like to point out in the court's prior decision, it stated that Intel has argued that who itself teaches determining whether to power down in response to the L2 access leak ratio.

[00:25:58] Speaker 01:
The board thus erred in determining that Hu does not use the L2 access leak ratio to determine whether to power down.

[00:26:05] Speaker 01:
And in its conclusion it stated, we reverse the board's determination that Hu does not teach the L2 access leak ratio to determine whether to power down a cache.

[00:26:14] Speaker 01:
that the court did not extend its finding to the entire determining limitation, but limited its finding to the narrow issue of what the board had done, which was find that L2 access leak was not used in connection with the determination of whether to power down.

[00:26:30] Speaker 01:
It said that it was.

[00:26:31] Speaker 01:
And we don't dispute that.

[00:26:33] Speaker 01:
I'd like to just respond to a couple of things that Mr. Massa said.

[00:26:37] Speaker 01:
He argued that this idea that the power saved by turning off is not the same as the power saved by resulting from powering down, that that's somehow our argument.

[00:26:46] Speaker 01:
That's not the case.

[00:26:47] Speaker 01:
I think, again, it's a little bit of a ham-fisted analogy, but the hamburger analogy is a good one.

[00:26:52] Speaker 01:
The idea is that an energy per access doesn't tell you how many accesses.

[00:26:56] Speaker 01:
It doesn't tell you how long you're going to shut it down for when you have an energy per cycle.

[00:27:01] Speaker 01:
Without that information, without estimates of that information, you just don't get there.

[00:27:05] Speaker 01:
You don't get revenue from price, you don't get cost from cost per hour, and you don't get energies from L2 axis and wheat.

[00:27:13] Speaker 01:
On the issue of reasonable expectation of success,

[00:27:17] Speaker 01:
Mr. Massa argued that both references teach deciding when to power down.

[00:27:23] Speaker 01:
I think we've shown the evidence is clear that Takahashi doesn't do that.

[00:27:26] Speaker 01:
It decides whether to power back up.

[00:27:29] Speaker 01:
So that's a different thing.

[00:27:31] Speaker 01:
How you would combine these two and whether you have success in doing that, that's not explained.

[00:27:35] Speaker 01:
There's no explanation.

[00:27:36] Speaker 01:
There's no factual foundation for that in the board's decision.

[00:27:40] Speaker 01:
He also argues that we made no differentiating arguments.

[00:27:44] Speaker 01:
Well, that's burden shifting.

[00:27:46] Speaker 01:
This is Intel's burden initially and the board's burden.

[00:27:50] Speaker 00:
Is there a further modification of the combination in order to have the combination satisfy the elements of the dependent claims?

[00:27:57] Speaker 01:
I believe there would be, Your Honor, and this is spelled out in, for example, in the Teva Farms case, where they said that the reasonable expectation of success analysis must be tied to the scope of the claims.

[00:28:09] Speaker 00:
I don't know that you've answered my question.

[00:28:11] Speaker 00:
So my question was,

[00:28:14] Speaker 00:
for in the petition after there was the original combination of the primary reference and you have the secondary reference where there are additional modifications that were proposed to be made by the one of ordinary skill in the art would have thought that they were obvious or not with respect to the dependent claims that you are arguing don't have a reasonable expectation for success well i don't believe that uh... intel or the board argued for additional modifications of the reference i didn't think that if any claims i would i would submit that they should have in order to uh...

[00:28:45] Speaker 00:
But do you think they argued that the modified, the modification of this Takahashi in view of Hugh taught the limitations in UD Penitentiary Claims, right?

[00:28:58] Speaker 01:
Yes.

[00:28:58] Speaker 00:
And so why do they additionally have to show, after they have already shown that they're reasonable, that's assumed for my purposes, that for the independent claim, they've shown there's a reasonable expectation of success.

[00:29:10] Speaker 00:
Why would they have to show it yet again for the dependent claims if there isn't a further modification?

[00:29:16] Speaker 01:
Well, I guess what I would argue is that based on cases like Teva Pharmaceuticals, the reasonable expectation of success showing, which is an independent showing from the teachings of the references, that requires some tie to the scope of the claim.

[00:29:29] Speaker 01:
That's what the case said.

[00:29:32] Speaker 00:
In general, the... So it's your view that a party would have to show, for every single dependent claim, they have to show reasonable expectation of success different from what they showed with respect to the independent claim?

[00:29:45] Speaker 01:
Well, there either has to be an argument about why the same argument extends to the dependent claims or some recognition of the further scope that's in the dependent claims.

[00:29:55] Speaker 01:
There is a difference in scope.

[00:29:56] Speaker 01:
It must be recognized.