[00:00:01] Speaker 01:
First case for argument this morning is Qualcomm Incorporated versus Intel Corporation 202023-1710.

[00:00:09] Speaker 01:
Council, whenever you're ready.

[00:00:16] Speaker 00:
Thank you, Your Honor.

[00:00:17] Speaker 00:
May it please the court.

[00:00:19] Speaker 00:
On a two to one decision, the board took away a property right based on a new claim construction that one does not comport with the board's original analysis.

[00:00:29] Speaker 00:
two contravenes this court's guidance in the original appeal, and three is not the broadest reasonable interpretation in light of the intrinsic record.

[00:00:39] Speaker 00:
The remand?

[00:00:47] Speaker 00:
Judge Toronto, I'm sorry, I'm having trouble hearing you.

[00:00:49] Speaker 01:
I am too.

[00:00:51] Speaker 01:
This is Judge Prost.

[00:00:52] Speaker 01:
I can't hear you, Judge Toronto.

[00:00:54] Speaker 03:
Me too.

[00:01:05] Speaker 01:
Still can't hear anything.

[00:01:12] Speaker 01:
I'm seeing your lips moving, but you've got only a faint voice off and on.

[00:01:19] Speaker 02:
Apologies, we'll pause the argument momentarily while we work to get this restored.

[00:01:28] Speaker 04:
Can you hear me now or no?

[00:01:30] Speaker 01:
Yes.

[00:01:31] Speaker 04:
Yes.

[00:01:32] Speaker 04:
OK.

[00:01:33] Speaker 04:
I switched the microphones.

[00:01:35] Speaker 04:
Thanks.

[00:01:35] Speaker 04:
Sorry for the pause.

[00:01:38] Speaker 04:
I was just asking you the question.

[00:01:39] Speaker 04:
You said a new claim construction.

[00:01:42] Speaker 04:
I knew you don't mean, do you, that this was not a construction that had been argued by Intel, just that it was different from the last time the board adopted a construction?

[00:01:54] Speaker 00:
Yes, Your Honor.

[00:01:54] Speaker 00:
I was referring to the board adopting a new construction that is

[00:01:58] Speaker 00:
For all purposes here, the issue before the court is the opposite of the construction it adopted in the original IPR proceedings.

[00:02:07] Speaker 04:
So why is it wrong?

[00:02:09] Speaker 00:
OK, let me jump to that.

[00:02:10] Speaker 00:
I'll jump straight to the evidence.

[00:02:14] Speaker 00:
This court directed the board to consider the specification in particular and some other pieces of evidence.

[00:02:19] Speaker 00:
And when you look at the specification, it consistently contrasts the temporary buffers of the prior art

[00:02:28] Speaker 00:
from the hardware buffer of the claim.

[00:02:31] Speaker 00:
So the hardware buffer has to be understood in the context of that consistent unequivocal contrast between temporary buffer on the one hand and hardware buffer on the other.

[00:02:42] Speaker 00:
And the board understood both times that a temporary buffer is something that is allocated during runtime.

[00:02:50] Speaker 00:
So what we mean by a permanent hardware buffer is something that's permanently assigned.

[00:02:55] Speaker 00:
It's not one of these temporary buffers.

[00:02:58] Speaker 00:
So that's point one, the consistent contract.

[00:03:00] Speaker 04:
I'm sorry.

[00:03:01] Speaker 04:
So one of the frustrations, at least I have experienced in this case, is that people use terms that are being defined in the definition.

[00:03:11] Speaker 04:
That's not helpful.

[00:03:13] Speaker 04:
So you've got to take the word permanent out of your definition.

[00:03:18] Speaker 04:
Here's what I think I understand you to mean.

[00:03:23] Speaker 04:
A buffer intrinsically performs a temporary function, which is as a waste station, as data is moving from one location to another, one residence to another.

[00:03:36] Speaker 04:
That's not what we're talking about about temporary, because all buffers are that.

[00:03:42] Speaker 04:
suggesting, I think, that there is a set of memory cells with lines going into and operating those memory cells that are used only for buffering and never including the next time the machine is booted up for this buffering function.

[00:04:02] Speaker 04:
Is that a correct understanding?

[00:04:05] Speaker 04:
You mean by permanent?

[00:04:07] Speaker 00:
Yes, I think largely so.

[00:04:10] Speaker 00:
Let me see if I can repeat it.

[00:04:13] Speaker 00:
avoiding the word permanent, but in the context of what you're asking, Judge Toronto, which is that, yes, in the context of this specification, the hardware buffer is there.

[00:04:24] Speaker 00:
It does not need to be allocated.

[00:04:26] Speaker 00:
The memory cells are devoted

[00:04:29] Speaker 04:
to this specific purpose of- Just so you can help me understand some things that I've had a little trouble understanding.

[00:04:40] Speaker 04:
The board and Intel both use the word dedicate.

[00:04:46] Speaker 04:
They use that word to mean something that I think is the crucial difference between the two sides position.

[00:04:55] Speaker 04:
They mean dedicated during the, let's just call it the computer session from the time you boot up until the time you turn the power off.

[00:05:06] Speaker 04:
Let's just call that a day, just to simplify the expression.

[00:05:10] Speaker 04:
During that day, this set of memory cells is dedicated to the buffering function in the sense that that set of memory cells is not being used as the final resting place

[00:05:24] Speaker 04:
of the software that's being put into the system memory.

[00:05:28] Speaker 04:
Is that, do I understand?

[00:05:32] Speaker 04:
So your understanding of permanent has to be broader than that.

[00:05:38] Speaker 00:
Yes.

[00:05:38] Speaker 04:
Or more extended than that, I'm sorry.

[00:05:41] Speaker 00:
That's correct.

[00:05:42] Speaker 00:
We also use dedicated in our construction to capture the notion that this set of memory cells, this thing, this portion of the hardware is set aside to be used

[00:05:54] Speaker 00:
for the purpose of buffering the data in the hardware and getting it in conjunction with the guidance.

[00:06:01] Speaker 04:
I'm sorry.

[00:06:02] Speaker 04:
I'm going through the same process right now that I went through both several years ago and when reading the briefs.

[00:06:11] Speaker 04:
Set aside doesn't help.

[00:06:13] Speaker 04:
The other side's view is these memory cells are also set aside.

[00:06:18] Speaker 04:
You have to keep focusing on what the difference is between your set aside and their set aside.

[00:06:24] Speaker 00:
The way I would say it in plain English, Judge Toronto, is that it's already been set aside.

[00:06:29] Speaker 00:
So in that day, the day wakes up, the day begins, it's already there.

[00:06:36] Speaker 00:
You don't have to go through the process like the prior art in creating, i.e.

[00:06:43] Speaker 00:
allocating the temporary buffer.

[00:06:45] Speaker 00:
And you see that at column two, where it's talking about the temporary buffer being allocated.

[00:06:52] Speaker 00:
You both have to use

[00:06:54] Speaker 00:
You take it out of a heap of a bunch of stuff.

[00:06:57] Speaker 00:
You could use these memory cells for this purpose or that purpose.

[00:07:01] Speaker 00:
The dedicated port is, I agree, not necessarily getting to the temporal limitation.

[00:07:07] Speaker 00:
It's saying, okay, we're going to set aside these cells to use as a buffer.

[00:07:12] Speaker 00:
But the permanent is reflecting that a temporary buffer not only is going to temporarily perform something, but it's temporarily in existence.

[00:07:21] Speaker 00:
It's allocated.

[00:07:23] Speaker 00:
after the day has begun and the patent contrasts that temporary buffer, the temporary in existence, the steps that the system memory has to take in grabbing today, it's going to be this piece out of the system memory and maybe this big or this small.

[00:07:40] Speaker 04:
Tomorrow- I think I recall that there are specification passages that do use the term temporary.

[00:07:48] Speaker 04:
There's some argument about whether that is the

[00:07:53] Speaker 04:
the point of contrast that is being made in those sentences, but where in the specification is there an explanation of your particular understanding of what temporary means in this context, which I think I understand, which is that there is a set of memory cells that today, tomorrow, the next

[00:08:16] Speaker 04:
the days after, will always be used for this function.

[00:08:20] Speaker 04:
And Intel's and now the board majority's view is, no, they are dedicated for the day, but who knows what they'll do tomorrow.

[00:08:31] Speaker 00:
Exactly, Your Honor.

[00:08:31] Speaker 00:
And I think I'll point you to a couple of specification passages and figure three.

[00:08:37] Speaker 00:
And my first top line point is that we, of course, have to read the patent

[00:08:41] Speaker 00:
as a whole.

[00:08:42] Speaker 00:
And I think one of the errors that the board and Intel did is looking at very discrete excerpts and demanding that we talk about temporary buffer in the same sentence as hardware buffer.

[00:08:53] Speaker 00:
And of course, the person of ordinary skill in the art here, a highly skilled artisan doesn't need that.

[00:08:57] Speaker 00:
They're going to look at the full context of the specification.

[00:09:00] Speaker 00:
So when the patent starts off by saying, we have problems with

[00:09:07] Speaker 00:
this extra memory copy operation that entails both allocating the temporary buffer and then using it.

[00:09:13] Speaker 00:
We wanna avoid that.

[00:09:14] Speaker 00:
We have to get to a really short, we're gonna be more efficient with our communications, get to a really short- I'm sorry, I'm gonna interrupt you again.

[00:09:22] Speaker 04:
Okay.

[00:09:22] Speaker 04:
Where in the spec is there a mention of allocating memory cells to perform the function?

[00:09:29] Speaker 00:
Sure, Your Honor, and I don't wanna misrepresent.

[00:09:31] Speaker 00:
There isn't exactly that phrase, but the term is allocate.

[00:09:35] Speaker 00:
So at column two,

[00:09:37] Speaker 00:
line 25.

[00:09:38] Speaker 00:
This is in the background section of the patent.

[00:09:48] Speaker 00:
And it's discussing these prior art systems in which a temporary buffer, one way of performing such loading, this is at line 25, is to allocate a temporary buffer.

[00:10:01] Speaker 04:
Okay, thank you.

[00:10:03] Speaker 00:
You're welcome.

[00:10:04] Speaker 00:
And

[00:10:06] Speaker 00:
What, you know, what that means to the person of ordinary skill in the art is that's the system memory, again, grabbing this set of cells this day, totally different set of cells another day, different sizes.

[00:10:16] Speaker 00:
There's flexibility in system memory.

[00:10:18] Speaker 00:
That's a benefit of doing it this way.

[00:10:20] Speaker 00:
But what our inventors came up with, and you see in the detailed description of the patent starting at column four, you see that for instance, at column four lines 46 to 47,

[00:10:34] Speaker 00:
direct scatter loading technique avoids use of a temporary buffer full stop.

[00:10:43] Speaker 00:
I will say there's a lot going on in these claims and the way it is able to get the data more efficiently isn't just the hardware buffer, we have to look at the claims as a whole, but what the hardware buffer does is enable

[00:10:57] Speaker 00:
but the inventors came up with, they wanted to avoid having to do this allocation of a temporary buffer and using that temporary buffer and all of that stuff that has to happen.

[00:11:06] Speaker 00:
They set up the hardware.

[00:11:09] Speaker 04:
So one of the things that was not clear to me is something that you at least mentioned, I think in maybe your reply brief, I don't remember, but that I don't think I saw in Dr. Renard's declarations, which is,

[00:11:27] Speaker 04:
Obviously, some time and resources are necessary to allocate a set of memory cells for this purpose.

[00:11:36] Speaker 04:
I don't remember Dr. Renard ever mentioning that as one of the efficiencies.

[00:11:45] Speaker 04:
As opposed to all of the rest, and then my natural follow-up to that question is all of the rest could, couldn't it, be in fact performed with a, your view is a temporary buffer, which is an allocation of memory cells

[00:12:04] Speaker 04:
buffering function today.

[00:12:05] Speaker 04:
All the rest could be done about sending the total batch of addresses at once instead of in each header of each data packet so that all of that kind of stuff.

[00:12:19] Speaker 04:
I didn't understand and I thought maybe there's no longer really a dispute that none of that is in any way dependent on whether the buffer is hardware or not.

[00:12:33] Speaker 00:
I'll start with Dr. Wren.

[00:12:35] Speaker 00:
I can go in the order you presented them, Your Honor.

[00:12:37] Speaker 00:
So in terms of Dr. Reinhardt's testimony, at appendix 3867, 3868, 3877, and 3878, you will see him consistently talking about the benefits of using a hardware buffer because it is implemented, it's permanently assigned.

[00:12:58] Speaker 04:
The person of ordinary skill in the art- Right, but he does that at an extremely high level of generality, right?

[00:13:04] Speaker 04:
The difference between designing something in a set of wires and having a set of wires that are used for different purposes, you can save certain kinds of expenses and you lose some flexibility.

[00:13:22] Speaker 04:
But for one thing, I don't think he ever mentions saving the cost of allocating and

[00:13:28] Speaker 04:
Even in addition to that, he never explains what the physical configuration specialization is of your so-called hardware buffer.

[00:13:42] Speaker 00:
In terms of the efficiency, it's not a narrow claim that's only about reclaiming a hardware buffer.

[00:13:50] Speaker 00:
That's not it.

[00:13:51] Speaker 00:
It's the hardware buffer being used at the direction of the

[00:13:56] Speaker 00:
scatter load controller so that the hardware buffer can get that data directly to system memory.

[00:14:03] Speaker 00:
So it's the coupling of using the hardware buffer.

[00:14:06] Speaker 00:
That was the solution.

[00:14:07] Speaker 00:
The problem was we've got extra memory operations, copy operations that make this take longer than we want.

[00:14:14] Speaker 00:
And this particular solution is to employ the hardware buffer, dedicate that permanently.

[00:14:20] Speaker 00:
You don't have to then create it and move it around.

[00:14:24] Speaker 00:
you read the claims the way the board did, you're just back to the prior art, or you're allowing a temporary buffer to be allocated and moving it around, which is what the patent over and over says we are avoiding.

[00:14:36] Speaker 03:
I don't think that's quite correct in the sense that it seems like when you read the patent, it's really trying to tell a story about changing up what the buffer operations are.

[00:14:50] Speaker 03:
It's less clear to me that

[00:14:52] Speaker 03:
the patent is trying to communicate that there's something about the physical nature of the buffer itself that is distinctive about the invention over the prior art.

[00:15:06] Speaker 03:
In other words, this invention seemed to be much more about how we wanna get the header information upfront and then send the data segments along separately and then feed individual data segments

[00:15:21] Speaker 03:
directly into the system memory from the buffer.

[00:15:26] Speaker 03:
I know it's called the hardware buffer, but let's just call it the buffer for now.

[00:15:31] Speaker 03:
And so it's that sequence of buffer operations that's distinctive.

[00:15:36] Speaker 03:
And what I'm trying to struggle with too is the idea of what is a hardware buffer.

[00:15:45] Speaker 03:
I am not familiar with that term, or at least I've

[00:15:51] Speaker 03:
don't think I've ever come across it before.

[00:15:54] Speaker 03:
Is it fair to say that it is not a term of art with a well-established meaning in the field of computers?

[00:16:04] Speaker 00:
Yes.

[00:16:05] Speaker 03:
OK.

[00:16:05] Speaker 03:
And so therefore, what it comes down to is what can we glean, what can we extract from the intrinsic evidence in terms of arriving at

[00:16:18] Speaker 03:
the most appropriate understanding of this term hardware buffer in the context of this patent?

[00:16:26] Speaker 00:
Yes, Your Honor, exactly.

[00:16:28] Speaker 03:
Obviously, taking into account brought us reasonable interpretation.

[00:16:32] Speaker 00:
Sure, of course.

[00:16:33] Speaker 00:
And it has to be reasonable under that interpretation.

[00:16:36] Speaker 00:
So let me see if I can unpack a few things here.

[00:16:38] Speaker 00:
Hardware has to have some meaning.

[00:16:42] Speaker 00:
This court already said, if we're just going to treat it as buffer, we're not giving the full term

[00:16:48] Speaker 00:
and all the words meaning.

[00:16:48] Speaker 00:
So hardware has to mean something.

[00:16:51] Speaker 00:
And in the context of this patent, when you have the contrast between the two word term temporary buffer that's allocated and hardware buffer key point one, that's a contrast.

[00:17:02] Speaker 00:
Judge Chen, in terms of the data segments, that is how the inventors were able to do it this way because they used the hardware buffer that's not gonna change.

[00:17:11] Speaker 00:
It's gonna stay the same, maybe not big enough sometimes

[00:17:15] Speaker 00:
the data can come in there and in conjunction with the way the data is divided up, then the scatter load controller can send it directly from that hardware to its final destination in system memory without having to do the copying of data around.

[00:17:30] Speaker 00:
So they have to be viewed in conjunction because that's how the inventors set it up in order to achieve what they first set out to do, which let's get rid of these extra

[00:17:41] Speaker 00:
copy operations that require creating the temporary buffer and using it.

[00:17:45] Speaker 03:
And just so I understand, I think you might have had this colloquy with Judge Toronto already.

[00:17:53] Speaker 03:
If your claimed invention wasn't just for your conception of a hardware buffer, but it also encompassed the use of a temporary buffer for the purposes of these novel buffer operations,

[00:18:11] Speaker 03:
Isn't it true that the invention would still work as intended?

[00:18:19] Speaker 03:
Perhaps it wouldn't work as fast as you believe it would work with a hardware buffer because then during runtime you have to set up the temporary buffer.

[00:18:29] Speaker 03:
But barring that, the invention would still operate just as planned while at the same time skipping over

[00:18:39] Speaker 03:
the additional copy operation that takes place in the prior art set of buffer operations?

[00:18:45] Speaker 03:
Is that fair to say?

[00:18:48] Speaker 00:
No, I don't think so, Your Honor.

[00:18:50] Speaker 00:
I think this dovetails with Judge Toronto's question.

[00:18:52] Speaker 00:
Number one, we're not talking about other ways to transfer data efficiently.

[00:18:57] Speaker 00:
We did not claim all such ways.

[00:19:00] Speaker 00:
We claimed a very specific configuration that employs the hardware buffer to replace and take out

[00:19:05] Speaker 00:
the step of the temporary buffer?

[00:19:08] Speaker 03:
Yeah, but what I'm asking is, isn't it true that assuming I accept your understanding of hardware buffer, why wouldn't it be that a temporary buffer instead of a hardware buffer could do the very buffer operations that are contemplated?

[00:19:23] Speaker 03:
Sending the data segments separately from the header information, header information comes first, directly scatter loading the data segments from the

[00:19:34] Speaker 03:
temporary buffer directly into the system memory.

[00:19:40] Speaker 00:
Your Honor, I think an excellent visual representation to respond to that comment is the descents figure at appendix 85, because it's showing we had a prior.

[00:19:49] Speaker 00:
The data still has to come in through something, right?

[00:19:52] Speaker 00:
So the hardware buffer, the hardware is there.

[00:19:54] Speaker 00:
What the prior art did was have it come in through the hardware, and you see that at appendix 85, then get copied into a temporary buffer

[00:20:04] Speaker 00:
because there's all sorts of speed issues with system memory, and you got to let it catch up once it's in the system memory before you can finally put it where you want it to go.

[00:20:12] Speaker 00:
What the inventors did is utilize this pre-existing hardware that's going to come in through this.

[00:20:18] Speaker 00:
And they said, we're going to use that.

[00:20:20] Speaker 00:
It has drawbacks.

[00:20:21] Speaker 00:
It's fixed.

[00:20:22] Speaker 00:
It might be too small at times.

[00:20:23] Speaker 00:
We can't change it.

[00:20:24] Speaker 00:
We don't have flexibility the way we do with temporary buffers.

[00:20:28] Speaker 00:
But we will employ that in conjunction with.

[00:20:32] Speaker 03:
Perhaps I'm not explaining myself

[00:20:34] Speaker 03:
completely.

[00:20:37] Speaker 03:
Are you saying that if you swapped out your version of a hardware buffer with a temporary buffer, the claimed invention would not work?

[00:20:48] Speaker 03:
The temporary buffer would not be able to scatter load from a temporary buffer, the individual data segments into the system memory.

[00:20:58] Speaker 03:
You wouldn't be able to receive the header information first before

[00:21:03] Speaker 03:
receiving the individual data segments into that temporary buffer?

[00:21:08] Speaker 03:
You're not saying that, right?

[00:21:10] Speaker 00:
No, but I'm saying that's not our invention.

[00:21:12] Speaker 00:
We're not going to accuse that.

[00:21:14] Speaker 00:
We're not covering that.

[00:21:15] Speaker 00:
That's the boundaries of our invention.

[00:21:18] Speaker 00:
We didn't claim just using a temporary buffer.

[00:21:20] Speaker 00:
And again, that goes back to the specification that unequivocally says problems with temporary buffer, they're bad.

[00:21:28] Speaker 00:
And we're going to employ the hardware buffer in this special purpose.

[00:21:31] Speaker 00:
You can think of the hardware buffer as special purpose.

[00:21:35] Speaker 00:
Temporary buffers are created out of general purpose system memory or memory.

[00:21:40] Speaker 00:
The hardware buffer has a special purpose and it wouldn't have a special purpose if you had to go through the step of creating it each time.

[00:21:46] Speaker 00:
And it might be something different from day one to day two to day 10.

[00:21:51] Speaker 00:
It's got a consistent special purpose and figure three.

[00:21:55] Speaker 00:
is a good visual representation of that because it shows the hardware buffer in the USB controller.

[00:22:02] Speaker 00:
We understand the board rejected our view that the term should be defined in the context of the USB controller as being too narrow.

[00:22:09] Speaker 00:
But as the descent makes clear, the person of ordinary skill in the art would understand the hardware buffer is in the nature of something like USB controller.

[00:22:17] Speaker 00:
We don't have to define the full context of that, but it's going to be something like that in the hardware.

[00:22:21] Speaker 00:
You're employing the hardware consistently from the beginning.

[00:22:27] Speaker 00:
it's permanently assigned to that role of transferring data.

[00:22:31] Speaker 00:
And then where it sends the data is at the direction of the scatter.

[00:22:35] Speaker 04:
Can I just ask, and perhaps you don't know the answer to this, but I will tell you this is what I've been spending weeks, in some sense, years trying to understand.

[00:22:46] Speaker 04:
Tell me something physical

[00:22:50] Speaker 04:
about the configuration of a forever dedicated buffer that's different from the configuration of the great mass of memory cells for executing programs.

[00:23:07] Speaker 04:
Just even one thing.

[00:23:09] Speaker 00:
Sure, Your Honor.

[00:23:11] Speaker 00:
It hearkens back to the last argument where we talked about etching.

[00:23:14] Speaker 00:
I don't know that etching is the right technological term, instantiate

[00:23:20] Speaker 00:
made part of, you'd see it in a circuit diagram, it's there.

[00:23:24] Speaker 00:
It's not something that can change and you can only sort of pictorially represent it.

[00:23:30] Speaker 04:
What's the it?

[00:23:33] Speaker 04:
No, no, because I sort of had the sense when I read the board opinion that what I was trying to get at, like repeatedly using the term substantive

[00:23:45] Speaker 04:
in the last opinion was get down to the level of wires.

[00:23:50] Speaker 04:
Tell me about wires.

[00:23:54] Speaker 00:
I appreciate that, Your Honor, and I know it's probably dissatisfying that we didn't do that.

[00:24:01] Speaker 00:
I think the board and the parties approached it with the perspective of what resolves the nature of the controversy here.

[00:24:07] Speaker 00:
And we have a

[00:24:10] Speaker 00:
Fuller construction that we attempted to offer in terms of a dedicated buffer.

[00:24:14] Speaker 00:
So it's got a specific purpose.

[00:24:16] Speaker 00:
You read it in the context of the rest of the claim in terms of what its dedicated purpose is.

[00:24:21] Speaker 00:
It's being operated on and used to scatter load data at the direction of the controller.

[00:24:27] Speaker 00:
And it's distinct from system memory.

[00:24:29] Speaker 00:
And the permanent was to address the dispute about whether it can be the type of prior art temporary buffer.

[00:24:36] Speaker 00:
So I appreciate that we didn't get down to the level of the wire.

[00:24:40] Speaker 00:
but because the board has consistently understood when we're looking at what is the asserted art, the board has always understood the asserted art has a temporary buffer.

[00:24:49] Speaker 04:
And so the nature of the dispute, unfortunately- And that's what intermediate storage area in one of the, I don't know, in two of the three RAM memories in Svensson is allocated temporarily during runtime.

[00:25:09] Speaker 00:
Exactly.

[00:25:11] Speaker 00:
Exactly.

[00:25:12] Speaker 00:
It's that allocation during runtime.

[00:25:14] Speaker 00:
That is the key difference between there's a lot going on in the hardware buffer and I'm not an engineer and they could probably describe it in considerable detail.

[00:25:23] Speaker 04:
But the key point for this, I will just say again, I think you get my point.

[00:25:27] Speaker 04:
Boy, was that what I was looking for.

[00:25:30] Speaker 00:
I understand your honor.

[00:25:34] Speaker 01:
Let's move on and we'll reserve some time for rebuttal.

[00:25:41] Speaker 02:
Thank you.

[00:25:47] Speaker 02:
I think there's a reason we are struggling with this term and don't have an established term here and that's the Qualcomm was keeping its options

[00:26:00] Speaker 02:
saying already set aside.

[00:26:03] Speaker 02:
It wasn't limiting itself to the buffer and the transport mechanism.

[00:26:06] Speaker 02:
It was repeatedly saying that the embodiments were merely exemplary.

[00:26:11] Speaker 02:
And when we step back and look at the sort of purported benefits here, none of those depend on the buffer already being set aside, right?

[00:26:22] Speaker 02:
The key of this patent, which is copying the

[00:26:30] Speaker 02:
The extra copy steps, we have a fact.

[00:26:33] Speaker 04:
Can I just say, I'm having a very hard time hearing you.

[00:26:35] Speaker 04:
Is there any way that you can amp up your mic or stand closer to it or something?

[00:26:43] Speaker 02:
That any better, Your Honor?

[00:26:44] Speaker 04:
Yeah, right then it was.

[00:26:45] Speaker 04:
Thank you.

[00:26:46] Speaker 04:
OK.

[00:26:47] Speaker 02:
I'll try to project more into it too.

[00:26:51] Speaker 04:
Thank you.

[00:26:51] Speaker 02:
So on the extra copy steps, we have a factual finding from the board at appendix 56 that

[00:27:07] Speaker 02:
with a push from the primary processor into that intermediate storage area, and then one memory copy operation by the secondary processor to the final destination.

[00:27:17] Speaker 02:
So that benefit in terms of eliminating extra copying doesn't depend on a buffer already being allocated.

[00:27:25] Speaker 02:
And to the extent there's anything about copying in or out of the same memory, we have that separation too, in the sense that we have an entirely different

[00:27:34] Speaker 02:
memory in which that intermediate storage area is allocated.

[00:27:39] Speaker 02:
I also want to highlight there are four things you're not going to see in the patent.

[00:27:44] Speaker 02:
You're not going to see any discussion of a benefit that can only be achieved with the buffer already being set aside.

[00:27:53] Speaker 02:
Patent doesn't tout any benefits of avoiding time with allocation or any efficiencies there.

[00:28:00] Speaker 02:
You're not gonna see any statement that's criticizing a so-called temporary buffer because it's only allocated for a day and wasn't sort of, you know, he's allocated at runtime.

[00:28:13] Speaker 02:
You're not gonna see anything that's limiting this to the USB embodiment or even a buffer in the hardware transport mechanism.

[00:28:20] Speaker 02:
And then it's just mentioned, you're not gonna see anything that's as direct.

[00:28:24] Speaker 02:
You're not gonna see anything that's more direct than what we have in the primary.

[00:28:29] Speaker 01:
Is there anything in the prosecution history that would guide us to determine whether or not there was an effort to just differentiate the permanent from the temporary or the dedicated from the temporary?

[00:28:42] Speaker 02:
Absolutely.

[00:28:42] Speaker 02:
We think the prosecution history is very helpful in this case.

[00:28:46] Speaker 02:
So we have the unusual situation in which the Svensson setup, but without the Bauer, which brought in the separate images and headers, was being looked at by the examiner.

[00:29:00] Speaker 02:
The examiner understood the term hardware buffer to encompass the intermediate storage area of Spencer, which is the same intermediate storage area we're relying on.

[00:29:14] Speaker 02:
The applicant then didn't say, oh, you're misinterpreting hardware buffer.

[00:29:20] Speaker 02:
It amended the claims to narrow them and its statements focused

[00:29:29] Speaker 02:
alleged breakthrough, which was transmitting the header first.

[00:29:34] Speaker 02:
The problem they have, of course, is that alleged breakthrough is exactly what is rendered obvious by Bauer, which wasn't before the patent office.

[00:29:42] Speaker 03:
Mr. Saunders, is there any precedent for the idea that if a patent owner does not resist a particular examiner mapping of a reference to a claim limitation, then therefore the

[00:29:57] Speaker 03:
the patent owner has essentially conceded that that given claim limitation necessarily means what the examiner said it meant?

[00:30:07] Speaker 02:
So I want to be clear that when you look at the lines of cases, there are lines of cases that are trying to apply sort of a stopover.

[00:30:16] Speaker 02:
And we're not saying it's an acquiescence or some undeniable concession.

[00:30:22] Speaker 02:
But this court has definitely looked at exchanges like this as informing

[00:30:27] Speaker 02:
the claim construction.

[00:30:28] Speaker 02:
So in the L'Oreal case that's discussed in the briefs, one part of that was the statement when the amendment was being made of, oh, we're just trying to capture the same thing we had before.

[00:30:42] Speaker 02:
The other thing the court's opinion noted though, was that in comparing the claims to the prior art, the applicant was distinguishing itself over prior art.

[00:30:55] Speaker 02:
It wouldn't have to,

[00:30:57] Speaker 02:
if there was a different meaning of where you measured the concentration.

[00:31:02] Speaker 02:
So that was telling.

[00:31:04] Speaker 02:
And then I think, you know, there's another case that this court has a case called Vitana.

[00:31:11] Speaker 02:
I think it's from 2006, but also it's sort of same thing.

[00:31:14] Speaker 02:
It's looking at the examiner understood this

[00:31:18] Speaker 02:
term to have a broader meaning and there's no dispute from the applicant and in fact it's amending its claims.

[00:31:26] Speaker 02:
So again it's not this isn't a rigid per se rule we're not saying you're a stopped on pain of silence but in this case where you have a claim language that doesn't provide us guidance.

[00:31:46] Speaker 02:
It matches the same understanding the examiner had during prosecution.

[00:31:52] Speaker 02:
I think it's very powerful evidence.

[00:31:53] Speaker 03:
Maybe that adds a little color.

[00:31:55] Speaker 03:
Okay.

[00:31:55] Speaker 03:
But let's get back to the real heart of the problem, which is the term hardware buffer, a term I'm not familiar with, but this patent could have just said buffer, or it could have said temporary buffer.

[00:32:15] Speaker 03:
And it does say temporary buffer in other places, which raises the inference that a hardware buffer, whatever it is, is something different than a temporary buffer.

[00:32:27] Speaker 03:
And so why shouldn't we be guided just by the word choice that we need to, in this search for the Holy Grail, at least understand that

[00:32:43] Speaker 03:
Hardware buffer is not a temporary buffer.

[00:32:47] Speaker 02:
Well, so the statements in the specification about a temporary buffer really aren't, they're not distinguishing it because it's temporary.

[00:32:58] Speaker 02:
You won't see a substantive criticism.

[00:33:00] Speaker 03:
No, I completely understand your point of view about the context of this patent distinguishing the prior art set of buffer operations from the inventive

[00:33:13] Speaker 03:
set of buffer operations, and where the temporary buffer was located before, and perhaps where, in your view, the temporary buffer can be located now.

[00:33:24] Speaker 03:
But nevertheless, I'm just more concerned about the word choice, which, again, raises the inference that hardware buffer, as that term is used not very often, but nevertheless used in this patent, including in the claims,

[00:33:41] Speaker 03:
should be understood by its nature being something different from this other noun called the temporary buffer.

[00:33:50] Speaker 03:
And so it feels a little unsatisfying to reach a construction of hardware buffer that, in terms of physical attributes, sweeps in temporary buffers.

[00:34:03] Speaker 03:
Could you please help me with that seemingly unsatisfying outcome?

[00:34:10] Speaker 02:
Right, so a few points.

[00:34:14] Speaker 02:
really setting up that hardware buffer, temporary buffer contrast.

[00:34:19] Speaker 02:
I think that the place to look for the contrast is between the hardware buffer and the system.

[00:34:25] Speaker 02:
So they're latching on to sort of this very broad term to try to draw a distinction there.

[00:34:33] Speaker 02:
And we've tried to accommodate that by saying, well, then let's think about what hardware is actually being used.

[00:34:42] Speaker 02:
And at least in a case, which is what we have in the prior art here, where you have physically separate hardware from where the image is going to ultimately be loaded, that can give meaning to the hardware buffer.

[00:34:56] Speaker 02:
It is, the ISA is a different hardware from the XRAM where it's going to be.

[00:35:04] Speaker 04:
That sentence really grates in my ear.

[00:35:09] Speaker 04:
It's in fact the identical set of memory cells, the ones that if not today, then tomorrow will serve as system memory.

[00:35:23] Speaker 04:
This is just, I think, the point that I understand, or at least I take from Judge Chen.

[00:35:30] Speaker 04:
I think you have a lot to say, but the

[00:35:35] Speaker 04:
very much hardest thing for you is to say why in the world is this word hardware being used in front of the noun buffer in some way that makes intuitive sense of that word, and not just by finding a way that a form of words can give it a non-zero meaning.

[00:35:59] Speaker 04:
If it's exactly the same memory cells that every other day of the week is being used as system memory and it is configured with the right wires and the right set of signals in exactly the same way, but it's just being used for this way station function, it's peculiar

[00:36:21] Speaker 04:
on its face to call that set of memory cells being used today for that purpose as a hardware buffer.

[00:36:31] Speaker 04:
Peculiar.

[00:36:32] Speaker 04:
Maybe so peculiar that everything else doesn't matter, or maybe not.

[00:36:37] Speaker 04:
But that, to me, is the hard thing for you.

[00:36:42] Speaker 02:
Right.

[00:36:43] Speaker 02:
I mean, one thing I'll say is there's no disagreement between the parties that a hardware buffer

[00:36:51] Speaker 02:
IDQ's buffer in the Apple trial was RAM.

[00:36:55] Speaker 02:
I'm afraid I'm deepening the mystery here, but I think we all are struggling with what Volgogne has drafted here.

[00:37:03] Speaker 02:
But on this level of the cell, for us, as you think about the Spensenbauer setup, the actual execution

[00:37:22] Speaker 02:
area will not be happening in that intermediate storage area on any day because it's being sent to the extreme.

[00:37:32] Speaker 04:
I'm sorry, but it will be, if not today, then tomorrow being happening in that very same set of memory cells.

[00:37:42] Speaker 02:
No, Your Honor.

[00:37:44] Speaker 02:
The execution of the image, the final destination,

[00:37:48] Speaker 02:
in Svensson and Bauer is the external memory, not the internal memory.

[00:37:55] Speaker 04:
This is why the concreteness matters so much to me.

[00:37:59] Speaker 04:
My understanding, and tell me if this is wrong, is that the set of memory cells

[00:38:05] Speaker 04:
labeled in Svensson, the intermediate storage area, is just a bunch of certain of the RAM cells.

[00:38:13] Speaker 04:
I forget the SA RAM and the DA RAM that are part of ... I want you to put aside your argument about what the system memory is and the X RAM over off on the side.

[00:38:25] Speaker 04:
Just please put that aside.

[00:38:27] Speaker 04:
That the intermediate storage area cells are just the very same cells that

[00:38:36] Speaker 04:
could be and the next day could easily be used for the SARAM and DARAM functions.

[00:38:46] Speaker 02:
They are cells within SARAM and DARAM.

[00:38:51] Speaker 02:
The destination where you are sending the image for execution is not SARAM and DARAM.

[00:39:01] Speaker 02:
It's the XRAM.

[00:39:02] Speaker 04:
It's a different- Right, but that, just tell me, I thought that that's actually a distinct argument from, and that's kind of your backup argument that the actual system memory is over an XRAM in Swensen, an as applied argument.

[00:39:19] Speaker 04:
And so it doesn't matter whether the ISA is temporary or not.

[00:39:25] Speaker 04:
Whatever else it is, it's never an XRAM.

[00:39:28] Speaker 04:
I think that's just a whole different argument.

[00:39:31] Speaker 02:
Right.

[00:39:31] Speaker 02:
So the other, I guess, uncertainty here is we, in terms of, oh, those same cells will be used for something else tomorrow, I'm really not sure that's right in the sense that they're allocated right at runtime.

[00:39:49] Speaker 02:
They're never deallocated.

[00:39:51] Speaker 02:
I'm sorry.

[00:39:52] Speaker 04:
Wait a minute.

[00:39:54] Speaker 04:
They're never deallocated until the power goes off.

[00:39:58] Speaker 04:
And then they're deallocated.

[00:40:01] Speaker 04:
No?

[00:40:02] Speaker 02:
You're right, correct.

[00:40:02] Speaker 02:
When the power goes off, any RAM, the accused RAM in the Apple trial, isn't storing anything.

[00:40:09] Speaker 02:
It isn't serving as memory when the power goes off.

[00:40:12] Speaker 02:
But what I was saying is, there's no, Spenson doesn't then say, oh, allocate somewhere else tomorrow.

[00:40:19] Speaker 02:
It may well be, and it's certainly vendor obvious, having software that's just allocating

[00:40:24] Speaker 02:
the same cells any time.

[00:40:27] Speaker 04:
But it could be, but there's no... I take it that the entire difference here is the difference between whether the same memory cells can be used for the place of execution of the software after it's finished with the buffer or cannot.

[00:40:50] Speaker 04:
And their position is cannot, your position is can.

[00:40:55] Speaker 02:
Right, Ryan.

[00:40:56] Speaker 02:
And maybe one way out of this dilemma is if we connect it to the purported benefits of the invention.

[00:41:06] Speaker 04:
Well, I guess I'm going to just, for purposes of this discussion, agree with you that the point that Judge Chen was discussing a good deal, which is that the copying and header and functions could just as easily be performed with an allocate

[00:41:26] Speaker 04:
standard RAM memory cells for the day.

[00:41:32] Speaker 04:
I want to accept that for this.

[00:41:35] Speaker 04:
Nevertheless, you would have all those benefits

[00:41:40] Speaker 04:
just from using the word buffer in the claim.

[00:41:44] Speaker 04:
And this claim doesn't say buffer, it says hardware buffer.

[00:41:48] Speaker 04:
So what is it that that adds?

[00:41:51] Speaker 04:
I don't mean what benefits it adds, presumably it does add an allocation benefit and maybe there's some extremely, there's no express calling out of that as a benefit, though there is a mention of allocation.

[00:42:04] Speaker 04:
And I don't think Dr. Reinhard

[00:42:08] Speaker 04:
touts that as a benefit, but just getting back to this point, it seems very peculiar to add the word adjective in front of that buffer if there isn't some, and now I'll just use the word permanent.

[00:42:26] Speaker 04:
Today, tomorrow, and every other day, limitation of the use of those memory cells to the buffer function to the exclusion of using them for the rest of system memory function.

[00:42:38] Speaker 02:
Well, let me take another shot at it.

[00:42:45] Speaker 02:
The reason it's important to link this to the purpose here is because that permanence isn't getting you those benefits.

[00:42:52] Speaker 02:
The physical separation, viewing it as, well, this is a hardware buffer in the sense that it's in different hardware from our final destination.

[00:43:04] Speaker 02:
That can be helpful.

[00:43:06] Speaker 02:
Because if you're trying to read out of a memory and copy into that same memory, there can be inefficiencies.

[00:43:15] Speaker 02:
But as our expert talked about in his remand reply declaration, as long as you're taking it and you have sort of one piece of hardware over here, whether it is already allocated or allocated at runtime set aside doesn't matter.

[00:43:32] Speaker 02:
But as long as that is one piece

[00:43:35] Speaker 02:
and your destination for execution is on another piece of hardware, you don't end up reading and writing in and out of the same memory, you can be very efficient moving it over from one piece of hardware to another.

[00:43:48] Speaker 02:
So we think that's in the search for meaning for hardware, that that is a way to give meaning to the term that actually comports with an efficiency

[00:44:02] Speaker 02:
comport with the column five disclosure about ram 112 which is about where it's talking about sort of the final destination and the temporary buffer there so it's copying in and out of that same ram note we're going from one ram the internal one to the other and that gives me harder but i can see your honor this is a struggle

[00:44:22] Speaker 04:
Just where, I think you referred to, is it Dr. Lin?

[00:44:27] Speaker 04:
Is that who your expert is?

[00:44:29] Speaker 04:
Correct, yes.

[00:44:30] Speaker 04:
You said, where did he explain that something concrete about the efficiency benefits of staying... I have an intuition about this one, but where did he explain about the efficiencies

[00:44:48] Speaker 04:
of having basically to address only some physically distinguished component, part of memory for one function and the other function.

[00:45:04] Speaker 04:
Skip a whole bunch of the bits on the address lines because you're always within the middle 12 bits or something.

[00:45:15] Speaker 02:
So on 2457 of the appendix, he talks about the efficiency is due to the separation.

[00:45:25] Speaker 02:
And then he follows up at 2470 of the appendix to elaborate on this same point and talk about how, you know, because you have the separation means instead of copying and writing back into the same memory, the data segments are copied to the distinct grain.

[00:45:46] Speaker 02:
And he says on that same page, two, four, seven, zero, it is just as fast and efficient.

[00:45:54] Speaker 02:
And he talks about it.

[00:45:55] Speaker 02:
So that's because with a terminal memory and balance fence, it's operating the speed of the processor.

[00:46:00] Speaker 02:
But so those are the relevant passages where he's saying, as long as you're maintaining this physical separation between those two memories, you're getting the efficiency.

[00:46:11] Speaker 02:
And my broader point is if we,

[00:46:15] Speaker 02:
are struggling this much with this term than under the VRI standard and under the public notice function.

[00:46:22] Speaker 02:
The fact that we can give a meaning to hardware that comports with an efficiency benefit and is consistent with the claims and specification and is the same understanding the examiner had in prosecution.

[00:46:37] Speaker 02:
surely has to mean that under the BRI standard, they can't come back and narrow this.

[00:46:42] Speaker 02:
And as I started, I think there was a deliberate attempt.

[00:46:45] Speaker 02:
Qualcomm was the one in control here.

[00:46:48] Speaker 02:
It chose the term to use, didn't use a narrower term.

[00:46:51] Speaker 02:
It didn't provide this additional guidance and the specification and distinguish its claims.

[00:46:55] Speaker 02:
It thought it could get these claims issued on other grounds.

[00:46:58] Speaker 02:
At every step along the way,

[00:47:01] Speaker 02:
to file a motion to amend in the IPR itself, it could have brought clarity here.

[00:47:08] Speaker 02:
And when it didn't, and when it tried to maintain that breadth there, then we should be okay to interpret the claims the same way the examiner did during prosecution and have them read on a priority system that, as we've been discussing, achieves every benefit that they are saying that they achieved.

[00:47:32] Speaker 01:
Any other questions?

[00:47:33] Speaker 01:
Thank you.

[00:47:41] Speaker 01:
Ms.

[00:47:41] Speaker 01:
Sweezy, will we store three minutes of rebuttal if you need it?

[00:47:46] Speaker 00:
Thank you, Your Honor.

[00:47:47] Speaker 00:
I will try to be quick and go through several points.

[00:47:49] Speaker 00:
First, Intel's expert Dr. Lin and Intel are overreaching on broadest

[00:47:55] Speaker 00:
reasonable interpretation.

[00:47:56] Speaker 00:
It's not hypotheticals.

[00:47:57] Speaker 00:
It's not what other systems could do.

[00:47:59] Speaker 00:
It's what our system does.

[00:48:01] Speaker 00:
You have to look at the intrinsic evidence.

[00:48:04] Speaker 00:
Intel says there's no guidance in the claim language or specification.

[00:48:07] Speaker 00:
This court already disagreed with that.

[00:48:10] Speaker 00:
And on that guidance, at column one, lines 30 to 31, the patent makes very clear, this is devoted to efficiently communicating data.

[00:48:20] Speaker 00:
And it achieves that this particular configuration, which is exactly what we accused and a jury had no trouble finding in the litigation, we accused a permanent dedicated hardware buffer in the sense that it is not a temporary buffer that is allocated.

[00:48:36] Speaker 00:
That buffer is the way that Qualcomm system achieves avoiding copying.

[00:48:43] Speaker 00:
If you let it be a temporary buffer, you're right back to what the prior art did and what Qualcomm distinguished.

[00:48:48] Speaker 03:
Miss Sweezy, sorry to interrupt, but your theory that a hardware buffer is a permanent buffer, can you explain why is it that a so-called permanent buffer is something that is limited to using the exact same memory cells forever?

[00:49:08] Speaker 03:
That's another little gap that I never completely understood.

[00:49:13] Speaker 03:
I mean, we seem to be assuming that the term permanent buffer

[00:49:18] Speaker 03:
necessarily means this particularized kind of configuration of a memory, of a buffer.

[00:49:26] Speaker 03:
And I didn't quite pick up on why that must necessarily be so.

[00:49:31] Speaker 03:
And I don't know.

[00:49:32] Speaker 03:
Why can't there be other kinds of hardware buffers that don't necessarily require that?

[00:49:39] Speaker 03:
Or why can't it be that there are other kinds of permanent buffers that aren't necessarily

[00:49:47] Speaker 03:
preordained in that same manner.

[00:49:50] Speaker 00:
I understand the point, Judge Chen.

[00:49:52] Speaker 00:
And we are defining hardware buffer in the context of this patent.

[00:49:55] Speaker 00:
So not suggesting that every hardware buffer has to be permanent.

[00:49:58] Speaker 00:
But here, as we've noted, as the court noted in the prior opinion, we didn't just say buffer.

[00:50:05] Speaker 00:
Both in the claims, we said hardware buffer.

[00:50:07] Speaker 00:
And we didn't just say buffer in discussing the prior art.

[00:50:11] Speaker 00:
We said temporary buffer.

[00:50:14] Speaker 00:
So in the context of this patent,

[00:50:16] Speaker 00:
This patent is using a set aside dedicated space for a dedicated purpose.

[00:50:23] Speaker 00:
I think of it as my desk.

[00:50:24] Speaker 00:
I can move things around on my desk and one day devote this area to this set of briefs and another move it around and another day devote another set.

[00:50:32] Speaker 00:
But I have a permanent drawer that's going to consistently store my pens and paperclips.

[00:50:39] Speaker 00:
And maybe it's not permanent in the sense of being philosophically, I'm not going to get into the mechanics of it, but that's what we're talking about here.

[00:50:46] Speaker 00:
using the permanent space so that you can avoid having to create a temporary buffer.

[00:50:51] Speaker 00:
So they are linked.

[00:50:52] Speaker 00:
And it's not just enough to say it's distinct from system memory as this court also already noted.

[00:50:59] Speaker 00:
Let me say there's nothing that turns on the type of memory here.

[00:51:01] Speaker 00:
There is no such argument by Intel in its briefing and no such finding by the fore.

[00:51:05] Speaker 00:
That is a red herring.

[00:51:07] Speaker 00:
What matters if you want to think about it in terms of type of memory, it's special purpose memory versus general purpose memory, system memory.

[00:51:14] Speaker 00:
everybody agrees, Dr. Lin included, Appendix 2172, that that's general purpose.

[00:51:20] Speaker 00:
This is a special purpose, and it can't be a special purpose if it could be something different every time.

[00:51:25] Speaker 00:
In terms of the number of copy operations, that is too abstract a level.

[00:51:29] Speaker 00:
Our claim is not about reducing from X to Y copy operations.

[00:51:33] Speaker 00:
It's the way we did it.

[00:51:34] Speaker 00:
that's in the specification and the intrinsic record and the prosecution history a whole host of reasons that did not directly address the term hardware buffer at appendix 1563 qualcomm itself said we're only giving some examples here why the way we uh read over the prior art not exhaustive and i would point the court to that i see most of the time thank you your honor thank you we thank both sides the case system