Ex Parte Chakraborty et al

Patent Trial and Appeal Board

Dec 27, 2016

12874235 (P.T.A.B. Dec. 27, 2016)

United States Patent and Trademark Office
UNITED STATES DEPARTMENT OF COMMERCE
United States Patent and Trademark Office
Address: COMMISSIONER FOR PATENTS
P.O.Box 1450
Alexandria, Virginia 22313-1450
www.uspto.gov
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR ATTORNEY DOCKET NO. CONFIRMATION NO.
12/874,235 09/02/2010 Parag Chakraborty 330051.01 6823
69316 7590 12/29/2016
MICROSOFT CORPORATION
ONE MICROSOFT WAY
REDMOND, WA 98052
EXAMINER
HILL, SCHILLER D
ART UNIT PAPER NUMBER
2614
NOTIFICATION DATE DELIVERY MODE
12/29/2016 ELECTRONIC
Please find below and/or attached an Office communication concerning this application or proceeding.
The time period for reply, if any, is set in the attached communication.
Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the
following e-mail address(es):
u sdocket @ micro soft .com
chriochs @microsoft.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte PARAG CHAKRABORTY, HAO ZHANG,
PAREEKSHIT SINGH, and PANDELE STANESCU1
Appeal 2016-000873
Application 12/874,235
Technology Center 2600
Before ALLEN R. MacDONALD, HUNG H. BUI, and
MICHAEL M. BARRY, Administrative Patent Judges.
BARRY, Administrative Patent Judge.
DECISION ON APPEAL
Appellants appeal under 35 U.S.C. § 134(a) from a Non-Final
Rejection of claims 1—20. We have jurisdiction under 35 U.S.C. § 6(b).
We AFFIRM.
Introduction
Appellants’ claims are directed to systems and methods in which a
virtual machine (VM) has a virtual graphics processing unit (VGPU) that
may be reset or restarted. (Abstract, Claims.) Claim 1 is illustrative:
1. A system comprising:
a processor;
1 Appellants identify the real party in interest as Microsoft Technology
Licensing, LLC. (App. Br. 1.)
Appeal 2016-000873
Application 12/874,235
a first virtual machine operable on said processor, said first
virtual machine having a first virtual graphics processing unit;
a hypervisor operable on said processor, said hypervisor
comprising a graphics processor unit thread configured to receive
instructions from a virtual graphics processing unit and cause
said instructions to be performed; and
said hypervisor further configured to detect that said first
virtual machine is being restored and cause said first virtual
graphics processing unit to be reset.
(App. Br. 8 (App’x A).)
References and Rejections
(1) Claims 1—9 and 17—19 stand rejected under 35 U.S.C. § 103(a)
as obvious over Micah Dowty et al., GPU Virtualization on VMware’s
Hosted I/O Architecture, 43(3) ACM SIGOPS Operating Systems Review,
73-82 (July 2009) (“Dowty”) and Belay (US 2010/0250824 Al; publ. Sept.
30, 2010). (Non-Final Act. 2-18.)
(2) Claims 11—16 stand rejected under 35 U.S.C. § 103(a) as
obvious over Belay and Dowty. (Non-Final Act. 19—24.)
(3) Claims 10 and 20 stand rejected under 35 U.S.C. § 103(a) as
obvious over Dowty, Belay, and Arndt et al. (US 2006/0195673 Al; publ.
Aug. 31, 2006) (“Arndt”). (Non-Final Act. 24-29.)
ANALYSIS
We have reviewed the Examiner’s rejections in light of Appellants’
contentions of reversible error. We disagree with Appellants’ conclusions.
We adopt the Examiner’s findings and reasons as set forth in the rejection
from which this appeal is taken and as set forth in the Answer. We highlight
the following for emphasis.
2
Appeal 2016-000873
Application 12/874,235
Motivation to Combine Dowty and Belay
Appellants argue the Examiner errs in finding motivation to combine
the teachings of Dowty and Belay because essentially it means “one of
ordinary skill in the art would be motivated to emulate already emulated
devices. ... If one were to emulate an already emulated device the result
would be a more complicated and less efficient system than already exists.”
(App. Br. 9.) We find this unpersuasive. The Examiner responds, and we
agree, that “the combination of Dowty with Belay would use the restore/
migrate functionality of Belay in the virtualization system of Dowty” so that
“when the system of Dowty performs the restore/migration function of
Belay, the reinitialization would be on the VGPU present in Dowty.” (Ans.
3; see also Non-Final Act. 4—5).) The Examiner also finds, and we also
agree, that by using Belay’s reinitialization/reset method during Dowty’s
migration or save/restore operation, “the hybrid implementation of the
mediated pass-through VGPU would be reinitialized in the same manner as
the pass-through device of Belay. This combination would result in
significant improvement in more advanced interposition features such as live
migration and migration across different models of physical GPUs.” (Ans. 5
(citing Belay 118).)
Regarding emulation of an already emulated device, the Examiner
responds by finding, and we agree, that:
Belay does not generate an emulator/emulated device for the
pass-through device. The mappings of the pass-through device
are replaced with mappings of an emulation handler, in effect
freezing the state of the pass-through device since the emulation
handler [responds] to requests as though the device is
experiencing a hardware failure. The state of the virtual machine
is then migrated, and the emulation handlers trigger a
3
Appeal 2016-000873
Application 12/874,235
reinitialisation of the pass-through device. The emulation
handlers are not emulated devices.”
(Ans. 8.)
Appellants reply that Belay’s emulation handlers are emulated
versions of the pass-through devices because, during migration, “the
emulated handlers while not being fully functional take the place of the
actual pass through device and report that they have failed.” (Reply Br. 2.)
We find this unpersuasive. Belay’s emulation handlers simulate the pass­
through device. (See, e.g., Belay Abstract (“the pass-through device is
substituted with an emulation handler that simulates the pass-through device
experiencing errors”).) Simulation is not emulation. (See, e.g., https://
en.wikipedia.org /wiki/Emulator (introducing the emulator topic with “/n]ot
to be confused with Simulator”).)
Claim 1
Appellants also argue the Examiner errs in finding Belay teaches or
suggests claim l’s requirement to “detect that said first virtual machine is
being restored and cause said first virtual graphics processing unit to be
reset.” (App. Br. 10.) Appellants contend “the process in Belay does not
reset the virtual device; rather, it simply removes the emulator from the
system and reconfigures the portion of the physical pass through device.”
(Id.) Appellants further contend:
in Belay the requests are initiated or controlled by the device
drivers and not by the hypervisor itself in response to an error
signal generated by the hypervisor. The hypervisor in Belay does
not have the ability to actually reset the device; it only produces
the signal that causes components in the virtual machine to
generate commands to reinitialize a pass through device.
(Id. at 11.)
4
Appeal 2016-000873
Application 12/874,235
Appellants do not persuade us. We agree with the Examiner’s finding
that “removal of the emulation handler [i.e., the emulator Appellants refer
to] is not considered to be a reset. The VGPU of Dowty would be the device
that is reinitialized [(i.e., reset in the vernacular of the claim)].” (Ans. 11.)
We also agree with the Examiner that “intercepting and forwarding of the
[reset] request as Belay teaches ([0005]) would be considered as ‘causing’
the [reset]” as claimed. {Id. at 12 (replacing “reinitialization” with “reset”;
bracketed paragraph number in original).)
We accordingly sustain the rejection of claim 1, and the rejections of
its dependent claims 2—10, for which Appellants provide no separate
arguments.
Claim 17
Independent claim 17 recites a method that requires, inter alia,
“detecting that said first virtual machine has resumed” and “causing said
first virtual graphics processing unit to be restarted based on said resume by
resetting the first virtual graphics processing unit without resetting another
virtual graphics processing unit associated with another virtual machine
operating on the hypervisor” (App. Br. 13). Appellants argue the
Examiner errs in finding Belay teaches resetting one VGPU without
resetting another associated VGPU as claimed because:
Belay does not have the capability of restarting only the portion
of the device associated with the virtual machine without
affecting the device associated with other virtual machines. In
Belay, if a restart or reset occurred on the pass through device
that restart or reset would impact other virtual machines
operating on the pass through device.
(App. Br. 14.)
5
Appeal 2016-000873
Application 12/874,235
The Examiner responds by finding basic hypervisor functionality
includes the capability to manage multiple VMs and selectively reset an
individual VM:
A hypervisor may manage many virtual machines which are
operating on many virtual hardware platforms. Each individual
VM interacts with emulated devices on its virtual hardware
platform, and therefore, any requests to restore/migrate a VM or
any requests to reinitialize, restart, or reset a virtual/emulated
device on the VM’s virtual hardware platform (VHP) would be
performed without [a]ffecting any other VM/VHP that [is]
concurrently running in the execution space.
(Ans. 16.)
We agree with the Examiner. We also note Appellants provide no
rebuttal to these findings (see Reply Br. 1 4). We accordingly sustain the
rejection of claim 17, and the rejections of its dependent claims 18—20, for
which Appellants provide no separate arguments.
Claim 11
Appellants argue the Examiner errs in rejecting independent method
claim 11 for reasons similar to claim 1. For the same reasons as for claim 1,
we sustain the rejection of claim 11, and the rejection of its dependent
claims 12—16, for which Appellants provide no separate arguments.
DECISION
For the above reasons, we affirm the rejections of claims 1—20. No
time period for taking any subsequent action in connection with this appeal
maybe extended under 37 C.F.R. § 1.136(a)(l)(iv).
AFFIRMED
6