DETAILED ACTION
This action is in response to communication filed on 4/6/2026.
Claims 1-20 are pending.
Claims 1, 9, 11, and 19 have been amended.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Arguments
Applicant's arguments filed 4/6/2026 have been fully considered but they are not persuasive.
In the communication filed, applicant argues in substance that:
Klenk does not teach the amended “primitive” limitation. Klenk disclose only MCR (multicast region) address-range setup and reduction tables via the fabric manager. These are “memory address configuration, not primitive.” A primitive must be “indicative of an analytical, computational, or transformative operations” and one of “a set of primitives” tied to a protocol supported by network device. Applicant contends the Office Action’s mapping of “MCR setup” to the primitive is incorrect.
In response to argument [a], Examiners respectfully disagrees.
Klenk [0046] states that packets addressed to MCRs “can be treated as primitives for implementing in-network computations”. The MCR setup received via the fabric manager (the control/configuration mechanism) both defines the multicast region and indicates the specific analytical, computational, or transformative operation (e.g., reduction) that the network device will perform on payloads when packets are addressed to that region. Thus, the MCR setup is not mere address configuration, it is the mechanism that configures and indicates the operation the network device executes, disclosing the amended limitation of claim 1, 11, and 19. The rejection is therefore maintained for the reason set forth in the detailed claim mapping below.
Double Patenting
Applicant's arguments filed 4/6/2026, with respect to the provisional nonstatutory double patenting rejection of claims 1, 11 and 19 have been fully considered but they are not persuasive. Applicant asserts only that the claims have been amended to further distinguish from the cited references of copending Application No. 18/401,309. This argument is purely conclusory. MPEP § 804(I)(B)(1) requires a complete response that addresses why the claims are now patentably distinct. Applicant has not explained how the new “set of primitives…protocol” language overcomes the copending claims, which the Non-Final Office Action already identified as covering the same in-network SDN acceleration for AI collectives via header-driven operation. The provisional rejection therefore stands until a terminal disclaimer is filed or substantive distinction is shown.
The nonstatutory double patenting rejection is based on a judicially created doctrine grounded in public policy (a policy reflected in the statute) so as to prevent the unjustified or improper timewise extension of the “right to exclude” granted by a patent and to prevent possible harassment by multiple assignees. A nonstatutory double patenting rejection is appropriate where the conflicting claims are not identical, but at least one examined application claim is not patentably distinct from the reference claim(s) because the examined application claim is either anticipated by, or would have been obvious over, the reference claim(s). See, e.g., In re Berg, 140 F.3d 1428, 46 USPQ2d 1226 (Fed. Cir. 1998); In re Goodman, 11 F.3d 1046, 29 USPQ2d 2010 (Fed. Cir. 1993); In re Longi, 759 F.2d 887, 225 USPQ 645 (Fed. Cir. 1985); In re Van Ornum, 686 F.2d 937, 214 USPQ 761 (CCPA 1982); In re Vogel, 422 F.2d 438, 164 USPQ 619 (CCPA 1970); In re Thorington, 418 F.2d 528, 163 USPQ 644 (CCPA 1969).
A timely filed terminal disclaimer in compliance with 37 CFR 1.321(c) or 1.321(d) may be used to overcome an actual or provisional rejection based on nonstatutory double patenting provided the reference application or patent either is shown to be commonly owned with the examined application, or claims an invention made as a result of activities undertaken within the scope of a joint research agreement. See MPEP § 717.02 for applications subject to examination under the first inventor to file provisions of the AIA as explained in MPEP § 2159. See MPEP § 2146 et seq. for applications not subject to examination under the first inventor to file provisions of the AIA . A terminal disclaimer must be signed in compliance with 37 CFR 1.321(b).
The filing of a terminal disclaimer by itself is not a complete reply to a nonstatutory double patenting (NSDP) rejection. A complete reply requires that the terminal disclaimer be accompanied by a reply requesting reconsideration of the prior Office action. Even where the NSDP rejection is provisional the reply must be complete. See MPEP § 804, subsection I.B.1. For a reply to a non-final Office action, see 37 CFR 1.111(a). For a reply to final Office action, see 37 CFR 1.113(c). A request for reconsideration while not provided for in 37 CFR 1.113(c) may be filed after final for consideration. See MPEP §§ 706.07(e) and 714.13.
The USPTO Internet website contains terminal disclaimer forms which may be used. Please visit www.uspto.gov/patent/patents-forms. The actual filing date of the application in which the form is filed determines what form (e.g., PTO/SB/25, PTO/SB/26, PTO/AIA /25, or PTO/AIA /26) should be used. A web-based eTerminal Disclaimer may be filled out completely online using web-screens. An eTerminal Disclaimer that meets all requirements is auto-processed and approved immediately upon submission. For more information about eTerminal Disclaimers, refer to www.uspto.gov/patents/apply/applying-online/eterminal-disclaimer.
At least claims 1, 11, and 19 are provisionally rejected on the ground of nonstatutory double patenting as being unpatentable over claims 1, 10, and 17 of copending Application No. 18/401/309 (reference application). Although the claims at issue are not identical, they are not patentably distinct from each other because the independent claims both target SDN for AI workloads (e.g., collectives lieks all-reduce), in-network operation (aggregation, broadcast), and acceleration by network devices (e.g., swtiches with DPU). The copending application VXLAN-based implementation would be an obvious variants of Kandula’s primitive/header-based approach to one skilled in the art. Both solve the same problem (offloading AI operations to network for latency reduction) using similar techniques (header-driven payload operation in SDN devices).
This is a provisional nonstatutory double patenting rejection because the patentably indistinct claims have not in fact been patented.
Claim Rejections - 35 USC § 101
The rejection presented in the previous Office Action is withdrawn in view of response filed on 4/6/2026.
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claims 1-6, 11-14, and 19 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Klenk et al. (US 2021/0036877).
Regarding claim 1, Klenk discloses a method for processing data packets in a computing network processing workloads in a network device operating in a software defined network (SDN), the network device configured to execute primitives for data payloads in the packets associated with the workloads in the network prior to forwarding the packets to hosts in the SDN, the method comprising:
receiving, by the network device via a control plane, a primitive indicative of an analytical, computational, or transformative operation to be performed on data payloads transmitted by data packets associated with a workload being processed in the SDN (Klenk discloses the network device receiving configuration (i.e., the primitive as MCR setup) via the fabric manger (functioning as a control plane), where the primitive defines regions for in-network computation (computational operations like reductions) on packet payload associated with workload (e.g., distributed ML training involving collective primitives); see [0046] “regions of addresses in the shared global address space can be defined as multicast regions, and packets addressed to addresses in these multicast regions can be treated as primitives for implementing in-network computations” and [0051] “the network 100 includes a fabric manager 150 that is connected to the network device 110. The fabric manager 150 is configured to setup and manage special MCRs within a global address space shared within the network 100”.)
wherein the primitive is associated with a protocol for configuring network devices to perform in-network acceleration of workloads in coordination with source and destination hosts in the SDN (Klenk discloses the primitive is tied to a protocol (shared global address space management, including endpoint registration and MRT programming) that configures devices for in-network acceleration (offloading reduction to reduce latency/bandwidth for workloads like ML parameter aggregation) in coordination with host (endpoints as source/destination register allocations and participate via load/store to MCR addresses). In other words, this discloses the association because the protocol enables coordinated acceleration between devices and hosts in the network; see [0049] “scalable in-network computations, such as a reduction operation, can be performed in the network 100 by off-loading the computation to the logic 130 in the network device 110 (or devices) rather than performing the computations on one of the endpoints” and [0153] “at step 702, the fabric manager 150 is configured to create an MCR. As part of step 702, each endpoint registers an existing memory allocation with the fabric manager 150”);
the primitive being one of a set of primitives that indicate analytical, computational, or transformative operations to be performed by network devices that support the protocol (Klenk discloses a set of such primitives (broadcast primitive, multicast primitive, and reduction/all reduced primitive) that indicate analytical, computational, or transformative operations (e.g., one-to-many transmission, subset transmission, or many-to-one aggregation/reduction of data such as gradients in ML training) to be performed by network devices that support the protocol (the MCR/shared global address space protocol); [0150] “a first collective communication primitive is a Broadcast primitive. The Broadcast primitive enables one endpoint to transmit data (e.g., a payload) to every other endpoint participating in the network. In an embodiment, a second collective communication primitive is a Multicast primitive. The Multicast primitive is narrower than a Broadcast primitive in that the Multicast primitive enables one endpoint to transmit data to a subset of endpoints participating in the network” and [0147] “Although many collective patterns are only concerned with data distribution (e.g., transmitting data from one endpoint to another endpoint), some collective patterns such as Reduce or All-Reduce operations also involve computations…This can be implemented as an All-Reduce operation that is called during every training iteration”. Thus, Klenk’s MCR setup via the fabric manager is not mere address configuration, it is the mechanism that both defines and indicates the specific operations the network device will perform on payloads when packets addressed to the MCR are received);
receiving, by the network device, a data packet associated with the workload being processed in the computing network (Klenk discloses a network device receiving a pull request associated with a collective communication primitive; see [0162] “A pull request refers to a load operation associated with a collective communication primitive. A load operation is essentially a request to read a memory location or range of memory addresses in the shared memory space of the network. In the case of an All-Reduce primitive associated with m data elements and p endpoints, each participating endpoint is configured to receive m/p pull requests from the network, each separate pull request is associated with a different network address in the range of network addresses associated with an MCR”);
determining, by the network device, that the packet is associated with the workload being processed in the computing network based on a transmission identifier and a parameter indicated by the primitive, the transmission identifier and parameter contained in a header of the data packet (Klenk discloses the network device decoding the packet header to read a network address (transmission identifier) and determine association with the MCR-based workload; the MCR setup (via fabric manager for primitives) indicates parameters like the reduction operator, which can be in the pull request (packet) or MRT queried from the header address, effectively basing determination on header-contained identifier and primitive-indicated parameter; see [0025] “a reduction operator is specified in at least one of the pull request or a multicast region table” and [0155] “the header can include a field for a destination address that specifies a network address included within the range of addresses allocated to the MCR. The network device 110 includes logic 130 that decodes the header of the packet and determines that the destination address is associated with a network address corresponding to the MCR. Responsive to determining the network address corresponds with the MCR, the logic 130 looks up an entry in the MCR table 132 corresponding to the MCR in order to identify the participating endpoints in the network”);
in response to the determining, performing, by the network device, the operation on a payload of the data packet (Klenk discloses performing a reduction operation (e.g., combine) on the payload in direct response to determining the packet’s association via header address and table checks; see [0170] “as each response is received and decoded by a network device, the network device checks the network address included in the response to determine if the network address is associated with an MCR and, if the network address is associated by the MCR, checks the reduction table to determine if the response is associated with an entry in the reduction table. If the response is associated with the entry in the reduction table, then the logic 130 decodes the payload of the response and combines the payload with the intermediate reduction result in the reduction table”); and
sending, by the network device, the processed data packet to a destination address identified in the header of the packet (Klenk [0190] at step 912, a pull response is generated by the network device and is forwarded to at least one participating endpoint. A payload of the pull response includes the intermediate result calculated based on one or more responses associated with the collective communication primitive received at the network device from two or more participating endpoints).
Regarding claim 2, Klenk discloses the method of claim 1, wherein the workload comprises an artificial intelligence (AI) workload (see Klenk [0147] “the All-Reduce operation is especially critical in parallel deep-learning training algorithms on large distributed systems. For example, after each endpoint adjusts the parameters of a neural network based on a loss function applied to results from a batch of training samples, the parameters are shared with all other endpoints associated with different batches of training samples. This can be implemented as an All-Reduce operation that is called during every training iteration”).
Regarding claim 3, Klenk discloses the method of claim 1, wherein the network device comprises one of a network interface card (NIC), a network switch, or a disaggregated pool of NICs (Klenk [0047] “each endpoint can also include a networking capability, such as a network interface controller (NIC) configured to communicate with the network device 110 via one or more communications protocols such as Ethernet”).
Regarding claim 4, Klenk discloses the method of claim 1, wherein the workload comprises a collective (Klenk [0150] “a collective communication primitive comprises a message (e.g., a data packet) associated with a network address”).
Regarding claim 5, Klenk discloses the method of claim 1, wherein the operations comprise multicast, in-network aggregation, or gradient compression (Klenk [0150] “a second collective communication primitive is a Multicast primitive. The Multicast primitive is narrower than a Broadcast primitive in that the Multicast primitive enables one endpoint to transmit data to a subset of endpoints participating in the network”).
Regarding claim 6, Klenk discloses the method of claim 1, wherein the workload comprises one or more of a one-to-one transformation, one-to-many transformation, or many-to-one transformation (Klenk [0150] “a first collective communication primitive is a Broadcast primitive. The Broadcast primitive enables one endpoint to transmit data (e.g., a payload) to every other endpoint participating in the network”).
Regarding claim(s) 11-14 and 19 do(es) not teach or further define over the limitation in claim(s) 1, 4-6 and 1 respectively. Therefore claim(s) 11-14 and 19 is/are rejected for the same rationale of rejection as set forth in claim(s) 1, 4-6 and 1 respectively.
Further, for claim 11, and 19, the limitation “wherein the primitive is associated with a protocol for configuring network devices to perform in-network acceleration of workloads in coordination with source and destination hosts in the SDN” Klenk discloses that this primitive/MCR setup is associated with a protocol (shared global address space management, including endpoint registration and MRT programming) for configuring network devices to perform in-network acceleration of workloads in coordination with source and destination hosts in the SDN (endpoints as source/destination register allocation and participate via load/store to MCR addresses; [0049] “Scalable in-network computations, such as a reduction operation, can be performed in the network 100 by off-loading the computation to the logic 130 in the network device 110 (or devices) rather than performing the computations on one of the endpoints or, as described in the conventional solutions using a ring-scheme algorithm, by spreading the computation around all of the participating endpoints” and [0153] “At step 702, the fabric manager 150 is configured to create an MCR. As part of step 702, each endpoint registers an existing memory allocation with the fabric manager 150”. In other words, the protocol enables coordinated acceleration between the network device and the host).
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102 of this title, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 7-10, 15-18 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Klenk et al. (US 2021/0036877) in view of Lee et al. (US 2020/0313999).
Regarding claim 7, Klenk discloses the method of claim 1, wherein the primitive is received by the network device via an application programming interface (API) operable to receive a message indicative of the primitive and instructions to the network device for executing the primitive (Klenk [0153] “a range of network addresses can be allocated to the fabric manager 150 and requests to map a memory allocation in the endpoint to a particular MCR can be made by a write request to one of the network addresses allocated to the fabric manager 150…. an application can invoke a driver to request a memory region be allocated as an MCR. The driver is a component associated with the fabric manager 150 and enables an application to call the fabric manager 150 to create the MCR in the shared network address space”).
Klenk teaches receiving the primitive configuration via a driver/API-like interface, but the driver is application-invoked rather than explicitly an API on the network device itself. Therefore, Klenk does not explicitly disclose wherein the primitive is received by the network device via an application programming interface (API).
Lee in the field of the same endeavor discloses techniques for utilizing programmable packet engines for cross-platform network testing, including test packets generation and recirculation, header modification, state injection, multi-table classification, and synchronized timestamping. In particular, Lee teaches the following:
wherein the primitive is received by the network device via an application programming interface (API) (Lee discloses the network device (programmable packet engine) receiving primitives via a control plane API using protocols like gRPC with protobuf, where the API receives messages for configuring operation (primitives) and instructions for execution in match-action units (MAUs); see [0030] “a programmable packet engine can be programmable by configuring software executed by the forwarding element. A device user, testing engineer, control plane, or device manufacturer can load the configuration software into the packet engine”).
Therefore, it would have been obvious to a person of ordinary skill in the art at the time the invention was effectively filed to modify the prior art with the teaching of Lee to incorporate techniques for utilizing programmable packet engines for cross-platform network testing. One would have been motivated to combine the prior art with Lee because Lee’s teachings would improve the flexibility, efficiency, and programmability of packet payload operation in software-defined networks for workloads like data analytics and machine learning.
Regarding claim 8, Klenk-Lee discloses the method of claim 7, wherein the API is configured to receive a control path and a data path for a primitive (Lee [0034] “network tester including a programmable packet engine that provides at least the following: Start/Stop sending/receiving traffic on a port”).
Regarding claim 9, Klenk-Lee discloses the method of claim 7, wherein the control path includes a transformation identifier, a match clause, and transformation information (Klenk [0010] ”the identifying step comprises: reading a network address from a header of the pull request, querying a multicast region table based on the network address, and identifying the one or more participating endpoints based on information included in a corresponding entry of the multicast region table. The multicast region table maps ranges of addresses in a shared global address space to multicast regions” and Lee [0159] “to generalize, the wide match key space can be divided into N number of sub-tables, e.g., t_1, t_2, . . . t_{N−1}, t_N, as long as the match type required for fields in t_1 to t_{N−1} is exact-match. Only the last table t_N does ternary or range match with different priorities. The chaining can be performed by returning a match index from previous t_{i−1} and exact-matches on the index in the next table t_i.”).
Regarding claim 10, Klenk-Lee discloses the method of claim 9, wherein the match clause includes one of an offset from which to obtain values of the transformation identifier, or a Boolean valued filter expression (Lee [0155] “referring to FIG. 10A, a received packet has fields A and B with respective values of 45 and 50. A first table, Table 1, includes a range of values of Field A (e.g., 40-50 and 0-50). In Table 1, a match on an entry with a range 40-50 occurs with an index of 20”).
Regarding claim(s) 15-18 and 20 do(es) not teach or further define over the limitation in claim(s) 7-10 and 7 respectively. Therefore claim(s) 15-18 and 20 is/are rejected for the same rationale of rejection as set forth in claim(s) 7-10 and 7 respectively.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
For the reason above, claims 1-20 have been rejected and remain pending.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JIMMY H TRAN whose telephone number is (571)270-5638. The examiner can normally be reached Monday-Friday 9am-5pm PST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Chris Parry can be reached at 571-272-8328. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
JIMMY H TRAN
Primary Examiner
Art Unit 2451
/JIMMY H TRAN/Primary Examiner, Art Unit 2451
Prosecution Insights