DETAILED ACTION
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 .
This Office Action is in response to the amendment filed on 8/19/2026. This action is made FINAL.
Claims 1, 5-8, 12-15, 19 and 20 are pending and they are presented for examinations.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claim(s) 1, 5-8, 12-15, 19 and 20 is/are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 1 (similarly claims 8 and 15) recites:
“wherein before the receiving the input data transmitted by the central processing unit, the method further comprises:
connecting to the central processing unit, wherein the central processing unit is configured to obtain the input data and transmit the input data to an accelerator card,
wherein the accelerator card comprises an accelerator core and a memory;
level 1 cache data is stored in the accelerator core; and the input data, the output data, and level 2 cache data are stored in the memory.”
The examiner is unclear if the limitation following “further comprises:” is within the wherein clause (“wherein before the receiving”) or if it is outside the scope of the wherein clause (“wherein before the receiving”).
The lack of indentation and the use of “:” adds to the ambiguity as to the mete-and-bounds of the claim.
Claim 6 (similarly claims 13 and 20) recite: “a parameter set”. The examiner is unclear what “a parameter set” is referring to. The term, parameter set, is not clearly defined.
The specification recite: “Specifically, the parameter set is preset”.
Therefore, the examiner is unclear if the parameter set is a set of parameters, a parameter that is preset and/or both.
Claims 5-7, 12-14 and 19-20 are rejected based on rejection of its corresponding dependent claim.
Response to Arguments
Applicant's arguments filed regarding claim 1-4 (page 10), “First, Lim has a fundamentally different hardware architecture and does not disclose any hardware-architectural features of the present application. Lim is a System-on-Chip (SoC) component residing inside a CPU, rather than a board-level accelerator card. The accelerator card of the present application is a board independent of the CPU and connects to a CPU server via a PCIE slot. The "on-chip SoC component" of Lim and the "board-level PCIE accelerator card" of the present application are completely different in physical form and integration hierarchy. The "buffer 122" in Lim is a 4-8 KB SRAM fully-associative cache, which is a tiny on-chip storage structure. The "memory" in the present application is a high-capacity on-board DDR memory configured to store complete input datasets, output datasets and level 2 cache data. They are not equivalent in terms of capacity, physical location or functionality. Lim never mentions the concept of an "accelerator core", and therefore there is no scenario where "an accelerator core provides level 1 cache.”
The examiner would like to point out to Sukhwani in view of Lim discloses the above limitation. As previously shown, Sukhwani discloses separate hardware accelerator card, input data, output data and data processing system having cache memory to store data.
Lim further discloses not just SoC but also specialized hardware accelerator(s) with accelerator core.
[Paragraph 13], Accelerators that provide acceleration of instructions executed by a processor, for example, for indexing, may be designated as indexing accelerators. Indexing accelerators may include both specialized hardware and dedicated buffers for targeting relatively large data workloads.
[Paragraph 25], FIG. 2 illustrates a memory hierarchy 200 including the indexing accelerator 100 of FIG. 1, according to an example of the present disclosure. The example of the memory hierarchy 200 may include the processor core 128, a level 1 (L1) cache 202, multiple indexing accelerators 204, which may include an arbitrary number of identical indexing accelerators 100 (three shown in the example) with an arbitrary number of additional configuration register contexts 206 (three shown with the shaded pattern in the example) corresponding to the configuration registers 106, and a L2 cache 208. During operation of the indexing accelerator 100, the processor core 128 may send a signal to the indexing accelerator 100 indicating, via execution of non-transitory machine readable instructions, that the indexing accelerator 100 is to index a certain location or search for specific data. After the various indexing tasks have been performed by the indexing accelerator 100, the indexing accelerator 100 may send an interrupt signal to the processor core 128 indicating that the indexing tasks are complete, and the indexing accelerator 100 is now available for other tasks.
In addition, applicants’ argument regarding size of memory is not applicable. If the applicant’s intention is to specifically claim a memory of certain size and/or type. The examiner recommends amending the claim to specify a memory of certain size and/or type.
Furthermore, applicants’ argument (page 11) regarding Lim hash join operation is not applicable since the rejection on hash join operation pertains to Sukhwani.
Therefore, argument(s) is/are not persuasive.
Applicant's arguments filed regarding claim 6 (page 8), “With respect to this issue, the Applicant clarifies that, as recited in Paragraph [0074] of the specification of the present application: "the parameter set includes a data length and an operation operator", it can be seen that the "parameter set" is a set of parameters including the data length and the operation operator. A person of ordinary skill in the art can understand from the recitation "calculating the hash value of the input data and the quantity of the hash values based on the input data and a parameter set" that the parameter set necessarily contains parameters capable of determining the quantity of the hash values.”
The examiner would like to point out to the instant specification:
PGPub paragraph 58: “Specifically, the parameter set is preset, and the hash value of the input data and the quantity of the hash values may be calculated based on the input data and the parameter set.”
PGPub paragraphs 74: “the parameter set includes a data length and an operation operator.”
The examiner is unclear if a parameter set is just a mere preset value plus data length and an operation operator, or just data length and operation operator, or a single preset value, etc.
The specification does not provide sufficient disclosure. If the applicants’ intention is to state “the parameter set” is a set of parameter which includes a data length and an operation operator”, the examiner recommends amending the claim.
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 (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 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, 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.
Claim(s) 1, 5-8, 12-15, 19 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Sukhwani et al. (Pub 20130318067) (hereafter Sukhwani) in view of Lim et al. (Pub 20160070701) (hereafter Lim).
As per claim 1, Sukhwani teaches:
A data processing method, comprising:
receiving input data transmitted by a central processing unit; ([Paragraph 28], For example, arrangements may have more than one hardware accelerator, system memory, CPU, source, etc. [Paragraph 27], The hardware accelerator may read a fact table 210 and one or more dimension tables 220 from source 203. Source 203 may be any type of local or remote storage memory, including a hard disk, RAM (random access memory), ROM (read only memory), data stored on an intranet or internet server or storage device, etc. [Paragraph 56], As used herein, a "server" includes a physical data processing system (for example, system 712 as shown in FIG. 7) running a server program. It will be understood that such a physical server may or may not include a display and keyboard. Further, it is to be understood that components may be implemented on one server or on more than one server.)
performing a hash join operation on the input data to obtain a target operation result, and using the target operation result as output data; and ([Paragraph 13], Advantageously, one or more embodiments of the invention allow for hardware-accelerated relational joins using a hash join algorithm. [Paragraph 29], Hash join is performed in a two-phase process, a build phase and a probe phase. FIG. 3 illustrates a build phase 300 and FIG. 4 illustrates a probe phase 400, according to an embodiment of the invention. A source 302 has a fact table 310 and a dimension table 320. In this embodiment, the fact table 310 is shown to be the larger table and the dimension table is 320 is shown to be the smaller table. It is important to note that this is not necessarily the case, and that in other embodiments the fact table 310 may be smaller than the dimension table 320. In addition, although only a single fact table 310 and dimension table 320 are shown in FIGS. 3-4 for clarity, there may be multiple fact and/or dimension tables to be joined.)
transmitting the output data to the central processing unit, ([Paragraph 34], The matching units 460 use the current hash value to read, from memory banks 330, all the rows of dimension table 320 which correspond to the current hash value. In each cycle, multiple rows from the dimension table 320 (up to one from each memory bank) are compared against the selected row from the fact table 310. If a match occurs, the join evaluates to true and the desired fields are selected from the row of the fact table 310 and the row of the dimension table 320 and are written to an output buffer (not shown in FIGS. 3-4). If a match does not occur, the join condition evaluates to false and the next rows are processed. The joined rows in the output buffer are then streamed from the hardware accelerator 301 to a system memory 102 or 202. In other embodiments, the joined rows in the output buffer are streamed directly to a CPU.)
wherein the input data and the output data are columnar data, and the input data and the output data are in the form of an array, ([Paragraph 34], The matching units 460 use the current hash value to read, from memory banks 330, all the rows of dimension table 320 which correspond to the current hash value. In each cycle, multiple rows from the dimension table 320 (up to one from each memory bank) are compared against the selected row from the fact table 310. If a match occurs, the join evaluates to true and the desired fields are selected from the row of the fact table 310 and the row of the dimension table 320 and are written to an output buffer (not shown in FIGS. 3-4). If a match does not occur, the join condition evaluates to false and the next rows are processed. The joined rows in the output buffer are then streamed from the hardware accelerator 301 to a system memory 102 or 202. In other embodiments, the joined rows in the output buffer are streamed directly to a CPU. [Paragraph 26], The hardware accelerator may read fact table 110 and one or more dimension tables 120 from the system memory, perform a join operation, and write a joined table 130 to the system memory. The CPU 103 is operative to read from and write to the system memory 102 as well.)
wherein before the receiving the input data transmitted by the central processing unit, the method further comprises:
connecting to the central processing unit, wherein the central processing unit is configured to obtain the input data and transmit the input data to an accelerator card, ([Paragraph 13], Advantageously, one or more embodiments of the invention allow for hardware-accelerated relational joins using a hash join algorithm. [Paragraph 29], Hash join is performed in a two-phase process, a build phase and a probe phase. FIG. 3 illustrates a build phase 300 and FIG. 4 illustrates a probe phase 400, according to an embodiment of the invention. A source 302 has a fact table 310 and a dimension table 320. In this embodiment, the fact table 310 is shown to be the larger table and the dimension table is 320 is shown to be the smaller table. It is important to note that this is not necessarily the case, and that in other embodiments the fact table 310 may be smaller than the dimension table 320. In addition, although only a single fact table 310 and dimension table 320 are shown in FIGS. 3-4 for clarity, there may be multiple fact and/or dimension tables to be joined. [Paragraph 27], FIG. 2 illustrates an alternate arrangement 200 of a hardware accelerator 201, system memory 202 and a source 203. The hardware accelerator may read a fact table 210 and one or more dimension tables 220 from source 203. Source 203 may be any type of local or remote storage memory, including a hard disk, RAM (random access memory), ROM (read only memory), data stored on an intranet or internet server or storage device, etc. The hardware accelerator is operative to perform a join operation on the fact table 210 and one or more dimension tables 220 and to write a joined table 230 to system memory 202. A CPU (not explicitly shown in FIG. 2) may read the joined table 230 from system memory 202.)
Sukhwani teaches accelerator card, input data, output data and data processing system having cache memory to store data.
However, Sukhwani does not explicitly disclose wherein: the accelerator card comprises an accelerator core and a memory; level 1 cache data is stored in the accelerator core; and the input data, the output data, and level 2 cache data are stored in the memory.
Lim discloses wherein: the accelerator card comprises an accelerator core and a memory; level 1 cache data is stored in the accelerator core; and the input data, the output data, and level 2 cache data are stored in the memory. ([Paragraph 61], The parallel fetching technique of FIG. 5 may be applicable to hash joins, such as, joins that combine two database tables into one table. In order to expedite performance of the join operation, a smaller table of the database tables may be converted into a hash table index, and then probed by entries (i.e., keys) in the larger table of the database tables… For every matching entry, a result buffer may be populated and eventually the entries that reside in both tables may be located. Given that the larger table may include thousands to millions of entries, which may need to probe an index independently, such a scenario may include a substantial amount of inter-probe parallelism. [Paragraph 28], According to another example, the indexing accelerator 100 may take the place of the L1 cache 202 and include a relatively larger buffer 122. [Paragraph 18], Referring to FIG. 1, the indexing accelerator 100 is depicted as including a request decoder 104 to receive a number of requests corresponding to the queries 102 from a central processing unit (CPU) or a higher level cache (e.g., the L2 cache 202 of FIG. 2). [Paragraph 13], Accelerators that provide acceleration of instructions executed by a processor, for example, for indexing, may be designated as indexing accelerators. Indexing accelerators may include both specialized hardware and dedicated buffers for targeting relatively large data workloads. Such large data workloads may include segments of execution that may not be ideally suited for standard processors due to relatively large amounts of time spent accessing data and waiting on dynamic random-access memory (DRAM) (e.g., time spent chasing pointers through indexing structures). The indexing accelerators may provide an alternate and more energy efficient option for executing these data segments, while also allowing the main processor core to be put into a low power mode.)
It would have been obvious to a person with ordinary skill in the art, before the effective filing date of the invention, to combine the teachings of Sukhwani wherein input data provided by a CPU is utilized to perform a hash join operation by an accelerator to provide an output (result) to the CPU, into teachings of Lim wherein level 1 cache memory is provided by the accelerator (accelerator core) and level 2 cache memory is provided by the memory, because this would enhance the teaching of Sukhwani wherein by utilizing hierarchy of cache memories, it allows acceleration of instruction results which provides performance and energy efficiency by providing memory level parallelism (MLP) utilizing cache levels. [Lim paragraph 1, 13, 14, 17]
As per claim 5, rejection of claim 1 is incorporated:
Sukhwani teaches wherein the performing the hash join operation on the input data to obtain the target operation result comprises: reading, by the accelerator core, the input data from the memory; calculating a hash value of the input data and a quantity of hash values; establishing a hash table based on the hash value and the quantity of the hash values; and performing the hash join operation on the input data based on the hash table to obtain the target operation result. ([Paragraph 23], Illustrative embodiments provide advantages relative to conventional techniques by utilizing hardware-accelerated hash joins. For example, hash join is a highly parallel algorithm which is well suited to hardware implementation. Hash join is a particularly effective method of performing table-join operations. Other techniques include nested loop join algorithms, which perform direct "all-to-all" comparisons of two or more tables and sort-merge join, which performs "compare and jump" on two or more tables. Nested loop join algorithms are computationally expensive and have O(N.sup.2) complexity. Sort-merge join requires expensive sorting of one of the tables and has O(NlogN) complexity. Hash joins are fast but approximate and false positives are possible. Hash join is the most efficient and commonly applied algorithm for equijoin operations, as it requires looking at one or a few positions to determine presence in an entire table. Hash join has O(N) complexity. Hash join may be implemented in software or hardware. [Paragraph 25], In one example of hardware implementation, hash joins are performed using bit-join based filtering on a FPGA. A bit-vector is generated based on the small table. As rows of the large table are streamed in to the FPGA, lower-order bits of the column of interest are used to look-up a bit in bit vector and decide if the row of the large table qualifies for further processing by the host or general purpose processor. In illustrative embodiments of the present invention, the actual join operations are performed in a hardware accelerator, which provides improved performance and lower power consumption. Moreover, in embodiments of the current invention, hash values are generated using one or more hash functions as opposed to simply using the lower order bits of the column as the hash, which may result in uneven hash distribution and potentially a large number of hash collisions.)
As per claim 6, rejection of claim 5 is incorporated:
Sukhwani teaches wherein the calculating the hash value of the input data and the quantity of the hash values comprises: calculating the hash value of the input data and the quantity of the hash values based on the input data and a parameter set. ([Paragraph 23], Illustrative embodiments provide advantages relative to conventional techniques by utilizing hardware-accelerated hash joins. For example, hash join is a highly parallel algorithm which is well suited to hardware implementation. Hash join is a particularly effective method of performing table-join operations. Other techniques include nested loop join algorithms, which perform direct "all-to-all" comparisons of two or more tables and sort-merge join, which performs "compare and jump" on two or more tables. Nested loop join algorithms are computationally expensive and have O(N.sup.2) complexity. Sort-merge join requires expensive sorting of one of the tables and has O(NlogN) complexity. Hash joins are fast but approximate and false positives are possible. Hash join is the most efficient and commonly applied algorithm for equijoin operations, as it requires looking at one or a few positions to determine presence in an entire table. Hash join has O(N) complexity. Hash join may be implemented in software or hardware. [Paragraph 25], In one example of hardware implementation, hash joins are performed using bit-join based filtering on a FPGA. A bit-vector is generated based on the small table. As rows of the large table are streamed in to the FPGA, lower-order bits of the column of interest are used to look-up a bit in bit vector and decide if the row of the large table qualifies for further processing by the host or general purpose processor. In illustrative embodiments of the present invention, the actual join operations are performed in a hardware accelerator, which provides improved performance and lower power consumption. Moreover, in embodiments of the current invention, hash values are generated using one or more hash functions as opposed to simply using the lower order bits of the column as the hash, which may result in uneven hash distribution and potentially a large number of hash collisions.)
Lim also teaches ([paragraph 25], FIG. 2 illustrates a memory hierarchy 200 including the indexing accelerator 100 of FIG. 1, according to an example of the present disclosure. The example of the memory hierarchy 200 may include the processor core 128, a level 1 (L1) cache 202, multiple indexing accelerators 204, which may include an arbitrary number of identical indexing accelerators 100 (three shown in the example) with an arbitrary number of additional configuration register contexts 206 (three shown with the shaded pattern in the example) corresponding to the configuration registers 106, and a L2 cache 208. During operation of the indexing accelerator 100, the processor core 128 may send a signal to the indexing accelerator 100 indicating, via execution of non-transitory machine readable instructions, that the indexing accelerator 100 is to index a certain location or search for specific data. After the various indexing tasks have been performed by the indexing accelerator 100, the indexing accelerator 100 may send an interrupt signal to the processor core 128 indicating that the indexing tasks are complete, and the indexing accelerator 100 is now available for other tasks. [Paragraph 36], Referring to FIG. 3, at block 302, the indexing accelerator 100 including a set of the configuration registers 106 (e.g., 8 configuration registers) may receive indexing requests (e.g., indexing requests 1 to 8) for multiple queries 102 for acceleration. As described herein, the memory hierarchy 200 may include multiple indexing accelerators 204. Moreover, each indexing accelerator 100 may include a plurality of the configuration registers 106 including corresponding configuration register contexts 206, such as the three configuration register contexts 206 shown in FIG. 2. [Paragraph 37], For example, the first entry of the index may be located by reading the base address of the index table and adding the base address with the length of each index entry, where these values may be sent to the indexing accelerator 100 during a configuration stage and reside in the configuration registers 106.)
As per claim 7, rejection of claim 5 is incorporated:
Lim teaches wherein the establishing the hash table based on the hash value and the quantity of the hash values comprises: determining a level 1 cache depth and a level 2 cache depth of the input data based on the hash value and the quantity of the hash values, wherein the level 1 cache depth is a first hash bucket corresponding to the level 1 cache data, and the level 2 cache depth is a second hash bucket corresponding to the level 2 cache data; and establishing the hash table based on the hash value, the level 1 cache depth, and the level 2 cache depth. ([paragraph 25], FIG. 2 illustrates a memory hierarchy 200 including the indexing accelerator 100 of FIG. 1, according to an example of the present disclosure. The example of the memory hierarchy 200 may include the processor core 128, a level 1 (L1) cache 202, multiple indexing accelerators 204, which may include an arbitrary number of identical indexing accelerators 100 (three shown in the example) with an arbitrary number of additional configuration register contexts 206 (three shown with the shaded pattern in the example) corresponding to the configuration registers 106, and a L2 cache 208. During operation of the indexing accelerator 100, the processor core 128 may send a signal to the indexing accelerator 100 indicating, via execution of non-transitory machine readable instructions, that the indexing accelerator 100 is to index a certain location or search for specific data. After the various indexing tasks have been performed by the indexing accelerator 100, the indexing accelerator 100 may send an interrupt signal to the processor core 128 indicating that the indexing tasks are complete, and the indexing accelerator 100 is now available for other tasks. [Paragraph 36], Referring to FIG. 3, at block 302, the indexing accelerator 100 including a set of the configuration registers 106 (e.g., 8 configuration registers) may receive indexing requests (e.g., indexing requests 1 to 8) for multiple queries 102 for acceleration. As described herein, the memory hierarchy 200 may include multiple indexing accelerators 204. Moreover, each indexing accelerator 100 may include a plurality of the configuration registers 106 including corresponding configuration register contexts 206, such as the three configuration register contexts 206 shown in FIG. 2. [Paragraph 37], For example, the first entry of the index may be located by reading the base address of the index table and adding the base address with the length of each index entry, where these values may be sent to the indexing accelerator 100 during a configuration stage and reside in the configuration registers 106. [Paragraph 51], Referring to FIG. 4, at block 402, during a configuration stage of indexing, in addition to a bucket array address (i.e., index table address), the array addresses and layout information (if different from a bucket array) for links may also be loaded to the configuration registers 106. [Paragraph 54], At block 408, the key may be compared against the null value (i.e., which means there is no such entry in the hash table) and the key used to calculate the bucket address. [Paragraph 60], To mitigate the first bucket header miss, the indexing accelerator 100 may exploit inter-probe parallelism as there may be a plurality (e.g., millions) of keys searched on a single index table for an indexing request (e.g., hash joins in data analytics workloads). To exploit such parallelism, the next probe key may be prefetched and the hash value may be calculated to issue the bucket header's corresponding entry in advance. Prefetching the next probe key may be performed based on the probe key access patterns as these keys are stored in an array in a DBMS and may follow a fixed stride pattern (e.g., add 8 bytes to the previous address). Prefetching the next probe key may be performed in advance so that the value may be hashed and the bucket entry may be prefetched.)
As per claims 8, 12-14, these are device claims corresponding to the method claims 1, 5-7. Therefore, rejected based on similar rationale.
As per claims 15, 19 and 20, these are non-transitory computer-readable storage medium claims corresponding to the method claims 1, 5 and 6. Therefore, rejected based on similar rationale.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Chang et al. (Pub 20110246667) discloses accelerator with multi level cache.
Dickie et al. (Pub 20160098439) discloses accelerator performing aggregation or joins on input data from cache to produce output data.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DONG U KIM whose telephone number is (571)270-1313. The examiner can normally be reached 9:00am - 5:00pm.
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, Bradley Teets can be reached at 5712723338. 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.
/DONG U KIM/Primary Examiner, Art Unit 2197