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 .
Status of Claims
The following is a Final Office action in response to applicant's amendment and response received 07/30/2026, responding to the 04/30/2026 non-final office action provided in rejection of claims 1-20.
3. Claims 1-2, 4-5, 6, 9-13, and 16-17 have been amended. Claims 1-20 are pending and are addressed in this office action. New grounds of rejection are presented in view of the newly presented limitation(s).
Examiner notes
(A). Limitations have been provided with the Bold fonts in order to distinguish from the cited part of the reference (Italic).
(B). Examiner has cited particular columns, line numbers, references, or figures in the references applied to the claims above for the convenience of the applicant. Although the specified citations are representative of the teachings of passages and figures may apply as well. It is respectfully requested from the applicant in preparing responses to fully consider the reference in entirety, as potentially teaching all or part of the claimed invention. See MPEP § 2141.02 VI and 2123.
The examiner requests, in response to this Office action, support be shown for language added to any original claims on amendment and any new claims. That is, indicate support for newly added claim language by specifically pointing to page(s) and line number(s) in the specification and/or drawing figure(s). This will assist the examiner in prosecuting the application.
When responding to this office action, Applicant is advised to clearly point out the patentable novelty which he or she thinks the claims present, in view of the state of the art disclosed by the references cited or the objections made. He or she must also show how the amendments avoid such references or objections See 37 CFR 1.111 (c).
Response to Amendments and Arguments
In light of the amendment of claims 1-2, 4-5, 6, 9-13, and 16-17, the Previous Action's rejections of those claims under 35 U.S.C. § 112 (b), 112 (a), claim objections and claim interpretation under 35 U.S.C. § 112 (f) are hereby withdrawn.
With respect to the rejections under 35 USC 101, Applicant argues that step 2A, Prong One - the claims are not a mental process. The claims train a neural network on resource costs obtained for configurations of a computing template … A process that operates over a search space of this magnitude, by training and evaluating a neural network, cannot practically be performed in the human mind, with or without pen and paper. A limitation that cannot practically be performed mentally is not a mental process. See MPEP 2106.04(a)(2)(III). (Remarks, page 8)
Examiner respectfully disagrees. In determining whether a claim falls within an excluded category, examiner guided by the Court's two-part framework, described in Mayo and Alice. Id. at217-18 (citing Mayo Collaborative Servs. v. Prometheus Labs., 566 U.S. 66, 75-77 (2012)). In accordance with that framework, we first determine what concept the claim is "directed to." See Alice, 573 U.S. at 219 ("On their face, the claims before us are drawn to the concept of Examiner follow the patent eligibility guidance to determine the eligibility. Examiner cited in the previous non-final office action that limitations are functions (i.e. estimating resource costs, obtaining resource costs and training a neural network using the resource costs) that can be reasonably carried out in the human mind with the aid of pen and paper, through observation, evaluation, judgment, opinion, thus it is reasonable to identify these limitation as reciting a mental process. Further, claim as they are do not reflect “the number of operators and array units” and do not exclude reasonable number of operators and array units can be done by human.
With respect to the rejections under 35 USC 101, Applicant argues that step 2A, Prong Two - practical application. Even if a mathematical concept is recited, the claims as a whole integrate it into a practical application: a specific improvement to the operation of a reconfigurable dataflow computing system and its compiler. … Claims directed to a specific improvement in the way computers operate are patent-eligible. See MPEP 2106.05(a); Enfish; McRO. The amended claims are further tied to a particular machine, a reconfigurable dataflow computing system comprising an array of computer and memory units configured by dataflow configuration information (claims 10-12), reinforcing integration under MPEP 2106.05(b). (Remarks, page 8)
Examiner respectfully disagrees. Applicant claimed that instance claims presented improvement to the operation of a reconfigurable dataflow computing system and its compiler. Examiner notes that the federal circuit in Enfish stated that certain claims directed to improvements in related technology are not necessarily abstract (Step A). The court noted that some improvements in computer-related technology such as chip architecture or an LED display, when appropriately claimed, are undoubtedly not abstract. In this case specification discloses an improvement to the functionality of the computer itself, but the claims does not contain necessary elements to realize that improvement. Finally, estimating resource costs, obtaining resource costs and training a neural network using the resource costs is ubiquitous across multiple technology. Further, applicant argues that claim recites an improvement in computer related McRO memo and that the claim does not recite an abstract idea. However, the claim is clearly directed to estimating resource cost i.e. fundamental economic practice. Applicant argues that the claimed invention is an improvement in computer-related technology because it the operation of a reconfigurable dataflow computing system and its compiler. However, the claim is solely directed obtain resource costs for a computing template, resource costs as training targets to produce a trained, produce estimated resource costs for the uncompiled configuration. Because claim is directed to the mathematical algorithm without any additional steps applying the results, the claim is clearly directed to an abstract idea with significantly more.
With respect to the rejections under 35 USC 101, Applicant argues that step 2B - inventive concept. The Examiner's assertion that the additional elements are "well- understood, routine, and conventional" is respectfully traversed as conclusory and unsupported under Berkheimer. Training a neural network on resource costs obtained for compiled configurations and using it to predict resource costs for uncompiled configurations of a reconfigurable dataflow template is not a well-understood, routine, or conventional activity; the specification describes it as, to the inventors' knowledge, the first data-driven cost model guiding compilation for a reconfigurable dataflow architecture. That ordered combination supplies an inventive concept. (Remarks, page 9)
Examiner respectfully disagrees. Examiner identified in step 2B additional element under Prong 2 as insignificant extra solution activity. Evidence is provided in MPEP 2106.05(d) for generic computer functions such as gathering, displaying, updating, transmitting and storing data. So, under step 2B, the additional elements merely recite generic computer and computer components, thus do not amount to significantly more than the judicial exception. The courts have identified functions such as gathering, displaying, updating, transmitting and storing data as well-understood, routine, conventional activity, thus do not amount to significantly more than the judicial exception. See MPEP 2106.05(d). Therefore, none of the additional elements recite an inventive concept, thus, the claimed invention is patent ineligible under 35 USC 101.
With respect to the rejections under 35 USC 103, Applicant argues that Jiang does not train any network using resource costs as training targets and does not produce a trained network costs that estimates resource. Jiang therefore does not disclose the claimed training and estimation operations. (Remarks, page 9)
Examiner respectfully disagrees. Jiang discloses at least in abstract, “the invention can fully exert [i.e. obtain] the performance of the computing device, reduce the cost of the model designer training model, …” . Further, at page 2, discloses “providing a convolutional neural network module training method and system based on containerized …”. Applicant's arguments have been considered but not persuasive. Gupta is relied on to teach costs that estimates resource.
With respect to the rejections under 35 USC 103, Applicant argues that Gupta does not train a neural network using resource costs (such as the memory-unit count, compute-unit count, and compute latency of claim 5) as training targets and does not use such a trained network to estimate resource costs for an uncompiled configuration for a computing template. The Office Action equates Gupta's dollar-denominated "end cost" with the claimed "resource costs," but these are different things; conflating them cannot supply the claimed limitations. (Remarks, page 10)
Examiner respectfully disagrees. Examiner cited in the previous Gupta discloses at least paragraphs 0023-0024, “computing tasks may be referred to as being compute-bound, in which the time needed to complete the computing task is determined mainly by the speed of the processor (or multiple processing cores in the case of a computing task … which the time needed to complete the computing task is determined mainly by the amount of memory required to hold the working data, … where the end cost to the user is calculated based on the amount of time the computing resource …”. Applicant’s arguments have been considered but not persuasive.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter.
Step 1: Claims 1-12 are directed to system and fall within the statutory category of machines; Claims 13-20 are directed to method and fall within the statutory category of processes. Therefore, “Are the claims to a process, machine, manufacture or composition of matter?” Yes.
In order to evaluate the Step 2A inquiry “Is the claim directed to a law of nature, a natural phenomenon or an abstract idea?” we must determine, at Step 2A Prong 1, whether the claim recites a law of nature, a natural phenomenon or an abstract idea and further whether the claim recites additional elements that integrate the judicial exception into a practical application.
As to claim 13: Under Step 2A, Prong 1, the claim recites multiple limitations that recite an abstract idea. The limitations “ estimating resource costs for computing tasks for a reconfigurable dataflow computing system the method comprising:” , “obtaining resource costs for a computing template for each configuration of a set of template configurations, the computing template corresponding to a computing task;” , “training a neural network using the resource costs as training targets to produce a trained neural network;” and “using the trained neural network to estimate the costs for an uncompiled configuration for the computing template”, is a concept that can be reasonably performed in the human mind (with the aid of pen and paper), judgement and/or opinion. The limitation The BRI of these limitations requires performing an arithmetic calculation (estimate), Neurons network (mathematical models) specifically “calculation” which requires “calculating a resource cost.” Therefore, since the BRI of the claim requires a mathematical calculation, the limitation is directed to a mathematical concept which is a judicial exception that is not patent eligible.
Under Step 2A, Prong 2, the additional elements “thereby produce estimated resource costs for the uncompiled configuration for the computing template.” are not indicative of integration into a practical application. The limitations “computing system includes” merely recites generic computer system or device to carry out or apply the judicial exception. MPEP 2106.05(f).
Under step 2B, the additional elements do not amount to significantly more than the abstract idea. As stated above, the claimed invention merely recites generic computer system for carrying out or applying the abstract idea. Furthermore, the courts have recognized that mere data obtain, such as those defined in the claim, are well-understood, routine, and convention computer functions which cannot serve as an inventive concept according to MPEP 21.06.05(d).
For the above reasons, the claims of this application are not patentable under 35 USC 101.
As to claim 1: Under Step 2A, Prong 1, the claim recites multiple limitations that recite an abstract idea. The limitations “obtain resource costs for a computing template for each configuration of a set of template configurations, the computing template corresponding to a computing task;” , “train a neural network using resource costs as training targets to produce a trained neural network” and “use the trained neural network to estimate the resources costs for an uncompiled configuration for the computing template”, mental process since “…a reconfigurable … defining” is a concept that can be reasonably performed in the human mind (with the aid of pen and paper), judgement and/or opinion. The limitation The BRI of these limitations requires performing an arithmetic calculation (estimate), Neurons network (mathematical models) specifically “calculation” which requires “calculating a resource cost.” Therefore, since the BRI of the claim requires a mathematical calculation, the limitation is directed to a mathematical concept which is a judicial exception that is not patent eligible.
Under Step 2A, Prong 2, the additional elements “the system comprising a memory and at least one processor coupled to the memory” and “thereby produce estimated resource costs for the uncompiled configuration for the computing template.” are not indicative of integration into a practical application. The limitations of “memory” , “processor” and “computing system includes” merely recites generic computer system or device to carry out or apply the judicial exception. MPEP 2106.05(f).
Under step 2B, the additional elements do not amount to significantly more than the abstract idea. As stated above, the claimed invention merely recites generic computer system for carrying out or applying the abstract idea. Furthermore, the courts have recognized that mere data obtain, such as those defined in the claim, are well-understood, routine, and convention computer functions which cannot serve as an inventive concept according to MPEP 21.06.05(d).
For the above reasons, the claims of this application are not patentable under 35 USC 101.
Claims 2-3 and 15 are not patent eligible for the same reasons given for claim 1, “wherein: the at least one processor configured to …” and “wherein each configuration comprises …” ; are functions that can be reasonably carried out in the human mind with the aid of pen and paper, through observation, evaluation, judgment, opinion, thus it is reasonable to identify these limitation as reciting a mental process.
Claims 4, 7, 16 and 19 are not patent eligible for the same reasons given for claim 1, “… comprises one or more of an input size, a filter size, a stride, and a base grid size.” and “wherein the plurality of proposed configure comprise a plurality of base grid sizes.” are functions that can be reasonably carried out in the human mind with the aid of pen and paper, through observation, evaluation, judgment, opinion, thus it is reasonable to identify these limitation as reciting a mental process.
Claims 5, 10, 11-12, 14 and 17 are not patent eligible for the same reasons given for claim 1, “… wherein the resources costs comprise a memory unit count, a compute unit ...” , “… configured to generate dataflow configuration information that enables the reconfigurable dataflow computing system …” , “… the runtime module configured to launch execution of the computing template …” and “...wherein the resources costs comprise a memory unit count, a compute unit count, and a compute latency. ” the additional elements are merely recite instructions to implement an abstract idea on a generic computer, or merely uses a generic computer or computer components as a tool to perform the abstract idea, thus is not a practical application under Prong 2. See MPEP 2106.05(f).
Claims 6, 8-9, 8 and 20 are not patent eligible for the same reasons given for claim 1, “… module configured to determine the estimated resource costs ...” , “… wherein the selected configuration is selected according …” , “ and “… configured to comprise: allocate resources ...” are functions that can be reasonably carried out in the human mind with the aid of pen and paper. Therefore, since the BRI of the claim requires a mathematical calculation. Additionally, the limitations “determining, selecting, based on the estimation and allocation, a compute resource of the available compute resources to perform at least one calculation of the plurality of calculations;” is a concept that can be reasonably performed in the human mind (with the aid of pen and paper), which include observation, evaluation, judgement and/or opinion, through observation, evaluation, judgment, opinion, thus it is reasonable to identify these limitation as reciting a mental process.
Claim Rejections - 35 USC § 103
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.
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.
Claims 1, 3, 6, 8-9, 13, 15, 18, and 20 are rejected under 35 U.S.C. 103 as being obvious over Jiang et al. (CN-114528070-A, hereinafter Jiang) in view of Gupta et al. (US 20230409387 A1, hereinafter Gupta).
As to claim 1, Jiang discloses a system for estimating resource costs for computing tasks for a reconfigurable dataflow computing system the system comprising a memory and at least one processor coupled to the memory and configured to (page. 3, … calculating [i.e. estimating] resource scale (memory, CPU, GPU, bandwidth), …):
(abstract, generating complete machine learning model code according to the template; abstract modular template, based on hierarchical characteristics of the convolutional neural network module, providing template of standard abstract convolutional neural network module … The invention can fully exert [i.e. obtain] the performance of the computing device, reduce the cost of the model designer training model, … ), the computing template corresponding to a computing task (page 9, The method according to the convolutional neural network module relationship between each module, designing a segmented machine learning frame and a set of template of abstract convolutional neural network module, providing the abstract according to the module needed by the calculation resource [i.e. computing task], …);
train a neural network using the (abstract, The invention can fully exert [i.e. obtain] the performance of the computing device, reduce the cost of the model designer training model, …. Further, page 2, providing a convolutional neural network module training method and system based on containerized); and
use the trained neural network (page 6, The calculation amount determines the calculation resource needed by the module, such as CPU, GPU and so on. based on the convolutional neural network module relationship between each module is obvious and each module parameter quantity and calculating quantity difference is large convolutional neural network module each module does not accord with the hardware requirement of the characteristic such that a large amount of storage resource and calculating resource is wasted, increasing the cost of the model training) for an uncompiled configuration for the computing template (page 8, using the method design model, there is no need to compile the related model code, only need in the declaring type template according to the type of the template) and thereby produce estimated (page 6, The calculation amount determines the calculation resource needed by the module, such as CPU, GPU and so on. based on the convolutional neural network module relationship between each module is obvious and each module parameter quantity and calculating quantity difference is large convolutional neural network module each module does not accord with the hardware requirement of the characteristic such that a large amount of storage resource and calculating resource is wasted, increasing the cost of the model training).
Jiang does not explicitly disclose the following limitations but,
Gupta disclose obtain resource costs for a computing (par. 0023-0024, computing tasks may be referred to as being compute-bound, in which the time needed to complete the computing task is determined mainly by the speed of the processor (or multiple processing cores in the case of a computing task … which the time needed to complete the computing task is determined mainly by the amount of memory required to hold the working data, … where the end cost to the user is calculated based on the amount of time the computing resource);
using the estimated resource costs (par. 0023, … The various types of computer hardware are typically offered to users at different prices (e.g., different hourly price rates), where more powerful computer systems are typically more expensive than less powerful computer systems, and where the end cost to the user is calculated [i.e. estimated] based on the amount of time the computing resource,
produce estimated resource costs (see pars. 0023-0024).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include obtain / producing resource costs for a computing, as disclosed by Gupta, for the purpose to determine / obtain the amount of memory / resource required to hold the working data. (see paragraph 0023 of Gupta).
As to claim 3, Jiang discloses the system wherein each configuration comprises a set of configuration parameters (page 3, the convolution layer template is mainly used for configuring parameters of the convolutional neural network module layer, the convolution layer template mainly comprises the following parameters).
As to claim 6, the system wherein the at least one processor is further configured to (page. 3, … calculating [i.e. estimating] resource scale (memory, CPU, GPU, bandwidth), …):
Jiang does not explicitly disclose the following limitation but,
Gupta discloses determine the estimated resource costs for a plurality of proposed configurations and select a selected configuration for the computing template (par. 0012, FIG. 2A is a flowchart of a method for generating a proposed computer hardware configuration according to one example of the present technology. Further, par. 0024, … training deep neural networks. The various types of computer hardware are typically offered to users at different prices (e.g., different hourly price rates), where more powerful computer systems are typically more expensive than less powerful computer systems, and where the end cost to the user is calculated based on the amount of time the computing resource … . Further, 0038, FIG. 1, the computing task management interface 130 may present the proposed computer hardware configuration to the user 102 via the client application 120 and may configure an instance of a computing resource in accordance with a selection of a computer hardware configuration made by the user 102. In some examples, the user makes a selection that is consistent with the proposed computer hardware configuration, but in other cases the user may make a selection that is different from the proposed computer hardware configuration. In other examples, the computing task management interface 130 automatically selects the proposed computer hardware configuration generated by the self-tuning computer hardware configuration proposal engine 115 without requesting authorization or confirmation from the user 102, in which case the proposed computer hardware configuration may be referred to herein as an updated computer hardware configuration. Further, par. 0032, the user 102 will typically be asked to specify the type of computer system that will be provisioned to execute the specified computing task, illustrated in FIG. 1 in the simplified form [i.e. template] of selecting one hardware configuration from among hardware configuration)).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include determine the estimated resource costs for a plurality of proposed configurations and select a selected configuration for the computing template, as disclosed by Gupta, for the purpose to automatically allocate an instance of a computing resource that is configured according to the proposed computer hardware configuration. (see paragraph 0048 of Gupta).
As to claim 8, Gupta discloses the system wherein the selected configuration is selected according to one or more optimization criteria (par. 0045, … a computer hardware configuration of a computer system, specifies computing hardware that will execute a task, such as a number of CPUs, amount of memory, an amount of storage bandwidth, a number of GPUs, or the like. As one concrete example, the proposed computer hardware configuration may be selection between a “general” compute type and a “memory-optimized” compute type, as well as a number of cores (e.g., selected from a collection of possible numbers of cores, such as: 8 cores, 16 cores, 32 cores, 48 cores, 80 cores, 144 cores, or 272 cores). …).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include the system wherein the selected configuration is selected according to one or more optimization criteria, as disclosed by Gupta, for the purpose to automatically allocate an instance of a computing resource that is configured according to the proposed computer hardware configuration. (see paragraph 0048 of Gupta).
As to claim 9, Jiang discloses the system wherein the at least one processor is further configured to (page 3):
Gupta disclose allocate resources according to the estimated resource costs for the selected configuration for the computing template to produce allocated resources (par. 0024, … the end cost to the user is calculated [i.e. estimated] based on the amount of time the computing resource is used … . Further, par. 0025, … configured upon request to implement the different types of computer system configurations. For example, a physical server may have 256 processor cores and 2048 GB of installed memory, and these computing resources may be allocated and dedicated to particular virtual machines in accordance with their configuration types. For example, a virtual machine providing a low performance computer system configuration may be allocated one processor core and 4 GB of memory, while a compute-optimized computer system … . Further, par. 0032, the user 102 will typically be asked to specify the type of computer system that will be provisioned to execute the specified computing task, illustrated in FIG. 1 in the simplified form [i.e. template] of selecting one hardware configuration from among hardware configuration).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include allocate resources according to the estimated resource costs for the selected configuration for the computing template to produce allocated resources, as disclosed by Gupta, for the purpose to automatically allocate an instance of a computing resource that is configured according to the proposed computer hardware configuration. (see paragraph 0048 of Gupta).
As to claim 13, Jiang discloses a computer-implemented method for
obtaining (abstract, generating complete machine learning model code according to the template; abstract modular template, based on hierarchical characteristics of the convolutional neural network module, providing template of standard abstract convolutional neural network module … The invention can fully exert [i.e. obtain] the performance of the computing device, reduce the cost of the model designer training model, … ), the computing template corresponding to a computing task (page 9, The method according to the convolutional neural network module relationship between each module, designing a segmented machine learning frame and a set of template of abstract convolutional neural network module, providing the abstract according to the module needed by the calculation resource [i.e. computing task], …);
training a neural network using the (abstract, The invention can fully exert [i.e. obtain] the performance of the computing device, reduce the cost of the model designer training model, …. Further, page 2, providing a convolutional neural network module training method and system based on containerized); and
using the trained neural network to estimate the resources costs for an uncompiled configuration for the computing template (page 8, using the method design model, there is no need to compile the related model code, only need in the declaring type template according to the type of the template) and thereby produce estimated (page 6, The calculation amount determines the calculation resource needed by the module, such as CPU, GPU and so on. based on the convolutional neural network module relationship between each module is obvious and each module parameter quantity and calculating quantity difference is large convolutional neural network module each module does not accord with the hardware requirement of the characteristic such that a large amount of storage resource and calculating resource is wasted, increasing the cost of the model training).
Gupta disclose method for estimating resource costs and obtaining resource costs (par. 0023-0024, computing tasks may be referred to as being compute-bound, in which the time needed to complete the computing task is determined mainly by the speed of the processor (or multiple processing cores in the case of a computing task … which the time needed to complete the computing task is determined mainly by the amount of memory required to hold the working data, … where the end cost to the user is calculated based on the amount of time the computing resource);
using the resource costs (par. 0023, … The various types of computer hardware are typically offered to users at different prices (e.g., different hourly price rates), where more powerful computer systems are typically more expensive than less powerful computer systems, and where the end cost to the user is calculated based on the amount of time the computing resource)
produce estimated resource costs (see pars. 0023-0024).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include obtain / producing resource costs for a computing, as disclosed by Gupta, for the purpose to determine / obtain the amount of memory / resource required to hold the working data. (see paragraph 0023 of Gupta).
As to claim 15, it is the method claim, having similar limitations of claim 3. Thus, claim 15 is also rejected under the same rationale as cited in the rejection of claim 3.
As to claim 18, Gupta discloses the computer-implemented method further including:
determining the estimated resource costs for a plurality of proposed configurations and selecting a selected configuration for the computing template (par. 0012, FIG. 2A is a flowchart of a method for generating a proposed computer hardware configuration according to one example of the present technology. Further, par. 0024, … training deep neural networks. The various types of computer hardware are typically offered to users at different prices (e.g., different hourly price rates), where more powerful computer systems are typically more expensive than less powerful computer systems, and where the end cost to the user is calculated based on the amount of time the computing resource … . Further, 0038, FIG. 1, the computing task management interface 130 may present the proposed computer hardware configuration to the user 102 via the client application 120 and may configure an instance of a computing resource in accordance with a selection of a computer hardware configuration made by the user 102. In some examples, the user makes a selection that is consistent with the proposed computer hardware configuration, but in other cases the user may make a selection that is different from the proposed computer hardware configuration. In other examples, the computing task management interface 130 automatically selects the proposed computer hardware configuration generated by the self-tuning computer hardware configuration proposal engine 115 without requesting authorization or confirmation from the user 102, in which case the proposed computer hardware configuration may be referred to herein as an updated computer hardware configuration. Further, par. 0032, the user 102 will typically be asked to specify the type of computer system that will be provisioned to execute the specified computing task, illustrated in FIG. 1 in the simplified form [i.e. template] of selecting one hardware configuration from among hardware configuration)).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include an optimization module configured to determine the estimated resource costs for a plurality of proposed configurations and select a selected configuration for the computing template, as disclosed by Gupta, for the purpose to automatically allocate an instance of a computing resource that is configured according to the proposed computer hardware configuration. (see paragraph 0048 of Gupta).
As to claim 20, it is the method claim, having similar limitations of claim 8. Thus, claim 20 is also rejected under the same rationale as cited in the rejection of claim 8.
Claims 2 and 14 are rejected under 35 U.S.C. 103 as being obvious over Jiang et al. and Gupta et al. as applied in the claims 1 and 13 in above and in view of Branson et al. (US 20150161289 A1, hereinafter Branson).
As to claim 2, Jiang discloses the system wherein:
the at least one processor (page 3) is further configured to (page 10, calculating resource is wasted, increasing the cost of the model training. By convolutional neural network module each module and forming a template method, it can solve the problem that the computer hardware resource not fully used in the model training process. The invention claims a method based on convolutional neural network module each module and made into template, providing a set of standard abstract convolutional neural network module template of each module. The template configures the parameters of each module in a declaring definition manner).
Gupta disclose method of resource costs (par. 0023-0024, computing tasks may be referred to as being compute-bound, in which the time needed to complete the computing task is determined mainly by the speed of the processor (or multiple processing cores in the case of a computing task … which the time needed to complete the computing task is determined mainly by the amount of memory required to hold the working data, … where the end cost to the user is calculated based on the amount of time the computing resource);
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include the method of resource cost, as disclosed by Gupta, for the purpose to automatically allocate an instance of a computing resource that is configured according to the proposed computer hardware configuration. (see paragraph 0048 of Gupta).
Jiang as modified by Gupta does not explicitly disclose the following limitations, but,
Branson discloses configured to initiate compilation (par. 0004, a system for initializing a streaming application for execution on compute nodes. In various embodiments, the system may include a compiler configured to receive a source code that includes an operator graph that includes a plurality of processing elements, each processing element having stream operators. In addition, the compiler may also be configured to parse, from the source code, a metadata tag describing a customization of at least one of the one or more stream operators having a windowing processing operation. Furthermore, the compiler may also be configured to compile the source code of the streaming application having the windowing processing operation based on the metadata tag. Further, 0041, The memory 425 may store a compiler 136. The compiler 136 compiles modules, which include source code or statements, into the object code, which includes machine instructions that execute on a processor. …).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include configured to initiate compilation, that can be used for selection of communication links, as disclosed by Branson, for the purpose to perform peephole optimizations, local optimizations, loop optimizations, inter-procedural or whole-program optimizations, machine code optimizations, or any other optimizations that reduce the amount of time required to execute the object code, to reduce the amount of memory required to execute the object code, or both (see paragraph 0041 of Branson).
As to claim 14, it is the method claim, having similar limitations of claim 2. Thus, claim 14 is also rejected under the same rationale as cited in the rejection of claim 2.
Claims 4-5, 10-12 and 16-17 are rejected under 35 U.S.C. 103 as being obvious over Jiang et al. and Gupta et al. as applied to claims 1, 3, 9, 13, 15, and 18 in the above and further in view of Farabet et al. (US 20120303932 A1, hereinafter Farabet).
As to claim 4, Jiang discloses the system wherein the set of configuration parameters comprises one or more of an input size, a filter size, a stride, (page 4, stride: step length, representing the moving step length in the convolution process, default is 1; the movement of the general convolution kernel on the input image is from left to right, from up to down [i.e. input size]… Stride: pooling window moving step length, default value is kernel-size; Padding: the layer number of O is supplemented to each side of the input; Dilation: a parameter of the element step amplitude in the control window; return - if it is True, returning the sequence number for outputting the maximum value; ceil mode: if it is True, calculating the output signal is too small will be integer, replacing the default downward whole operation [i.e. filter size]).
Jiang as modified by Gupta does not explicitly disclose the following claim limitations, but,
Farabet discloses the system comprises a base grid size (Fig. 1, par. 0019, … The dataflow processor 100 can be configured into various data grids. A dataflow grid or grid is a particular configuration of a dataflow processor 100 …).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include the system comprises a base grid size, as disclosed by Farabet, for the purpose to implement of the dataflow grid, what is the sequence of grid configurations that yields the shortest computation time (see paragraph 0045 of Farabet).
As to claim 5, Jiang discloses the system wherein the resource costs comprise a memory unit count, a compute unit count (abstract, … The invention can fully exert the performance of the computing device, reduce the cost of the model designer training model, improve the repeated utilization rate of the cloud service provider device, simplify the design process of the convolutional neural network module. Further, page 9, calculating resource scale (memory [i.e. memory unit]),
Farabet discloses the system wherein a compute latency unit count (par. 0023, The dataflow processor 100 [i.e. latency unit] can be reconfigured at runtime. The time to reconfigure the dataflow processor 100 is in the order of the latency of the dataflow processor 100 …).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include the system wherein a compute latency unit count, as disclosed by Farabet, for the purpose to comparing to the time needed to perform one such operation (see paragraph 0023 of Farabet).
As to claim 10, Gupta discloses the system wherein the at least one processor is further configured to (par. 0023, … the computing task is determined mainly by the speed of the processor (or multiple processing cores in the case of a computing task …):
conduct the computing template according to the allocated resources (par. 0024, … the end cost to the user is calculated [i.e. estimated] based on the amount of time the computing resource is used … . Further, par. 0025, … configured upon request to implement the different types of computer system configurations. For example, a physical server may have 256 processor cores and 2048 GB of installed memory, and these computing resources may be allocated and dedicated to particular virtual machines in accordance with their configuration types. For example, a virtual machine providing a low performance computer system configuration may be allocated one processor core and 4 GB of memory, while a compute-optimized computer system … . Further, par. 0032, the user 102 will typically be asked to specify the type of computer system that will be provisioned to execute the specified computing task, illustrated in FIG. 1 in the simplified form [i.e. template] of selecting one hardware configuration from among hardware configuration).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include conduct the computing template according to the allocated resources, as disclosed by Gupta, for the purpose to automatically allocate an instance of a computing resource that is configured according to the proposed computer hardware configuration. (see paragraph 0048 of Gupta).
Jiang as modified by Gupta does not explicitly disclose the following limitations but,
Farabet generate dataflow configuration information (par. 0019, ] FIG. 1 illustrates a dataflow architecture in accordance with an illustrative embodiment. The dataflow architecture can process homogeneous streams of data in parallel. A dataflow processor 100 includes numerous processing tiles 110, a controller or control unit 120, and a memory access module 130 … ) that enables the reconfigurable dataflow computing system t(par. 0023, The dataflow processor 100 can be reconfigured at runtime. The time to reconfigure the dataflow processor 100 is in the order of the latency of the dataflow processor 100. This allows the dataflow processor to be reconfigured between two kinds of operations, where the time to reconfigure the dataflow processor is negligible compared to the time needed to perform one such operation).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include generate dataflow configuration information that enables the reconfigurable dataflow computing system, as disclosed by Farabet, for the purpose to reconfigure the dataflow processor 100 is in the order of the latency of the dataflow processor 100 (see paragraph 0023 of Farabet).
As to claim 11, Farabet discloses the system wherein the at least one processor is further configured to (abstract, A processor includes a plurality of processing tiles, wherein each tile is configured at runtime to perform a configurable operation):
configure the reconfigurable dataflow computing system using the dataflow configuration information (par. 0025, The runtime configuration bus 160 allows the dataflow processor 100 to be configured at runtime. In one embodiment, each module in the design, such as the processing tiles and the memory access module, has a set of configurable parameters, routes or settings (depicted as squares on FIG. 1), and possesses a unique address on the network. Groups of similar modules can also share a broadcast address, which dramatically speeds up reconfiguration of elements that need to perform similar tasks. As a non-limiting example, the controller 120 can broadcast a configuration packet to a group of processing tiles, that cause the group. … ).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include configure the reconfigurable dataflow computing system using the dataflow configuration information, as disclosed by Farabet, for the purpose to reconfigure the dataflow processor 100 is in the order of the latency of the dataflow processor 100 (see paragraph 0023 of Farabet).
As to claim 12, Jiang disclose the system wherein:
at least one processor is further (page 3) configured to launch execution of the computing template (page 4, an order for defining the name of the template and the execution step of the template; Further, page 13, a step of defining the name of the template and the execution step sequence of the template).
Farabet discloses configured to launch execution of the computinginformation (par. 0021, The dataflow processor 100 also includes a runtime configuration bus 160 that can reconfigure many aspects of the processing tiles 110 and the memory access module 130 at runtime. Configurable aspects include, but are not limited to, connections, operators, and memory access module modes. Configurable elements are depicted as squares in FIG. 1. The runtime configuration bus 160 is operably connected to the memory access module 130 and to each of the processing tiles 110. The controller 120 uses the runtime configuration bus 160 to reconfigure the processing tiles 110).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include configured to launch execution of the computing template with the reconfigurable dataflow computing system according to the dataflow configuration information, as disclosed by Farabet, for the purpose to reconfigure the dataflow processor 100 is in the order of the latency of the dataflow processor 100 (see paragraph 0023 of Farabet).
As to claim 16, it is the method claim, having similar limitations of claim 4. Thus, claim 16 is also rejected under the same rationale as cited in the rejection of claim 4.
As to claim 17, it is the method claim, having similar limitations of claim 5. Thus, claim 17 is also rejected under the same rationale as cited in the rejection of claim 5.
Claims 7 and 19 are rejected under 35 U.S.C. 103 as being obvious over Jiang et al. and Gupta et al. as applied to claims 6 and 18 in the above and further in view of Ghazvinan et al. (US 20230186476 A1, hereinafter Ghazvinan) .
As to claim 7, Jiang as modified by Gupta does not explicitly disclose the following limitations but,
Ghazvinan discloses the system wherein the plurality of proposed configurations comprise a plurality of base grid sizes (par. 0058, The object detector 106 may include a deep neural network system configured to generate object proposals. An object proposal may include a volume in the 3D space of the point cloud that includes a set of points that has a high probability of representing a certain object. ... FIG. 2B the 3D grid may define nodes 207 of a certain density in the 3D space of the point cloud. The nodes of the 3D grid may define centers for object proposals in the 3D space. This way, the deep neural network system of the object detector may define an object proposal network configured to determine object proposals, e.g. 3D bounding boxes, located on the nodes of the 3D grid, wherein the 3D grid comprises a plurality of nodes which spans the 3D space of the point cloud.).
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the system disclosed by Jiang to include the system wherein the plurality of proposed configurations comprise a plurality of base grid sizes, as disclosed by Ghazvinan, for the purpose to includes a set of points that has a high probability of representing a certain object. (see paragraph 0058 of Gupta).
As to claim 19, it is the method claim, having similar limitations of claim 7. Thus, claim 19 is also rejected under the same rationale as cited in the rejection of claim 7.
Conclusion
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 MOHAMMAD H KABIR whose telephone number is (571)270-1341. The examiner can normally be reached M-F, 8:00 am - 5:00 pm.
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, Sam Sough can be reached at 571-272-6799. 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.
/Mohammad Kabir/
Examiner, Art Unit 2192
/S. Sough/SPE, Art Unit 2192
Prosecution Insights