“I had the pleasure of working with Carlos at Schlumberger software technology innovation center (STIC) . I was hired by Carlos and he was my direct manager for around 1 year. Carlos is that combination of outstanding technical acumen and brilliant leadership qualities, which is extremely rare to find. On the technical front, he is an expert and sharp thinker in the field of computer systems and architecture in general; and HPC and cloud in particular. He likes challenging problems and will always have interesting opinions and innovative ideas on technical problems. On the leadership front, he was great at managing the HPC and 3D viz team and also made sure that the work is extremely visible to stakeholders. Carlos ensured that a major part of the innovation that happened in team eventually became commercial success for Schlumberger. With such diverse capabilities, Carlos is well positioned to succeed in challenging jobs. ”
About
Activity
-
We are looking for a SLURM expert! :) If you are or know of someone who would like to join the Cloud HPC team in Google, we are hiring:…
We are looking for a SLURM expert! :) If you are or know of someone who would like to join the Cloud HPC team in Google, we are hiring:…
Shared by Carlos Boneti
-
Voilà ce que j'appelle une entame d'article dégueulasse, et je pèse mes mots. Je viens de lire le début de l'article du Monde sur l'attrait de…
Voilà ce que j'appelle une entame d'article dégueulasse, et je pèse mes mots. Je viens de lire le début de l'article du Monde sur l'attrait de…
Liked by Carlos Boneti
Experience & Education
-
Google
View Carlos’s full experience
See their title, tenure and more.
or
Licenses & Certifications
-
Perl
Brainbench
Issued -
Compaq Accredited System Engineer (ASE)
-
Issued Expires
Publications
-
Automatic denoising by 2-D continuous wavelet transform
Society of Exploration Geophysicists
We present a seismic processing workflow for identification, separation, and removal of specific wave modes, combining feature extraction by two-dimensional continuous wavelet transforms (2-D CWT) with machine learning algorithms. It addresses the challenges arising from temporally and spatially transient phases, which cannot be effectively addressed using conventional stationary filtering. We first transform the seismic data into a domain in which the signal of interest and the noise are…
We present a seismic processing workflow for identification, separation, and removal of specific wave modes, combining feature extraction by two-dimensional continuous wavelet transforms (2-D CWT) with machine learning algorithms. It addresses the challenges arising from temporally and spatially transient phases, which cannot be effectively addressed using conventional stationary filtering. We first transform the seismic data into a domain in which the signal of interest and the noise are well-separated. We establish a representation of the 2-D CWT output that is intuitive to understand, visualize and label. We characterize the noise in this domain and use a machine learning classifier to automate the noise identification process. We then design a filter to remove the unwanted noise modes and transform the data back to its original time domain. We demonstrate the effectiveness of the method by applying it to noisy marine acquisition shot gathers. The described method is computationally robust and its theory can be extended to higher dimensions. As a consequence, the methodology is applicable to any temporally and spatially continuous seismic dataset, both pre-stack and after imaging
Other authors -
-
SMT Malleability in IBM POWER5 and POWER6 Processors
IEEE Transactions on Computers, Volume: PP , Issue: 99
-
[Paper] Characterizing Power and Temperature Behavior of POWER6Based System
IEEE Journal on Emerging and Selected Topics in Circuits and Systems - JETCAS , vol. 1, no. 3, pp. 228-241, 2011
-
Power and thermal characterization of POWER6 system
19th International Conference on Parallel Architecture and Compilation Techniques (PACT 2010), Vienna, Austria
Other authors -
-
A dynamic scheduler for balancing HPC applications
Proceeding SC '08 Proceedings of the 2008 ACM/IEEE conference on Supercomputing Article No. 41
Load imbalance cause significant performance degradation in High Performance Computing applications. In our previous work we showed that load imbalance can be alleviated by modern MT processors that provide mechanisms for controlling the allocation of processors internal resources. In that work, we applied static, hand-tuned resource allocations to balance HPC applications, providing improvements for benchmarks and real applications.
In this paper we propose a dynamic process scheduler…Load imbalance cause significant performance degradation in High Performance Computing applications. In our previous work we showed that load imbalance can be alleviated by modern MT processors that provide mechanisms for controlling the allocation of processors internal resources. In that work, we applied static, hand-tuned resource allocations to balance HPC applications, providing improvements for benchmarks and real applications.
In this paper we propose a dynamic process scheduler for the Linux kernel that automatically and transparently balances HPC applications according to their behavior. We tested our new scheduler on an IBM POWER5 machine, which provides a software-controlled prioritization mechanism that allows us to bias the processor resource allocation. Our experiments show that the scheduler reduces the imbalance of HPC applications, achieving results similar to the ones obtained by hand-tuning the applications (up to 16%). Moreover, our solution reduces the application's execution time combining effect of load balance and high responsive scheduling.Other authors -
-
Balancing HPC applications through smart allocation of resources in MT processors
IEEE International Symposium on Parallel and Distributed Processing, 2008. IPDPS 2008.
Many studies have shown that load imbalancing causes significant performance degradation in high performance computing (HPC) applications. Nowadays, multi-threaded (MT1) processors are widely used in HPC for their good performance/energy consumption and performance/cost ratios achieved sharing internal resources, like the instruction window or the physical register. Some of these processors provide the software hardware mechanisms for controlling the allocation of processor's internal…
Many studies have shown that load imbalancing causes significant performance degradation in high performance computing (HPC) applications. Nowadays, multi-threaded (MT1) processors are widely used in HPC for their good performance/energy consumption and performance/cost ratios achieved sharing internal resources, like the instruction window or the physical register. Some of these processors provide the software hardware mechanisms for controlling the allocation of processor's internal resources. In this paper, we show, for the first time, that by appropriately using these mechanisms, we are able to control the tasks speed, reducing the imbalance in parallel applications transparently to the user and, hence, reducing the total execution time. Our results show that our proposal leads to a performance improvement up to 18% for one of the NAS benchmark. For a real HPC application (much more dynamic than the benchmark) the performance improvement is 8.1%. Our results also show that, if resource allocation is not used properly, the imbalance of applications is worsened causing performance loss.
Other authors -
-
Soft Real-Time papers Scheduling on SMT Processors with Explicit Resource Allocation.
Soft Real-Time papers Scheduling on SMT Processors with Explicit Resource Allocation
Several software or hardware approaches have been proposed to reduce the execution time variability of SMT processors. Software solutions rely on profiling the schedulable tasks to determine the effects of resource sharing over their performance, while the hardware approaches consider a fixed small number of tasks, avoiding the global system-scheduling problem. Both approaches lack of generality or do not take into account architectural details.
This work targets the scheduling of soft…Several software or hardware approaches have been proposed to reduce the execution time variability of SMT processors. Software solutions rely on profiling the schedulable tasks to determine the effects of resource sharing over their performance, while the hardware approaches consider a fixed small number of tasks, avoiding the global system-scheduling problem. Both approaches lack of generality or do not take into account architectural details.
This work targets the scheduling of soft real-time tasks on an explicit resource allocation processor, where the system is able to enforce hardware allocation decisions. We propose a simple extension to the Earliest Deadline First scheduler: Resource Aware EDF. RA-EDF uses resource allocation mechanisms to ensure at least the minimum amount of resources needed by a task to meet its deadline. It yields improvements on every case when compared to previous task schedulers: 8% better on average and up to 18%, requiring no additional profiling.Other authors -
-
Software-controlled priority characterization of POWER5 processor
Proceedings of the 35th International Symposium on Computer Architecture (ISCA), 2008.
Due to the limitations of instruction-level parallelism, thread-level parallelism has become a popular way to improve processor performance. One example is the IBM POWER5TM processor, a two-context simultaneous-multithreaded dual-core chip. In each SMT core, the IBM POWER5 features two levels of thread resource balancing and prioritization. The first level provides automatic in-hardware resource balancing, while the second level is a software-controlled priority mechanism that presents eight…
Due to the limitations of instruction-level parallelism, thread-level parallelism has become a popular way to improve processor performance. One example is the IBM POWER5TM processor, a two-context simultaneous-multithreaded dual-core chip. In each SMT core, the IBM POWER5 features two levels of thread resource balancing and prioritization. The first level provides automatic in-hardware resource balancing, while the second level is a software-controlled priority mechanism that presents eight levels of thread priorities. Currently, software-controlled prioritization is only used in limited number of cases in the software platforms due to lack of performance characterization of the effects of this mechanism. In this work, we characterize the effects of the software-based prioritization on several different workloads. We show that the impact of the prioritization significantly depends on the workloads coscheduled on a core. By prioritizing the right task, it is possible to obtain more than two times of throughput improvement for synthetic workloads compared to the baseline. We also present two application case studies targeting two different performance metrics: the first case study improves overall throughput by 23.7% and the second case study reduces the total execution time by 9.3%. In addition, we show the circumstances when a background thread can be run transparently without affecting the performance of the foreground thread.
Other authors -
Patents
-
Efficient simulation of oilfield production systems
Filed US US20160275217A1
Systems and methods for simulating fluid properties in a production system include receiving data from a sensor. The data represents a first property of a fluid that is measured by the sensor. A second property of the fluid is estimated based at least in part on the measured first property. A third property of the fluid is calculated using the estimated second property. Calculating the third property includes running a first simulation on a first simulator and running a second simulation on a…
Systems and methods for simulating fluid properties in a production system include receiving data from a sensor. The data represents a first property of a fluid that is measured by the sensor. A second property of the fluid is estimated based at least in part on the measured first property. A third property of the fluid is calculated using the estimated second property. Calculating the third property includes running a first simulation on a first simulator and running a second simulation on a second simulator. The first and second simulations at least partially overlap in the time domain. It is then determined whether results of the first and second simulations converge to a common value.
Other inventors -
-
Blockchain Ledger for Persisting and Verifying Oil and Gas Events
Filed US20200162260A1
Methods, apparatuses, and computer-readable media are set forth for logging oil and gas events using a combination of a primary database that stores event data for an oil and gas event along with a blockchain ledger that stores one or more identifying characteristics of the event (e.g., a hash of the event data), thereby enabling the event data to be verified. By utilizing the combination of a primary database and a blockchain ledger, the blockchain ledger is not required to store all of the…
Methods, apparatuses, and computer-readable media are set forth for logging oil and gas events using a combination of a primary database that stores event data for an oil and gas event along with a blockchain ledger that stores one or more identifying characteristics of the event (e.g., a hash of the event data), thereby enabling the event data to be verified. By utilizing the combination of a primary database and a blockchain ledger, the blockchain ledger is not required to store all of the data associated with an event. As such, where an oil and gas event is associated with a large volume of data, the size and thus the processing overhead associated with the blockchain ledger may be reduced.
Other inventors -
-
Geological Imaging and Inversion Using Object Storage
Filed US20190005637A1
Prestack images from the object store are hierarchically combined to generate a hierarchically stacked image. The hierarchically stacked image library stored in an object store, a set of prestack image is generated by combining stacked images that includes a stacked image. The stacked image is generated by combining at least the prestack images. Based at least on the hierarchically stacked image, a quality measure of a prestack image is generated. Prior to deleting at least a subset of the…
Prestack images from the object store are hierarchically combined to generate a hierarchically stacked image. The hierarchically stacked image library stored in an object store, a set of prestack image is generated by combining stacked images that includes a stacked image. The stacked image is generated by combining at least the prestack images. Based at least on the hierarchically stacked image, a quality measure of a prestack image is generated. Prior to deleting at least a subset of the prestack images from the object store and based at least on the quality measure, the prestack images are further combined to generate an enhanced stacked image. The stacked image is substituted using the enhanced stacked image. Subsequent to the substituting and prior to deleting at least the subset of the stacked images from the object store, the stacked images are combined to generate an enhanced hierarchically stacked image. The enhanced stacked image and the enhanced hierarchically stacked image are generated using failure recovery metadata. The enhanced hierarchically stacked image is presented.
Other inventors -
-
Multiscale Method for Reservoir Models
Filed US 20160018562
Methods, computing systems, and computer-readable media for multi-scale modeling. The method includes determining a first matrix for a plurality of fine cells of a model based at least in part on a physical property value represented by respective fine cells, identifying one or more overlapped cells of the plurality of fine cells that are part of at least two of the plurality of subdomains, and determining a second matrix. Determining the second matrix includes determining an intermediate…
Methods, computing systems, and computer-readable media for multi-scale modeling. The method includes determining a first matrix for a plurality of fine cells of a model based at least in part on a physical property value represented by respective fine cells, identifying one or more overlapped cells of the plurality of fine cells that are part of at least two of the plurality of subdomains, and determining a second matrix. Determining the second matrix includes determining an intermediate product by multiplying the first matrix by a prolongation matrix, which includes predicting a row of zeros in the intermediate product for the plurality of fine cells that are not the one or more overlapped cells and are not part of the at least two of the plurality of subdomains that include the one or more overlapped cells. Determining the second matrix also includes multiplying the intermediate product by a restriction matrix.
Other inventors -
-
Predictive user interface
Filed US US 14/449,954
Methods, computing systems, and computer-readable media for organizing options in an interface. The method includes receiving data representing a type of domain object, and predicting one or more generic predicted events based at least in part on the type of domain object. The one or more generic predicted events are predicted using a generic predictor. The method also includes receiving data representing an actual next event, and updating the generic predictor based on the actual next event.
Other inventors -
-
Integrated system for production design assistance
Filed US 20140303951
A method can include rendering terrain and equipment locations to a display; assigning one or more domain constraints; analyzing an objective function associated with the terrain and the equipment locations subject to at least one of the one or more domain constraints; and rendering information to the display based at least in part on the analyzing. Various other technologies, techniques, etc., are also disclosed.
Other inventors -
-
Simulation of production systems
Filed US 20140303949
A method can include, via an interface of a computing system, receiving information representing a change to a network model of a production system; via the computing system, simulating at least a portion of the production system based at least in part on a portion of the information and at least a portion of the network model to generate simulation results; and, via the interface, transmitting at least a portion of the simulation results. Various other technologies, techniques, etc., are also…
A method can include, via an interface of a computing system, receiving information representing a change to a network model of a production system; via the computing system, simulating at least a portion of the production system based at least in part on a portion of the information and at least a portion of the network model to generate simulation results; and, via the interface, transmitting at least a portion of the simulation results. Various other technologies, techniques, etc., are also disclosed.
Other inventors -
-
Method and system for accessing a virtual seismic cube
US US 13/425,207
See patentA method for processing requests includes receiving, from a requestor, a first read request to read a portion of a seismic cube. The first read request includes a virtual location of the portion. The method further includes querying a seismic cube index to obtain a mapping parameter and a storage location of a section including the portion of the seismic cube. The mapping parameter maps virtual locations in the seismic cube with data locations in the section. The section is identified using the…
A method for processing requests includes receiving, from a requestor, a first read request to read a portion of a seismic cube. The first read request includes a virtual location of the portion. The method further includes querying a seismic cube index to obtain a mapping parameter and a storage location of a section including the portion of the seismic cube. The mapping parameter maps virtual locations in the seismic cube with data locations in the section. The section is identified using the virtual location of the portion. The method further includes calculating, on a computer processor and using the mapping parameter, a data location in the section corresponding to the virtual location of the portion, and transmitting a second read request to the storage location of the section. The second read request includes the data location. The requestor receives the portion from the storage location.
Courses
-
Mastering .NET Debugging and .NET Performance, by John Robbins (Wintellect)
-
-
NCT-500 PGI Accelerator with OpenACC Directives
-
-
Software Developer Boot Camp, by Craig Larman
-
-
Two-Phase Flow in Pipes by Dr. James P. Brill
-
Honors & Awards
-
HiPEAC Paper Award
European Network of Excellence on High Performance and Embedded Architecture and Compilation
Award granted for the following paper:
Software-controlled priority characterization of POWER5 processor
Proceedings of the 35th International Symposium on Computer Architecture (ISCA), 2008.
2008
Languages
-
Portuguese
-
-
French
-
-
Spanish
-
-
English
-
-
Catalan
Elementary proficiency
Recommendations received
7 people have recommended Carlos
Join now to viewOther similar profiles
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore MoreOthers named Carlos Boneti
-
Antonio Carlos Boneti, PMP, MBA
Project Management, Team Leadership, PMI, Engineering, Scrum Master
-
Carlos Boneti
Estudiante en Universidad Nacional San Luis Gonzaga de Ica
-
João Carlos Boneti
Engenheiro Agrônomo🚜🌾🌱
-
evandro carlos boneti
Dentista
5 others named Carlos Boneti are on LinkedIn
See others named Carlos Boneti