The Experts below are selected from a list of 119595 Experts worldwide ranked by ideXlab platform
Josep Torrellas - One of the best experts on this subject based on the ideXlab platform.
-
Design of efficient simple Coma architectures
2000Co-Authors: Josep Torrellas, S. BasuAbstract:Distributed shared-memory machines that are designed as Simple Cache-Only Memory Architecture (S-Coma) allow automatic replication and migration of data in the main memory like traditional Coma organizations. S-Coma performs in software many of the tasks that the latter did in hardware. Even then the cost-effectiveness of the S-Coma proposal is hindered by some performance bottlenecks and hardware overheads. Those issues are addressed in this work. Since the allocation unit of data in S-Coma is a page, there is a possibility of memory fragmentation. This can lead to poor utilization of memory, and frequent page faults. To solve the memory fragmentation problem, this work proposes an extension to S-Coma called Multiplexed Simple Coma (MS-Coma). The idea is to allow multiple virtual pages to map into the same physical page at the same time, therefore compressing the page working set of the application. MS-Coma requires very little support over S-Coma and reduces its execution time by about 40%. We find that MS-Coma can be implemented with off-the-shelf processors. S-Coma has a second drawback which has been addressed in this work. To enforce coherence efficiently in hardware, S-Coma requires a global address space; SRAM tables for physical-to-global and global-to-physical address translations are to be provided. However for large systems with few gigabytes of memory per node, these tables will be few megabytes of SRAM. The additional cost of S-Coma implementation comes primarily from these translation tables and the two bits of states per line of memory required to maintain coherence, as has been observed in the implementation of the WildFire prototype from Sun Microsystems. In this work, we propose a novel technique of combining the two translation tables. Further benefit could be obtained by caching the resulting data structure. The reduction in cost is significant and is attained with negligible impact on performance. Overall, we believe this work has improved the cost-effectiveness of the S-Coma architecture. Our address translation hardware is then used to improve on existing approaches for managing pages between S-Coma and CC-NUMA modes in hybrid systems.
-
HPCA - Enhancing memory use in Simple Coma: Multiplexed Simple Coma
Proceedings 1998 Fourth International Symposium on High-Performance Computer Architecture, 1Co-Authors: S. Basu, Josep TorrellasAbstract:Scalable shared-memory multiprocessors that are designed as cache-only memory architectures (Coma) allow automatic replication and migration of data in the main memory. This enhances programmability by hopefully eliminating the need for data distribution strategies and page migration schemes. A variant of Coma called Simple Coma has been proposed as a lower-cost alternative to hardware-intensive systems like Flat Coma. However, we find that Simple Coma is quite slower than Flat Coma. The main reason is the high page mapping, unmapping, and transfer overhead caused by memory fragmentation in Simple Coma. We propose a solution to the memory fragmentation problem that we call Multiplexed Simple Coma. The idea is to allow multiple virtual pages to map into the same physical page at the same time, therefore compressing the page working set of the application. Multiplexed Simple Coma requires very little support over Simple Coma and reduces its execution time by about 40%. We find that Multiplexed Simple Coma can be very easily implemented with off-the-shelf processors. In addition, there is no need to be selective when choosing what virtual pages are to share the same physical page. Overall, although Multiplexed Simple Coma is still slower than Flat Coma, since it is cheaper to implement, it represents a good cost-performance design point.
David Stuckler - One of the best experts on this subject based on the ideXlab platform.
-
always read the small print a case study of commercial research funding disclosure and agreements with coca cola
Journal of Public Health Policy, 2019Co-Authors: Sarah Steele, Gary Ruskin, Martin Mckee, David StucklerAbstract:Concerns about conflicts of interest in commercially funded research have generated increasing disclosure requirements, but are these enough to assess influence? Using the Coca-Cola Company as an example, we explore its research agreements to understand influence. Freedom of Information requests identified 87,013 pages of documents, including five agreements between Coca-Cola and public institutions in the United States, and Canada. We assess whether they allowed Coca-Cola to exercise control or influence. Provisions gave Coca-Cola the right to review research in advance of publication as well as control over (1) study data, (2) disclosure of results and (3) acknowledgement of Coca-Cola funding. Some agreements specified that Coca-Cola has the ultimate decision about any publication of peer-reviewed papers prior to its approval of the researchers' final report. If so desired, Coca-Cola can thus prevent publication of unfavourable research, but we found no evidence of this to date in the emails we received. The documents also reveal researchers can negotiate with funders successfully to remove restrictive clauses on their research. We recommend journals supplement funding disclosures and conflict-of-interest statements by requiring authors to attach funder agreements.
-
Coca-Cola - a model of transparency in research partnerships? A network analysis of Coca-Cola's research funding (2008-2016).
Public Health Nutrition, 2018Co-Authors: Paulo Serodio, Martin Mckee, David StucklerAbstract:OBJECTIVE: To (i) evaluate the extent to which Coca-Cola's 'Transparency Lists' of 218 researchers that it funds are comprehensive; (ii) map all scientific research acknowledging funding from Coca-Cola; (iii) identify those institutions, authors and research topics funded by Coca-Cola; and (iv) use Coca-Cola's disclosure to gauge whether its funded researchers acknowledge the source of funding. DESIGN: Using Web of Science Core Collection database, we retrieved all studies declaring receipt of direct funding from the Coca-Cola brand, published between 2008 and 2016. Using conservative eligibility criteria, we iteratively removed studies and recreated Coca-Cola's transparency lists using our data. We used network analysis and structural topic modelling to assess the structure, organization and thematic focus of Coca-Cola's research enterprise, and string matching to evaluate the completeness of Coca-Cola's transparency lists. RESULTS: Three hundred and eighty-nine articles, published in 169 different journals, and authored by 907 researchers, cite funding from The Coca-Cola Company. Of these, Coca-Cola acknowledges funding forty-two authors (
S. Basu - One of the best experts on this subject based on the ideXlab platform.
-
Design of efficient simple Coma architectures
2000Co-Authors: Josep Torrellas, S. BasuAbstract:Distributed shared-memory machines that are designed as Simple Cache-Only Memory Architecture (S-Coma) allow automatic replication and migration of data in the main memory like traditional Coma organizations. S-Coma performs in software many of the tasks that the latter did in hardware. Even then the cost-effectiveness of the S-Coma proposal is hindered by some performance bottlenecks and hardware overheads. Those issues are addressed in this work. Since the allocation unit of data in S-Coma is a page, there is a possibility of memory fragmentation. This can lead to poor utilization of memory, and frequent page faults. To solve the memory fragmentation problem, this work proposes an extension to S-Coma called Multiplexed Simple Coma (MS-Coma). The idea is to allow multiple virtual pages to map into the same physical page at the same time, therefore compressing the page working set of the application. MS-Coma requires very little support over S-Coma and reduces its execution time by about 40%. We find that MS-Coma can be implemented with off-the-shelf processors. S-Coma has a second drawback which has been addressed in this work. To enforce coherence efficiently in hardware, S-Coma requires a global address space; SRAM tables for physical-to-global and global-to-physical address translations are to be provided. However for large systems with few gigabytes of memory per node, these tables will be few megabytes of SRAM. The additional cost of S-Coma implementation comes primarily from these translation tables and the two bits of states per line of memory required to maintain coherence, as has been observed in the implementation of the WildFire prototype from Sun Microsystems. In this work, we propose a novel technique of combining the two translation tables. Further benefit could be obtained by caching the resulting data structure. The reduction in cost is significant and is attained with negligible impact on performance. Overall, we believe this work has improved the cost-effectiveness of the S-Coma architecture. Our address translation hardware is then used to improve on existing approaches for managing pages between S-Coma and CC-NUMA modes in hybrid systems.
-
HPCA - Enhancing memory use in Simple Coma: Multiplexed Simple Coma
Proceedings 1998 Fourth International Symposium on High-Performance Computer Architecture, 1Co-Authors: S. Basu, Josep TorrellasAbstract:Scalable shared-memory multiprocessors that are designed as cache-only memory architectures (Coma) allow automatic replication and migration of data in the main memory. This enhances programmability by hopefully eliminating the need for data distribution strategies and page migration schemes. A variant of Coma called Simple Coma has been proposed as a lower-cost alternative to hardware-intensive systems like Flat Coma. However, we find that Simple Coma is quite slower than Flat Coma. The main reason is the high page mapping, unmapping, and transfer overhead caused by memory fragmentation in Simple Coma. We propose a solution to the memory fragmentation problem that we call Multiplexed Simple Coma. The idea is to allow multiple virtual pages to map into the same physical page at the same time, therefore compressing the page working set of the application. Multiplexed Simple Coma requires very little support over Simple Coma and reduces its execution time by about 40%. We find that Multiplexed Simple Coma can be very easily implemented with off-the-shelf processors. In addition, there is no need to be selective when choosing what virtual pages are to share the same physical page. Overall, although Multiplexed Simple Coma is still slower than Flat Coma, since it is cheaper to implement, it represents a good cost-performance design point.
John Grahame - One of the best experts on this subject based on the ideXlab platform.
-
The nature of heat Coma in Littorina littorea (Mollusca: Gastropoda)
Marine Biology, 2000Co-Authors: A. P. Clarke, P. J. Mill, John GrahameAbstract:The nature of heat Coma was examined in the edible periwinkle Littorina littorea (L.). Duration of acclimation did not influence heat-Coma temperature at 12 °C, although other acclimation temperatures were important in influencing thermal tolerance, with positive shifts in Coma temperature observed in response to elevated temperatures. Previous thermal history also influenced heat-Coma temperatures. Individuals subjected to repeat heat-Coma events on a daily basis showed significant declines (P < 0.05) in Coma-temperature; in contrast individuals exposed to repeat heat-Coma events on a weekly basis showed no decline in thermal tolerance. Size-effects occurred at selected sites, where decreased heat-Coma temperatures were recorded in large individuals.
Martin Mckee - One of the best experts on this subject based on the ideXlab platform.
-
always read the small print a case study of commercial research funding disclosure and agreements with coca cola
Journal of Public Health Policy, 2019Co-Authors: Sarah Steele, Gary Ruskin, Martin Mckee, David StucklerAbstract:Concerns about conflicts of interest in commercially funded research have generated increasing disclosure requirements, but are these enough to assess influence? Using the Coca-Cola Company as an example, we explore its research agreements to understand influence. Freedom of Information requests identified 87,013 pages of documents, including five agreements between Coca-Cola and public institutions in the United States, and Canada. We assess whether they allowed Coca-Cola to exercise control or influence. Provisions gave Coca-Cola the right to review research in advance of publication as well as control over (1) study data, (2) disclosure of results and (3) acknowledgement of Coca-Cola funding. Some agreements specified that Coca-Cola has the ultimate decision about any publication of peer-reviewed papers prior to its approval of the researchers' final report. If so desired, Coca-Cola can thus prevent publication of unfavourable research, but we found no evidence of this to date in the emails we received. The documents also reveal researchers can negotiate with funders successfully to remove restrictive clauses on their research. We recommend journals supplement funding disclosures and conflict-of-interest statements by requiring authors to attach funder agreements.
-
Coca-Cola - a model of transparency in research partnerships? A network analysis of Coca-Cola's research funding (2008-2016).
Public Health Nutrition, 2018Co-Authors: Paulo Serodio, Martin Mckee, David StucklerAbstract:OBJECTIVE: To (i) evaluate the extent to which Coca-Cola's 'Transparency Lists' of 218 researchers that it funds are comprehensive; (ii) map all scientific research acknowledging funding from Coca-Cola; (iii) identify those institutions, authors and research topics funded by Coca-Cola; and (iv) use Coca-Cola's disclosure to gauge whether its funded researchers acknowledge the source of funding. DESIGN: Using Web of Science Core Collection database, we retrieved all studies declaring receipt of direct funding from the Coca-Cola brand, published between 2008 and 2016. Using conservative eligibility criteria, we iteratively removed studies and recreated Coca-Cola's transparency lists using our data. We used network analysis and structural topic modelling to assess the structure, organization and thematic focus of Coca-Cola's research enterprise, and string matching to evaluate the completeness of Coca-Cola's transparency lists. RESULTS: Three hundred and eighty-nine articles, published in 169 different journals, and authored by 907 researchers, cite funding from The Coca-Cola Company. Of these, Coca-Cola acknowledges funding forty-two authors (
