David Patterson et al. Intelligent RAM(IRAM): the Industrial Setting, Applications, and Architectures. In International Conference on COmputer Design, Oct 1997.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....are used to bridge the speed gap of fast 1 processor and slow memory, but they are not always e ective. Modern processors sometimes dedicate more than 80 of the chip area to caches, we still see programs with high miss rates. Naturally, people start to think of putting logic inside the memory [11] to exploit the low latency and high bandwidth of memory macro cells. This thesis summarizes some of the results of an ongoing research project that quantitatively evaluates the design trade o s in implementing advanced intelligent memory. The design of the architecture is driven by these ....

....processor architecture [23] This architecture combines scalar CPU, caches, vector unit and DRAM on a single chip. Vector unit can better utilize the potential bandwidth of onchip DRAM, and it is also claimed vectorization is a very mature compiler technique and many application can be vertorized [11]. The reasons for choosing a vector approach also include consideration of energy eciency and yield of the chip. While we do agree with these observations we still believe vector approach is more limited than the MIMD alternative. In retrospect of the past architectures we can see that nearly ....

[Article contains additional citation context not shown here]

David Patterson et al. Intelligent RAM(IRAM): the Industrial Setting, Applications, and Architectures. In International Conference on COmputer Design, Oct 1997.

.... a personal digital assistant (PDA) with multimedia capabilities, including speech recognition, handwriting recognition, video, digital photography and audio [10] Another application, called I STORE for Intelligent Storage, envisions building low cost clusters using IRAM chips as building blocks [11]. 3 These two applications will have very different I O requirements. Based on these requirements, we will examine the different options available for a high speed interconnect. In Section 2.2 we will determine the bandwidth of a single I O module which can then be replicated to achieve more ....

D. Patterson et al. Intelligent RAM (IRAM): the industrial setting, applications, and architecture. In ICCD '97 International Conference on Computer Design, 1997.

....example, the Sears Roebuck and Co decision support database grew from 1.3 TB in 1997 to 4.6 TB in 1998. The usage trends indicate that there is a change in user expectations regarding large databases from primarily archival storage to frequent re processing in their entirety. Patterson et al. [28] quote an observation by Greg Papadopolous while processors are doubling performance every 18 months, customers are doubling data storage every nine to twelve months and would like to mine this data overnight to shape their business practices [27] To meet this need, several researchers have ....

....architectures. Keeton et al. [22] and Hsu et al. [19] analyzed technological trends and individual applications to evaluate the expected performance of Active Disk architectures for decision support tasks. Keeton et al. [22] proposed an architecture (IDISK) in which a processor in memory chip (IRAM [28]) is integrated into the disk unit, the disk units are connected by a crossbar and can communicate directly. They argue that IDISK architectures offer several potential price and performance advantages over traditional server architectures for decision support. Our results regarding the impact of ....

D. Patterson et al. Intelligent RAM (IRAM): the Industrial Setting, Applications, and Architectures. In Proceedings of the International Conference on Computer Design, 1997.

....placed online is outstripping the growth in disk capacity as well as the improvement in performance of commodity processors. Furthermore, there is a change in user expectations regarding large datasets from primarily archival storage to frequent reprocessing in their entirety. Patterson et al. [25] quote an observation by Greg Papadopolous while processors are doubling performance every 18 months, customers are doubling data storage every five months and would like to mine this data overnight to shape their business practices [24] Jim Gray argues that satellite data repositories will ....

....propose a model much similar to ours and evaluate its performance for data mining and image processing algorithms. They show that these algorithms can achieve significant gains from the use of Active Disks. Keeton et al. [18] propose an architecture (IDISK) in which a processor in memory chip (IRAM [25]) is integrated into the disk unit and the disk units are connected by a crossbar. They compare this architecture with a conventional 5 http: www.nsic.org nasd 0 0.5 1 1.5 2 2.5 3 3.5 Select Groupby Cube Sort Conv Earth 40M(CD) 200M(CD) 400M(CD) 0 0.2 0.4 0.6 0.8 1 1.2 Select Groupby Cube ....

[Article contains additional citation context not shown here]

D. Patterson et al. Intelligent RAM (IRAM): the Industrial Setting, Applications, and Architectures. In Proceedings of the International Conference on Computer Design, 1997.

....placed online is outstripping the growth in disk capacity as well as the improvement in performance of commodity processors. Furthermore, there is a change in user expectations regarding large datasets from primarily archival storage to frequent reprocessing in their entirety. Patterson et al. [11] quote an observation by Greg Papadopolous while processors are doubling performance every 18 months, customers are doubling data storage every five months and would like to mine this data overnight to shape their business practices [24] Jim Gray argues that satellite data repositories will ....

....and the utility of downloading application specific code to disk processors. The idea of moving computation closer to the data it processes is also being explored for other devices. The growing processor memory gap [30] has lead to several proposals for integrating processing logic into DRAM [23, 11]. Chong et al. [23] have proposed that data intensive programs should be partitioned into processing intensive and memory intensive components and the memoryintensive component should be offloaded into logic placed in the memory system. In this paper, we have assumed a traditional processor memory ....

[Article contains additional citation context not shown here]

D. Patterson et al. Intelligent RAM (IRAM): the Industrial Setting, Applications, and Architectures. In Proceedings of the International Conference on Computer Design, 1997.

....For example, the Sears Roebuck and Co decision support database grew from 1.3 TB in 1997 to 4.6 TB in 1998. The usage trends indicate that there is a change in user expectations regarding large databases from primarily archival storage to frequent reprocessing in their entirety. Patterson et al. [37] quote an observation by Greg Papadopolous while processors are doubling performance every 18 months, customers are doubling data storage every nine to twelve months and would like to mine this data overnight to shape their business practices [36] To meet this need, several researchers have ....

....data and for edgedetection algorithms. Their results indicate that these algorithms can achieve significant gains from the use of Active Disks. Based on an analysis of several technological trends, Keeton et al. [27] propose an architecture (IDISK) in which a processor in memory chip (IRAM [37]) is integrated into the disk unit and the disk units are connected by a crossbar. They argue that IDISK architectures offer several potential price and performance advantages over traditional server architectures for decision support. Given the volume of data processed and the cost of fetching ....

D. Patterson et al. Intelligent RAM (IRAM): the Industrial Setting, Applications, and Architectures. In Proceedings of the International Conference on Computer Design, 1997.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC