D.A.Patterson et al., "Intelligent RAM (IRAM): chips that remember and compute", Proc. IEEE Int. Solid-State Circ. Conf.,San Francisco CA, pp.224-225, Feb. 1997.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the memory bandwidth can be achieved by integrating the cache and the main memory into the same chip, or merged DRAM logic LSI. Eliminating the chip boundary between the cache and the main memory solves the I O pin bottleneck problem, thereby improving dramatically the memory bandwidth [42] [48], 53] 4. Low Energy Memory Access Techniques From the energy definition explained in Section 2, it can be understood that there are at least three approaches to reducing the average memory access energy (AMAE) as follows: Reducing the cache access energy (ECache ) and maintaining the ....

D. A. Patterson, T. Anderson, N. Cardwell, R. Fromm, K. Keeton, C. Kozyrakis, R. Thomas, and K. Yelick,"Intelligent RAM(IRAM) Chips that Remember and Compute," Proc. of the 1997.

....of 0.1m technologies it will be possible to integrate billions of transistors on one single device. Different approaches have been proposed to fill the gap between the tremendous new possibilities and the resulting design complexity. These were, e.g. the integration of large memory blocks [16] or the use of pre verified intellectual property (IP) components in SoC designs. The standardization of IP deliverables has been a first step to simplify the integration of powerful macro blocks into custom designs. However, an efficient on chip interconnect structure is necessary to use the ....

D. Patterson, T. Anderson, N. Cardwell, R. Fromm, K. Keeton, C. Kozyrakis, R. Thomas, and K. Yelick, "Intelligent RAM (IRAM): Chips that Remember and Compute," in Proceedings of the 1997.

....memory bandwidth requirement of these. And we focus on one of the architectures, and evaluate the impact of the actual memory bandwidth and the DRAM access latency upon the performance. 1 Introduction Merged DRAM logic LSIs, such as Parallel Processing RAM (PPRAM) 5] and Intelligent RAM (IRAM)[6], would provide much freedom for the design of their memorypath architectures, as well as higher memory bandwidth, than conventional separate DRAM logic type systems. Namely, most of conventional separate DRAM logic type systems employ a traditional memorypath architecture: i.e. the memory ....

....1 Datapath Datapath Datapath Registers Registers Cache Main Memory Main Memory Main Memory (a) DRCM (b) DRM (c) DM Figure 1: On Chip Memorypath Architectures would be exploited between the registers and the main memory on load store operations. Examples include U .C. Berkeley s Vector IRAM[6]. 3) DM architecture (Figure 1(c) DM (Datapath Main memory) renews the obsolete vector architectures with no vector registers such as CDC Cyber. As shown in Figure 1(c) the on chip memorypath consists of datapath and DRAM main memory only. High on chip memory bandwidth would be exploited ....

[Article contains additional citation context not shown here]

Patterson, D., Anderson, T., Cardwell, N., Fromm, R., Keeton, K., Kozyrakis, C., Thomas, R., and Yelick, K., "Intelligent RAM (IRAM): Chips that remember and compute,"

....of SRAM is much lower than DRAM. Thus, the researches on the integration of MPU and DRAM in a single chip becomes active. The researches that merge the DRAM and MPU(or DSP) in a single chip are focused on the high performance system such as MPPs and vector processors by architecture level approach [3] [4] The architecture in [3] is focused on getting the high performance using vector processors and the architecture in [4] is similar to the DRAM USED MEMORY REQUIRED 1M 4M 16M 64M 20 40 60 MEMORY WASTED MEMORY REQUIRED Figure 3: Silicon Wasted In Systems using a NonStandard Memory 1 ....

....Thus, the researches on the integration of MPU and DRAM in a single chip becomes active. The researches that merge the DRAM and MPU(or DSP) in a single chip are focused on the high performance system such as MPPs and vector processors by architecture level approach [3] 4] The architecture in [3] is focused on getting the high performance using vector processors and the architecture in [4] is similar to the DRAM USED MEMORY REQUIRED 1M 4M 16M 64M 20 40 60 MEMORY WASTED MEMORY REQUIRED Figure 3: Silicon Wasted In Systems using a NonStandard Memory 1 10 100 1000 10000 1e 00 ....

[Article contains additional citation context not shown here]

D. Patterson and et al., "Intelligent RAM(IRAM):Chips that Remember and Compute," in ISSCC Digest of Technical Papers, pp. 224--226, Feb. 1996.

....cache has no large tables for storing the memory access history. This scheme does not require any modification of instruction set architectures. The goal of D VLS cache is to improve the system performance of merged DRAM logic LSIs such as PPRAM(Parallel Processing RAM) 8] or IRAM(Intelligent RAM)[9] by making good use of the high on chip memory bandwidth. The rest of this paper is organized as follows. Section 2 describes the concept and principle of the VLS cache. Section 3 discusses the D VLS cache architecture. Section 4 presents some simulation results and shows the performance ....

Patterson, D., Anderson, T., Cardwell, N., Fromm, R., Keeton, K., Kozyrakis, C., Thomas, R., and Yelick, K., "Intelligent RAM (IRAM): Chips that remember and compute," 1997 ISSCC Digest of Technical Papers, pp.224--225, Feb. 1997.

....the current memory technology can support the gigabit DRAMs, a single memory chip would cover the memory volume needed for the computer systems in the future. A number of studies for the memory logic integration have utilized both high internal memory bandwidth and the available chip density [1, 2, 3, 4, 5, 6] In this paper an effective memory processor integrated architecture, called memory based processor array (MPA) for computer vision is proposed. It is an effective SIMD array which is based on the memory processor integration structure. Thus, it can be easily attached into any host system from ....

....c] then O[r; c] 0 34 else O[r; c] 1; 35 parend fline 31 parbeging 36 endfor fline 29 for g 37 endprocedure fThresholding MPAg Figure 3: An algorithm for TBT implemented on the MPA system. as in lines 1727 of Figure 3. Finding the BT in this procedure is based on Otsu s algorithm [8] Let R[1]; R[G] represent the histogram probabilities of the observed gray values 1; G, respectively. Here, the BT is the threshold which is chosen in such a way that the weighted sum of the group variances should be minimized. Therefore, the number FBT HP of computation steps in finding the BT ....

D. Patterson, et. al, "Intelligent RAM(IRAM): Chips that remember and compute," ISSCC97.

....of proposals for single chip processor memory integration use a very simple processor and fill the remaining area on the chip with DRAM. On the other hand, the proposed vector processor or quad processors integrated with DRAM are proposed in order to exploit the wide on chip DRAM bandwidth [2] [4] In this paper we evaluate the performance of a powerful microprocessor architecture, such as a multiprocessor, when the integrated DRAM is organized as on chip main memory and as on chip cache, comparing with the performance of a more conventional chip which only has SRAMbased on chip ....

....high data transfer rates for serial accesses. This speed up offers a large benefit for bandwidthintensive applications. However, improvements in memory latency have not kept up with the almost exponential increase in processor speed, so system performance is often limited by the memory access time [2]. Recently, very simple microprocessors integrated with DRAM main memory have been developed for the embedded systems [1] In this architecture, the DRAM and the processor are connected using a wide internal data bus on a single die. The high speed data transfer through the internal data bus can ....

D.Patterson, T.Anderson, N.Cardwell, R.Fromm, K.Keeton, C.Kozyrakis, R.Thomas, and K.Yelick, "Intelligent RAM (IRAM): Chips that remember and compute," Digest of Technical Papers, 1997 IEEE International Solid State Circuits Conference, pp.224-225, San Francisco, CA 1997.

....high internal memory bandwidth and the available chip density. For computer graphics, a large amount of DRAMs and a small number of logic circuits are integrated into a 3 D DRAM chip [11] A processor and memory integration onto a chip, for example U.C. Berkeley Intelligent RAM (IRAM) project [21], Mitsubishi M32R D [23] and SUN S3mp project [22] and multiple instruction stream multiple data stream (MIMD) multiprocessors with their on chip local memories, for example Kyushu Univ. Parallel Processing RAM (PPRAM) project [18, 17] and Stanford Univ. Hydra project [27, 28] were proposed in ....

D. Patterson, et. al, "Intelligent RAM(IRAM): Chips that remember and compute," Dig. Tech. Papers IEEE Int'l Solid--State Circuit Conf., (1997) 224 -- 225.

....These results demonstrate that a multiprocessor takes better advantage of the large bandwidth provided by the on chip DRAM than a uniprocessor. 1 Introduction Recently, microprocessor chips with integrated DRAM have been developed [3] to close the speed gap between processors and memory [1,2]. In these chips, the DRAM and the processor are connected using a wide internal data bus. The high speed data transfer over the internal data bus improves memory latency, because the load capacitance of the internal data bus is small compared to that of an external data bus. The majority of ....

D. Patterson, T. Anderson, N. Cardwell, R. Fromm, K. Keeton, C. Kozyrakis, R. Thomas, and K. Yelick, "Intelligent RAM (IRAM): Chips that Remember and Compute," Digest of Technical Papers, 1997 IEEE International Solid State Circuits Conference, pp. 224-225, San Francisco, CA 1997.

....than a uniprocessor. Keywords DRAM, on chip DRAM, embedded DRAM, L2 caches, on chip L2 caches, SRAM caches, multiprocessors, multiprocessor on a chip 2 1 Introduction Recently, microprocessor chips with integrated DRAM have been developed [1] to close the speed gap between processors and memory [2,3]. In these chips, the DRAM and the processor are connected using a wide internal data bus. High speed data transfer over the internal data bus improves memory latency, because the load capacitance of the internal data bus is small compared to that of an external data bus. The majority of proposals ....

D. Patterson, T. Anderson, N. Cardwell, R. Fromm, K. Keeton, C. Kozyrakis, R. Thomas, and K. Yelick, "Intelligent RAM (IRAM): Chips that Remember and Compute," Digest of Technical Papers, 1997 IEEE International Solid State Circuits Conference, pp. 224-225, San Francisco, CA 1997.

No context found.

D. Patterson et al., "Intelligent RAM (IRAM): Chips That Remember and Compute," Dig. Technical Papers, 1997 IEEE Int'l Solid-State Circuits Conf., IEEE, Piscataway, N.J., 1997, pp. 224-225.

....style of computer than those based on conventional microprocessors. IRAM technology offers the following potential: Improve memory latency by factors of 5 to 10 and memory bandwidth by factors of 50 to 100, by redesigning the memory interface and exploiting the proximity of on chip memory [1][2] Improve energy efficiency of memory by factors of 2 to 4, primarily by going off chip less frequently [3] 4] Reduce design effort tenfold by filling the die with replicated memory rather than with custom logic [5] Make the memory size and organization fit the intended workload; ....

Patterson, D.; Anderson, T.; Cardwell, N.; Fromm, R.; Keeton, K.; Kozyrakis, C.; Thomas, R.; Yelick, K. "Intelligent RAM (IRAM): chips that remember and compute," 1997 IEEE International Solids-State Circuits Conference. Digest of Technical Papers, San Francisco, CA, USA, 6-8 Feb. 1997. p.224-5.

....the processor memory performance gap. We call the percent of die area and transistors dedicated to caches and other memory latency hiding hardware the Memory Gap Penalty . Table 3 quantifies the penalty; it has grown to 60 of the area and almost 90 of the transistors in several microprocessors[Pat97]. In fact, the Pentium Pro offers a package with two dies, with the larger die being the 512 KB second level cache. While the Processor Memory Performance Gap has widened to the point where it is dominating performance for many applications, the cumulative effect of two decades of 60 per year ....

....of testing during manufacturing is significant for DRAMs. Adding a processor would significantly increase the test time on conventional DRAM testers. 3. Quantifying the Potential Advantages of IRAM This section looks at three early attempts to quantify what might be done with IRAM technology [Pat97] [Fro97] Estimating Performance of an IRAM Alpha The fastest current microprocessor, the Alpha 21164, was described in sufficient detail [Cve96] to allow us to estimate performance of an IRAM using a similar organization. The Alpha 21164 has three caches on chip: an 8 KB instruction cache, an 8 ....

[Article contains additional citation context not shown here]

Patterson, D.; Cardwell, N.; Anderson, T.; Cardwell, N.; Fromm, R.; Keeton, K.; Kozyrakis,K.; Thomas,R.; and Yelick,.K. "Intelligent RAM (IRAM): Chips that remember and compute". Digest of Technical Papers, 1997 IEEE International Solid-State Circuits Conference, San Francisco, CA, USA, Feb. 1997.

No context found.

D.A.Patterson et al., "Intelligent RAM (IRAM): chips that remember and compute", Proc. IEEE Int. Solid-State Circ. Conf.,San Francisco CA, pp.224-225, Feb. 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