Mca ppt
Download as PPTX, PDF1 like1,879 views
Virtual memory maps virtual addresses to physical addresses using hardware and software. Recent work proposed Redundant Memory Mappings (RMM) to improve virtual memory performance. RMM represents ranges of virtually and physically contiguous pages using a single translation entry. This reduces TLB misses by expanding the TLB's reach. RMM requires modest changes and delivers a high-performance virtual memory system transparent to applications. Evaluations show RMM improves performance over various workloads by reducing the overhead of virtual memory translations.
Mca ppt
- 1. October 21 RANGE TRANSLATIONS FOR FAST VIRTUAL MEMORY Rahul &Sunil Mtech-CSE Sem-I ASE Bangalore 1
- 2. October 21 Introduction to Virtual memory and TLB • virtual memory is a memory management technique that is implemented using both hardware and software. It maps memory addresses used by a program, called virtual addresses into physical addresses in computer memory. • Each process has a large private virtual address space which are typically 4K bytes in size. • Memory Manager divides the virtual address space into equal sized pieces called pages • Memory Manager divides the physical address space into equal sized pieces called frames 2
- 3. MMU with TLB 3 • TLB hit. • TLB miss. • Logical address. • Physical address. Size of(page)=size of(frame) October 21
- 4. October 21 4 Problem Statement • Physical memory is growing exponentially • But TLB size have grown slowly Limited TLB reach
- 7. October 21 7 Components for RMM
- 8. October 21 8 TLB operations in RMM design
- 9. October 21 9 RMM node design with CR-RT register
- 10. • If memory is requested on demand, it may be assigned out of order. • Demand paging allocates single pages even when large amount of physical memory is available. • Used in place of demand paging. • Allocates contiguous memory. • Similar to buddy paging. Eager Paging October 21 10
- 11. 11 Pseudocode of RMM memory allocator October 21
- 12. • Implemented in Linux kernel v3.15.5. • RMM is defined with respect to Intel x86-64 Snady Bridge dual-socket Xeon E5- 2430 core. • Chosen a 32-entry fully associative range TLB accessed in parallel with the L2 page TLB. • Workloads with poor TLB performance from SPEC 2006, BioBench, Parsec, and big-memory workload were selected. October 21 12
- 13. 13 Evaluating RMM with different workloads October 21
- 14. 14 Summary Redundant memory mappings is a robust translation mechanism that improves performance by reducing cost of virtual memory. RMM efficiently represents ranges of many pages that are virtually and physically contiguous. RMM requires modest changes to existing hardware and oprating system. The resulting system delivers a virtual memory system that gives high performance, flexibility and completely transparent to applications. October 21
- 15. 15October 21