Memory management using dynamically allocated dirty mask space
A technology of memory and memory system, applied in memory system, memory architecture access/allocation, memory address/allocation/relocation, etc., can solve the problems of consuming memory space and weakening system performance, etc.
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment Construction
[0018] One scheme to maintain data coherency is sometimes referred to as a "read-assign-write" scheme. Upon receiving a write request, the processor may first read the target cache line from system memory, and then the processor may write the selected data unit (eg, byte) to the cache. Unwritten data cells have the same value as system memory. When a cache line is evicted from cache, the entire cache line is sent to system memory. Any unaffected data cells can be written with the same value. With this scheme, any write to a cache line results in a read to system memory. This results in extra traffic to system memory and undesirable latency for write requests. In modern digital systems, memory bandwidth can often be the bottleneck of system performance. This may be especially true for graphics processing units (GPUs). Therefore, this approach may not be preferable due to the increased traffic of the system memory.
[0019] Another scheme to maintain data coherency involve...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


