Arm CoreLink System Controllers orchestrate critical AMBA system tasks such as hardware virtualization support, interrupt management, L3 cache operation, DMA, TrustZone security and peripheral operation. Designed for optimal compatibility with Arm processors and Multimedia IP, they are the natural complements to the System IP Interconnect and Memory Controller product lines.
System Controllers are used for many functions within an SoC, including:
Virtualization is the ability of a system to support multiple virtual machines each running its own guest operating system with its own private access to memory and IO peripherals without any interference whatsoever from the others. Arm's SMMU enables hardware virtualization by performing stage1 and/or stage2 address translation for IO devices, thereby providing them with the same view of memory as the CPU/GPU.
Another interesting use case of SMMU is to perform memory mapping to enable a programmable view of memory for IO peripherals. As an example, SMMU supports scatter-gather operations whereby an IO peripheral can access disjoint locations of physical memory as a single contiguous block of virtual memory due to the SMMU translation. This results in improved utilization of memory while ensuring that the IO peripheral performance is not affected.
CPU to off-chip memory communication has become the performance bottleneck in many SoCs. Level 2 Cache Controllers improve CPU performance by keeping memory access on-chip with a typical latency 10-25% of accessing the data off-chip. At the same time, the reduced CPU demands on the off-chip memory bandwidth free up that resource for other masters. Level 2 Cache Controllers also contribute significantly to power efficiency as on-chip accesses are typically an order of magnitude lower in power versus going off-chip. Typically these are embedded inside the CPU or delivered as standalone components optimized to match the processor requirements and integrate easily into the AMBA interconnect.
Interrupt Management & Translation
Arm GICs perform the key function of interrupt management and translation. Management functions include detection, masking, prioritization and routing of interrupts to the appropriate cores. Interrupt Translation Service (ITS) modules in the GIC perform the task of device isolation and ID translation for incoming message-based interrupts, thereby enabling virtual machines to program peripherals directly.
Arm's TrustZone security platform enables hardware-based secure access to regions of both off-chip memory and on-chip SRAM. The memory space is divided into a configurable number of regions each with its own access permissions. This setup ensures that malicious software cannot compromise system operation by latching onto data or code it shouldn't have access to.
Direct Access to Memory
Arm's CoreLink DMA controllers perform critical functions of moving streams of data between a peripheral and memory without overloading the CPU. Software programs typically write the data transfer instructions to memory and trigger the DMA engine which then takes over and performs the data move without incurring any additional overhead on the CPU.
Get support with Arm Training courses and design reviews. You can also open a support case or manage existing cases.
|Answered||Forum FAQs||0 votes||5557 views||0 replies||Started 8 months ago by Annie||Answer this|
|Answered||Forum FAQs||0 votes||4567 views||0 replies||Started 8 months ago by Annie||Answer this|
|Not answered||Unaligned transfer pattern in AXI4||0 votes||122 views||0 replies||Started yesterday by Supal||Answer this|
|Not answered||AHB - continue the transfer after an error response||0 votes||276 views||0 replies||Started 2 days ago by Joon Hong||Answer this|
|Not answered||Flash Patching in Cortex M7||0 votes||366 views||0 replies||Started 2 days ago by SaiGautamJP||Answer this|
|Answered||HSELx behavior for One master to two slave transfer (back to back) for address A (slave1) and address B (slave2)||0 votes||557 views||2 replies||Latest 3 days ago by Tapas||Answer this|
|Answered||Forum FAQs Started 8 months ago by Annie||0 replies 5557 views|
|Answered||Forum FAQs Started 8 months ago by Annie||0 replies 4567 views|
|Not answered||Unaligned transfer pattern in AXI4 Started yesterday by Supal||0 replies 122 views|
|Not answered||AHB - continue the transfer after an error response Started 2 days ago by Joon Hong||0 replies 276 views|
|Not answered||Flash Patching in Cortex M7 Started 2 days ago by SaiGautamJP||0 replies 366 views|
|Answered||HSELx behavior for One master to two slave transfer (back to back) for address A (slave1) and address B (slave2) Latest 3 days ago by Tapas||2 replies 557 views|