Direct memory access
Direct memory access is a feature of computer systems that allows certain hardware subsystems to access main system memory independent of the central processing unit.
Without DMA, when the CPU is using programmed input/output, it is typically fully occupied for the entire duration of the read or write operation, and is thus unavailable to perform other work. With DMA, the CPU first initiates the transfer, then it does other operations while the transfer is in progress, and it finally receives an interrupt from the DMA controller when the operation is done. This feature is useful at any time that the CPU cannot keep up with the rate of data transfer, or when the CPU needs to perform work while waiting for a relatively slow I/O data transfer. Many hardware systems use DMA, including disk drive controllers, graphics cards, network cards and sound cards. DMA is also used for intra-chip data transfer in multi-core processors. Computers that have DMA channels can transfer data to and from devices with much less CPU overhead than computers without DMA channels. Similarly, a processing element inside a multi-core processor can transfer data to and from its local memory without occupying its processor time, allowing computation and data transfer to proceed in parallel.
DMA can also be used for "memory to memory" copying or moving of data within memory. DMA can offload expensive memory operations, such as large copies or scatter-gather operations, from the CPU to a dedicated DMA engine. An implementation example is the I/O Acceleration Technology. DMA is of interest in network-on-chip and in-memory computing architectures.
Principles
Third-party
Standard DMA, also called third-party DMA, uses a DMA controller. A DMA controller can generate memory addresses and initiate memory read or write cycles. It contains several hardware registers that can be written and read by the CPU. These include a memory address register, a byte count register, and one or more control registers. Depending on what features the DMA controller provides, these control registers might specify some combination of the source, the destination, the direction of the transfer, the size of the transfer unit, and/or the number of bytes to transfer in one burst.To carry out an input, output or memory-to-memory operation, the host processor initializes the DMA controller with a count of the number of words to transfer, and the memory address to use. The CPU then commands the peripheral device to initiate a data transfer. The DMA controller then provides addresses and read/write control lines to the system memory. Each time a byte of data is ready to be transferred between the peripheral device and memory, the DMA controller increments its internal address register until the full block of data is transferred.
Bus mastering
In a bus mastering system, also known as a first-party DMA system, the CPU and peripherals can each be granted control of the memory bus. Where a peripheral can become a bus master, it can directly write to system memory without the involvement of the CPU, providing memory address and control signals as required. Some measures must be provided to put the processor into a hold condition so that bus contention does not occur.Modes of operation
Burst mode
In burst mode, an entire block of data is transferred in one contiguous sequence. Once the DMA controller is granted access to the system bus by the CPU, it transfers all bytes of data in the data block before releasing control of the system buses back to the CPU, but renders the CPU inactive for relatively long periods of time. The mode is also called "Block Transfer Mode".Cycle stealing mode
The cycle stealing mode is used in systems in which the CPU should not be disabled for the length of time needed for burst transfer modes. In the cycle stealing mode, the DMA controller obtains access to the system bus the same way as in burst mode, using BR and BG signals, which are the two signals controlling the interface between the CPU and the DMA controller. However, in cycle stealing mode, after one byte of data transfer, the control of the system bus is deasserted to the CPU via BG. It is then continually requested again via BR, transferring one byte of data per request, until the entire block of data has been transferred. By continually obtaining and releasing the control of the system bus, the DMA controller essentially interleaves instruction and data transfers. The CPU processes an instruction, then the DMA controller transfers one data value, and so on. On the one hand, the data block is not transferred as quickly in cycle stealing mode as in burst mode, but on the other hand the CPU is not idled for as long as in burst mode. Cycle stealing mode is useful for controllers that monitor data in real time.Transparent mode
Transparent mode takes the most time to transfer a block of data, yet it is also the most efficient mode in terms of overall system performance. In transparent mode, the DMA controller transfers data only when the CPU is performing operations that do not use the system buses. The primary advantage of transparent mode is that the CPU never stops executing its programs and the DMA transfer is free in terms of time, while the disadvantage is that the hardware needs to determine when the CPU is not using the system buses, which can be complex. This is also called as "Hidden DMA data transfer mode".Cache coherency
DMA can lead to cache coherency problems. Imagine a CPU equipped with a cache and an external memory that can be accessed directly by devices using DMA. When the CPU accesses location X in the memory, the current value will be stored in the cache. Subsequent operations on X will update the cached copy of X, but not the external memory version of X, assuming a write-back cache. If the cache is not flushed to the memory before the next time a device tries to access X, the device will receive a stale value of X.Similarly, if the cached copy of X is not invalidated when a device writes a new value to the memory, then the CPU will operate on a stale value of X.
This issue can be addressed in one of two ways in system design: Cache-coherent systems implement a method in hardware, called bus snooping, whereby external writes are signaled to the cache controller which then performs a cache invalidation for DMA writes or cache flush for DMA reads. Non-coherent systems leave this to software, where the OS must then ensure that the cache lines are flushed before an outgoing DMA transfer is started and invalidated before a memory range affected by an incoming DMA transfer is accessed. The OS must make sure that the memory range is not accessed by any running threads in the meantime. The latter approach introduces some overhead to the DMA operation, as most hardware requires a loop to invalidate each cache line individually.
Hybrids also exist, where the secondary L2 cache is coherent while the L1 cache is managed by software.
Examples
ISA
In the original IBM PC, there was only one Intel 8237 DMA controller capable of providing four DMA channels. These DMA channels performed 8-bit transfers, could only address the first megabyte of RAM, and were limited to addressing single 64 kB segments within that space. Additionally, the controller could only be used for transfers to, from or between expansion bus I/O devices, as the 8237 could only perform memory-to-memory transfers using channels 0 & 1, of which channel 0 in the PC was dedicated to dynamic memory refresh. This prevented it from being used as a general-purpose "Blitter", and consequently block memory moves in the PC, limited by the general PIO speed of the CPU, were very slow.With the IBM PC/AT, the enhanced AT Bus added a second 8237 DMA controller to provide three additional, and as highlighted by resource clashes with the XT's additional expandability over the original PC, much-needed channels. The page register was also rewired to address the full 16 MB memory address space of the 80286 CPU. This second controller was also integrated in a way capable of performing 16-bit transfers when an I/O device is used as the data source and/or destination, doubling data throughput when the upper three channels are used. For compatibility, the lower four DMA channels were still limited to 8-bit transfers only, and whilst memory-to-memory transfers were now technically possible due to the freeing up of channel 0 from having to handle DRAM refresh, from a practical standpoint they were of limited value because of the controller's consequent low throughput compared to what the CPU could now achieve. In both cases, the 64 kB segment boundary issue remained, with individual transfers unable to cross segments even in 16-bit mode, although this was in practice more a problem of programming complexity than performance as the continued need for DRAM refresh to monopolise the bus approximately every 15 μs prevented use of large block transfers.
Due to their lagging performance and unavailability of any speed grades that would allow installation of direct replacements operating at speeds higher than the original PC's standard 4.77 MHz clock, these devices have been effectively obsolete since the late 1980s. Particularly, the advent of the 80386 processor in 1985 and its capacity for 32-bit transfers, as well as the development of further evolutions to or replacements for the "ISA" bus with their own much higher-performance DMA subsystems made the original DMA controllers seem more of a performance millstone than a booster. They were supported to the extent they are required to support built-in legacy PC hardware on later machines. The pieces of legacy hardware that continued to use ISA DMA after 32-bit expansion buses became common were Sound Blaster cards that needed to maintain full hardware compatibility with the Sound Blaster standard; and Super I/O devices on motherboards that often integrated a built-in floppy disk controller, an IrDA infrared controller when FIR mode is selected, and an IEEE 1284 parallel port controller when ECP mode is selected. In cases where an original 8237s or direct compatibles were still used, transfer to or from these devices may still be limited to the first 16 MB of main RAM regardless of the system's actual address space or amount of installed memory.
Each DMA channel has a 16-bit address register and a 16-bit count register associated with it. To initiate a data transfer the device driver sets up the DMA channel's address and count registers together with the direction of the data transfer, read or write. It then instructs the DMA hardware to begin the transfer. When the transfer is complete, the device interrupts the CPU.
Scatter-gather or vectored I/O DMA allows the transfer of data to and from multiple memory areas in a single DMA transaction. It is equivalent to the chaining together of multiple simple DMA requests. The motivation is to off-load multiple input/output interrupt and data copy tasks from the CPU.
DRQ stands for Data request; DACK for Data acknowledge. These symbols, seen on hardware schematics of computer systems with DMA functionality, represent electronic signaling lines between the CPU and DMA controller. Each DMA channel has one Request and one Acknowledge line. A device that uses DMA must be configured to use both lines of the assigned DMA channel.
16-bit ISA permitted bus mastering.
Standard ISA DMA assignments:
PCI
A PCI architecture has no central DMA controller, unlike ISA. Instead, any PCI component can request control of the bus and request to read from and write to system memory. More precisely, a PCI component requests bus ownership from the PCI bus controller, which will arbitrate if several devices request bus ownership simultaneously, since there can only be one bus master at one time. When the component is granted ownership, it will issue normal read and write commands on the PCI bus, which will be claimed by the bus controller and will be forwarded to the memory controller using a scheme which is specific to every chipset.As an example, on a modern AMD Socket AM2-based PC, the southbridge will forward the transactions to the northbridge using HyperTransport, which will in turn convert them to DDR2 operations and send them out on the DDR2 memory bus. As a result, there are quite a number of steps involved in a PCI DMA transfer; however, that poses little problem, since the PCI device or PCI bus itself are an order of magnitude slower than the rest of the components.
A modern x86 CPU may use more than 4 GB of memory, utilizing Physical Address Extension, a 36-bit addressing mode, or the native 64-bit mode of x86-64 CPUs. In such a case, a device using DMA with a 32-bit address bus is unable to address memory above the 4 GB line. The new Double Address Cycle mechanism, if implemented on both the PCI bus and the device itself, enables 64-bit DMA addressing. Otherwise, the operating system would need to work around the problem by either using costly double buffers also known as bounce buffers, or it could use an IOMMU to provide address translation services if one is present.
I/OAT
As an example of DMA engine incorporated in a general-purpose CPU, newer Intel Xeon chipsets include a DMA engine called I/O Acceleration Technology, which can offload memory copying from the main CPU, freeing it to do other work. In 2006, Intel's Linux kernel developer Andrew Grover performed benchmarks using I/OAT to offload network traffic copies and found no more than 10% improvement in CPU utilization with receiving workloads.DDIO
Further performance-oriented enhancements to the DMA mechanism have been introduced in Intel Xeon E5 processors with their Data Direct I/O feature, allowing the DMA "windows" to reside within CPU caches instead of system RAM. As a result, CPU caches are used as the primary source and destination for I/O, allowing network interface controllers to DMA directly to the Last level cache of local CPUs and avoid costly fetching of the I/O data from system RAM. As a result, DDIO reduces the overall I/O processing latency, allows processing of the I/O to be performed entirely in-cache, prevents the available RAM bandwidth/latency from becoming a performance bottleneck, and may lower the power consumption by allowing RAM to remain longer in low-powered state.AHB
In systems-on-a-chip and embedded systems, typical system bus infrastructure is a complex on-chip bus such as AMBA High-performance Bus. AMBA defines two kinds of AHB components: master and slave. A slave interface is similar to programmed I/O through which the software can write/read I/O registers or local memory blocks inside the device. A master interface can be used by the device to perform DMA transactions to/from system memory without heavily loading the CPU.Therefore, high bandwidth devices such as network controllers that need to transfer huge amounts of data to/from system memory will have two interface adapters to the AHB: a master and a slave interface. This is because on-chip buses like AHB do not support tri-stating the bus or alternating the direction of any line on the bus. Like PCI, no central DMA controller is required since the DMA is bus-mastering, but an arbiter is required in case of multiple masters present on the system.
Internally, a multichannel DMA engine is usually present in the device to perform multiple concurrent scatter-gather operations as programmed by the software.
Cell
As an example usage of DMA in a multiprocessor-system-on-chip, IBM/Sony/Toshiba's Cell processor incorporates a DMA engine for each of its 9 processing elements including one Power processor element and eight synergistic processor elements. Since the SPE's load/store instructions can read/write only its own local memory, an SPE entirely depends on DMAs to transfer data to and from the main memory and local memories of other SPEs. Thus the DMA acts as a primary means of data transfer among cores inside this CPU.DMA in Cell is fully [|cache coherent]. In both read and write, a DMA command can transfer either a single block area of size up to 16 KB, or a list of 2 to 2048 such blocks. The DMA command is issued by specifying a pair of a local address and a remote address: for example when a SPE program issues a put DMA command, it specifies an address of its own local memory as the source and a virtual memory address as the target, together with a block size. According to an experiment, an effective peak performance of DMA in Cell reaches 200 GB per second.