Modern Processor Units

                                          Modern processor units

Understanding the modern processor units with the timeline is more convenient. The letter starts with the time of 1960-1970. Processes of this time used to take multiple cycles for the instructions. They could take only one instruction in flight at once. Also, Optimization means minimizing the number of instructions executed. Sometimes replacing expensive general-purpose instructions with specialized sequences of cheaper ones used to happen.

From 1980, the CPUs became pipelined and the meaning of Optimization became minimizing pipeline stalls. Dependency ordering such results wasn't needed within the next instruction. Also, the computed branches became very expensive when they were not predicted correctly.

CPU became much faster than memory in the early 1990. Caches hid some latency. The maximizing locality of reference and prefetching became Optimization. Sometimes, recalculating results is quicker than fetching from memory. Large caches consume a lot of power is the noticeable thing here.

From mid-1990 the processor unit became superscalar. Also, independent instructions were executed in parallel. The use of recorded instructions was made to reduce dependencies. Structuring code for the highest possible ILP was Optimization. Till this time look and rolling was no longer such a big win.

Till 2000, Programmable GPUs became mainstream, and hardware was optimized for stream processing in parallel. Processes became very fast for massively-parallel floating-point operations. The cost of moving data between CPUs became very high. Offloading operations to the GPU was Optimization.

In the modern time after 2010, the modern processes came with multiple CPU and GPU courses. All these cores are behind the same memory interface and the cost of moving data between them is very low. They Increasingly contain specialized accelerators.  

We can understand the major part of the modern processor units which are:

1) CPU 2) GPU 3) DPU

CPU

CPU is taken into account because of the brain of the pc. It performs all kinds of knowledge processing operations. It stores data, intermediate results, and directions (program) and also controls the operation of all parts of the pc. The CPU itself has the subsequent three components.

i) Memory or Storage Unit -

Stores all the data and thus the instructions required for processing.Stores intermediate results of processing.

All inputs and outputs are transmitted through the foremost memory.

ii) Control Unit

Controls the transfer of knowledge and directions among other units of a computer.

Manages and coordinates all the units of the pc.

iii) ALU(Arithmetic Logic Unit)

Perform arithmetic operations like addition, subtraction, multiplication, and division.

Perform logic operations like comparing, selecting, matching, and merging of knowledge.

GPU

GPU is a graphical processing unit that enables you to run high-definition graphics on your PC, which is the end of modern computing. Like the CPU (Central Processing Unit), it's a single-chip processor. The GPU has many cores as compared to the 4 or 8 within the latest CPUs.

The primary job of the GPU is to compute 3D functions.

Following are major features of GPU:

• Control hardware dominates processors

• Complex, difficult to build and verify

• Takes a substantial fraction of die Scales poorly

• Pay for max throughput,

• sustain average throughput

• Quadratic dependency checking

• Control hardware doesn’t do any math!

In the past few years, the GPU has evolved from a fixed-function special-purpose processor into a fully parallel programmable processor with additional fixed-function special-purpose functionality.

DPU

Data processing units, commonly referred to as DPUs, are a replacement class of reprogrammable high-performance processors combined with high-performance network interfaces that are optimized to perform and accelerate networks.

DPUs will become the third component in data center servers alongside CPU (central processing units) and GPUs (graphics processing units) due to their ability to accelerate and perform network and storage functions. following are major features of DPU:

• Very High-speed connectivity via one or multiple 100 Gigabit to 200 Gigabit interfaces

• Higher-speed packet processing

• Memory controllers offering support for DDR4 and DDR5 RAM

• Accelerators

• PCI Express Gen 4 Support

• Security features

• Custom operating system separated from the host system’s OS

Thank You for reading the Blog!

Blog by - Rachana Gaikwad | Siddhida Hulwane | Harsh Iramani | Isha Venikar

Department of EnTC | Vishwakarma Institute of Technology, Pune.