Latency and throughput – the two measures of system performance

Performance of the system is one of the most stringent criteria for its success. While performance increases the desirability among customers, cost is what makes it affordable. This is the reason why system designers aim for maximum performance with available resources such as power and area constraints. There are two related parameters that determine the performance output of a system –

Throughput - Throughput is a measure of the productivity of the system. In electronic/communication systems, throughput refers to rate at which output data is produced. Higher the throughput, more productive is the system. In most of the cases, it is measured as time difference between two consecutive outputs (nth and n+1th). Throughput also refers to the rate at which input data can be applied to system.

Let us discuss with the help of an example:

Above figure depicts the throughput of 3 number adder. Result of input set applied at 1st clock cycle appears at output at 3rd clock cycle and in 4th clock cycle next input set is applied and output comes in 6th clock cycle.  Hence, throughput of above design is ⅓ per clock cycle. As we can see from diagram, first input is applied in first clock cycle and 2nd input is applied in 4th clock cycle. Hence we can also say that throughput is rate at which input data can be applied to system.

Latency- Latency is the time taken by a system to produce output after input is applied. It is a measure of delay response of a design. Higher the latency value, slower is the system. in synchronous designs, it is measured in terms of number of clock cycles. In combinational designs, latency is basically propagation delay of circuit. In non pipelined designs, latency improvement is major area of concern. In more general terms, it is time difference between output and input time.

Relationship between throughput and latency: Both latency and throughput are inter-related. It is desired to have maximum throughput and minimum latency. Increasing latency and/or throughput might make the system costly. Let us take an example. Consider a park with 3 rides and it takes 5 minutes for a ride.  A child can take sequentially these rides; i.e, ride 1, ride 2 and then ride 3. Firstly, let us assume that only one child at a time is allowed to enter park at a time. While he is taking a ride, no one is allowed to enter the park. Thus, the throughput of the park is 15 minutes per child and latency is 15 minutes. Now, let us assume that while a child has finished taking ride1, another child is allowed to enter park. Thus, in this case, throughput will be 5 minutes per child whereas latency is still 15 minutes. Thus, we have increased the throughput of the system without affecting latency and at the same cost.

What is Logic Built-in Self Test (LBIST)

LBIST stands for Logic Built-In Self Test. As VLSI marches to deep sub-micron technologies, LBIST is gaining importance due to the unique advantages it provides. LBIST refers to a self-test mechanism for testing random logic. The logic can be tested with no intervention from the outside world. In other words, a piece of hardware and/or software is inbuilt into an integrated circuit to test itself. By random logic, is meant any form of hardware (logic gates, memories etc.) that can form a part or whole of the chip. A generic LBIST system is implemented using STUMPS (Self-Test Using MISR and PRPG) architecture. A typical LBIST system is as shown in the figure below:

Figure 1: A typical LBIST system

Components of an LBIST system: A typical LBIST system comprises following:

1. Logic to be tested, or, as is called Circuit Under Test (CUT): In case of LBIST, the logic to be tested through LBIST is the Circuit under Test (CUT). Any random logic residing on the chip can be brought under LBIST following a certain procedure.
2. PRPG (Pseudo-Random Pattern Generator): A PRPG generates input patterns that are applied to internal scan chains of the CUT for LBIST testing. In other words, PRPG acts as a Test Pattern Generator (TPG) for LBIST. A PRPG can either use a counter or an LFSR for pattern generation.
3. MISR (Multi-Input Signature Register): MISR obtains the response of the device to the test patterns applied. An incorrect MISR output indicates a defect in the CUT. In classical language, MISR acts as a ORA (Output Response Analyzer) for LBIST testing.
4. A master (LBIST controller): The controller controls the functioning of the LBIST; i.e. clocks propagation, initialization and scan patterns flow in and out of the LBIST scan chains.

One of the most stringent requirements in LBIST testing is the prohibition of X-sources. There cannot be any source of ‘X’ during LBIST testing. By ‘X’, is meant a definite, but unknown value. It might be either ‘0’ or ‘1’, but it is not known what value is being propagated. All X-sources are masked and a known value is allowed to be propagated in LBIST.

Why ‘X’ is prohibited in LBIST: As stated above, there cannot be any ‘X’ propagating during LBIST testing. The reason behind this is that LBIST involves MISR to calculate the signature of the LBIST patterns. Since, the resulting signal is unique, any unknown value can result in the corruption of the signature. So, there cannot be any ‘X’ in LBIST testing.

Advantages of LBIST: As stated above, there are many unique advantages of LBIST that make it desirable, especially in safety critical designs such as those used in automobiles and aeroplanes. LBIST offers many advantages as listed below:

• LBIST provides self-test capability to logic inside chip; thus, the chip can test itself without any external control and interference.
• This provides the ability to be tested at higher frequencies reducing test time considerably.
• LBIST can run while the chip is on field running functionally. Thus, it is very useful in safety critical applications wherein faults developed on field can be easily detectable at startup before chip goes into functional mode.

Overheads due to LBIST: Along with many advantages, there are some overheads due to LBIST as mentioned below:

(i)                  The LBIST implementation involves some hardware on-chip to control LBIST. So, there are area and power impacts due to these. In other words, the cost of chip increases.

(ii)                Also, ‘X’-masking involves addition of extra logic gates in already timing critical functional signals causing impact on timing as well.

(iii)               Another disadvantage of using LBIST is that even the on-chip test equipment may fail. This is not the problem with testing using outside equipment with proven test circuitry

References:

1. Identification and reduction of safe-stating points in LBIST designs 
2. Logic built-in self-test
3. Challenges in LBIST verification of high reliability SoCs

Also read:

• MBIST (Memory Built-In Self Test)
• Lockup latch - principle, application and timing
• Controllability and observability - basics of DFT
• X-propagation through different logic gates
• Meyer's singleton pattern