MEMORY ERRORS Flashcards
What are the two types of memory errors?
Hard failures
Soft Failures
Describe Hard Failures
- Permanent physical defect
- Damaged memory cell(s) cannot reliably store data:
- become stuck at 0 or 1 or switch erratically between 0 and 1
- Caused by:
- Harsh environmental abuse
- Manufacturing defects
- Wear and tear
Describe soft failures
- Random, nondestructive event
- Alters the contents of one or more memory cells
- Without damaging the memory
- Caused by:
- Power supply problems
- Alpha particles (radioactive decay)
Describe and draw an illustration of the process of error detection
M-bit data is to be written into memory;
- Computation f (M) is performed on the data (K check bits);
- Both the data and the code (M + K bits) are stored in memory
- Eventually, the stored word will be read out;
- Computation f (M) is again performed on the data (K check bits);
- A comparison is made between both check bits;
*See notes for pic
What are the three possible outcomes of the error detection process?
- No errors are detected:
- Fetched data bits are sent out;
- An error is detected, and it is possible to correct the error:
- Data bits plus error correction bits are fed into a corrector
- Producing a corrected set of M bits to be sent out.
- An error is detected, but it is not possible to correct it:
- =’( This condition is reported.
What are error detection codes
Error detection codes − are used to detect the error(s) present in the received data (bit stream).
• These codes contain some bit(s), which are included (appended) to the original bit stream.
• These codes detect the error, if it is occurred during transmission of the original data (bit stream).
• Example − Parity code, Hamming code.
What are error correction codes
.• Error correction codes − are used to correct the error(s) present in the received data (bit stream) so that, we will get the original data.
• Error correction codes also use the similar strategy of error detection codes. Example: Hamming code.
• Therefore, to detect and correct the errors, additional bit(s) are appended to the data bits at the time of transmission.
Why would having one or more levels of high-speed SRAM cache not be the best way to improve DRAM performance?
- SRAM is costlier than DRAM
* Expanding cache size beyond a certain point yields diminishes returns;
What enhancements have been developed to improve the DRAM?
- SDRAM;
- DDR-SDRAM;
- RDRAM
Describe SDRAM
• SDRAM works in a synchronous manner:
• Data exchanges with the processor are synchronized with system clock
• Unlike DRAM which is asynchronous and imposes wait states.
• Processor issues data and address information to the SDRAM which responds after a set number of clock cycles allowing processor to perform other tasks.
-Another advantage of SDRAM over DRAM is that several units of data are synchronously transferred onto the bus without having to specify their address or needing to refresh MAR multiple times. This is performed by specifying a burst length variable
What is a wait state?
- Processor presents addresses and control signals to the memory;
- Indicating that data at an address should be read/written from/into;
- After a delay (access time) the DRAM either writes or reads the data
- During the access-time delay DRAM performs various tasks, e.g.:
- Activating high capacitance of the row and column lines;
- Sensing the data;
- Routing data out through the output buffers
- Processor waits through this delay, slowing performance.
- Thus the name wait states
Describe DDR - SDRAM
Double Data Rate SDRAM (DDR-SDRAM) transfers are synchronized to the system clock:
• Same as SDRAM, however:
• SDRAM only transfers on the rising edge of the clock;
• DDR-SDRAM transfers on the rising and the falling edge of the clock:
• Twice the bandwidth based on the same clock rate and data width
*See notes for picture
Describe GDDR-SDRAM
Graphics Double Data Rate SDRAM (GDDR-SDRAM) is the same as DDR-SDRAM, however:
• Lower voltage requirements, thus lower heat dissipation;
• Higher data width bus allows for higher transfer speeds;
• Made specifically for the texture requirements of video games;
Give two different reasons why increasing the die (chip) size of a
microprocessor increases the cost of the microprocessor?
- The bigger the CPU, the less yield they get from the silicon, and not all silicon is made equal. Minor imperfections occur in the process, and a number of chips are tossed out right from the get go due to unavoidable defects during manufacture. The larger you go, the more likely a given chip will have a defect that affects its performance, or makes it unusable. (more bad chips per chip made).
- Turning silicon into chips is expensive, and that cost is roughly proportional to the surface area of the silicon (assuming the same process generation). Double the area of the die and you double the cost to make it
