Sorting networks and their applications
read more
Citations
A Survey of General-Purpose Computation on Graphics Hardware
Software protection and simulation on oblivious RAMs
A Survey of General-Purpose Computation on Graphics Hardware.
Billion-scale similarity search with GPUs
GPU Computing
References
On the Synthesis of Signal Switching Networks with Transient Blocking
Related Papers (5)
Frequently Asked Questions (16)
Q2. What are other applications of sorting memories?
Other applications of sorting memories are as a switching network with bu ering, a multiaccess memory, a multiaccess content-addressable memory and as a multiprocessor.
Q3. What is the simplest way to sort a bitonic sequence?
Since any two monotonic sequences can be put together to form a bitonic sequence a network which rearranges a bitonic sequence into monotonic order (a bitonic sorter) can be used as a merging network.
Q4. How many bits can be placed between different levels?
Parallel-input-parallel-output registers of 1024 bits each can be placed between certain levels to perform this task or the re-clocking may be incorporated within each comparison element with a pair of ipops on the outputs.
Q5. What is the function of a multi-processor?
Multi-processorBy adding processing logic to perform additions, subtractions, etc., on groups of adjacent words of a sorting memory one can implement a multi-processor.
Q6. What is the function of the adjacent word transfer?
The adjacent word transfer sends back signals over each path to signal each input and output line whether or not a connection has been established.
Q7. What is the effect of doubling the size of a merge?
Doubling the size of a merge only increases the longest path by unity so the merging time increases slowly with the size of the network.
Q8. What is the advantage of a sorting network over a normal crossbar?
It has the advantages over a normal crossbar of requiring less hardware (an n-input n-output switching network can be built with approximately ( 14 )n(log2n) 2 elements versus n2 in a normal crossbar) and of having a constant fan-in and a fan-out requirement on its elements.
Q9. What is the function of the multi-access content addressable memory?
Multi-access content addressable memoryBy adding facilities for shifting the bits within the words in the aforementioned memory di erent elds of the words can be brought into the more-signi cant positions which govern the ordering of the words.
Q10. How many inputs and outputs can be accommodated in a complete cycle?
While a complete cycle may be long in this memory (50-bit words at 100 nanoseconds/bit = 5 microseconds/recirculation = 10 microseconds/complete cycle) many inputs and outputs can be accommodated in each cycle.
Q11. What is the problem in the design of a computing system?
A major problem in the design of such a computing system is the connecting together of the various parts of the system (the I/O devices, memories, processing units, etc.) in such a way that all the required data transfers can be accommodated.
Q12. What is the function of the m-item sorting network?
Each input line inserts a word containing the output address desired (or zeroes if the line is inactive), a control bit equal to 1 and a priority number into an m-item sorting network with bi-directional elements.
Q13. What is the simplest way to build a s by t" merging network?
An \\s by t" merging network can be built by presenting the odd-indexed numbers of the two input lists to one small merging network (the odd merge), presenting the even-indexed number to another small merging network (the even merge) and then comparing the outputs of these small merges with a row of comparison elements.
Q14. What is the use of fast sorting capability?
Such fast sorting capability can be used to manipulate large sets of data quickly and solve some of the communications problems associated with large-scale computing systems.
Q15. What is the s+t output of the merging network?
The s + t outputs of the merging network present the s+t numbers of the merged lists in ascending order, c2; c2; :::; cs+t.A \\1 by 1" merging network is simply one comparison element.
Q16. How many elements are needed to sort a number?
Each path goes through ( 1 2 )p(p+ 1) levels.8 numbers4-number bitonic sortersA sorter of 1024 numbers will have 55 levels and 24,063 elements with odd-even merges or 28,160 elements with bitonic sorters.