eltonjohn
Advanced Member level 4
HI ..
I'm designing a massively parallel search engine in a FPGA (VIRTEX PRO)
The idea is to search by brute force a particular patern or a combination of patterns .It is possible to do multicriteria search .. like search "names of people with 40 years old or more ,and blind living in brussels "
This implies searching on data that is not indexed .Also another application of a typical seacrh would be .. looking for ..the ocurrence of "AGGGCTTTAAAAGCGCGCGCGCGCG" on a genome data base ..
Now according to my first evaluation it can be possible to search 50 gigabytes per second using 8 FPGAs .. Now this implies big reservoirs of DRAM memories ..But i'm out of touch with SDRAM or DDR memories ..
So i need to evaluate what are the current read access time with dram .
i don't want to use banking ,want to keep my solution very inexpensive ..
Also i see a big application of this technology as internet search engines
Does any body knows how is this done by let's say yahoo .. what kind of equipment they use .. and how fast they do it .. also what is the cost of their equipment .. Please if you have some thing useful to comment on the above ..please do
summerizing
1) DRAM reservoir specs
2) Yahoo typs of search equipment
3) Yahoo equipment prices
4) how fast is their equipment
I'm designing a massively parallel search engine in a FPGA (VIRTEX PRO)
The idea is to search by brute force a particular patern or a combination of patterns .It is possible to do multicriteria search .. like search "names of people with 40 years old or more ,and blind living in brussels "
This implies searching on data that is not indexed .Also another application of a typical seacrh would be .. looking for ..the ocurrence of "AGGGCTTTAAAAGCGCGCGCGCGCG" on a genome data base ..
Now according to my first evaluation it can be possible to search 50 gigabytes per second using 8 FPGAs .. Now this implies big reservoirs of DRAM memories ..But i'm out of touch with SDRAM or DDR memories ..
So i need to evaluate what are the current read access time with dram .
i don't want to use banking ,want to keep my solution very inexpensive ..
Also i see a big application of this technology as internet search engines
Does any body knows how is this done by let's say yahoo .. what kind of equipment they use .. and how fast they do it .. also what is the cost of their equipment .. Please if you have some thing useful to comment on the above ..please do
summerizing
1) DRAM reservoir specs
2) Yahoo typs of search equipment
3) Yahoo equipment prices
4) how fast is their equipment