GTX 470 benchmark

For those who are interested, here is a benchmark of my new GTX 470:

0 abc quick brown fox jumped over a lazy dog
1 quick brown fox jumped over a lazy dog
2 abc quick brown fox jumped over a lazy dog
3 quick abc brown fox jumped over a lazy dog
4 abc quick brown fox jumped over a lazy dog
5 quick brown fox jumped over a lazy dog
6 quick brown abc fox jumped over a lazy dog
7 quick brown fox jumped over a lazy dog
8 ybcquick brown fox jumped over a lazy dog
9 ybcquick brown fox jumped over a lazy dog
My needles:
Time Taken: 7456.942871 Result is : 0
Printing results:
Found 10 hits and 3899998 misses in 3900008 lines
Brute force comparison of 3900008 * 3 * 125 * 50 char lines, 73.1 bln byte comparisons: 9.75 bln per sec

I have a text with lines as above, there are 3900008 lines total 50 chars each (including spaces), I have 125 3 character needles, and the file is my haystack. I am using brute force byte comparison as seen above. The CPU time taken is 7.5 seconds, and the speed of comparisons is 9.7 billion per second. Take that CPU! (which can only do 333ml/sec in single threaded fashion, almost feels like cheating, but oh well).

-Jack

PS. each file text line, and each needle string is loaded into each thread, and executed without any loops whatsoever.

For those who are interested, here is a benchmark of my new GTX 470:

0 abc quick brown fox jumped over a lazy dog
1 quick brown fox jumped over a lazy dog
2 abc quick brown fox jumped over a lazy dog
3 quick abc brown fox jumped over a lazy dog
4 abc quick brown fox jumped over a lazy dog
5 quick brown fox jumped over a lazy dog
6 quick brown abc fox jumped over a lazy dog
7 quick brown fox jumped over a lazy dog
8 ybcquick brown fox jumped over a lazy dog
9 ybcquick brown fox jumped over a lazy dog
My needles:
Time Taken: 7456.942871 Result is : 0
Printing results:
Found 10 hits and 3899998 misses in 3900008 lines
Brute force comparison of 3900008 * 3 * 125 * 50 char lines, 73.1 bln byte comparisons: 9.75 bln per sec

I have a text with lines as above, there are 3900008 lines total 50 chars each (including spaces), I have 125 3 character needles, and the file is my haystack. I am using brute force byte comparison as seen above. The CPU time taken is 7.5 seconds, and the speed of comparisons is 9.7 billion per second. Take that CPU! (which can only do 333ml/sec in single threaded fashion, almost feels like cheating, but oh well).

-Jack

PS. each file text line, and each needle string is loaded into each thread, and executed without any loops whatsoever.

Making sense out of your post is like searching for a needle in a haystack.

“CPU Time is 7.5 seconds” == Time measured using CPU cycles for a GPU execution == 7.5 seconds ?

Making sense out of your post is like searching for a needle in a haystack.

“CPU Time is 7.5 seconds” == Time measured using CPU cycles for a GPU execution == 7.5 seconds ?

Sorry that meant to say GPU (Time Taken: 7456.942871, this is basically from profiler report).

Yes, its a needle in a haystack problem. I have many needles, and huge haystack, so I map both arrays of needles and haystack into a thread, and each thread simply searches the present of needles in each line, then reports the match of how many needles found.

-Jack

Sorry that meant to say GPU (Time Taken: 7456.942871, this is basically from profiler report).

Yes, its a needle in a haystack problem. I have many needles, and huge haystack, so I map both arrays of needles and haystack into a thread, and each thread simply searches the present of needles in each line, then reports the match of how many needles found.

-Jack