performance measurements

Each table row shows performance measurements for this Clojure program with a particular command-line input value N.

 N  CPU secs Elapsed secs Memory KB Code B ≈ CPU Load
250,0004.444.45111,5802793  1% 0% 0% 100%
2,500,00013.4213.44413,7442793  0% 1% 1% 100%
25,000,00082.2382.331,002,6522793  1% 0% 0% 100%

Read the ↓ make, command line, and program output logs to see how this program was run.

Read k-nucleotide benchmark to see what this program should do.


java version "1.8.0_45"
Java(TM) SE Runtime Environment (build 1.8.0_45-b14)
Java HotSpot(TM) Server VM (build 25.45-b02, mixed mode)

Clojure 1.7.0

 k-nucleotide Clojure #6 program source code

;;   The Computer Language Benchmarks Game

;; contributed by Andy Fingerhut
;; modified by Marko Kocic
;; modified by Mike Anderson to make better use of primitive operations

(ns knucleotide

(set! *warn-on-reflection* true)

;; Handle slight difference in function name between Clojure 1.2.0 and

;; 1.3.0-alpha* ability to use type hints to infer fast bit

;; operations.

(defmacro my-unchecked-inc-int [& args]
  (if (and (== (*clojure-version* :major) 1)
           (== (*clojure-version* :minor) 2))
    `(unchecked-inc ~@args)
    `(unchecked-inc-int ~@args)))

(defmacro key-type [num]
  (if (and (== (*clojure-version* :major) 1)
           (== (*clojure-version* :minor) 2))
    `(long ~num)))

(definterface ITallyCounter
  (^int get_count [])
  (inc_BANG_ []))

(deftype TallyCounter [^{:unsynchronized-mutable true :tag int} cnt]
  (get-count [this] cnt)
  (inc! [this]
    (set! cnt (my-unchecked-inc-int cnt))))


;; Return true when the line l is a FASTA description line

(defn fasta-description-line [l]
  (= \> (first (seq l))))

;; Return true when the line l is a FASTA description line that begins

;; with the string desc-str.

(defn fasta-description-line-beginning [desc-str l]
  (and (fasta-description-line l)
       (= desc-str (subs l 1 (min (count l) (inc (count desc-str)))))))

;; Take a sequence of lines from a FASTA format file, and a string

;; desc-str.  Look for a FASTA record with a description that begins

;; with desc-str, and if one is found, return its DNA sequence as a

;; single (potentially quite long) string.  If input file is big,

;; you'll save lots of memory if you call this function in a with-open

;; for the file, and don't hold on to the head of the lines parameter.

(defn fasta-dna-str-with-desc-beginning [desc-str lines]
  (when-let [x (drop-while
		(fn [l] (not (fasta-description-line-beginning desc-str l)))
    (when-let [x (seq x)]
      (let [y (take-while (fn [l] (not (fasta-description-line l)))
                          (map (fn [#^java.lang.String s] (.toUpperCase s))
                               (rest x)))]
        (apply str y)))))

(def dna-char-to-code-val-map {\A 0, \C 1, \T 2, \G 3})
(def code-val-to-dna-char {0 \A, 1 \C, 2 \T, 3 \G})

(defmacro dna-char-to-code-val [ch]
  `(case ~ch
     ~@(flatten (seq dna-char-to-code-val-map))))

;; In the hash map 'tally' in tally-dna-subs-with-len, it is more

;; straightforward to use a Clojure string (same as a Java string) as

;; the key, but such a key is significantly bigger than it needs to

;; be, increasing memory and time required to hash the value.  By

;; converting a string of A, C, T, and G characters down to an integer

;; that contains only 2 bits for each character, we make a value that

;; is significantly smaller and faster to use as a key in the map.

;;    most                 least

;; significant          significant

;; bits of int          bits of int

;;  |                         |

;;  V                         V

;; code code code ....  code code

;;  ^                         ^

;;  |                         |

;; code for               code for

;; *latest*               *earliest*

;; char in                char in

;; sequence               sequence

;; Note: Given Clojure 1.2's implementation of bit-shift-left/right

;; operations, when the value being shifted is larger than a 32-bit

;; int, they are faster when the shift amount is a compile time

;; constant.

(defn ^:static dna-str-to-key 
  (^long [^String s] (dna-str-to-key s 0 (count s)))
  (^long [^String s ^long start ^long length]
  ;; Accessing a local let binding is much faster than accessing a var

    (loop [key (long 0)
	         offset (int (+ start length -1))]
      (if (< offset start)
				(let [c (.charAt s offset)
			        code (int (dna-char-to-code-val c))
				      new-key (+ (bit-shift-left key 2) code)]
				  (recur new-key (dec offset)))))))

(defn key-to-dna-str [k len]
  (apply str (map code-val-to-dna-char
		  (map (fn [pos] (bit-and 3 (bit-shift-right k pos)))
		       (range 0 (* 2 len) 2)))))

;; required function : "to update a hashtable of k-nucleotide keys and count values, for a particular reading-frame"
(defn tally-dna-subs-with-len [len dna-str]
  (let [len (int len)
        dna-str ^String dna-str
        mask-width (* 2 len)
	      mask (key-type (dec (bit-shift-left 1 mask-width)))]
     (loop [offset (int (- (count dna-str) len))
            key (key-type (dna-str-to-key dna-str offset len))
            tally (let [h (java.util.HashMap.)
                        one (TallyCounter. (int 1))]
                   (.put h key one)
      (if (<= offset 0)
				(let [new-offset (unchecked-dec offset)
				      new-first-char-code (dna-char-to-code-val
			                                   (.charAt dna-str new-offset))
				      new-key (key-type (bit-and mask (unchecked-add (bit-shift-left key 2)
		          (if-let [^TallyCounter cur-count (get tally new-key)]
		            (.inc! cur-count)
		            (let [one (TallyCounter. (int 1))]
		              (.put tally new-key one)))
            (recur new-offset new-key tally))))))

(defn ^:static getcnt ^long [^TallyCounter tc]
  (.get-count tc))

(defn ^:static tally-total [tally]
  (loop [acc (long 0)
         s (vals tally)]
    (if-let [v (first s)]
      (recur (+ acc (getcnt v)) (next s))

(defn all-tally-to-str [tally fn-key-to-str]
    (let [total (tally-total tally)
          cmp-keys (fn [k1 k2]
                     ;; Return negative integer if k1 should come earlier

                     ;; in the sort order than k2, 0 if they are equal,

                     ;; otherwise a positive integer.

                     (let [cnt1 (int (getcnt (get tally k1)))
                           cnt2 (int (getcnt (get tally k2)))]
                       (if (not= cnt1 cnt2)
                         (- cnt2 cnt1)
                         (let [^String s1 (fn-key-to-str k1)
                               ^String s2 (fn-key-to-str k2)]
                           (.compareTo s1 s2)))))]
      (doseq [k (sort cmp-keys (keys tally))]
        (printf "%s %.3f\n" (fn-key-to-str k)
                (double (* 100 (/ (getcnt (get tally k)) total))))))))

(defn one-tally-to-str [dna-str tally]
  (let [zerotc (TallyCounter. 0)]
    (format "%d\t%s" (getcnt (get tally (dna-str-to-key dna-str) zerotc))

(defn compute-one-part [dna-str part]
   (condp = part
       0 (all-tally-to-str (tally-dna-subs-with-len 1 dna-str)
                           (fn [k] (key-to-dna-str k 1)))
       1 (all-tally-to-str (tally-dna-subs-with-len 2 dna-str)
                           (fn [k] (key-to-dna-str k 2)))
       2 (one-tally-to-str "GGT"
                           (tally-dna-subs-with-len 3 dna-str))
       3 (one-tally-to-str "GGTA"
                           (tally-dna-subs-with-len 4 dna-str))
       4 (one-tally-to-str "GGTATT"
                           (tally-dna-subs-with-len 6 dna-str))
       5 (one-tally-to-str "GGTATTTTAATT"
                           (tally-dna-subs-with-len 12 dna-str))
       6 (one-tally-to-str "GGTATTTTAATTTATAGT"
                           (tally-dna-subs-with-len 18 dna-str)))])

(defn run 
    (run ( ( "knucleotide-input.txt"))))
    (let [dna-str (fasta-dna-str-with-desc-beginning "THREE" (line-seq br))
          ;; Select the order of computing parts such that it is

          ;; unlikely that parts 5 and 6 will be computed concurrently.

          ;; Those are the two that take the most memory.  It would be

          ;; nice if we could specify a DAG for which jobs should finish

          ;; before others begin -- then we could prevent those two

          ;; parts from running simultaneously.

          results (map second
                       (sort #(< (first %1) (first %2))
                              #(compute-one-part dna-str %)
                              '(0 5 6 1 2 3 4)
      (doseq [r results]
        (println r)

(defn -main [& args]
  (with-open [br ( *in*)]
    (run br))  
  (System/exit 0))

 make, command-line, and program output logs

Wed, 01 Jul 2015 00:54:59 GMT

mv knucleotide.clojure-6.clojure knucleotide.clj
/usr/local/src/jdk1.8.0_45/bin/java -Dclojure.compile.path=. -cp .:/usr/local/src/clojure/clojure-1.7.0.jar: clojure.lang.Compile knucleotide
Picked up JAVA_TOOL_OPTIONS: -javaagent:/usr/share/java/jayatanaag.jar 
Compiling knucleotide to .
2.62s to complete and log all make actions

/usr/local/src/jdk1.8.0_45/bin/java -server -XX:+TieredCompilation -XX:+AggressiveOpts -Xmx1024m -cp .:/usr/local/src/clojure/clojure-1.7.0.jar: knucleotide 0 < knucleotide-input25000000.txt

A 30.295
T 30.151
C 19.800
G 19.754

AA 9.177
TA 9.132
AT 9.131
TT 9.091
CA 6.002
AC 6.001
AG 5.987
GA 5.984
CT 5.971
TC 5.971
GT 5.957
TG 5.956
CC 3.917
GC 3.911
CG 3.909
GG 3.902

1471758	GGT
446535	GGTA
47336	GGTATT

Picked up JAVA_TOOL_OPTIONS: -javaagent:/usr/share/java/jayatanaag.jar 

Revised BSD license

  Home   Conclusions   License   Play