Great illusions under heavy fire ...

D:\_KAZE_new-stuff\Leprechaun_quadrupleton_r14_minus>Leprechaun_quadrupleton.exe GRAFFITH_2048.lst GRAFFITH_2048.wrd 98000000 z
Leprechaun(Fast Greedy Word-Ripper), rev. 14_minus_quadrupleton, written by Svalqyatchx.
Leprechaun: 'Oh, well, didn't you hear? Bigger is good, but jumbo is dear.'
Kaze: Let's see what a 3-way hash + 6,602,752 Binary-Search-Trees can give us,
      also the performance of a 3-way hash + 6,602,752 B-Trees of order 3,
      also the performance of a 1-way hash + 134,217,728 external B-Trees of order 3.
Size of input file with files for Leprechauning: 42140
Allocating HASH memory 1,073,741,889 bytes ... OK
Allocating/ZEROing 100,352,000,014 bytes swap file ... OK
Size of Input TEXTual file: 33,470,581
|; Word count: 3,045,077 of them 2,597,942 distinct; Done: 64/64
...

[Manual log for Toshiba Satellite 2166MHz Merom Windows XP 32bit:]
Day#1: June 02, 2011, 8:37:16 AM - START
Day#2: June 03, 2011, 11:13:50 PM - Stopwatch

Formula:
(Day#2-Day#1)*24*3600 - (Military time#1) + (Military time#2) = elapsed seconds

Stopwatch #1:
June 02, 2011, 8:37:16
June 03, 2011, 23:13:50
(3-2)*24*3600 - (8*3600 + 37*60 + 16) + (23*3600 + 13*60 + 50) = 138994
; Word count: 69,468,160 of them 47,277,037 distinct; Done: 1/64
69,468,160/138,994 = 499 4-grams per second

Stopwatch #2:
June 02, 2011, 8:37:16
June 04, 2011, 3:39:38
(4-2)*24*3600 - (8*3600 + 37*60 + 16) + (3*3600 + 39*60 + 38) = 154942
/; Word count: 73,400,320 of them 48,850,248 distinct; Done: 54/64
73,400,320/154,942 = 473 4-grams per second

Stopwatch #3:
June 02, 2011, 8:37:16
June 04, 2011, 23:32:38
(4-2)*24*3600 - (8*3600 + 37*60 + 16) + (23*3600 + 32*60 + 38) = 226522
|; Word count: 86,769,664 of them 57,423,134 distinct; Done: 42/64
86,769,664/226,522 = 383 4-grams per second

Stopwatch #4:
June 02, 2011, 8:37:16
June 05, 2011, 23:07:38
(5-2)*24*3600 - (8*3600 + 37*60 + 16) + (23*3600 + 7*60 + 38) = 311422
\; Word count: 102,236,160 of them 63,549,665 distinct; Done: 53/64
102,236,160/311,422 = 328 4-grams per second

Stopwatch #5:
June 02, 2011, 8:37:16
June 07, 2011, 4:22:38
(7-2)*24*3600 - (8*3600 + 37*60 + 16) + (4*3600 + 22*60 + 38) = 416722
-; Word count: 123,469,824 of them 71,368,602 distinct; Done: 3/64
123,469,824/416,722 = 296 4-grams per second

Stopwatch #6:
June 02, 2011, 8:37:16
June 08, 2011, 2:03:38
(8-2)*24*3600 - (8*3600 + 37*60 + 16) + (2*3600 + 3*60 + 38) = 494782
-; Word count: 137,625,600 of them 76,907,298 distinct; Done: 43/64
137,625,600/494,782 = 278 4-grams per second

Stopwatch #7:
June 02, 2011, 8:37:16
June 09, 2011, 3:15:38
(9-2)*24*3600 - (8*3600 + 37*60 + 16) + (3*3600 + 15*60 + 38) = 585502
\; Word count: 152,567,808 of them 82,252,842 distinct; Done: 12/64
152,567,808/585,502 = 260 4-grams per second

Stopwatch #8:
June 02, 2011, 8:37:16
June 10, 2011, 1:56:38
(10-2)*24*3600 - (8*3600 + 37*60 + 16) + (1*3600 + 56*60 + 38) = 667162
\; Word count: 165,412,864 of them 88,555,337 distinct; Done: 9/64
165,412,864/667,162 = 247 4-grams per second

Stopwatch #9:
June 02, 2011, 8:37:16
June 11, 2011, 1:35:38
(11-2)*24*3600 - (8*3600 + 37*60 + 16) + (1*3600 + 35*60 + 38) = 752302
|; Word count: 177,995,776 of them 95,007,353 distinct; Done: 27/64
177,995,776/752,302 = 236 4-grams per second

Stopwatch #A:
June 02, 2011, 8:37:16
June 11, 2011, 23:11:38
(11-2)*24*3600 - (8*3600 + 37*60 + 16) + (23*3600 + 11*60 + 38) = 830062
|; Word count: 187,957,248 of them 100,213,903 distinct; Done: 13/64
187,957,248/830,062 = 226 4-grams per second

['Hard Disk Sentinel' Disk Information:]
Hard Disk Model ID,SAMSUNG HM640JJ
Native Command Queuing (NCQ),Supported
Queue Length,32
Disk Family,SpinPoint MP4
Capacity,"640 GB (640 x 1,000,000,000 bytes)"
Number Of Disks,2
Number Of Heads,4
Rotational Speed,7200 RPM
Rotation Time,8.33 ms
Average Rotational Latency,4.17 ms
Buffer Size,16384 KB
Average Seek Time,11.0 ms

[A dummy calculation:]
With seek time of 1/100 of a second (or 10 miliseconds) the number of accesses to the root of the trees are 100+.
On SSD drives with seek time of 1/10,000 of a second (or 100 microseconds) I expect 10,000+ accesses!?
On DDR2 with seek time of 1/10,000,000 of a second (or 100 nanoseconds) I expect 10,000,000+ accesses, in fact ten times slower - the tests show 1,000,000 4-grams per second.
So the dummy prediction:
HDD allows 100 4-grams per second.
SSD allows 10,000 4-grams per second.
DDR2 allows 1,000,000 4-grams per second.
Or the bottomline: HDD 100 times slower than SSD 100 times slower than DDR2!?

[The final conclusion:]
More tests needed on different drives (including Virtual RAM drives, PCI RAM drives, SSD) ...