Bioequivalence and Bioavailability Forum • Unrolled example

Unrolled example [Two-Stage / GS Designs]

posted by Helmut – Vienna, Austria, 2015-02-21 02:50 (3710 d 10:33 ago) – Posting: # 14483
Views: 12,410

Hi ElMaestro,

❝ I get about a 10% speedup difference if I unroll 4x:

Confirmed. ;-)

;-)

❝ 10% isn't bad, actually. One could probably even optimise the number of unrolls.

Yeah, but at the end of the day it’s not necessarily the loop-overhead that counts. Such a code is terrible to maintain.

PS: In your unrolled code set runs <- 5 and after execution type length(v2). Kinetica-style.

—
Dif-tor heh smusma 🖖🏼 Довге життя Україна!
Helmut Schütz

[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes

Complete thread:

Gimmick & survey Helmut 2015-02-20 01:47
- Gimmick & survey nobody 2015-02-20 08:25
  - Gimmick & survey Helmut 2015-02-20 13:01
    - Gimmick & survey nobody 2015-02-20 14:18
      - Speeed! Helmut 2015-02-20 15:38
        
        Speeed! nobody 2015-02-20 16:37
        
        Speeed! Helmut 2015-02-20 16:58
        
        Speeed! nobody 2015-02-20 17:01
        
        Speeed! ElMaestro 2015-02-20 23:48
        
        Speeed! Helmut 2015-02-21 01:11
        
        Unrolled example ElMaestro 2015-02-21 01:32
        
        Unrolled exampleHelmut 2015-02-21 01:50
        
        Unrolled example ElMaestro 2015-02-21 02:27
        
        Increasing vector & another example Helmut 2015-02-21 13:13
        
        Speeed! nobody 2015-02-21 21:02

Admin contact
23,424 posts in 4,927 threads, 1,699 registered users;
25 visitors (0 registered, 25 guests [including 7 identified bots]).
Forum time: 14:23 CEST (Europe/Vienna)

Do not put your faith in what statistics say until you have carefully
considered what they do not say. William W. Watt

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC

Ing. Helmut Schütz