[parsec-users] Seems VIPS results differs in 1 thread and 2 threads?

Yungang Bao ybao at CS.Princeton.EDU
Mon Jul 16 03:07:10 EDT 2012


Hi Zhunping,

I compared the results of the 1-thread run and 2-threa run and saw the difference too.

I further compared the output files in hex and found actually the difference is pretty tiny, only three bytes for 44MB. Yet, I have no idea why this happens. But I think this tiny difference should not affect the program's behavior significantly.  

-------------------------------------------
[~]$ cat run-1/output.v | od -x > run-1.output

[~]$ cat run-2/output.v | od -x > run-2.output

[~]$ diff run-1.output run-2.output > diff.log

[~]$ cat diff.log

2743305,2743306c2743305,2743306
< 247433620 6f4d 206e 754a 206c 3631 3020 3a32 3931
< 247433640 333a 2039 3032 3231 3c0a 662f 6569 646c
---
> 247433620 6f4d 206e 754a 206c 3631 3020 3a32 3831
> 247433640 343a 2037 3032 3231 3c0a 662f 6569 646c
-------------------------------------------


Yungang

----- Original Message -----
From: "Zhunping Zhang" <zhunping at csail.mit.edu>
To: "PARSEC Users" <parsec-users at lists.cs.princeton.edu>
Sent: Monday, July 16, 2012 1:16:14 PM
Subject: [parsec-users] Seems VIPS results differs in 1 thread and 2 threads?

Hi, its weird that I got different output from VIPs using 1 thread and 2 threads, as follows:

 > export IM_CONCURRENCY=1
 > vips im_benchmark ../data/bigben_2662x5500.v output1.v
...
> export IM_CONCURRENCY=2
> vips im_benchmark ../data/bigben_2662x5500.v output2.v
> diff output1.v output2.v 
Binary files output1.v and output2.v differ

any ideas?

cheers,
justin

_______________________________________________
parsec-users mailing list
parsec-users at lists.cs.princeton.edu
https://lists.cs.princeton.edu/mailman/listinfo/parsec-users


More information about the parsec-users mailing list