Showing entries tagged [ipcc]

Does (Cache) Size Matter, Continued...

Now for this file, I get results that make sense:

bash-3.2$ ./tst_ar4 -h pr_A1_50_16_64.nc 
cs[0] cs[1]  cs[2] cache(MB)  deflate  shuffle  read_hor(us)  read_time_ser(us)
50    16    64    4        1    0        157932        155803
50    16     64    16         1    0        3473         155237
50    16    64    32         1    0        3479        146510
50    16    64    64        1    0        3487        120306
50    16     64    128        1    0        3499         64149

Now the best performance comes from the largest cache.

Does (Cache) Size Matter?

Some cache size tests for netcdf-4 and ar4 data.

Oddly, increasing the cache here seems to hurt:

./tst_ar4 -h pr_A1_256_128_128.nc
cs[0]  cs[1]  cs[2]  cache(MB) deflate  shuffle  read_hor(us)  read_time_ser(us)
256    128    128    4        0    0        218         1611
256    128    128    16        0    0        9352        34872
256    128    128    32        0    0        134         32464
256    128    128    64        0    0        133        32303
256    128    128    128       0    0        146        12202

The best read time for the time series is a 4 MB chunk cache. Why?

Unidata Developer's Blog
A weblog about software development by Unidata developers*
Unidata Developer's Blog
A weblog about software development by Unidata developers*

Welcome

FAQs

News@Unidata blog

Take a poll!

What if we had an ongoing user poll in here?

Browse By Topic
Browse by Topic
« July 2024
SunMonTueWedThuFriSat
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
   
       
Today