1 |
On Sun, 26 Jun 2011 13:30:41 +0200 |
2 |
Volker Armin Hemmann <volkerarmin@××××××××××.com> wrote: |
3 |
|
4 |
> Believe it or not, this tarball is straight from hell. |
5 |
> |
6 |
> 1. do not tar-bomb. Pack directories not files. |
7 |
|
8 |
This one was my fault, but I didn't want to complicate the instructions |
9 |
for those who may not have much background in these things. That is |
10 |
why I specified to first create a directory and then to unpack within |
11 |
that directory. |
12 |
|
13 |
> 2. the permissions are a nightmare. |
14 |
> 3. I accidentally unpacked in /tmp and your tarball tried to set UTIME and the |
15 |
> permissions of the parent directory. WTF? |
16 |
|
17 |
These are the fault of the original programmers. Such messes were common |
18 |
back in 1995. |
19 |
|
20 |
> 4. have you checked that the errors you are seeing are really errors? |
21 |
> Just remember that stupidity about 80bit/64bit/128bit floats all around. |
22 |
> 5. http://gcc.gnu.org/wiki/FloatingPointMath |
23 |
|
24 |
You forgot to mention decimal floating point which is yet another |
25 |
variant. |
26 |
|
27 |
This particular test is configured to test double precision (64-bit) in |
28 |
C only which seems to be the standard on Linux. On Intel Linux at least, |
29 |
64-bit FP is done entirely in hardware. |
30 |
|
31 |
The IEEE 754 standard was set way back in 1985. Already it may becoming |
32 |
obsolete (just like IPv4). Hardware is getting more capable and can |
33 |
allow for greater precisions. Also, we are now seeing hardware for |
34 |
decimal floating point which will be a boon to financial software. |
35 |
|
36 |
It should also be noted that the emerging field of computation |
37 |
with GPUs does not have any floating point standards at all. |
38 |
With GPU programming, floating point operations are similar |
39 |
to the chaos that ruled in the CPU world before IEEE 754. |
40 |
|
41 |
> |
42 |
> without additional settings gcc 4.5.2: |
43 |
> |
44 |
|
45 |
> |
46 |
> /tmp/usbtest/clib_DP.output: ucbtest UCBFAIL in cabsd at line 701 for double |
47 |
> |
48 |
|
49 |
Thanks for this! It does seem to verify that gcc-4.5.2 does mangle the code |
50 |
for this particular test. |
51 |
|
52 |
Frank Peters |