K6-II+, 20010418 CVS tree passed all testsuites

[Byeong-ryeol Kim]

On Fri, 20 Apr 2001, Stephen L Moshier wrote:

>
> > This time, I tested 20010418 snapshot of CVS tree(2.2.3)
>
> The AMD K6 here also does a successful make check using this morning's
> CVS sources and default configuration settings.  Since no bug was
> found, maybe the pre3 tar file was broken.
>

I tested 20010420 snapshot of CVS tree.
I see very strange thing.
Though 'make check' was ok, compiled 'test-ifloat' emits folowing erros
while being executed under glibc-20010418 snapshot. It sometimes emits errors
and some times doesn't. The ratio of fail to success is approximately 50%.
Environements are the almost same as before. Now
kernel-2.4.3-ac10(compile with 'CONFIG_MK6=y') and glibc-20010418 runs on my
system.
If there were no hardware problems on my system(like deficiency of K6 or
K6-II+ CPU), this might show problem of test-ifloat code, or Might this
problem related with kernel version?

1) $ ./test-ifloat

testing float (inline functions)

Test suite completed:
  2410 test cases plus 2208 tests for exception flags executed.
  All tests passed successfully.
testing float (inline functions)
Failure: Test: jn (0, 8.0) == 0.17165080713755390609
Result:
 is:          1.19443722069263458252e-01   0x1.e93dd200000000000000p-4
 should be:   1.71650812029838562012e-01   0x1.5f8a7600000000000000p-3
 difference:  5.22070899605751037598e-02   0x1.abae3400000000000000p-5
 ulp       :  3503558.5000
 max.ulp   :  1.0000
Maximal error of `jn'
 is      :  3503558.5000 ulp
 accepted:  2.0000 ulp


2) $ ./test-ifloat

Test suite completed:
  2410 test cases plus 2208 tests for exception flags executed.
  All tests passed successfully.

-- 
 "Where there is a will, there is a way."  jinbo21@hananet.net
  For the future of you and me!            hitel: jinbo21