[openssl-users] OpenSSL 1.0.2h reports speed test results as 0 secs and Infk ops/sec

[Jeff Wieland]

Dennis Clarke wrote:
> On 09/11/2016 03:44 PM, Jeff Wieland wrote:
>> I see the same thing on Sun Blade 150 (650Mhz), with OpenSSL 1.0.2h
>> compiled with Studio 12.2 -- and with a Sun Fire V100 (550Mhz).
>>
>> It works correctly on a Sun Fire V240 (1.5Ghz), a Sun Ultra 10 (440Mhz),
>> a Sun Fire T1000, and Sun Enterprise M3000.
>>
>> I see these results with both 32 bit and 64 bit builds.
>>
>> It looks like you're building and running this on an UltraSPARC 2e
>> architecture system -- this is what the SB150 and the V100 are.
>
> Hrmmm .. not sure what that means. Given that I run Configure with the
> "no-asm" option then I would think that cross platform consistency
> would happen. I would have to go diving into the source to find out
> where the timings are happening. On Solaris the best timer is the
> gethrtime() function as it works down to the nanosec and is accurate to
> the millisec at least.
>
> Dennis
>
>
I do build with the no-asm option, and I'm seeing the problem.

I'm suspicious that somehow the compiler optimization is generating code 
that doesn't work
quite right on the UltraSPARC 2e.

--
Jeff Wieland, UNIX/Network Systems Administrator
Purdue University IT Infrastructure Services UNIX Platforms