On 14/11/2014 13:06, Thomas de Groot wrote:
> On 14-11-2014 12:59, Le_Forgeron wrote:
>> Le 14/11/2014 00:36, Cousin Ricky a écrit :
>>> Le_Forgeron <jgr### [at] freefr> wrote:
>>>> On 13/11/2014 21:21, Cousin Ricky wrote:
>>>>> My hardware is a Dell Inspiron 17R with Intel Core i7 (8 cores).
>>>>
>>>> Isn't that 4 cores with HT ? (for povray, it would be worth 5 cores,
>>>> YMMV)
>>>
>>> I've seen "quad core" in literature not attached to this particular
>>> model, but
>>> in the POV-Ray statistics, there is a clear 8-to-1 ratio between CPU
>>> times and
>>> elapsed times.  Does this mean that there are 2 CPUs in each core?
>>>
>>>
>> Sort of... there is two full set of registers in each core, allowing to
>> run two code in parallel... as long as they do not need the same
>> resources from the core.
>>
>> at 100% cpu like povray, yes there is a 8:1 ratio between cpu times and
>> wall-clock time. Yet, if the scheduler of the operating system is not
>> too bad, if you force povray to run on less threads (with -WT<number>
>> option), you can get some surprise like -WT8 is not twice faster that
>> -WT4.
>>
>> With good schedulers, povray -WT4 would run one thread per core, leaving
>> the other set of registers idle or used by something else.
>>
>>
> Talking of which, is there a difference between -WT and +WT? I get the
> impression that both are used indiscriminately.
> 
> Thomas

Actually, for some option there is a difference between + and -
(usually one form disable the option, and the other enable it, with some
value).

For instance +KC and -KC are opposite of each other.

But +H and -H are happily setting both the height of the output.

Now, if I could find a mnemonic to remember which one is enable and
which one is disable... that would be a great year.



-- 
IQ of crossposters with FU: 100 / (number of groups)
IQ of crossposters without FU: 100 / (1 + number of groups)
IQ of multiposters: 100 / ( (number of groups) * (number of groups))

Post a reply to this message