Coming up often and just noted that again in Ger's answer in the thread 
above.

What is the meaning of percentages higher than 100? I don't get it.

If my bottle of wine is considered 100% full, what would be 200%? Two 
bottles of wine? :-)

-- 
Thomas

Post a reply to this message

Thomas de Groot wrote:

> Coming up often and just noted that again in Ger's answer in the thread
> above.
> 
> What is the meaning of percentages higher than 100? I don't get it.
> 
> If my bottle of wine is considered 100% full, what would be 200%? Two
> bottles of wine? :-)
> 

The answer is pretty simple. 
If 1 (one) core is working on povray that would be 100% load. If 2 cores are 
working on povray that would be 2 x 100%. So, in my case that would be 8 
cores, each working at (or close to) 100%, so 800%.
The percentage load is taken from the "1 core" or "1 proc" view.
-- 

Ger

Post a reply to this message

On 1-9-2015 11:32, Ger wrote:
> Thomas de Groot wrote:
>
>> Coming up often and just noted that again in Ger's answer in the thread
>> above.
>>
>> What is the meaning of percentages higher than 100? I don't get it.
>>
>> If my bottle of wine is considered 100% full, what would be 200%? Two
>> bottles of wine? :-)
>>
>
> The answer is pretty simple.
> If 1 (one) core is working on povray that would be 100% load. If 2 cores are
> working on povray that would be 2 x 100%. So, in my case that would be 8
> cores, each working at (or close to) 100%, so 800%.
> The percentage load is taken from the "1 core" or "1 proc" view.
>

OK. I missed the 8 cores obviously. it makes sense, but

Still, I feel uncomfortable with the notion itself and would rather 
think that povray is rendering at 100%, divided over 8 cores... each 
working at 100%... for me, that does not add up to 800%, only to faster 
rendering, which is why I guess this is a semantic question.

-- 
Thomas

Post a reply to this message

> OK. I missed the 8 cores obviously. it makes sense, but
>
> Still, I feel uncomfortable with the notion itself and would rather
> think that povray is rendering at 100%, divided over 8 cores... each
> working at 100%... for me, that does not add up to 800%, only to faster
> rendering, which is why I guess this is a semantic question.

Yes, typically % is defined as % of some (obvious) maximum. Defining CPU 
load as a % of only 1 core being fully loaded seems a bit 
counter-intuitive. But so long as everyone is aware of this, and they 
know exactly how many cores are in the CPU then it shouldn't be a huge 
problem. If we didn't know there were 8 cores, 800% could mean anything.

Also, in general, when you combine percentages from various bits you 
need to sum the numerators and denominators separately, otherwise you 
get a wrong result.

eg

Bag 1 has 6 apples out of space for 10 : 6/10 = 60% full
Bag 2 has 3 apples out of space for 4  : 3/4  = 75% full

Combined there are 9 apples out of space for 14, so 64% full. NOT 135% full!

Post a reply to this message

On 1-9-2015 15:24, scott wrote:
>> OK. I missed the 8 cores obviously. it makes sense, but
>>
>> Still, I feel uncomfortable with the notion itself and would rather
>> think that povray is rendering at 100%, divided over 8 cores... each
>> working at 100%... for me, that does not add up to 800%, only to faster
>> rendering, which is why I guess this is a semantic question.
>
> Yes, typically % is defined as % of some (obvious) maximum. Defining CPU
> load as a % of only 1 core being fully loaded seems a bit
> counter-intuitive. But so long as everyone is aware of this, and they
> know exactly how many cores are in the CPU then it shouldn't be a huge
> problem. If we didn't know there were 8 cores, 800% could mean anything.
>
> Also, in general, when you combine percentages from various bits you
> need to sum the numerators and denominators separately, otherwise you
> get a wrong result.
>
> eg
>
> Bag 1 has 6 apples out of space for 10 : 6/10 = 60% full
> Bag 2 has 3 apples out of space for 4  : 3/4  = 75% full
>
> Combined there are 9 apples out of space for 14, so 64% full. NOT 135%
> full!
>

Thanks indeed. That sums it up nicely :-)

The case of Ger's which triggered my thread is pretty obvious and 
acceptable in the context. However, the misuse nowadays in all kinds of 
situations seems to be growing and can be pretty annoying sometimes.

-- 
Thomas

Post a reply to this message

> The case of Ger's which triggered my thread is pretty obvious and
> acceptable in the context. However, the misuse nowadays in all kinds of
> situations seems to be growing and can be pretty annoying sometimes.

Yes, whenever an athlete/sportsperson says they "gave it 110%" they 
should be immediately investigated for doping :-)

Post a reply to this message

On 2-9-2015 12:35, scott wrote:
>> The case of Ger's which triggered my thread is pretty obvious and
>> acceptable in the context. However, the misuse nowadays in all kinds of
>> situations seems to be growing and can be pretty annoying sometimes.
>
> Yes, whenever an athlete/sportsperson says they "gave it 110%" they
> should be immediately investigated for doping :-)

That's a fair assumption indeed ;-)

-- 
Thomas

Post a reply to this message

Am 02.09.2015 um 09:15 schrieb Thomas de Groot:

> The case of Ger's which triggered my thread is pretty obvious and
> acceptable in the context. However, the misuse nowadays in all kinds of
> situations seems to be growing and can be pretty annoying sometimes.

I think there are plenty of cases where values >100% are perfectly ok,
such as:

"This year, our budget has increased by 250% [compared to last year]."

"Let's see for how long we can push the engine to 120% [nominal] thrust."

"This amplifier has a gain of 420%."

"This telescope rod can extend to 550% [of its retracted length]."

Post a reply to this message

> "This year, our budget has increased by 250% [compared to last year]."

Reminds me of my last company, where the profit was always hovering 
around zero (and sometimes negative). The column "profit compared to 
last year" was always an amusing one, it could be 1000000% or negative, 
or the year after a negative amount they didn't know what to do so left 
it blank that year :-)

Post a reply to this message

On 2-9-2015 16:21, clipka wrote:
> Am 02.09.2015 um 09:15 schrieb Thomas de Groot:
>
>> The case of Ger's which triggered my thread is pretty obvious and
>> acceptable in the context. However, the misuse nowadays in all kinds of
>> situations seems to be growing and can be pretty annoying sometimes.
>
> I think there are plenty of cases where values >100% are perfectly ok,
> such as:
>
> "This year, our budget has increased by 250% [compared to last year]."
>
> "Let's see for how long we can push the engine to 120% [nominal] thrust."
>
> "This amplifier has a gain of 420%."
>
> "This telescope rod can extend to 550% [of its retracted length]."
>

Yes, I suppose so but more often than not the bracketed portion is 
omitted which results in confusion, at least for me.

-- 
Thomas

Post a reply to this message

On 9/1/2015 8:06 AM, Thomas de Groot wrote:
> If my bottle of wine is considered 100% full, what would be 200%? Two
> bottles of wine? :-)

One full bottle and a lot of spillage. ;-)

-- 

Regards
     Stephen

Post a reply to this message

On 9/2/2015 11:35 AM, scott wrote:
>> The case of Ger's which triggered my thread is pretty obvious and
>> acceptable in the context. However, the misuse nowadays in all kinds of
>> situations seems to be growing and can be pretty annoying sometimes.
>
> Yes, whenever an athlete/sportsperson says they "gave it 110%" they
> should be immediately investigated for doping :-)

I think that the media percentage is different from normal percentage.

-- 

Regards
     Stephen

Post a reply to this message

On 5-9-2015 18:53, Stephen wrote:
> On 9/1/2015 8:06 AM, Thomas de Groot wrote:
>> If my bottle of wine is considered 100% full, what would be 200%? Two
>> bottles of wine? :-)
>
> One full bottle and a lot of spillage. ;-)
>

That was my conclusion too...

[finishes mopping up the mess left on the table]

-- 
Thomas

Post a reply to this message

On 01/09/2015 02:24 PM, scott wrote:
> Yes, typically % is defined as % of some (obvious) maximum. Defining CPU
> load as a % of only 1 core being fully loaded seems a bit
> counter-intuitive. But so long as everyone is aware of this, and they
> know exactly how many cores are in the CPU then it shouldn't be a huge
> problem. If we didn't know there were 8 cores, 800% could mean anything.

The fun part is when you have 4 cores with hyperthreading, and a task 
manages to consume more than 800% CPU...

In other news, the Windows "Process Explorer" tool considers my computer 
to have 8 physical cores, when in fact it has 4 cores with 2x 
hyperthreading. Which means that maximum utilisation is reported as 50%...

Post a reply to this message

On 18-9-2015 19:23, Orchid Win7 v1 wrote:
> The fun part is when you have 4 cores with hyperthreading, and a task
> manages to consume more than 800% CPU...
>
> In other news, the Windows "Process Explorer" tool considers my computer
> to have 8 physical cores, when in fact it has 4 cores with 2x
> hyperthreading. Which means that maximum utilisation is reported as 50%...

Running the marathon with one body / two legs is hyperthreading without 
knowing it... ;-)

-- 
Thomas

Post a reply to this message

Am 19.09.2015 um 09:21 schrieb Thomas de Groot:
> On 18-9-2015 19:23, Orchid Win7 v1 wrote:
>> The fun part is when you have 4 cores with hyperthreading, and a task
>> manages to consume more than 800% CPU...
>>
>> In other news, the Windows "Process Explorer" tool considers my computer
>> to have 8 physical cores, when in fact it has 4 cores with 2x
>> hyperthreading. Which means that maximum utilisation is reported as
>> 50%...
> 
> Running the marathon with one body / two legs is hyperthreading without
> knowing it... ;-)

I /think/ that analogy limps a bit, as we'd say in German ;)

Post a reply to this message

On 20-9-2015 1:26, clipka wrote:
> Am 19.09.2015 um 09:21 schrieb Thomas de Groot:
>> On 18-9-2015 19:23, Orchid Win7 v1 wrote:
>>> The fun part is when you have 4 cores with hyperthreading, and a task
>>> manages to consume more than 800% CPU...
>>>
>>> In other news, the Windows "Process Explorer" tool considers my computer
>>> to have 8 physical cores, when in fact it has 4 cores with 2x
>>> hyperthreading. Which means that maximum utilisation is reported as
>>> 50%...
>>
>> Running the marathon with one body / two legs is hyperthreading without
>> knowing it... ;-)
>
> I /think/ that analogy limps a bit, as we'd say in German ;)
>

Of course, but I couldn't resist.

-- 
Thomas

Post a reply to this message

> In other news, the Windows "Process Explorer" tool considers my computer
> to have 8 physical cores, when in fact it has 4 cores with 2x
> hyperthreading. Which means that maximum utilisation is reported as 50%...

Ditto here, but all CPU-intensive apps I've got manage to get to a 
reported 100% CPU (POV, commerical simulation tools, multi-threaded C# 
apps, etc).

Post a reply to this message

On 21/09/2015 08:00 AM, scott wrote:
>> In other news, the Windows "Process Explorer" tool considers my computer
>> to have 8 physical cores, when in fact it has 4 cores with 2x
>> hyperthreading. Which means that maximum utilisation is reported as
>> 50%...
>
> Ditto here, but all CPU-intensive apps I've got manage to get to a
> reported 100% CPU (POV, commerical simulation tools, multi-threaded C#
> apps, etc).

Are you looking at the Windows Task Manager rather than the System 
Internals Process Explorer?

Post a reply to this message

>>> In other news, the Windows "Process Explorer" tool considers my computer
>>> to have 8 physical cores, when in fact it has 4 cores with 2x
>>> hyperthreading. Which means that maximum utilisation is reported as
>>> 50%...
>>
>> Ditto here, but all CPU-intensive apps I've got manage to get to a
>> reported 100% CPU (POV, commerical simulation tools, multi-threaded C#
>> apps, etc).
>
> Are you looking at the Windows Task Manager rather than the System
> Internals Process Explorer?

Both:

Post a reply to this message

Attachments:
Download 'croppercapture[40].png' (54 KB)

Preview of image 'croppercapture[40].png'

On 22/09/2015 07:41 AM, scott wrote:
>>> Ditto here, but all CPU-intensive apps I've got manage to get to a
>>> reported 100% CPU (POV, commerical simulation tools, multi-threaded C#
>>> apps, etc).
>>
>> Are you looking at the Windows Task Manager rather than the System
>> Internals Process Explorer?
>
> Both:

Interesting. Perhaps you have a newer version of Process Explorer? Or 
maybe there's some setting somewhere to configure this that I haven't 
noticed...

Post a reply to this message

> Interesting. Perhaps you have a newer version of Process Explorer? Or
> maybe there's some setting somewhere to configure this that I haven't
> noticed...

v16.05 here, can't see any relevant setting.

If I tell POV to render with 6 threads I only get 50% usage, but with 12 
threads I get 100% usage.

However the elapsed times for the benchmark are:
6 threads  : 192s
12 threads : 136s

So 50% CPU is actually more like 70% CPU, but I guess it depends on the 
exact algorithm (how nicely it can hyperthread).

Post a reply to this message

Le 22/09/2015 12:09, scott a écrit :
>> Interesting. Perhaps you have a newer version of Process Explorer? Or
>> maybe there's some setting somewhere to configure this that I haven't
>> noticed...
>
> v16.05 here, can't see any relevant setting.
>
> If I tell POV to render with 6 threads I only get 50% usage, but with 12
> threads I get 100% usage.
>
> However the elapsed times for the benchmark are:
> 6 threads  : 192s
> 12 threads : 136s
>
> So 50% CPU is actually more like 70% CPU, but I guess it depends on the
> exact algorithm (how nicely it can hyperthread).
>
 From my personal bench with povray, a HT-core is worth between 20 to 
25% of a real core. (tests done on i7 980X, at nominal speed, with 
benchmark scene)

If the OS is not total crap, the threads are allocated first on distinct 
real cores.

beware, when comparing with time, the performance is 1/time.

your 6 threads has a performance index of 5.20833, about 0.868 per thread
your 12 threads has one of 7.35, about .613 per threads.
(and if you remove the 5.208 from the 7.35, the additional HT-core only 
provides 2.142, that's 0.357 per HT-core, so a HT-core thread is in your 
system about 41% of a real core thread, for povray-binary and your scene).

Post a reply to this message

>  From my personal bench with povray, a HT-core is worth between 20 to
> 25% of a real core. (tests done on i7 980X, at nominal speed, with
> benchmark scene)

This is on Win 7 with the standard POV binary, the benchmark scene and a 
Xeon E5-1650 @ nominal speed.

> beware, when comparing with time, the performance is 1/time.
>
> your 6 threads has a performance index of 5.20833, about 0.868 per thread
> your 12 threads has one of 7.35, about .613 per threads.
> (and if you remove the 5.208 from the 7.35, the additional HT-core only
> provides 2.142, that's 0.357 per HT-core, so a HT-core thread is in your
> system about 41% of a real core thread, for povray-binary and your scene).

Yes, I think the additional HT'd threads being worth 41% of a real core 
thread, is the same thing as saying that if you only use the real cores 
you're utilising about 70% of the total CPU performance available. I 
assume that modern compilers optimise code to be hyperthread nicely?

Post a reply to this message

Am 22.09.2015 um 15:06 schrieb scott:

> Yes, I think the additional HT'd threads being worth 41% of a real core
> thread, is the same thing as saying that if you only use the real cores
> you're utilising about 70% of the total CPU performance available. I
> assume that modern compilers optimise code to be hyperthread nicely?

Most unlikely - there's nothing much to optimize there.

In a hyperthreaded CPU, you actually /do/ have the number of cores that
the task manager claims you have, but they're not independent; instead,
they are organized in pairs that share various resources; for instance,
there might only be one floating-point arithmetics unit for each pair of
cores.

Any thread that needs any such shared resource /will/ compete for it
with any other thread running on the pair's other core. And as you
typically can't optimize away such resource requirements, you can't
optimize away the competition. In addition, different CPU types might
share different resources among core pairs, so you don't even know what
resources to try to avoid.

There might be ways for an application developer to optimize an
application for hyperthreading; the strategy would be to separate the
application's workload into two groups of threads that use different
sets of CPU resources, and coax the OS to run one thread from each group
in each pair of HT cores; this way, competition for resources will be
reduced. However, it might not be easy to partition the workload in such
a manner.

Post a reply to this message

On 22/09/2015 11:09 AM, scott wrote:
>> Interesting. Perhaps you have a newer version of Process Explorer? Or
>> maybe there's some setting somewhere to configure this that I haven't
>> noticed...
>
> v16.05 here, can't see any relevant setting.
>
> If I tell POV to render with 6 threads I only get 50% usage, but with 12
> threads I get 100% usage.
>
> However the elapsed times for the benchmark are:
> 6 threads : 192s
> 12 threads : 136s
>
> So 50% CPU is actually more like 70% CPU, but I guess it depends on the
> exact algorithm (how nicely it can hyperthread).

Oh, I see. I'm still running the single-threaded POV-Ray binary. So if 
it uses 1 core, it says 12.5% CPU. I thought you were saying your copy 
says 100% for each CPU core used. ;-)

Post a reply to this message

> Oh, I see. I'm still running the single-threaded POV-Ray binary.

Now why doesn't that surprise me? :-)

Post a reply to this message

On 9/23/2015 7:36 AM, scott wrote:
>> Oh, I see. I'm still running the single-threaded POV-Ray binary.
>
> Now why doesn't that surprise me? :-)
>
>
It surprised me. Ver 3.7 is shiny.


-- 

Regards
     Stephen

Post a reply to this message

Am 22.09.2015 um 22:58 schrieb Orchid Win7 v1:

> Oh, I see. I'm still running the single-threaded POV-Ray binary. So if
> it uses 1 core, it says 12.5% CPU. I thought you were saying your copy
> says 100% for each CPU core used. ;-)

Okay, now is officially the time for you to upgrade to POV-Ray 3.7. If
you refuse to, I /will/ consider that as a personal insult.

I've put a lot of effort into that version to make it awesome. You don't
need to try UberPOV (which is even more awesome), but you absolutely
/must/ install version 3.7. Now. No excuses.


(And yes, I /am/ at least semi-serious about the personal insult thing.)

Post a reply to this message

On 23/09/2015 08:04 AM, Stephen wrote:
> On 9/23/2015 7:36 AM, scott wrote:
>>> Oh, I see. I'm still running the single-threaded POV-Ray binary.
>>
>> Now why doesn't that surprise me? :-)
>>
>>
> It surprised me. Ver 3.7 is shiny.

I remember it being in beta for a long time, and having lots of 
interesting-sounding stuff. I wasn't aware it had actually made it to 
release version yet.

Then again, when was the last time I did much with POV-Ray? Sure, it's 
running 24x7 working on an animation, but I wrote that years ago. I 
haven't really "used" it much recently...

Post a reply to this message

On 9/23/2015 5:56 PM, Orchid Win7 v1 wrote:
> On 23/09/2015 08:04 AM, Stephen wrote:
>> On 9/23/2015 7:36 AM, scott wrote:
>>>> Oh, I see. I'm still running the single-threaded POV-Ray binary.
>>>
>>> Now why doesn't that surprise me? :-)
>>>
>>>
>> It surprised me. Ver 3.7 is shiny.
>
> I remember it being in beta for a long time, and having lots of
> interesting-sounding stuff. I wasn't aware it had actually made it to
> release version yet.
>
> Then again, when was the last time I did much with POV-Ray? Sure, it's
> running 24x7 working on an animation, but I wrote that years ago. I
> haven't really "used" it much recently...

An animation! I would get the latest version just for the speed up.

You've mentioned this animation before. Refresh my memory, if you would?

-- 

Regards
     Stephen

Post a reply to this message

On 23/09/2015 06:03 PM, Stephen wrote:
> On 9/23/2015 5:56 PM, Orchid Win7 v1 wrote:
>> Then again, when was the last time I did much with POV-Ray? Sure, it's
>> running 24x7 working on an animation, but I wrote that years ago. I
>> haven't really "used" it much recently...
>
> An animation! I would get the latest version just for the speed up.
>
> You've mentioned this animation before. Refresh my memory, if you would?

https://www.youtube.com/watch?v=8s7I80H96CM

That took about 2 years to render. I'm currently rendering the sequel, 
which is a similar thing inside a Cornell box.

Post a reply to this message