Confusing as to why Multicore isnt much better?

[FatRakoon]

Doing a load more AVI to DVD Convertions ( Yes, legally )

Assigned my Intel 9550 to the task.

Running Convert X to DVD v4 to do it. ( Only just upped from v3 )

Now, I have a load to do, and I normally load up ConvertX 4 times, go into the TaskManager and set the affinity to one core each, and click on go for all 4 and then, once done, I get 4 more folders in my destination.

Sounds ok right?

Anyway, I noticed one thing when upgrading to ConvertX v4 and that is that it can set the number of Cores.

Now, I had a quick look at the speeds that it does the convertion and I am a little confused.

Firstly, if I run one instance of the program and convert one AVI File, using only one core, I get roughly around 2.8x convertion.

However, if I do the same AVI and set it to 4 core, its only converting at 5.3 ?

How is that possible?

Now, I take 4 instances of ConvertXtoDVD, and in each instance, I set the affinity to one... ( One different one for each core clearly ) and I convert 4 instances all at the same time and guess what? - Each instance is convering at about 2.7 or 2.8

So, if it can convert 4 and about 2.8 Each... Why is is unable to convert more than 5... Surely I should be able to convert a single file at over 10 instead of just 5 ?

Thanks.

 

[FatRakoon]

It wont have anything to do with the HD.

I use multiple HDs and I convert the AVIs from E: to D:

E: and D: are physically on different controlers to help speed thigns up a little.

Also When converting 4 instances of AVIs into 4 different DVD VOBs, I am doing them roughly around 2.6x each and thats over 10x but then when doing only the one AVI to DVD but using 4 Cores it does it at half of that...

So, impossible for it to be the HD for me.

 

[FatRakoon]

It won't be the hard drive, as it would also affect the 4 single threads to a greater amount.

OFFS! Im late again!

 

[FatRakoon]

+3

Shame.

I mean, dont get me wrong... I am happy that it does 4 at a time at a great speed cos we all put our footage together and I Cut & Paste the stuff together so doing 4 @ 2.6x is for the most part better than 1 @ 10x but it would be nice for doing the odd one or two wouldnt it?

 

[FatRakoon]

Yes, thats certainly a plausable reason... Ill go with that.

 

[FatRakoon]

When encoding with .avi you will never get any better then 2 threads no matter what container you use or software app you work with. If you really want to push your quad core encode with H.264 which will run your CPU at 100% if not I suggest you look at overclocking if you want to speed up the process.

No, no no... Its not the speed thats the concern.. Hell, Im more than happy with it as it is... I dont even overclock anymore as I feel that its pretty pointless now.

Its just more that I felt that if one core gave me 25% and 4 cores gave me 4x25% then why was it adding up to only 50%.

This also goes a fair way to explain why me moving from 2 cores to 4 cores wasnt that big of a jump I suppose?

 

[FatRakoon]

I thought it was quite well documented that when performing single tasks as opposed to multiple tasks the CPU is often unable to achieve super fast speeds due to limitations in software.

Yes, thats the same as saying that Software must also be written to be multi-core or hyperthreading aware to be able to benefit from multi-cores.

However...

v4 of the software I am using does in fact have the option to select how many cores to use, and, it does what other software does when using multiple cores...

IT NEVER USES THEM FULLY!

For example...

When the PC is idle, in the Task manager, all 4 charts are floating around the bottom as they should be.

Now, if I select "1" in the number of cores this program is to use, then in the task manager, all 4 are floating about the 25% mark.
Now, if I set the afinity to one in the task manager, then sure enough, the one core I set it to shoots up to 100% and the rest drop off ( As Expected )

Similarly, if I chose "2 cores" then all 4 float up to 50%, however when watching them, the actual pointer goes from anywhere between 25% and 75% and choosing 3 or 4 cores does not make any difference.

And this is pratty much the same with every app I have thats supposed to be multicore aware.

The one and only app that does seem to be doing things as they should be is Folding At Home.. Even then I have to set the affinity and then everything runs at 100% throughout all 4 cores
The affinity opition in the config does not work either... I have to manually set it on a restart. Not a bad thing as I never restart my PC unless I have to.

 

[FatRakoon]

WinAvi ?

Im uinsing ConvertXtoDVD v4

Not used WinAvi for some time as I found the output files were often a little jerky.

I will have a look however.

 

[FatRakoon]

Well, ok, I have tried both WINAVI and AVS

Ok, both use Multicores much much better than ConvertXtoDVD do

However...

WinAVI was converting at about the 85% mark and this allowed me to also play Dawn Of War SoulStorm while I waited and I found no noticeable loss of speed in Disk access.

While AVS however was churning the AVI to DVD at about 98% and made using the PC a little irritating... I could not even cycle through my videos that I had already done and this was simply not a good thing for me.

Now, Im loking at a few things here

The interface, the way I can drag & Drop AVI Files, the CPU useage in converting 1,2,3 or 4 files as well as how much this afects the use of the PC when I am in deed converting and its clear to me, that since I fold on all 4 cores, I also convert AVI to DVD on all 4 cores, and I burn 4 DVDs at a time, plus I like to play games while I am converting and burning, that ConvertXtoDVD is still by far the better chioce, for much more than just one or two reasons.

However, if there is just one massive file that I need converting quickly and I dont need to use the PC while its doing it, then AVS might well be the better option... I suppose I could just knock it on to use 3 cores maybe but thats also defeating the object and thats what I do with ConvertXtoDVD anyway, so...

Ah well...

I wil also have a play on the Phenom in the next couple of days to see if the results are any different.