Handbrake encoding Sandy Bridge *Updated with several results*

[odedia]

Ok, so I reapplied the thermal paste on my CPU/GPU. The results are the same (3 seconds better), but my CPU is getting hotter and the fans are running faster . I have no idea why, I guess I did something wrong.

http://att.macrumors.com/showpost.php?p=12421517&postcount=303

 

[daneoni]

^
Maybe you screwed up a thermal sensor cable somewhere, applied too much TP or just need to wait till the new compound 'breaks in'

 

[odedia]

^
Maybe you screwed up a thermal sensor cable somewhere, applied too much TP or just need to wait till the new compound 'breaks in'

No, I'm pretty sure all cables were put back correctly... I wonder if the thermal compound I've used is at all that great (Zalman STG1). I will try out an MX-2 or MX-4 tomorrow.

 

[superericla]

I find it odd that my base 15" Pro is encoding video at the same rate or faster in comparison to most of the top of the line models with faster CPUs. This may be because I'm using the latest nightly build of Handbrake.

 

[Mr. Retrofire]

I also ran it on two Windows 7 x64 systems - although Handbrake is only 32bit on Windows - the results seem right in line with the OSX results...

All modern CPU-based H.264-encoders use Streaming SIMD Extensions (SSE) instructions (Intel x86 & x64) or AltiVec instructions (PowerPC G4 & G5), which work both with 128-Bit registers. 64-Bit versions of these encoders are only 10 percent faster than the 32-Bit versions, because all time consuming code is already "128-Bit optimized", i.e. optimized for vector calculations.

 

[cambookpro]

Just to compare with some of the newer MBPs

15" MBP (Late '08) 2.4GHz Intel Core 2 Duo 4GB RAM - 21 mins 01 secs @ 11.5 fps

New MBP to arrive in 6 days

 

[rw3]

2010 Core i7 2.93Ghz iMac 16GB DDR3-1333Mhz - 5:30 on my encode

 

[cluthz]

I didn't like the new i5 13" MBP beat my 2010 MBP!
Seems like the new SB Dual Cores are almost 20% faster than similar clocked Arrandales.

2010:
15" MBP 2.40 GHz Dual Core i5 8GB RAM - 12 mins 24 secs @ 19.14 FPS

 

[adrian1480]

17" MBP 2.2GHz Quad Core i7 8GB RAM - 5 mins 54 secs @ 40.45 fps

had a couple of things running in the background, which may have added the extra 10 seconds or so over what similar systems are getting.

 

[lykranian]

q

 

[lykranian]

q

 

[cirus]

15"

15" MBP 2.0GHz Quad Core i7 4GB RAM - 6 mins 44 secs @ 35.31 fps
15" MBP 2.0GHz Quad Core i7 8GB RAM - 6 mins 7 secs @ 39.23 fps

15" MBP 2.2GHz Quad Core i7 4GB RAM - 5 mins 39 secs @ 42.27 fps
15" MBP 2.2GHz Quad Core i7 8GB RAM - 5 mins 39 secs @ 42.31 fps

15" MBP 2.3GHz Quad Core i7 4GB RAM - 5 mins 19 secs @ 44.78 fps
15" MBP 2.3GHz Quad Core i7 8GB RAM - 5 mins 25 secs @ 44.20 fps

17"

17" MBP 2.2GHz Quad Core i7 4GB RAM - 5 mins 37 secs @ 42.48 fps
17" MBP 2.2GHz Quad Core i7 8GB RAM - 5 mins 40 secs @ 42.03 fps

17" MBP 2.3GHz Quad Core i7 4GB RAM - TBD
17" MBP 2.3GHz Quad Core i7 8GB RAM - 5 mins 49 secs @ 41.12 fps

This should clear it up for those who are unsure whether or not to get the 2.3 GHz. Practically no difference and not worth the money (20 seconds maximum for the 15 inch and nothing for the 17 inch).

 

[xxBURT0Nxx]

15"

15" MBP 2.0GHz Quad Core i7 4GB RAM - 6 mins 44 secs @ 35.31 fps
15" MBP 2.0GHz Quad Core i7 8GB RAM - 6 mins 7 secs @ 39.23 fps

15" MBP 2.2GHz Quad Core i7 4GB RAM - 5 mins 39 secs @ 42.27 fps
15" MBP 2.2GHz Quad Core i7 8GB RAM - 5 mins 39 secs @ 42.31 fps

15" MBP 2.3GHz Quad Core i7 4GB RAM - 5 mins 19 secs @ 44.78 fps
15" MBP 2.3GHz Quad Core i7 8GB RAM - 5 mins 25 secs @ 44.20 fps

17"

17" MBP 2.2GHz Quad Core i7 4GB RAM - 5 mins 37 secs @ 42.48 fps
17" MBP 2.2GHz Quad Core i7 8GB RAM - 5 mins 40 secs @ 42.03 fps

17" MBP 2.3GHz Quad Core i7 4GB RAM - TBD
17" MBP 2.3GHz Quad Core i7 8GB RAM - 5 mins 49 secs @ 41.12 fps

This should clear it up for those who are unsure whether or not to get the 2.3 GHz. Practically no difference and not worth the money (20 seconds maximum for the 15 inch and nothing for the 17 inch).

while i agree for most everyone there is no need for the 2.3, this is a short video, ~10 mins. If you were encoding much longer movies at higher quality and bitrates with full audio (full length movies, etc.) the difference would be much more noticeable, could end up saving a half hour or so off encoding time.... in day to day use most people wouldn't ever even notice the difference between the 2.0 and 2.3.. the high end processors are usually only beneficial to "pros" or people maxing out and really taxing their systems.

 

[lykranian]

q

 

[xxBURT0Nxx]

I think the comparison was for the 2.2 and the 2.3. Pretty much everyone agrees the 2.2 is worth it over the 2.0.

I would say people prefer the 2.2 because of the HD 6750M gpu, not for the 200mhz speed increase...

 

[brentsg]

while i agree for most everyone there is no need for the 2.3, this is a short video, ~10 mins. If you were encoding much longer movies at higher quality and bitrates with full audio (full length movies, etc.) the difference would be much more noticeable, could end up saving a half hour or so off encoding time.... in day to day use most people wouldn't ever even notice the difference between the 2.0 and 2.3.. the high end processors are usually only beneficial to "pros" or people maxing out and really taxing their systems.

It would have to be HD encodes queue'd one after the other to save 30 minutes. You're going to have to encode for nearly a full day straight to save 1 hour with the 2.3.

If you're doing that regularly enough for it to be an issue then a MBP's probably the wrong tool for the job.

 

[Uffish Thought]

15"

15" MBP 2.0GHz Quad Core i7 4GB RAM - 6 mins 44 secs @ 35.31 fps
15" MBP 2.0GHz Quad Core i7 8GB RAM - 6 mins 7 secs @ 39.23 fps

15" MBP 2.2GHz Quad Core i7 4GB RAM - 5 mins 39 secs @ 42.27 fps
15" MBP 2.2GHz Quad Core i7 8GB RAM - 5 mins 39 secs @ 42.31 fps

15" MBP 2.3GHz Quad Core i7 4GB RAM - 5 mins 19 secs @ 44.78 fps
15" MBP 2.3GHz Quad Core i7 8GB RAM - 5 mins 25 secs @ 44.20 fps

17"

17" MBP 2.2GHz Quad Core i7 4GB RAM - 5 mins 37 secs @ 42.48 fps
17" MBP 2.2GHz Quad Core i7 8GB RAM - 5 mins 40 secs @ 42.03 fps

17" MBP 2.3GHz Quad Core i7 4GB RAM - TBD
17" MBP 2.3GHz Quad Core i7 8GB RAM - 5 mins 49 secs @ 41.12 fps

This should clear it up for those who are unsure whether or not to get the 2.3 GHz. Practically no difference and not worth the money (20 seconds maximum for the 15 inch and nothing for the 17 inch).

Here is my 17" MBP 2.3GHz Quad Core i7 4GB RAM Data:
Encode time: 5:28 min
Average FPS: 43.8
CPU Temp: 83 C
Fans: 3850 RPM
For what it's worth, that is 3% faster in time and fps than the 2.2GHz machine.

 

[adrian1480]

while i agree for most everyone there is no need for the 2.3, this is a short video, ~10 mins. If you were encoding much longer movies at higher quality and bitrates with full audio (full length movies, etc.) the difference would be much more noticeable, could end up saving a half hour or so off encoding time.... in day to day use most people wouldn't ever even notice the difference between the 2.0 and 2.3.. the high end processors are usually only beneficial to "pros" or people maxing out and really taxing their systems.

not really. after 10 minutes, the difference is approximately 2 seconds.

if you put that into a simple equation (and I mean simple):

(total seconds of difference between the 2 transcodes / total length of the transcoded video) * (number of seconds in a 2-hour movie) = amount of time saved with the $300 more expensive processor in 2 hours of uninterrupted transcoding.

2/(10*60) * 7200 =

0.0033333 * 7200 = 23.9999976 seconds saved.

in short, you're saving 24 seconds on each 2-hour movie you encode/transcode. Or, approximately 12 seconds saved per hour of transcoded material that utilizes 100% of the CPU. Yea, you'd have to do a helluva lot of transcoding to appreciate or work at 100% of the CPU to appreciate that 0.3% advantage.

In order to save just 1 hour of your lifetime, you'd have to work at 100% cpu for:

60/12 * 60 = 300 hours.

yea, it'll take you around 300 hours to save yourself one hour of real time if your purpose is transcoding. other applications may feel the boost a little more in real-time. maybe 3D design applications. maybe games will get a couple extra FPS of performance boost.

overall? I'd strongly suggest people putting that $300 into a nice SSD drive.

 

[polbit]

Actually, looking at best case of saving 20 seconds over 5 min encode on the 15", that's 4 minutes saved over an hour.

That means that every 15 hours you save an hour. I'd say that's substantial enough.

Polbit

 

[acedickson]

I didn't like the new i5 13" MBP beat my 2010 MBP!
Seems like the new SB Dual Cores are almost 20% faster than similar clocked Arrandales.

2010:
15" MBP 2.40 GHz Dual Core i5 8GB RAM - 12 mins 24 secs @ 19.14 FPS

Ha, at least you have a nicer screen? I dunno, I tried to lessen the sting.

 

[imahawki]

Did I download the right file? It said MP4 .torrent but the file it downloaded is an .avi.

 

[imahawki]

Well assuming I used the right file I got
19min 57 seconds at 11.998235 fps average on a 2.4Ghz C2D Mac Mini with 8GB RAM.

 

[xxBURT0Nxx]

not really. after 10 minutes, the difference is approximately 2 seconds.

if you put that into a simple equation (and I mean simple):

(total seconds of difference between the 2 transcodes / total length of the transcoded video) * (number of seconds in a 2-hour movie) = amount of time saved with the $300 more expensive processor in 2 hours of uninterrupted transcoding.

2/(10*60) * 7200 =

0.0033333 * 7200 = 23.9999976 seconds saved.

in short, you're saving 24 seconds on each 2-hour movie you encode/transcode. Or, approximately 12 seconds saved per hour of transcoded material that utilizes 100% of the CPU. Yea, you'd have to do a helluva lot of transcoding to appreciate or work at 100% of the CPU to appreciate that 0.3% advantage.

In order to save just 1 hour of your lifetime, you'd have to work at 100% cpu for:

60/12 * 60 = 300 hours.

yea, it'll take you around 300 hours to save yourself one hour of real time if your purpose is transcoding. other applications may feel the boost a little more in real-time. maybe 3D design applications. maybe games will get a couple extra FPS of performance boost.

overall? I'd strongly suggest people putting that $300 into a nice SSD drive.

2 seconds? not sure where you got that but it is around a 20-25 second difference between the 2.2 and 2.3 which is going to change your numbers, closer to 4 or 5 minutes than 24 seconds a movie...

I also assume that if you are encoding at higher bitrates, resolutions, audio codecs, etc. that the time differences would be more drastic... but maybe not

 

[adrian1480]

2 seconds? not sure where you got that but it is around a 20-25 second difference between the 2.2 and 2.3 which is going to change your numbers, closer to 4 or 5 minutes than 24 seconds a movie...

yea, you're right. I was moving too fast and grabbed the FPS instead of the time.

Taking the best 2.3 time and the best 2.2 time on the 15"ers, we have a 6% difference (20 seconds). with the same, very basic math:

20/(10*60) * 7200 =

0.03 * 7200 = 240 seconds, or 4 minutes saved per 2 hours.

or, for every 1 hour of transcoding you do, you save about 2 minutes of life time.

Continuing on with the simple calculations:

In order to save 1 hour of lifetime, you'd have to transcode at max speed for 30 hours.

Not nearly as much as 300 hours...but still, how many people are going to spend even 30 hours transcoding video or rendering 3D? The same concept applies as before: if you're doing 3D CAD-type work or do video encoding/transcoding for a living, you could potentially save a few hours a year by going with the $300 CPU upgrade. Whether that is worth the investment is purely a function of how much transoding you calculate that you do annually and how much your time is worth. And if you're doing *that much* transcoding, you're probably doing it from a desktop anyway, right?

For the other 99% of Mac owners who aren't transcoding hundreds of of audio/video or rendering 3D scenes...your gain is going to be next to nothing. $300 by far better spent on a reliable SSD.

 

[enklined]

Well, ran it twice. Here are the results:

[19:08:18] starting job
[19:11:10] work: average encoding speed for job is 46.659645 fps

[19:23:34] starting job
[19:26:29] work: average encoding speed for job is 46.115345 fps

If my math is right, the first job finished in 2 minutes and 52 seconds, while the second finished in 2 minutes and 55 seconds.

2011 15" Macbook Pro Quad-Core 2.3GHz i7 8GB PC3-16000 G.Skill RAM running the 64 bit version of Handbrake.

My times blow all the over MBP times out of the water...not sure what I'm doing wrong here. Unmolested version of 64bit Handbrake, using the AppleTV2 preset. Is this because I'm using the 64bit version? That's the only thing I can think of...odd.

Edit: Just watched the video. It's only 5:32 long. Heh.

Edit 2: Re-downloaded the test file again. Now it's only 7:02 long. Can't seem to get the whole file downloaded! Server must be having issues. Will try again later...maybe.