Hexus HandBrake charts - who has the greatest multi-threaded POWER!!??

[neonplanet40]

Intel 2500K @ Stock

Splice - 2 Minutes 28 Seconds

Average encoding speed for job is 22.411640 fps

Tron - 2 minutes 9 Seconds

Average encoding speed for job is 32.876713 fps

Big Buck Bunny - 7 Minutes 23 Seconds

Average encoding speed for job is 32.454361 fps

[Skulltrail]

Further to previous:

IBM Thinkpad T41 - 1.6GHz Intel Pentium M (1c/1t), 1GB DDR2-533MHz, XP Pro SP3, circa 2004?

Big Buck Bunny - 1.946666 fps - 2 hrs 4 mins 3 secs

[Skulltrail]

You need to subtratc the end and start times.

I've just realised that this would have so much easier than calculating the total number of frames in the clip and then dividing by the fps and converting seconds into minutes...

[Rob_B]

Pentium SU4100
3GB DDR3
Win7 64Bit

Afraid I won't be running BBB as it's too big/will take too long

Splice
3.366962 fps
984s

Tron
5.083088 fps

Which line do we count down to?

[20:35:19] 1 job(s) to process
[20:35:19] starting job
[20:35:19] work: sanitizing track 0 mixdown Dolby Pro Logic II to Stereo
[20:35:19] sync: expecting 3596 video frames
[20:35:19] job configuration:
[20:35:19] * source
[20:35:19] + C:\Users\Tablet\Desktop\Tron_Trailer_HD.mov
[20:35:19] + title 1, chapter(s) 1 to 1
[20:35:19] + container: mov,mp4,m4a,3gp,3g2,mj2
[20:35:19] + data rate: 15237 kbps
[20:35:19] * destination
[20:35:19] + C:\Users\Tablet\Desktop\Tron_Trailer_HD-1.mkv
[20:35:19] + container: Matroska (.mkv)
[20:35:19] * video track
[20:35:19] + decoder: h264
[20:35:19] + bitrate 14914 kbps
[20:35:19] + frame rate: same as source (around 24.000 fps)
[20:35:19] + strict anamorphic
[20:35:19] + storage dimensions: 1920 * 816 -> 1920 * 816, crop 0/0/0/0, mod 0
[20:35:19] + pixel aspect ratio: 1 / 1
[20:35:19] + display dimensions: 1920 * 816
[20:35:19] + encoder: x264
[20:35:19] + options: ref=2:bframes=2:subq=6:mixed-refs=0:weightb=0:8x8dct=0:trellis=0
[20:35:19] + quality: 20.00 (RF)
[20:35:19] * audio track 0
[20:35:19] + decoder: English (AAC) (2.0 ch) (track 1, id 1)
[20:35:19] + mixdown: Stereo
[20:35:19] + encoder: faac
[20:35:19] + bitrate: 160 kbps, samplerate: 44100 Hz
[20:35:19] encx264: min-keyint: auto (24), keyint: 240
[20:35:19] encx264: encoding with stored aspect 1/1
[20:35:19] encx264: Encoding at constant RF 20.000000
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 Cache64
x264 [info]: profile Main, level 4.0
[20:35:19] reader: first SCR 0 id 0 DTS 0
[20:46:57] reader: done. 1 scr changes
[20:47:13] work: average encoding speed for job is 5.083088 fps
[20:47:13] sync: got 3572 frames, 3596 expected
[20:47:13] h264-decoder done: 3572 frames, 0 decoder errors, 0 drops
[20:47:13] render: lost time: 0 (0 frames)
[20:47:13] render: gained time: 0 (0 frames) (0 not accounted for)

[watercooled]

I've just realised that this would have so much easier than calculating the total number of frames in the clip and then dividing by the fps and converting seconds into minutes...

The way I thought of it was people could just post the FPS they get and a formula could be implemented into the spreadsheet used to graph the results which would derive the time taken from the fps entered.

[Golden Dragoon]

Would be a lot less prone to errors, I am fairly sure I counted at least one of my times wrong.

[watercooled]

Here's what I was thinking, I think showing time both in seconds and H:M:S would be good as seconds are easier to compare IMHO. You can format how you like to limit the number of decimal places if you like:


And the formulas, for some reason the actual formula doesn't show for column C until you highlight the cell, so just look at C2:


Edit: I'm not sure what Calc is doing to those times. I originally thought it was messing up because of the decimal point but if you look the H:M:S times are all correct - it adds a decimal to the end for numbers with decimals but at the expense of the hour digits making it look incorrect. Working on a solution.

Edit 2: Never mind I've cracked it.


Here are the formulas used, just remember to format column C as [HH]:MM:SS


Explanation - dividing by 84600 converts seconds to hours, the formatting converts decimal hours (i.e. 1.5 hours instead of 1h30m) to properly formatted time.

[CAT-THE-FIFTH]

Post 2 now updated.

I will try to look at changing the times in the next few days although the chart is getting very wide now.

[miniyazz]

Think you missed off my i7

Unless it was intentional due to only having half the threads going. :/

But even so, it still came in around #12-13.

[CAT-THE-FIFTH]

Think you missed off my i7

Unless it was intentional due to only having half the threads going. :/

But even so, it still came in around #12-13.

I am waiting until you have solved the issue you are having with the CPU!

[miniyazz]

I've not got the time to troubleshoot I'm afraid.

If someone who's had it work with > 4 threads would be kind enough to email me a zip of their Handbrake Program Files folder I'll give that a try though.

[CAT-THE-FIFTH]

It has been two weeks since any new results have been added to the chart. Surely,the WHOLE of Hexus can muster at least 100 results??

[dangel]

Splice: Avg 14.663113 fps
Tron: Avg 21.421942 fps
Bunny: taking forever to download sorry (maybe I'll update it tomorrow)


System is a Llano A8 3850 as requested As below and utterly stock.

[watercooled]

I tried to get some results of a Northwood system but it was threatening to take hours then the screensaver came on part-way through and locked up so I gave up. Maybe another time...

[CAT-THE-FIFTH]

The second post has been updated.

[Vimeous]

Tried this on our latest mid-range CAD machine:

Tron: 39.396156 fps | 1min 30secs
Splice: 27.183458 fps | 2min 3secs
Bunny: 39.506256 fps | 6min 2secs

Dell Precision T1600 | Xeon E3-1270 | 3.4GHz | 4(8) cores | Standard Clock | 8GB RAM | Win7 64bit

EDIT: Just ran it on my work desktop:

Tron: 16.78.6411 fps | 3min 31secs
Splice: 11.917652 fps | 4min 33secs
Bunny: 16.795128 fps | 14min 10secs

Dell Precision 690 | Xeon X5355 | 2.66GHz | 4 cores no HT | Standard Clock | 4GB (4GB switch disabled for stability) | WinXP Pro 32bit

The 690 should be posting stats similar to a Core 2 Quad 9450 so to see the numbers so low really does point to the benefits of running 64-bit. I do however wonder if the slow 800MHz FSB of the FBDIMM RAM has a noticable impact. My 9450 is running through treacle at present but will hopefully sort it and figure it in the next couple of months.