new HD x264 video encoding benchmark

[graysky]

Techarp is hosting a new HD x264 benchmark. Have a look and run it on your system to help populate the data table.

Basically, you run the test encode of an HD sample (1280x720) and it will report back frames-per-second values for your machine @ it's clock/overclock level. You can run it at your stock settings and at your overclock settings to see how your machine compares to others in the database.

My goal is to have a representative set of data for many different chips and chipsets. Please just report your results here in this thread. I will keep the data at that url to keep things simple.

Thanks all and enjoy!

The following image will be updated automatically as new data comes in. It doesn't display any results, but it does show the number of data points collected so far and how they break down by CPU type:

[hnzw_rui]

Corrections on the Avisynth script:

Code: Select all

LoadPlugin("DGDecode.dll")
DGDecode_mpeg2source("test-720p.d2v")
loop(2)

Too lazy to go back to stock but here are OC'ed settings. Maybe I'll do a stock benchmark before I swap it out for a Q6600.

Pentium DC E2160 @ 2.40
Gigabyte P35-DS3R
9x266, 2,394 MHz
2x2GB 5-4-4-11 @ 667
XP Pro SP2

Code: Select all

---------- RUN1PASS1.LOG
encoded 1442 frames, 25.89 fps, 3905.08 kb/s

---------- RUN2PASS1.LOG
encoded 1442 frames, 25.89 fps, 3905.08 kb/s

---------- RUN3PASS1.LOG
encoded 1442 frames, 25.92 fps, 3905.08 kb/s

---------- RUN4PASS1.LOG
encoded 1442 frames, 25.79 fps, 3905.08 kb/s

---------- RUN1PASS2.LOG
encoded 1442 frames, 6.38 fps, 3942.92 kb/s

---------- RUN2PASS2.LOG
encoded 1442 frames, 6.33 fps, 3942.92 kb/s

---------- RUN3PASS2.LOG
encoded 1442 frames, 6.38 fps, 3942.92 kb/s

---------- RUN4PASS2.LOG
encoded 1442 frames, 6.41 fps, 3942.92 kb/s

[graysky]

Killer man, thanks for the results. You don't actually need the loadplugin line with this version of avisynth but it doesn't hurt either.

[hnzw_rui]

You don't need that line if the DGDecode.dll is located inside the Avisynth plugins folder (which is what you're trying to avoid here to lessen the steps the user would need to do). If you put it elsewhere, then Avisynth would have no idea where to find the function unless you use LoadPlugin.

[graysky]

Interesting... so if the dll resides in the same dir as the batch file, how can one specify that location in the avs without knowing the abs path to the dll? Since I have no way of knowing what drive letter and directory the user will unrar the benchmark into, it makes that tough. If you look at my batch file you'll see the logic.

[DMB2000uk]

Overclocked:

---------- RUN1PASS1.LOG
encoded 1442 frames, 32.95 fps, 3905.08 kb/s

---------- RUN2PASS1.LOG
encoded 1442 frames, 33.13 fps, 3905.08 kb/s

---------- RUN3PASS1.LOG
encoded 1442 frames, 33.14 fps, 3905.08 kb/s

---------- RUN4PASS1.LOG
encoded 1442 frames, 34.40 fps, 3905.08 kb/s

---------- RUN1PASS2.LOG
encoded 1442 frames, 8.23 fps, 3942.92 kb/s

---------- RUN2PASS2.LOG
encoded 1442 frames, 8.17 fps, 3942.92 kb/s

---------- RUN3PASS2.LOG
encoded 1442 frames, 8.32 fps, 3942.92 kb/s

---------- RUN4PASS2.LOG
encoded 1442 frames, 8.19 fps, 3942.92 kb/s

E6600 @ 3024Mhz (336*9)
AW9D Max (975x chipset)
2x1GB OCZ Reaper 840Mhz 3-4-4-11
XP Pro SP2

Dan

[DMB2000uk]

Stock resutls:

---------- RUN1PASS1.LOG
encoded 1442 frames, 16.49 fps, 3905.08 kb/s

---------- RUN2PASS1.LOG
encoded 1442 frames, 19.59 fps, 3905.08 kb/s

---------- RUN3PASS1.LOG
encoded 1442 frames, 24.72 fps, 3905.08 kb/s

---------- RUN4PASS1.LOG
encoded 1442 frames, 22.63 fps, 3905.08 kb/s

---------- RUN1PASS2.LOG
encoded 1442 frames, 4.43 fps, 3942.92 kb/s

---------- RUN2PASS2.LOG
encoded 1442 frames, 4.87 fps, 3942.92 kb/s

---------- RUN3PASS2.LOG
encoded 1442 frames, 5.01 fps, 3942.92 kb/s

---------- RUN4PASS2.LOG
encoded 1442 frames, 5.85 fps, 3942.92 kb/s

E6600 @ 2403Mhz (267*9)
AW9D Max (975x chipset)
2x1GB OCZ Reaper 801Mhz 3-4-4-11
XP Pro SP2

Dan

[DMB2000uk]

Just looked at the results now that I've rebooted back to the overclock, and a 25% increase in corespeed is giving me a 60% improvement in the benchmark. That's messed up!

Just checked and superpi gives me the expected 25% improvement from stock to overclocked, so its not my system being freaky.

Dan

[hnzw_rui]

Interesting... so if the dll resides in the same dir as the batch file, how can one specify that location in the avs without knowing the abs path to the dll? Since I have no way of knowing what drive letter and directory the user will unrar the benchmark into, it makes that tough. If you look at my batch file you'll see the logic.

Location of dll plugin is relative to the location of the Avisynth script and not the batch file. It's Avisynth that does the MPEG2Source function call, not the batch. Hence, as long as the dll is in the same directory as the Avisynth script, LoadPlugin("filename.dll") should suffice.

Just looked at the results now that I've rebooted back to the overclock, and a 25% increase in corespeed is giving me a 60% improvement in the benchmark. That's messed up!

Just checked and superpi gives me the expected 25% improvement from stock to overclocked, so its not my system being freaky.

Dan

Your results for stock seem a bit low. I'm running my E2160 @ 2.40GHz (9*266) and I'm averaging ~25.87 for first pass and ~6.38 for second pass. Granted, I had 4GB RAM, but still, your processor is better, RAM speed was faster and latencies were lower. Perhaps you had something running in the background for the stock test or maybe C1E and EIST were enabled?

[graysky]

Location of dll plugin is relative to the location of the Avisynth script and not the batch file. It's Avisynth that does the MPEG2Source function call, not the batch. Hence, as long as the dll is in the same directory as the Avisynth script, LoadPlugin("filename.dll") should suffice.

Cool man, I actually forgot that the dll can sit anywhere on the FS so long as the script calls it as you described. That's a more elegant way to go about it. Thanx!