x264 video encoding benchmark

[directhex]

Did you compile a 64bit version of x264 for win64?

i don't have windows x64 installed anymore, and don't have an environment set up to do it

i'm not even sure gcc 3.4.6 CAN do it, and switching compilers changes the game entirely

[Kilmatead]

QX6800 stock 11x266, p965, 4GB PC2-8500-1067 5-5-5-15, Vista x64

---------- RUN1PASS1.LOG
encoded 1749 frames, 147.27 fps, 1850.94 kb/s

---------- RUN2PASS1.LOG
encoded 1749 frames, 147.15 fps, 1850.94 kb/s

---------- RUN3PASS1.LOG
encoded 1749 frames, 148.26 fps, 1850.94 kb/s

---------- RUN4PASS1.LOG
encoded 1749 frames, 147.58 fps, 1850.94 kb/s

---------- RUN5PASS1.LOG
encoded 1749 frames, 147.66 fps, 1850.94 kb/s

---------- RUN1PASS2.LOG
encoded 1749 frames, 39.78 fps, 1829.33 kb/s

---------- RUN2PASS2.LOG
encoded 1749 frames, 40.25 fps, 1829.03 kb/s

---------- RUN3PASS2.LOG
encoded 1749 frames, 40.36 fps, 1829.37 kb/s

---------- RUN4PASS2.LOG
encoded 1749 frames, 40.14 fps, 1829.54 kb/s

---------- RUN5PASS2.LOG
encoded 1749 frames, 40.03 fps, 1829.54 kb/s

and QX6800 12x266 @3.2, p965, 4GB PC2-8500-1067 5-5-5-15, Vista x64

---------- RUN1PASS1.LOG
encoded 1749 frames, 161.76 fps, 1850.94 kb/s

---------- RUN2PASS1.LOG
encoded 1749 frames, 161.44 fps, 1850.94 kb/s

---------- RUN3PASS1.LOG
encoded 1749 frames, 161.62 fps, 1850.94 kb/s

---------- RUN4PASS1.LOG
encoded 1749 frames, 160.11 fps, 1850.94 kb/s

---------- RUN5PASS1.LOG
encoded 1749 frames, 162.03 fps, 1850.94 kb/s

---------- RUN1PASS2.LOG
encoded 1749 frames, 44.31 fps, 1829.49 kb/s

---------- RUN2PASS2.LOG
encoded 1749 frames, 44.10 fps, 1829.19 kb/s

---------- RUN3PASS2.LOG
encoded 1749 frames, 44.53 fps, 1829.31 kb/s

---------- RUN4PASS2.LOG
encoded 1749 frames, 44.49 fps, 1829.34 kb/s

---------- RUN5PASS2.LOG
encoded 1749 frames, 44.45 fps, 1829.41 kb/s

Yes, I know just upping the multi by one isn't much of an overclock, but 3.2 is fast enough for the temps I get...

[graysky]

Cool man, thanks for the data... what speed did you run your RAM @ for both those runs?

Also just curious what sort of temps you get under load @ 3.2... also what is your cooling?

[Kilmatead]

For simplicity I just leave the ram at 1067, or at least so the mobo tells me. Used to test my own system using a volumetrics rendering test in the Blender design program, and was genuinely surprised at the simple difference switching from 800 to 1067 made. More, in my opinion, than 2.93 to 3.2 did, so I tend to look at overclocking somewhat skeptically now. But that's me.

Cooling is a Tuniq 120 using Arctic Ceramique... again, perhaps slightly over-rated. Temps at full are a disappointing 70-75 per core. I've read (and followed) Clunk's guide on these things, but got frustrated at shredding my fingers on heatsinks for 1 degree here and 1 degree there. I have an Amanda TEC cooler (for a rainy day project), but haven't tried to fit it yet as it requires a bit of modding for use on new 775 boards.

Ironically, after switching to Vista x64 the only thing that didn't work first thing was CoreTemp, which I used religiously. Switched to Speedfan, which at least interfaces with LcdStudio & G15.

Sorry for rambling off-topic there...

[graysky]

70-75 is pretty hot in my opinion. Have you considered lapping the IHS on the chip and the base of your HS?

[Kilmatead]

Amusingly enough, I remember reading those posts awhile ago thinking to myself "this guy's a nutcase", though I can't argue with your results. There are some people in the world who think it's great fun to jump out of airplanes; I like my feet on the ground. Likewise, attacking a bit of kit that cost me a few week's wages with stuff that belongs on the beach just strikes me as... um... nuts?

Having recently fried a Commando mobo by letting my curiosity get the better of me, a nice healthy conservative (fear) respect of "if it ain't broke don't fix it" comes to mind.

I'm not mad about the 70 degrees myself, but your pictures of a million and one scratches are just unhealthy mate! I would have nightmares of a million and one metallic dust particles floating about the place...

[graysky]

Kilmatead: Haha. That's what the water is for... it immobilizes those particles you're talking about into a metallic/silicon mud really. I think you just had a bad experience frying your MB. As long as your careful, I think lapping chips/heat sinks is pretty safe. I'm sure once you pass that initial hill, you'll agree. It took me 2 days of thinking about it before I finally did.

[graysky]

Updated the Intel table. It now contains several Yorkfield ES chips including:

Xeon E5330 (Dual board)
Q9550
Q9350

[graysky]

Updated the tables with another 45 nm chip: the QX9650 -- both at stock levels and @ overclocked to 4.2 GHz! With it, and the others (Xeon E5330 (Dual board), Q9550, and Q9350) there is now data on 4 different 45 nm chips.

One thing that I found striking about these new chips is that they are only marginally faster than their 65 nm counterparts when encoding x264 (about 5-6 % faster with all other factors being equal or close to equal). Have a look at the general trends table for the Kentsfield vs. Yorkfield comparison at the official host.

[graysky]

First off, thanks to all who contributed data.

24-Feb-2008 - Finally updated the data tables on the x264 benchmark page. They are now html based (not .gif images) which makes my life updating them much easier and I will keep this tables up-to-date daily as people post results. Have a look at the 'Data Tends' table that contains a look at the Phenom quad vs. both Kentfield and Yorkfield quads. There are also some comparisons of Wolfdale dual vs. Conroe dual, and some other good stuff.