AMD X2 / Pentium D & AutoGK DivX Encoding

[stevty2889]

Yeah, I"d like to to see the 1 hour barrier broken as well..and you are darn close..which drive did you run it from? My Presler was running from a 250GB WD SATA 3G 16mb cache drive, and the X2 from raid-0 WD 160 SATA 3G's.

 

[stevty2889]

Originally posted by: Technonut
stevty2889: If you have the time, would you run FOTR through DVD Decrypter in file mode, then use DVD Shrink on it. After it is through, post the total DVD Shrink processing time shown....

TIA...

Yep, will do when I get a chance.

EDIT: Running DVD shrink on my 920 now, decrypting in file mode on my X2. Getting 100% cpu usage on the 920, so doesn't appear to be an I/O limit.

 

[stevty2889]

Left all the settings on default.

Pentium-D 920 @3.7ghz......7m 58s
X2 4200+ @2.618ghz..........8m 59s

DVD decrypt times were bascily the same, both using NEC 3540A's
17m 41s on the X2
17m 44s on the 920

 

[PetNorth]

stevty, it would be interesting to see some face to face between yours X2 and Pressler. Would you can to test them with AGK and new DivX 6.1? AGK with new XviD 1.2 beta? (it has initial SMP support) http://www.koepi.org/XviD-1.2.-127-07012006.exe

And with some other stuff, like 3dsmax rendering, iTunes mp3 encoding, TMPGEnc, Premiere Pro, Lightwave rendering, WME9... thanks

 

[Technonut]

Well, I decided to bump my OC up a notch (4.33GHz) and make a run... Note: This OC has not been stability tested, (even though I am 95% sure it will be dual SP-2004 stable) but did complete this run just fine...

[1/29/2006 8:23:52 AM] AutoGK 1.96
[1/29/2006 8:23:52 AM] OS: WinXP (5.1.2600).2
[1/29/2006 8:23:52 AM] Job started.
[1/29/2006 8:23:52 AM] Input dir: F:\FELLOWSHIP\VIDEO_TS
[1/29/2006 8:23:52 AM] Output file: F:\Fellowship AGK\VIDEO_TS.avi
[1/29/2006 8:23:52 AM] Audio: English
[1/29/2006 8:23:52 AM] Subtitles: none
[1/29/2006 8:23:52 AM] Codec: DivX
[1/29/2006 8:23:52 AM] Target size: 1400Mb
[1/29/2006 8:23:52 AM] Started encoding.
[1/29/2006 8:23:52 AM] Demuxing and indexing.
[1/29/2006 8:27:16 AM] Processing file: F:\FELLOWSHIP\VIDEO_TS\VTS_01_PGC_01_1.VOB
[1/29/2006 8:27:16 AM] Processing file: F:\FELLOWSHIP\VIDEO_TS\VTS_01_PGC_01_2.VOB
[1/29/2006 8:27:16 AM] Processing file: F:\FELLOWSHIP\VIDEO_TS\VTS_01_PGC_01_3.VOB
[1/29/2006 8:27:16 AM] Processing file: F:\FELLOWSHIP\VIDEO_TS\VTS_01_PGC_01_4.VOB
[1/29/2006 8:27:16 AM] Processing file: F:\FELLOWSHIP\VIDEO_TS\VTS_01_PGC_01_5.VOB
[1/29/2006 8:27:16 AM] Processing file: F:\FELLOWSHIP\VIDEO_TS\VTS_01_PGC_01_6.VOB
[1/29/2006 8:27:16 AM] Processing file: F:\FELLOWSHIP\VIDEO_TS\VTS_01_PGC_01_7.VOB
[1/29/2006 8:27:16 AM] Source aspect ratio: 16:9
[1/29/2006 8:27:16 AM] Source resolution: 720x480
[1/29/2006 8:27:16 AM] Found NTSC source.
[1/29/2006 8:27:16 AM] Source seems to be pure FILM.
[1/29/2006 8:27:16 AM] Output will contain 256563 frames
[1/29/2006 8:27:16 AM] Audio size: 599,243,008 bytes (571.48 Mb)
[1/29/2006 8:27:16 AM] Overhead: 8,072,192 bytes (7.70 Mb)
[1/29/2006 8:27:16 AM] Video size: 860,691,200 bytes (820.82 Mb)
[1/29/2006 8:27:16 AM] Target bitrate is: 643kbps
[1/29/2006 8:27:16 AM] Running compressibility test.
[1/29/2006 8:29:46 AM] Duration was: 2 minutes 29 seconds
[1/29/2006 8:29:46 AM] Speed was: 85.69 fps.
[1/29/2006 8:29:46 AM] Compressibility percentage is: 32.57
[1/29/2006 8:29:46 AM] Using softer resizer.
[1/29/2006 8:29:46 AM] Chosen resolution is: 528x224 ( AR: 2.36 )
[1/29/2006 8:29:46 AM] Predicted comptest value is: 63.24
[1/29/2006 8:29:46 AM] Running first pass.
[1/29/2006 8:54:47 AM] Duration was: 25 minutes 0 seconds
[1/29/2006 8:54:47 AM] Speed was: 170.93 fps.
[1/29/2006 8:54:47 AM] Running second pass.
[1/29/2006 9:21:08 AM] Duration was: 26 minutes 21 seconds
[1/29/2006 9:21:08 AM] Speed was: 162.25 fps.
[1/29/2006 9:21:09 AM] Splitting video into: F:\Fellowship AGK\VIDEO_TS.cd1.avi
[1/29/2006 9:22:01 AM] Splitting video into: F:\Fellowship AGK\VIDEO_TS.cd2.avi
[1/29/2006 9:22:54 AM] Job finished. Total time: 59 minutes 2 seconds

Note that pass 1 and 2 completed in 51m 21s

stevty2889: I am using my 200GB Seagate SATA... I have not defragged it yet. I need to get rid of a bunch of stuff on my Raptor, then use it...

Thanks for the DVD Shrink results... I also noticed 100% processor usage on both cores. This would make another interesting "real world" test.

 

[stevty2889]

Awsome, nice to see the 1 hour barrier broken. Presler is scaling much better than I would have expected.

 

[stevty2889]

Running TMPGEnc on the file created from AutoGK to convert it back to MPEG. Will post results from both machines when finished. Seems to be some I/O limit there as well, not getting 100% CPU usage.

EDIT: on my X2 with raid-0 drives, I am getting 100% CPU usage, but bouncing around between 65-97% on my 920. I'll have to put the drive from the 920 system in my X2 system to get a more fair comparison.

 

[stevty2889]

Strange, using the same HD I have in my presler system, in the X2, I still get higher CPU usage on the X2. Mostly staying at 100%, but occasionaly dropping in to the 80%'s instead of staying at a constant 100% like it did from the raid-0 drives.

Took 2h 2m 51s on the Presler
Took 1h 25m 20s on the X2 with raid-0
Looks like it may take 1-2 minutes longer using the same drive from the Presler system. Not sure why I can't get 100% CPU usage with the presler.

 

[Technonut]

stevty2889: That is strange... I am fairly certain that I observed 100% processor usage (both cores) with my 920 in TMPGEnc. I will fire it up and edit this in a few.....

EDIT: I started ROTK, and both cores pretty much stay between 95-100% usage on average... As for time, it looks as if it will finish in around 1h 20m.. (4.33GHz) I will keep it going to get a final time, and post the result.

 

[Furen]

Originally posted by: stevty2889
Awsome, nice to see the 1 hour barrier broken. Presler is scaling much better than I would have expected.

Well, considering that Prescott's pipeline was lengthened to 31 stages just to hit 5GHz I think it's not very surprising that it can actually hit 4.26GHz+. Remember that many of the drawbacks to Prescott also go away once you start hitting higher clocks so the performance potential is there (and since you're increasing the FSB to reach these clocks you're even helping it out even more). Intel's problem with Prescott was not that it sucked but that temperatures kept it from reaching its clock speed potential. AMD would have nothing on a 5GHz Prescott (well, maybe it could match it with a 3GHz A64) but the damn thing would probably melt on air.

 

[dmens]

Intel's problem with Prescott was not that it sucked but that temperatures kept it from reaching its clock speed potential

prescott was carelessly designed and hence was much hotter than it should have been. in fact, if the project was better managed from the beginning, imo prescott would have hit 5ghz on 90nm within reasonable thermals. the primary reason for cedarmill/presler's existence is to ramp the 65nm process... but inheriting the prescott database meant it would be a salvage operation from a design pov.

 

[stevty2889]

Originally posted by: Furen

Originally posted by: stevty2889
Awsome, nice to see the 1 hour barrier broken. Presler is scaling much better than I would have expected.

Well, considering that Prescott's pipeline was lengthened to 31 stages just to hit 5GHz I think it's not very surprising that it can actually hit 4.26GHz+. Remember that many of the drawbacks to Prescott also go away once you start hitting higher clocks so the performance potential is there (and since you're increasing the FSB to reach these clocks you're even helping it out even more). Intel's problem with Prescott was not that it sucked but that temperatures kept it from reaching its clock speed potential. AMD would have nothing on a 5GHz Prescott (well, maybe it could match it with a 3GHz A64) but the damn thing would probably melt on air.

I was really refering more to it's performance. By my original calculations based off of my P4's and my 830, I was figuring around 4.4ghz for it to catch up to my X2@2.618ghz. But it caught up at 3.64 and passed it at 3.7ghz for video encoding(other than this odd TMPGEnc issue). As for Prescott, in a lot of cases my northwood at 3.45ghz was able to keep up or surpass my prescott @3.82ghz. Video encoding or SSE3 optimized apps, the prescott was ahead, but at much higher temps, and I agree, the temps were what limted the prescotts. It would be interesting to see what a netburst chip could do on the 45nm proccess, but I don't think there are any plans for that. Netburst seems to really shine for video encoding.

 

[dmens]

steve, any idea if the program you're using was compiled on a p4 optimized compiler?

 

[stevty2889]

Originally posted by: dmens
steve, any idea if the program you're using was compiled on a p4 optimized compiler?

no idea, I am going to try installing the latest IAA drivers though and see if that helps.

 

[Technonut]

Here is a shot of ROTK/TMPGEnc 50% through... both cores still averaging 95-100% load.

Which TMPGEnc version are you using stevty2889?

 

[stevty2889]

Originally posted by: Technonut
Here is a shot of ROTK/TMPGEnc 50% through... both cores still averaging 95-100% load.

Which TMPGEnc version are you using stevty2889?

I'm using the free version(2.524.63.181).

Using the same Hard drive from my Presler system with the X2, I got 100% cpu usage, 1h 25m 48s so it doesn't seem to be an I/O limit as I originaly suspected..gonna try updating some drivers on my presler system and see if it turns out better.

EDIT: still having the same problem, can't get 100% cpu usage with my 920.

 

[Markfw]

Where to download that version ? I see 2.26 when I search for "autogk download" in google. I want to be sure I am using the right version if I run some benchmarks, so please link the the exact version I should use. I ripped to hard drive first using dvd dycrypter (I think thats what you said to do). And I downloaded that divx 6.1 above. Will be trying that soon.

Edit: I downloaded from the link at the top of this post, but it says:
" I am curious about how well the X2 and Pentium D is doing with Auto Gordian Knot 1.96"

But that version is 2.26. Whats up ? andI saw stevety was using "I'm using the free version(2.524.63.181). "

OK, so what version should I be using ? 1.96, 2.26, 2.254.63.181 ? This is all very confusing.

 

[Technonut]

Originally posted by: Markfw900
Where to download that version ? I see 2.26 when I search for "autogk download" in google. I want to be sure I am using the right version if I run some benchmarks, so please link the the exact version I should use. I ripped to hard drive first using dvd dycrypter (I think thats what you said to do). And I downloaded that divx 6.1 above. Will be trying that soon.

Good point.... version 1.96 has gotten long in the tooth since I started this thread last August. PM me your e-mail address, and I will get it to you.

I should change the thread to reflect the latest AutoGK version...

 

[Technonut]

and I saw stevety was using "I'm using the free version(2.524.63.181).

That is TMPGEnc....

 

[Technonut]

Markfw900: Just go ahead and download the latest AutoGK version and run it... We will all just need to start fresh with another run using version 2.26 and the DivX 6.1.1 VFW codec....

EDIT: You can get the DivX 6.1.1 VFW codec (codec only) Here...

 

[Markfw]

OK, did that, but it wants input file, not directory (I did rip using the IFO mode, and setting linked to above, what am I missing ? I called my directory fellowship2, since the fellowship is the default mode directory)

Can you put up a screen scot of what it should look like in 2.26 before I hit start ?

And I can;t hit start if I pick the IFO file. Do I have to add every single

... wait, It was waiting for me to accept some damn terms window....

 

[Technonut]

Can you put up a screen scot of what it should look like in 2.26 before I hit start ?

NP... I'm running it myself right now....

Screenie

EDIT: Damn.. I just noticed that I am running it in XvID... I'll need to start over with DivX....

 

[Markfw]

Yes, I am in divx, but stuck in the compressability test, using no cpu !!! Whats up ?

waiting on an accept message. Have to start over...

Well, I can;t see the autogk window anymore, there is a virual dub window, and I;m not at 100% cpu, so must be io bound. A single 60 gig maxtor. I ran this on my opty 170@2550.

Would IO make that much difference ?

 

[stevty2889]

Originally posted by: Markfw900
Yes, I am in divx, but stuck in the compressability test, using no cpu !!! Whats up ?

waiting on an accept message. Have to start over...

Well, I can;t see the autogk window anymore, there is a virual dub window, and I;m not at 100% cpu, so must be io bound. A single 60 gig maxtor. I ran this on my opty 170@2550.

Would IO make that much difference ?

I/O seems to make a pretty big differance. It took a lot longer when I was using an older PATA ATA100 7200rpm 8mb cache drive that with my 160gb 3g, or 250 3G 16mb cache drives, and raid-0 shaved off a few minutes as well. Gonna download the newer version of AutoGK now.

 

[Markfw]

Well, it finished in one hour 31 minutes, but it was only using ~85% cpu. I will put in my 200 gig maxtor pata 133, and wirte to that, and tead from the pata 133 60 gig. Maybe that will speed it up. (tomorrow)