VideoHelp.com Video Benchmarking

[jagabo]

26 seconds is a very, very quick time indeed.

And that was the slowest of the Core 2 Duo processors on one of the worst possible motherboards!

I'll be benchmarking on the Asus P5W64 WS Professional using the E6400 and 2GB of DDR2 PC6400 (800MHz) RAM.

That should do significantly better than my E6300.

This test is actually poorly controlled since there are several TMPGEnc features that are not set by the MCF file. Note my earlier post:

https://forum.videohelp.com/viewtopic.php?p=1479223#1479223

Differences between uncontrolled TMPGEnc settings are greater than those between vastly different processors!

Just for kicks, I converted the file with CCE SP Trial Version, 2-pass, similar settings, C2D E6300: 7 seconds.

[Cobra]

We'll need to sort out those settings then. Are they stored in the registry of a machine so if you've used TMPGEnc before it will remember the settings? If so, we'll need to go through each setting and define what it is for the test to be fair. If not, then if users don't change anything when they unzip it we'd be OK.

[jagabo]

The VBR cache option is saved in TMPGENC.INI, in the same folder as the EXE.

[CPUOption]
EnableMultipassVBRCache=1
MaxMultipassVBRCacheSize=8589934592

You can't just include an INI file as part of the benchmark though, many other settings are saved there including SSE flags that are particular to different CPUs.

You set it within the program via

Option -> Environmental Setting -> CPU

Then enable "Saves analyzing result of multipass VBR to cache" and set the desired cache size.

The other options that can have an effect on encoding time are under the Option -> Preview Option, and can be found in the INI file as:

[Option]
PreviewFrequency=0
PreviewAutoResizeWindow=1
Config_Preview_Size.Width=320
Config_Preview_Size.Height=240

Fastest results are with "Do not display" selected (where the rest of the related options don't matter).

There are many other options in the Enivironmental Setting dialog that look like they will effect encoding speed but I haven't examined them closely to see how large an effect they have.

[rallynavvie]

Not having the preview enabled does indeed speed it up, I always have that silly thing disabled.

I'm thinking of changing the encoding benchmark to H.264 soon, unfortunately QT Pro isn't free.

[Cobra]

x264 is, though, and if there's a way to script Gordian Knot then you're well away!

[jagabo]

You could just use a batch file and the CLI version of x264. Although, I wonder if x264 (or any h.264 encoder) is really of interest that many people at this time?

[Cobra]

A fair few of my movies are in x264. I don't watch them any more because I'm back at home with the originals. I used to copy my favourite DVDs to DivX to take to uni with me (saved a fair bit on the insurance - doesn't matter if my backups are stolen) but switched to x264 when Gordian Knot started supporting it. It's an outstanding codec.

[Cobra]

1) Core 2 Duo E6300, 1 GB DDR 3200, TMPGEnc -> Option -> Environmental Setting -> CPU -> Saves analyzing result of multipass VBR to cache -> enabled: 26 seconds.

jagabo,

I couldn't find that option in the version of TMPGEnc we used for this benchmark (see the guide sticky in the Computer Forum). This is all I get:



I performed the benchmark under the conditions stated in the guide, and came up with 31 seconds:

Start: 20/10/2006 21:16:04
End: 20/10/2006 21:16:36
Dealing time: 00:00:31
-------------------------------------------------------------
C:\Documents and Settings\X\Desktop\DIVX_Archive_NTSC.m1v
Format: MPEG-1 720x480 29.97fps VBR 7000kbps
Video: C:\Documents and Settings\X\Desktop\VideoHelp Benchmark\DIVX_Archive_NTSC.avi
Audio: none

Result: Succeeded
Dealing time: 00:00:31
-------------------------------------------------------------

My system details are here:

cobra_new.htm

[jagabo]

It was TMGPEnc Plus 2.5. The About screen says:

Version 2.521.58.169
Core version 1.98.152



Multipass caching must be part of the "Plus".

I downloaded version 2.524 (the last of the free versions?) from the TMPGEnc.net site and the dialog had the same blank area as in your post. It rendered at the same speed as Plus with multipass caching disabled. Curiously, I noticed the free version's INI file had the same variables but enabling mulitpass caching there had no effect on rendering time.

[glockjs]

not too bad for a budget machine i guess :/($200 worth of hardware) and reading and writing to the same hd(Seagate Barracuda 7200.9 80gig)

[budz]

Intel 2 Core Duo processor overclocked as much as I can for now. I would need a higher watt power supply to overclock it more. Using a BIOSTAR 945P-A7A (8.0) LGA 775 Intel 945P ATX motherboard, COOLER MASTER RR-LCH-P9E1 92mm UFO CPU Cooler and 1gb Kingston DDR 667 RAM.

[Capmaster]

No log file popped up, so I did a screen grab when it was done. Strange, but only two processors show up for the MMX extensions (see bot. pic):

[Dr_Layne]

Amd 64x2 5200+

[Cobra]

Nice result! 8)

[Heywould3]

re ran test with tighter mem timings and turned off preview. went from 1:07 stock to .40 over clocked

[träskmannen]

I decided to try out my new system:



First using the version of TMPGEnc supposed to be used in this test.
Result: 25s.


Interesting that I get the same result as Dr_Layne - obviously there are more things than pure processor speed that influence this.

Then using TMPGEnc Plus 2.5 (just for fun - to see how big difference it made)
Result: 18s.

[/img]

[vhelp]

Good evening everyone.

I'd like to attempt another go at this, with my new XP Home sytem, (see my
new file spec) but I'm seeing different CPU_Test versions, and this is making
my attempts confusing and unsatisfactory. And, to add, on my older AMD
XP 1800+ computer, when I run v1.26 or ealier, the text readout is completely
distorted and you can't read it. So, runing CPU_Test on my XP 1800+ is no
longer working -- I wanted to compare output reports of both pc's.

--> What's the latest ver of tools (and ?? to D/L) for the testings/reporting ?

Thanks,
-vhelp 4288

[rhegedus]

18 seconds on a Q6600

cpuz.htm

[rallynavvie]

I'm thinking that TMPGEnc can only handle 2 logical CPUs. We need to find a better encoding bench for when I build my dual-quad later this year.

[rhegedus]

We need to find a better encoding bench for when I build my dual-quad later this year.

Which mobo are you planning to use?

[jagabo]

x264 is well multithreaded. Running from the command line (via a preconfigured batch file) also removes a lot of variables.

[rallynavvie]

Which mobo are you planning to use?

One that isn't out yet I'm sure. I'm waiting for the Penryns to come out.

[Bondiablo]

Intel Q6600 16s at stock speeds, 13s overclocked to 3GHz.

CPU-Z 1.45 report file
Processor(s)
Number of processors 1
Number of cores 4 per processor
Number of threads 4 per processor
Name Intel Core 2 Quad Q6600
Code Name Kentsfield
Specification Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz
Package Socket 775 LGA
Family/Model/Stepping 6.F.B
Extended Family/Model 6.F
Core Stepping G0
Technology 65 nm
Core Speed 2400.5 MHz
Multiplier x Bus speed 9.0 x 266.7 MHz
Rated Bus speed 1066.9 MHz
Stock frequency 2400 MHz
Instruction sets MMX, SSE, SSE2, SSE3, SSSE3, EM64T
L1 Data cache (per processor) 4 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache (per processor) 4 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache (per processor) 2 x 4096 KBytes, 16-way set associative, 64-byte line size

Chipset & Memory
Northbridge Intel P35/G33/G31 rev. A2
Southbridge Intel 82801IB (ICH9) rev. 02
Graphic Interface PCI-Express
PCI-E Link Width x16
PCI-E Max Link Width x16
Memory Type DDR2
Memory Size 2048 MBytes
Memory Frequency 400.1 MHz (2:3)
CAS# Latency (tCL) 5.0 clocks
RAS# to CAS# (tRCD) 5 clocks
RAS# Precharge (tRP) 5 clocks
Cycle Time (tRAS) 18 clocks
Command Rate (CR) 2T

System
System Manufacturer Gigabyte Technology Co., Ltd.
System Name P35-DS3L
System S/N
Mainboard Vendor Gigabyte Technology Co., Ltd.
Mainboard Model P35-DS3L
BIOS Vendor Award Software International, Inc.
BIOS Version F7
BIOS Date 11/29/2007
Memory SPD
Module 1 DDR2, PC2-6400 (400 MHz), 1024 MBytes, Crucial Technology
Module 2 DDR2, PC2-6400 (400 MHz), 1024 MBytes, Crucial Technology

Software
Windows Version Microsoft Windows XP Professional Service Pack 2 (Build 2600)
DirectX Version 9.0c

CPU-Z 1.45 report file
Processor(s)
Number of processors 1
Number of cores 4 per processor
Number of threads 4 per processor
Name Intel Core 2 Quad Q6600
Code Name Kentsfield
Specification Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz
Package Socket 775 LGA
Family/Model/Stepping 6.F.B
Extended Family/Model 6.F
Core Stepping G0
Technology 65 nm
Core Speed 2997.5 MHz
Multiplier x Bus speed 9.0 x 333.1 MHz
Rated Bus speed 1332.2 MHz
Stock frequency 2400 MHz
Instruction sets MMX, SSE, SSE2, SSE3, SSSE3, EM64T
L1 Data cache (per processor) 4 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache (per processor) 4 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache (per processor) 2 x 4096 KBytes, 16-way set associative, 64-byte line size

Chipset & Memory
Northbridge Intel P35/G33/G31 rev. A2
Southbridge Intel 82801IB (ICH9) rev. 02
Graphic Interface PCI-Express
PCI-E Link Width x16
PCI-E Max Link Width x16
Memory Type DDR2
Memory Size 2048 MBytes
Memory Frequency 333.1 MHz (1:1)
CAS# Latency (tCL) 4.0 clocks
RAS# to CAS# (tRCD) 4 clocks
RAS# Precharge (tRP) 4 clocks
Cycle Time (tRAS) 12 clocks
Command Rate (CR) 2T

System
System Manufacturer Gigabyte Technology Co., Ltd.
System Name P35-DS3L
System S/N
Mainboard Vendor Gigabyte Technology Co., Ltd.
Mainboard Model P35-DS3L
BIOS Vendor Award Software International, Inc.
BIOS Version F7
BIOS Date 11/29/2007
Memory SPD
Module 1 DDR2, PC2-6400 (400 MHz), 1024 MBytes, Crucial Technology
Module 2 DDR2, PC2-6400 (400 MHz), 1024 MBytes, Crucial Technology

Software
Windows Version Microsoft Windows XP Professional Service Pack 2 (Build 2600)
DirectX Version 9.0c

[rallynavvie]

I'm curious as to why those Q6600s are pegging such high times when compared to my quads. Granted this bench only hits two cores so a dual-core is going to be just as fast as a quad-core (or my 8).

[cyclometric]

I realize this thread seems to have slowed down but thought I'd post anyway, having just found this benchmark test.

I was pleased with the results... having started to OC my system at all yet, and am planning on upgrading my GPU & adding another 4 GB RAM in the next couple of weeks, so I'll run this again to see the difference.

Cheers,

[rallynavvie]

Yeah, it has. We need a new benchmarking thread because my system won't run this test properly anymore.

FYI the extra RAM is going to do zilch for you encoding speeds. OCing certainly will. Unless you're utilizing all your 4GB now there is no good reason to double it except for bragging rights.

[cyclometric]

Hi, rallynavvie,

I am running Vista x64 Home Premium SP1. I thought that in 64 bit OS's, more than 4 GB is not a problem. In fact at one point I did have 8 GB (4 x 2 GB chips) installed and the OS recognized all 8, though I couldn't tell you for sure whether the system was using the RAM effectively or not, but it definitely recognized all 8 GB.

Anyway, I am a bit dazed at the prospect of having to decide whether to:

• A. Overclock my CPU, Memory, and GPU the hard way, in BIOS, which is what I am restricted to with my Mobo (Asus M3A78-CM/780V Chipset);
  
  B. Upgrade the mobo to enable Overdrive and DDR3, etc. (this would also require me to replace my microATX form factor case, which I love (Thermaltake Lanbox) with a full ATX something... it would be a lot more work)
  
  C. Upgrade the GPU (leaning against that now, I think the HD3870 is pretty good for my purposes);
  
  and/or
  
  D. Double RAM from one dual channel @ 2x2GB @ 800mhz DDR2, to two dual-channels @ 2 channels, each composed to 2x2GB @ 800mhz DDR2
  *** OR ***
  Swap my current memory chips with a single dual-channel 2x2GB DDR2 @ 1066mhz. Another limitation of this board: installing 4 memory 1066 chips means only 1 1066 chip will run at that speed per channel, and if all 4 slots contain 1066 chips, they all 4 will run @ 800mhz, for "system stability", according to ASUS.

I've got a thread on this subject over @ Tom's Hardware, so hopefully will figure things out while my mobo is still under my 30-day hassle-free return timeline!

thanks,

[redwudz]

You can post a link to your Tom's Hardware thread if you like. RAM doesn't help with encoding, but it may help if you use programs that can cache files in RAM like Photoshop or a few others. Check your memory usage while encoding. It will probably be less than 3GB. That's why more RAM won't make much difference. A 64bit OS can address a lot more RAM, but that doesn't mean it will use it for anything.

With most AMD CPUs, you can just increase the FSB speed to overclock. It runs at 200Mhz stock. I ran my FSB up to 236Mhz for my 920 Phenom II to get it from 2.8Ghz to 3.3Ghz. With a unlocked 940 you can also change your multiplier to OC your CPU. Either way works, as long as you don't want to go too far. Much above 3.3Ghz, you may have to raise CPU/RAM/NB/SB voltages and adjust some multipliers before you really push it. The CPU is likely capable of 3.7Ghz or more, but it requires a fair amount of 'fine tuning'.

You really want to use BIOS settings for a OC. You can do it in Windows with the MBs OC programs, but you may not be able to get much in the way of system adjustments that way.

You may be able to OC your video card also, though I don't really recommend that. Video cards run very warm at stock settings, and OC'ing them may be a bit too much stress. But check the OC forums for video cards and see if it will accept some speed ups.

EDIT: Just my opinion, but Gigabyte MBs work better for me with AMD CPUs.

EDIT2: BTW, your CPU-Z screen shots show that your CPU is running at 800Mhz. You have installed some MB programs to change your CPU speed to keep the CPU running cooler. You will need to disable all that if you want to do some serious OCing.

[cyclometric]

You're right in that I had installed MB monitoring/adjustment programs to tinker with my CPU and other settings, but then I read up on my 780V chipset and learned that unless I unlocked the settings in BIOS, nothing would actually be changed. I started to get a clue that the entire system was prevented from Overclocking at the OS level when I was running AMD Overdrive and even though I had changed the Performance mode, from "0" to "3", the maximum speed the CPU ever ran at was right around 3000, never more. Then it dawned on me that it wasn't just the CPU that was locked, when I ran Catalyst Control Center and found that the ATI Overdrive functions to Overclock the GPU were entire disabled as well. There are BIOS settings for the CPU, GPU, NB, HT and Memory which I can unlock by changing from 'Auto', so i'm going to have to study up on just how to do all this, and be careful about it.

That's cool, maybe I won't even need more RAM... and though I had a Gigabyte GA-MA78GM-S2H on my last build of the same case, with an Athlon X2 5000+ BE, and liked it, there were issues (it ran hot, for one, but being cooped up in the tight Thermaltake Lanbox didn't help. When I bought the Phenom II X4 940 BE originally the guy @ Fry's sold me a GA-MA78GM-US2H (my board & PSU had been fried by 5 intense power surges in the course of a few hours) -- anyway, that board was very much not compatible with this CPU -- temps @ idle were in the low 50's! Just swapping boards to the ASUS my idle temps are a little high, but right around 38, much more friendly, and they have never gone about 57 even under full load for 2 hours. Maybe I had a dud. I have about a week left on my "30 days hassle free return" for the Mobo, so I might still upgrade to a full ATX board with the 790GX, FX or X chip, but ugh, that means a case upgrade, too, which would be a drag.

I would consider water cooling the CPU in my beloved Lanbox case if the OC temps get high before I achieve great results... I'm going to try to stick with this board, I think.

Anyway, the Tom's thread is at http://www.tomshardware.com/forum/264445-31-just-realized-board-help

As far as my CPU running at 800, no no, it was just coincidence that it was @ that speed when I took the first shot. the Speed is being controlled by the BIOS -- I took 2 more shots of CPUZ... one under normal conditions, another under full load.

Thanks so much for your help, guys!

[deadrats]

just to add another reference point, my E7400 overclocked to 3.33ghz encodes the combined clip in 25 seconds.