My question below was triggered by a comment in this forum thread that "rendertest" was rendered in 30-40 seconds on an i7 machine.

My PC has Dual Xeon L5640 chips. Each chip has 6 "real cores" and 6 "hyper threaded" cores, which means 24 logical cores (threads) are available to Vegas. In addition, the machine has 24 GB DDR3 RAM, 12GB for each CPU.

The rendertest project runs in 125 seconds on my machine. When I hear a single i7 machine runs rendertest in 30-40 seconds, I question the results. Either my PC is setup incorrectly, I am rendering with wrong setting, or there is something about the i7 I am missing.

VP10 has a 16 thread limit which may mean my 24 cores are not all working, however, when I look at the Performance Monitor in Windows 7 64bit Professional, all 24 threads are busy during rendering.

Forgive me, but I just don't understand how an i7 can be 400% faster than a dual Xeon machine with 24 logical cores. I don't understand how the 30-40 second rendertest was performed and why it is so much faster.

Other factor besides threads are relevant such as HD read/write speed, but no matter how I look at it, I just find a 400% difference difficult to fathom. Please help me understand.

thanks,
Rich

30 - 40sec?

where / who?

my i7/920/6gb times at 268secs - which is plenty fast enough for price/speed ratio.....

ushere,

Regarding your question where did 30-40 seconds come from, I refer you to the initial post with contains the following in the second sentence ... "Recently, the render times have been in the 30-40 second range using the latest Intel i7 processors".

Rich

Rich, the way i read it, he's refering to the old rendertest not the latest one.
And the newest one is and i quote "The new RENDERTEST-2010.VEG should take four times longer".
You're getting it in 125 seconds? - thats excellent - twice as fast as my i7

Paul

i7 950 @3,9GHz
System and Vegas on SSD
Seven 64 Pro

VP9.0e 64 bits -> 171s !

Paul W,

Opps, I misread the comment about 30-40 second renders. Thanks for correction.

Rich

Rich,

As you were told, it was the old test. On the old one, you should be in the 27 sec range with your box.

Your box is only a few seconds behind mine and nobody has posted anything faster. Your machine is doing its job very well and will take a licking and keep on.... oh you know the rest.

5:46 or 4X the prior render test with 350mb of preview ram (default)
4:58 with 2gb of preview ram.
4:56 with 4gb of preview ram

Phenom II X6 with 8gb DDR3 dual-channel ram and dedicated Sata 3 7,200 rpm drive with 32mb cache.
No overclocking and no CUDAS..... yet:-)

P.S. The poster who flagged the need to restart Vegas after renders and Preview Ram changes was right on the money, thx,

I participated in a discussion over at DVinfo.net regarding render times and preview RAM ..

Like others, I found that increasing my preview RAM for this test decreased my time significantly, as much as 20%.

Thinking this was an across the board improvement, I had been leaving my preview RAM at 1g ..

However, after questions were raised, I found that actual rendering with .avi and .m2t media to .mpg ( DVDA video stream ), raising the preview RAM above the default, slowed my render times significantly; as much as 20%. ( with 10 minute test clips )

So, just an FYI for anyone interested..

"i7 950 @3,9GHz

I've been looking at an i7 950 to replace my Q9450, but have decided to focus on other aspects of the workflow and table the system upgrade for a while. A two to one improvement in render times doesn't seem to justify the forklift upgrade; motherboard, CPU, memory and complete rebuild of OS and applications. I did replace that nvidea 7600GS with an EVGA GTS450 to go with Vegas Pro 10.0a. That Dell U2410 is bubbling up as the next place to spend money.

John, what kind of speed increase do you see now with your new video card compared rendering using GPU vs your CPU?

I have been doing only MPEG since I installed the card. The Windows Experience number for video went from 4.6 to 7.1. The preview window seems a little more responsive but I haven't worked with AVC, yet. Perhaps this weekend.

Edit: 2010-12-23 _ I rendered 59 seconds of AVCHD (1440x1080-60i, 5.1 sound) from my son's Sony SDR-SR1 camera to the Sony AVC HD 1440-60i template.

Vegas Pro 9.0e (64 bit) ............................ 259 seconds
Vegas Pro 10.0a (64 bit) .......................... 274 seconds
Vegas Pro 10.0a (64 bit) (with CUDA) ...... 207 seconds

By my Microsoft calculator that's about a 20% improvement over the Vegas Pro 9.0e performance.

@ Stringer " regarding render times and preview RAM "

Right - as discussed above here also (in august) you can improve your render time by raising preview RAM.
When you do that, however, you should look at the CPU utilization in the Windows Task Manager. As you raise the preview RAM the CPU utilization will go up - possibly to near 100%.
Raising the preview RAM further will not improve render time - it will only block an area in your real RAM which might decrease the render times again.

The optimal setting for preview RAM is different from machine to machine and depending on CPU, total RAM etc. - probably even depending on the in/output video formats you use.

Here I actually use a low preview RAM to choke the rendering down to a CPU utilization of some 85-90 percent - it gives me a better response if I do some editing while rendering another job - without changing that jobs priority.

My 1090T AMD X6 with 8GIG took 251 secs to render the 2010 test at the specified settings. My preview was set to 1024 and my render threads was set to 6. CPU utilization went to 98%.

A couple of notes: the wep page with the results needs some adjustment to display correct. The table is not rendering correctly.

I started looking at this because I noticed with a nested project a similar render settings would only go to about 28% cpu utilization which I find bizarre. And this was no matter what I set the preview ram or threads at. Even at one render thread it was exactly the same. I did close down and restart Vegas each time after making the changes. It also made no difference whether it was 32 or 64 bit versions. Seems to be a real performance bottleneck with nested projects that destroys performance - comments anyone? It almost seems to be single threaded with a nested project.

189 s.
Vegas 10.0b 64-bit
Windows 7 Ult 64-bit
i7-965 3.4 GHz 6 GB DDR3-1800 MHz
Rendered to SSD.

Just FWIW, on my new laptop (i7-740QM), in HDV I'm getting 471 secs...

PS Interestingly, the CPU-only AVCHD render takes 480 secs ,so is not that much faster on VP 10 as it is on VP 9 (at least on my desktop, QX6700 machine). Engaging the laptop's GT 435M's GPU actually slows it marginally down to 474 seconds.

Piotr

NEW SYSTEM:

Gigabyte X58A-UD5 motherboard
12GB DDR3 RAM
Win7-64bit
Intel 980X 3.33Ghz 6-core CPU

Rendertest-2010 = 150 seconds
Original Rendertest-HDV = 35 seconds

Editing 24 Mbps 1920x1080 AVCHD from my NX5 is absolutely "butter smooth."

New system:

Asus P8P67 Deluxe with 3.0 Gb/s C disk on the 6.0 Gb/s port
i7 2600 (not K)
16GB DDR3 RAM
Win7-64bit

Vegas 10c
Rendertest-2010 = 163 seconds rendering from my C drive to my C drive.
Same results rendering from my Rosewill RSV-S4-X 4 Bay SATA to eSATA pcie Port Multiplier to my C drive. (Very happy about that, these units has been extremely reliable and a cheap alternative for expansion)


Overclock: i7 2600 (not K) to +27% (4362MHz)
Rendertest-2010 = 141 seconds rendering from my C drive to my C drive.

I was surprised that you can OC a i7 2600 , I could of pushed it one more step but I didn't since I don't have proper ventilation for it.

Old quad 6700
6GB DDR2 RAM
Win7-64bit
Vegas 10c
Rendertest-2010 = 395 seconds

Very happy with those results and the cost building the unit!

Only problem is that I will need to get the motherboard replaced due to the Intel recall.

Rob

i7-970 @ 4171 mHz:

Vegas 9c, 32-bit: 5:57
Vegas 9e, 64-bit: 1:49
Vegas 10c, 64-bit: 1:48

done with several other programs open, including Linux (Ubuntu) running in VMware virtual machine with Firefox - which I'm using as I type). Additional open programs seem to have little effect with this 6-core CPU.

In the past have pushed it past 4.3 gHz, but sometimes hit 73 degrees on one or another core momentarily. That's probably OK from what I've read (foolhardy overclockers often report temps of 80+), but slow degradation could be a problem for the CPU/motherboard at combination of high voltage/high temp.

I would have chosen a 2600k as it is cheaper if only it had more PCI lanes.

I see that the results pages at the website have been hijacked and corrupted. My system is Win 7 64 Bit, Vegas 10d 64 bit, ASUS P8P67 mobo, 16GB RAM, SSD boot drive, 11 TB of storage on spinning drives - mostly SATAIII, i7 2600K overclocked to 4.4GHz, GTX460 2GB graphics.

My source files are on one spinning SATA III drive and I render to a separate SATA drive. Temp files are on a separate drive again.

My results were:
155 seconds with 256MB preview RAM
125 seconds with 1.5GB preview RAM (any more RAM made no difference.)

These results seem to be in line with what others are getting with i7 2600 CPUs. I did overclock to 5.1 GHz but wasn't happy with the CPU temp. At 4.4 GHz it sits at 60 to 65 degrees Celsius.

Cheers

Russ

I looked at the database and I deleted a few nonsense records, but everything looks pretty good.

Here's some interesting data from the database:

Keith Kolbo still holds the record for the fastest rendertime of 104-seconds. Following closely at 114-seconds is Gregory Gatz. One thing that stands out is that the Intel i7-980X is predominate in the fastest 30 PC's with the notable exception of number-three in the list at 125-seconds is Rich Parry with an BYO Dual Xeon system and two MacPros with quad Xeons. Overclocking seems to help, but only about half of the top 30 are overclocked, including the top two performers, followed by the next two which are not overclocked. In the slowest-30, only three are overclocked including the third-slowest (at 1126-seconds on a BYO Pentium D - the only Pentium PC in the database).

All of the 30 fastest results are running Windows 7, 64-bit. Most are running Vegas Version 10, a few Version 9, and one, in the second-fastest slot, Gregory Gatz is using Version 8.1

In the fastest 30, there are two Mac Pro's (152- and 191-seconds), three HP/Compaq's (159-, 163- and 180-seconds), one toshiba (the fastest factory PC with 125-seconds), one Asus (number 30 at 212-seconds), one ASRock(? never heard of them, but a respectable 166-seconds).

For the BYO's, various ASUS boards predominate with only four gigabyte boards in the top 30. Of the ASUS boards, none stand out, but Keith is using the Asus P6X58D Premium motherboard.

Only one PC in the fastest 30 has a single hard-disk drive. Rob ? clocks in at 163- and 141-seconds (apparently before and after overclocking).

Looking at the other end, the slowest renders, doesn't reveal much of a pattern. I still see a few i7's, mostly Vegas Version 9. Here's something curious. There were no Dell computers in the fastest 30, and there are four in the slowest 30. Also, there's more "factory-built" than BYO's in the slowest 30 PCs.

Steve Mann

I continue to be a little skeptical of some of the results. In particular, there is a huge difference between rendering with the "best" and "good" setting. It has been a long time since I installed Vegas 10.0, so I don't know if the installation still defaults to "good" (it used to in earlier versions). I have a pretty darned fast (3.2 GHz) i7, and can't get even close to the top numbers (3x the render time) when using Best, but can get very close using Good.

My only reason for posting is that those who do this test and don't get a result that is close to the top marks might not have to feel too disappointed. I know that when the first render test came out, I initially had a fantastic time, but then realized I'd grabbed the wrong render template.

I think it is absolutely great to have these render tests, and we should keep doing them, but unless they are all done by one person or team, operating with some sort of cross check, we cannot be certain of the results. I am trying to think of a way to create some sort of way to either check or to ensure all the settings, but I'm not sure how to get all the settings which affect render speed (including preview RAM) into some standard state, except perhaps by creating a REG file that would modify the registry (after first saving the existing state) to some set of standard settings.