Caveats, caveats.

Preview RAM takes memory *away* from Vegas and Windows. It's a frame cache. Uncompressed frames get cached to RAM and make things on the timeline that'd need rendering play back smoothly. That memory can't be used for other rendering and processing purposes, so it's a trade-off. Sometimes it speeds things up, until it runs out and slows everything else down. All this is to say that you need to leave adequate RAM available for Vegas to actually do its job. At 64GB total, letting the cache consume half your RAM is no big deal, but as a rule of thumb, the more preview RAM you allocate, the less is left for actual operations.

That, and users who test this often say that renders are faster when preview RAM is set below 512 MB, sometimes even better set at zero.

So what about threads? I just looked at task manager and noticed that there are over a thousand threads allocated at this moment on my computer. Vegas is not running.

Would Vegas really need more than 16 threads? How much time would be spent managing threads versus actual rendering? Are Vegas and the two codecs you're probably using (a video codec and an audio codec) really *that* multithreaded? Would it make any sense to kill other processes to try to reduce the number of running threads Vegas has to compete with?

If you have 64 cores (I'm assuming thats what you mean) then I guess you're running something like a quad-processor setup or opterons. Or something similar. Maybe you'd be better off running 4 instances of Vegas and have them all doing renders at once.

Rob

Thank you for the info. This is my new computer specs:


2 Processors Intel(R) Xeon(R) CPU E5-2698 v3 @ 2.30GHz
Total amount of system memory 64.0 GB RAM
System type 64-bit operating system
Number of processor cores 64

Storage
--------------------------------------------------------------------------------

Total size of hard disk(s) 12668 GB
Disk partition (C:) 1210 GB Free (1490 GB Total)

Graphics
--------------------------------------------------------------------------------

Display adapter type AMD Radeon (TM) R9 390 Series
Total available graphics memory 12032 MB
Dedicated graphics memory 8192 MB
Dedicated system memory 0 MB
Shared system memory 3840 MB
Display adapter driver version 15.300.1025.1001
Primary monitor resolution 3840x2160
Secondary monitor resolution 3840x2160
DirectX version DirectX 10

[I]2 Processors Intel(R) Xeon(R) CPU E5-2698 v3 @ 2.30GHz[/I]

It has been stated many times in this forum that higher clock speeds are preferable over core counts. You would have been better off with 5960X on water running at 4+GHz.
You can try and increase the max. number of threads in the internal tab of preferences. Hold down the SHIFT key and open preferences. You should see an additional tab called "Internal". At the bottom you type in "threads" to find the setting faster. Change it, save it and restart Vegas.

Thank you for the suggestion but I have question. I did as you said and got into the Internal settings. There is not just one so please tell me what I should tweak. This is what I see:

Max Video Render threads Value 4 Default 4
Max Video Render threads Value 16 Default 16
Minimum Render threads Value 1 Default 1

Max for maximum video render Value 4 Default 4
Max for maximum video render Value 16 Default 16

Mxfxavc: Threads per video deco... value 0 Default 0

"Max Video Render threads Value 16 Default 16" is truncated in your view. The actual text is "Max Video Render threads (64-bit) Value 16 Default 16"

Change the value to something larger. However, there are probably some things to think about. Set up a test project of something you would normally work on. Time the render with several preview RAM settings. Do a render to ram to fill the buffer before doing a render. Basically set up several test scenarios.

How much are your cores being utilized before raising the thread count? Are they maxed out? Are some unused? Does this change after raising the threads setting? More importantly, do you get a faster render? Or is it slower? Speed is the only real benchmark here. Busier cores with slower renders aren't much good.

Also, do you know how to flush your preview RAM? A Vegas restart should do it but another method is to change your preview quality setting. Watch memory usage in Taskman to make sure Vegas gave up a bunch of memory.

@ OldSmoke

You Wrote: You would have been better off with 5960X on water running at 4+GHz.

Just a question, wouldn't one be much better off with a pair of something like 5820Ks at 4GHz for much less money? In my one use of a 5960 I found it extremely hot and not very impressive for the price. But, that may have just been the particular set-up.

[SWAG]

5820 and other desktop processors in the same class will not run in a dual socket configuration like Xeon.

[/SWAG]

Somebody check my work, but I think the ability to run multiple physical processors has been an intel market differentiator for a decade or more.

There are AMD server boards with two CPUs.

http://www.newegg.com/Product/Product.aspx?Item=N82E16813131643

And yes, Intel desktop processors are single CPU operation only.

But I do wonder how much additional performance a 5960X brings over a 5930K when it comes to Vegas and whether the price difference is justified.

Sorry, I didn't make myself very clear.

I was simply trying to say that it seems as if there is a definite decrease in bang-for-buck as you go up the processor line, then I pulled the wrong processor number from memory which made my comment meaningless.

I apologize for the mistake.

I did test with 32 and 64 threads in the Max Video Render Thread and all test were the same speed. In fact this computer which should is 4 times faster then my 6 year old computer takes longer to render old projects. I did 30 minute projects in 14 minutes and now this new computer is doing 20 minutes!
Are there any other tweaks I can make? How about my video card? I have an AMD.
Or is Vegas very limited on speeding up renders with very fast computers?

As I stated in my earlier post, it's not your GPU, the R9 390 is a good card, it's the CPUs that are too slow. You have plenty of threads but Vegas and the codecs don't make use of it and clock speed becomes more important.

Randini,
I am looking at getting a similar system to your but with the E5-2667 v3 chips @3.2Ghz (only 8 core vs. your 16 per chip).
I am very interested to find out what you discover as you try to tweak Vegas to get the best performance.

Our current system (ASUS Laptop running i7-4200HQ @2.4Ghz) is usable but not optimal. We have rendered up to 3 videos at once using it, but I am eager to see how this would work on a system like yours.
Perhaps a single video will not be a drastic difference, but being able to render one video (or more) while editing another with no loss of performance would be nice!

Keep us posted as to your results.
Good luck!

Andrew

ps - Have you tried any other video applications on your new system like Premiere, After Effects (older versions that still allow multi-frame rendering), etc? Have you noticed a performance difference with them?

Would the poster with the xeon board post .speccy snap shot of the machine? I would like to look at some of the details on the configuration.

As has been said in this post and other posts before. Threads are not CPU or HT CPUs. Windows regulates thread service, Vegas is simply presenting the threads that the programmers designed into the software. 16 threads are going to be serviced more often by 32 cores than 64 threads.

My understanding of the way video is handled by Vegas, I would look at:

Memory bandwidth - is it optimized for the motherboard, speed over quantity here. WinSAT mem from an admin command prompt will give memory speed. Does that result match the bandwidth you should be getting.

GPU GFLOPs - the 390x is good for the configuration, is the card placed in the best slot on the motherboard for the installed chipset, and is it running at full interface speed?

Disk latency - this is probably a lessor thing due to the amount of disk cache windows would be doing with 64GB of ram. SSD and NVME drives would be optimal due to the lower latency.

Codec choice - Sony codecs are more optimized in Vegas, also in terms of rendering. Run some tests with the project source files in XAVC-I, HDCAM, or XDCAM. Then try rendering to those same codecs 1:1 as well as Sony AVC with GPU profile in auto. There really was no mention of your workflow and the codec types you are using.

CPU speed - the Xeon chips installed is very impressive due to the expense and core counts, however, Vegas seems to be very time sensitive. Playback, effects, and composting that require heavy computations are slowed the most by clocks speeds all over the system. Sort of like an assembly line where one guy completes his task slower than the rest, the over all production line suffers. High core counts seems like a simple gauge, but computational speed is better. Sort of like engine cylinders and HP, actual top speed on the track is a better gauge of performance.

Multi-CPU - there is inherent latency in negotiating multiple CPUs, vs a single CPU with direct memory channels. 64GB on an x99 is going to be a different than 64GB on a multi CPU Xeon board. Big core count Xeon chips are designed more for virtual machines, and processing large amounts of SQL users for example.

Vegas does not work h.264/AVC material in AVC form. Vegas decompresses/unzips the video into uncompressed RGBA frames, then does all the effects and composites in that format. Very fat RGB information is tossed about the memory, CPU and back and forth to the GPU for results. That information is then conformed to the display standard, or sent to the encoder for compression and file writing. Your system has to do all these things in what we perceive as the same time, or be able to do this faster than 30-60fps. A machine not capable of faster than 30-60fps is a machine that drops frames.

If handbrake uses all the cores and does so with high utilization, then there is certainly a component that is holding back Vegas processing.

Hi Andrew,

I will definitely let you know if I discover a tweak. I was told by my egghead friend that built the computer that if he knew of the limitations of Vegas he would have built me a fast consumer computer instead. Vegas really limits you on renders. The big upside is my timeline action is great! When I would show my work inside Vegas before a render at Preview certain effects would skip frames but I'm not getting that.
No I have not tried other software for I have been using Vegas since 2.0.
Cheers,
Randy

Why when I do a render of Main Concept mp4 with openGPU it starts out very fast then about the 1/4 mark slows down?

The main drive is an Intel P3700 1.6TB

Hi Randini!

Any update?

It really looks like you have the best of everything in this machine.
We are also looking at the intel solid state drives for our OS so I am curious how well that is working for you.

Have you had that chance to run any render tests yet?

Andrew

Hi Andrew,

I just replaced the Mother Board (asrock) and both processors with one (Intel i7-596ox CPU 3 GHZ and Ram 32. Now my computer renders are twice as fast. My computer friend that built it said that having two processors was confusing Vegas causing slowness. I believe he also tweaked my processor to 4.0 Ghz but I have to find out that.
The main thing is I'm very happy!

[I]I just replaced the Mother Board (asrock) and both processors with one (Intel i7-5960X CPU 3 GHZ and Ram 32.[/I]

Good choice!

A water cold enhanced 4.5 GHZ processor lintel 5960x.

Since my scheduled system upgrade is due in October, I've been reading the spam from Newegg and Fry's more closely. I've been noticing adds for the 5930 and 5960 weekly of late. My uninformed perception is that there will be price pressure on these processors to keep the fabs running at full tilt now that the specs of the Broadwell-E are on the street. Whether I choose a bleeding edge processor or look for the best deal on current tech, my summer will probably be interesting.

Randini,

Congratz on the performance improvement!
I wonder if the double rendering speed is because you are now running a chip with double the clock speed.

I would love to see a comparison with a fast XEON chip - I know none are up to 4+Ghz, but at least something around 3.7 or so.

Has anyone heard of overclocking the XEON chips?
I thought I read about a motherboard that would allow you some flexibility, but it might just be wishful thinking.