Rendering threads, page file size question

MTuggy wrote on 3/31/2009, 10:17 PM
As one of those with memory issues leading to crashes or lockup with HD video, I have uncovered another "solution" but I am curious about others experience. I am running 8.0c on Vista 64 Business system with 8 GB of RAM on an Intel 6600 Quad. Recently had to reinstall all Vegas software trying to fix a bug in 8.1 that still renders it useless but 8.0c is working fine.

Anyway, I discovered that if I setup Windows page file to 12 GB (1.5 x the RAM) that helped with rendering stability. BUT if I lower the rendering threads from 4 to 2 the rendering speed TRIPLED. Can anyone explain that to me? I am thrilled that I can again render 1920x1080 WMV files again (used to crash all the time), but I sure wish I understood the memory usage and how to tweak it better.

Mike

Comments

megabit wrote on 4/1/2009, 1:15 AM
I really don't have a clue what you guys have wrong with your Vegas installations, but I'll repeat that on mine (almost identical to MTuggy's - Vista x64, QX6700 quad, 8GB RAM) setting the max number of rendering threads to 2 instead of 4 almost exactly DOUBLES rendering times (with CPU usage dropping down from 100% to roughly 50-60%).

AMD TR 2990WX CPU | MSI X399 CARBON AC | 64GB RAM@XMP2933  | 2x RTX 2080Ti GPU | 4x 3TB WD Black RAID0 media drive | 3x 1TB NVMe RAID0 cache drive | SSD SATA system drive | AX1600i PSU | Decklink 12G Extreme | Samsung UHD reference monitor (calibrated)

farss wrote on 4/1/2009, 4:25 AM
You really need to compare apples to apples.
Decoding source video can use up a lot of the CPU time as can a zillion other things influence what is required of the code and CPUs.
Even variations in content can have an impact with some codecs.

Bob.
blink3times wrote on 4/1/2009, 5:39 AM
"I really don't have a clue what you guys have wrong with your Vegas installations,"

Agreed.

no offense Mike I don't devalue your circumstance at all and I know your probably a bit frustrated (as well as a few other here).... but it always seems to be the same relatively small handful of people complaining about these "memory handling issues".

Either you guys are pushing Vegas to depths that are far beyond what others do (which I don't believe) or there is SOMETHING uniquely wrong with your install/machine/hardware.

For the record though, I run fine with no page file at all and have for at least a year now.
megabit wrote on 4/1/2009, 5:57 AM
FWIW, I also tried no page file and everything was stable - but since I'm using more than default RAM preview buffer, as well as usually run more than a single instance of Vegas, I set up a page file of the fixed size as recommended by Vista.

AMD TR 2990WX CPU | MSI X399 CARBON AC | 64GB RAM@XMP2933  | 2x RTX 2080Ti GPU | 4x 3TB WD Black RAID0 media drive | 3x 1TB NVMe RAID0 cache drive | SSD SATA system drive | AX1600i PSU | Decklink 12G Extreme | Samsung UHD reference monitor (calibrated)

rmack350 wrote on 4/1/2009, 9:23 AM
Just to clarify,

Did the rendering SPEED triple (3 minute render becomes 1 minute)

or did the rendering DURATION triple (3 minute render comes 9 minutes)

Rob Mack
MTuggy wrote on 4/1/2009, 7:50 PM
Clarification:
The rendering speed tripled when I reduced the number of rendering threads down to 2. (shorter duration).

As far as my system, I am not running anything terribly unusual - an MSI motherboard, a 1GB ASUS video card, hyper-X DDR3 RAM. I am working only in HD. I never have any issues rendering to DV so if most users are rendering to DV or working with DV video files they probably aren't going see issues until they output to full HD file formats (unless we can figure out the optimal settings to avoid locking up Vegas).

I have yet to try no page file - sounds intriguing and worth testing. I was hoping with the number a people that have had problems we might get a sense of what the optimal settings are within Vegas but also within the Vista OS.
blink3times wrote on 4/1/2009, 8:33 PM
Mike:

I'm just sort of grabbing at straws here but if you get a chance... go into your bios and slow your ram down a little and see of that has any effect on crashing
farss wrote on 4/1/2009, 10:40 PM
To the best of my knowledge there's simply no advantage to having no page file. You may well run OK without one, Windows (or any similar OS) will only use the page file when it runs out of physical RAM. The assumption is if that doesn't happen then OK, you don't need one. However according to one technical article I've read there is a scenario where Windows will resort to using the page file for its own internal purposes even though there is physical RAM free and if there isn't a page file things get ugly.

Bob.
Andy E wrote on 4/2/2009, 12:27 AM
There is conflicting information on this. I've read several articles stating that removing the pagefile gives no performance gain and may cause problems.

However it is true that it will get used even when not needed. Here's a quote from Microsoft Technet:

The system is always using the pagefile even if there is enough RAM to support all of the applications.

One thing you can do is use PerfMon and check some of the memory counters.

You can also set a registry entry to prevent memory occupied by kernel and driver files being swapped out to the pagefile (which may well be where things get ugly).

My own anecdotal evidence is that 9 months on Windows XP 32-bit with 4Gb RAM and no pagefile is 0 problems and a subjective impression of improved performance.
blink3times wrote on 4/2/2009, 2:49 AM
I've been running with no page file for well over a year now with no issues what so ever. This is not just Vegas either... but with every other program I have as well.

Windows 7 BTW defaulted with no pagefile when I installed on my system.
blink3times wrote on 4/2/2009, 2:57 AM
"Windows will resort to using the page file for its own internal purposes even though there is physical RAM free and if there isn't a page file things get ugly."

And how ugly do things get when you run out of page file?

My experience when you rung out of memory (page or otherwise) is that you get a popup saying to the effect "Sorry, not enough memory to complete your request" This BTW has happened to me maybe once in 10 years.

It's absolutely true that pagefile gets used.... even when you have empty physical ram. The only way to stop using pagefile is run without one.
farss wrote on 4/2/2009, 5:53 AM
OK, but what's the advantage to not having one and are you 100% certain you don't actually have one?

If an OS has to dump a code page, no big drama, it can reload it. What about a pages full of data, e.g. buffered video. Do the maths, HD can gobble up a heck of a lot of RAM. By my quick and dirty calcs and assumptions to hold a 15 frame GOP's worth of 32bit video you need 370MB of RAM. A few tracks of that and I see a problem looming. With multiple cores running in parallel I haven't a clue what could / does happen. Suffice to say that all the poeple I've spoken with who really push the envelope with video are running 16 or 32GB of ECC RAM.

I'm only running 2GB at the moment, big mistake I know. AE hangs in there but only just. Once Win 7 gets sorted I'll upgrade to around that amount of RAM, subject to budget :)

Bob.
Andy E wrote on 4/2/2009, 6:03 AM
OK, but what's the advantage to not having one and are you 100% certain you don't actually have one?

Speed. If the OS is never going to disk for any memory pages, I've got no speed hit while it does.

I'm 100% certain I don't have one - no pagefil.sys file, no page file.
cliff_622 wrote on 4/2/2009, 7:33 AM
Blink;

.... but it always seems to be the same relatively small handful of people complaining about these "memory handling issues".

I have no factual number to back this up; But I'd venture to say that Vaegs crashing because of memory errors is THE most widely complained bug that I have ever seen on these boards. (if not that, than at least top 5 greatest Vegas bugs) The biggest competator might be the dreaded "red frame" issue.

Other than that, can anybody think of a more widely reported "bug"?

CT
rmack350 wrote on 4/2/2009, 11:33 AM
I guess I'd take it at face value that some codecs work better on 1 or 2 cores than 4. It's not beyond imagination.

The question of how to optimize Vegas for XYZ starts to get especially tricky if some things run better on 4 cores, some on 2 cores, and some on 1 core. It means you can't optimize for one thing without a detrimental effect on something else.

So, for what you're doing you get a threefold increase in performance by limiting Vegas' render threads to two, and Megabit gets a big decrease in performance because what he's doing can use more render threads. The conclusion is that there's a limit to how much you can tweak things.
LReavis wrote on 4/2/2009, 11:39 AM
I don't think it's memory errors - I've run memtest several times (sometimes over the weekend) on both machines (Q6600 & P4) and have never seen an error. Moreover, I recently swapped out my 4 half-gig memory pieces on the Q6600 machine in favor of 4 1-gig pieces, with no change in the pattern of crashes. Incidentally, the two machines are entirely different thoughout - different type of memory sticks, different video & sound cards, etc.; but still the exact same pattern of crashes on the very same projects and sets of files (I either switch the SATA cables to the hard disks or else work from the backup USB disks re-connected to the P4).
cliff_622 wrote on 4/2/2009, 12:28 PM
Your memory is fine. Dont drive yourself crazy trying to "fix" your PC, it's not that.

Just drop you cores in your BIOS to 2 and cut your Vegas rendering threads to 1 in your options. Make sure that any still graphics inside any track are downsized to as small as they can be. dont toss in a 6 megapixel .jpg or anything like that.

You will get it rendered, you just need to fight and wrestle with it to do it.

CT
blink3times wrote on 4/2/2009, 12:52 PM
" OK, but what's the advantage to not having one and are you 100% certain you don't actually have one?"

Well Bob I assume that "no page file" written for all drives means just that ....0 page file. not withstanding my defragger which maps out the drives graphically show no page file on all drive (it shows one when you are running page files)
blink3times wrote on 4/2/2009, 12:55 PM
"I don't think it's memory errors - I've run memtest several times"

I'm not looking to see if there are memory errors with slower ram. I'm more interested in seeing if it has the same effect as dropping cores, that's all.
MTuggy wrote on 4/2/2009, 1:04 PM
OK, I can now say my system and Vegas 8.0 and 8.1 are happily married once again. No crashes, no lockups with lots of HD footage with video FX added being rendered into every HD output format now for the last couple days without a hitch. I am in rendering Nirvana...

Ahhh... the wonder of 12 GB page files, 2 rendering threads and a bit of random luck.

:)

Mike
blink3times wrote on 4/3/2009, 3:15 AM
"Other than that, can anybody think of a more widely reported "bug"?"

Prove to me it's a bug.

A bug is something that is reproducible by all. If you import 5.1 sound to 8c you will lose the lfe track. This is steadfast reproducible by all others... it's a bug.

If you import 5.1 sound to 8.1 none of the panning will be not correct. This is steadfast reproducible by all others. It's a bug.

Sorry Cliff... I can not reproduce your crashing.

Now maybe this is a bug when running x memory at y speed with z cpu.... maybe it's a bug with x transition on the time line and rendering to y format at the same time. Maybe it's a bug when you have x program installed on your machine along with Vegas... but until such time as others can reliably reproduce under the same circumstances then it's not a bug. It's merely an irritating problem that is unique to your machine.

As I said... I have problems running DVDa and neo scene on the same OS. As far as I know others are NOT have these issues except for one other person. This is not a bug.... it's something unique with my machine.
Christian de Godzinsky wrote on 4/3/2009, 5:20 AM
Hi,

I don't get it blink?! My project renders fine to SD in Vegas 8.1 on two different computers (both running Vista 64 bit). This same project renders fine to SD in 8.0c on the same two different computers. No crashing - ever.

Trying to render the same project out in HD crashes 8.0c on both computers. 8.1 still works ok on both computers. Reducing the number of cores (from 4 down) solves the issue. Only Vegas 8.0c has this behavior/bug/error/shortcoming/issue. All other software that is capable of running on 4 cores works ok on both computers perfectly. I have NO issues with memory or overheating, non whatsoever. Do you still think seriously that its not an SCS issue?

The only common nominator is that both computers have the same QX9650 intel CPU running Vista 64 Ultimate.

Is this a bug in Vegas 8.0c or not? OK - I know that officially Vegas 8.0x is NOT supported on Vista 64, but so is a bunch of other softtware that runs ok. Othervise also Vegas 8.0c runs perfectly on Vista 64.

I would call this a bug (!) in Vegas 8.0c, seriously....

Christian

PS: I'm quite confident that this will be fixed in either VP9 or 8.0d... Probably it is not SCS to blame direclty (might be originally an MS issue), but only SCS can fix their own code to be compliant with the MS-OS... Independently. it is a sw-issue, not hw.

WIN10 Pro 64-bit | Version 1903 | OS build 18362.535 | Studio 16.1.2 | Vegas Pro 17 b387
CPU i9-7940C 14-core @4.4GHz | 64GB DDR4@XMP3600 | ASUS X299M1
GPU 2 x GTX1080Ti (2x11G GBDDR) | 442.19 nVidia driver | Intensity Pro 4K (BlackMagic)
4x Spyder calibrated monitors (1x4K, 1xUHD, 2xHD)
SSD 500GB system | 2x1TB HD | Internal 4x1TB HD's @RAID10 | Raid1 HDD array via 1Gb ethernet
Steinberg UR2 USB audio Interface (24bit/192kHz)
ShuttlePro2 controller

blink3times wrote on 4/3/2009, 5:38 AM
"Only Vegas 8.0c has this behavior/bug/error/shortcoming/issue. "

It does?
Then I along with many others must be doing something wrong because we're not crashing. I can run HD on all 4 cores just fine. I have stated this as well as others and it KEEPS getting ignored. People are not interested in the fact that 8c IS WORKING fine for others

What I'm saying here is that this "behavior/bug/error/shortcoming/issue" is occurring on YOUR machine. Now... find out what is different and we may have an answer.... but until then....
farss wrote on 4/3/2009, 6:07 AM
Seeing as how SCS can reproduce these problems your protestations are kind of questionable. The term "agent provocateur" regularly comes to mind when I read most of your posts on this subject.

Bob.