VP11 revisisted (+)

Former user wrote on 1/14/2012, 3:42 PM
My experience with VP11 (so far) is that I've had as more problems than many and fewer than some. (note: for mission critical projects (I'm still using VP10).

That said, the system I'm using now which is an ASUS P8x68 v-pro gen3, i7 2600 with 16GB of RAM on Win7 Ultimate has been reasonably stable, but not as stable as I'm used to.

The system has been fine, but very occasionally I was experiencing spontaneous, random reboot. Whatever I was doing, the system would just power down (and not a shut down - just suddenly shut off - and then power back on and reboot).

While disconcerting, it wasn't bad enough to really make me rethink what I was doing.

It finally occurred to me that I might have a power supply issue. I replaced the power supply, which was marginal considering the GTX590 video card. The shut down problem persisted.

I finally had a pow wow with a couple of my geek friends and tried to think back the last time I had the problem. One particularly clever fellow (a network admin for a major publishing firm) said, "replace your memory."

I was reticent to replace my memory because memory is expens...doh. Right. $99 for 16GB of premium RAM. Not a big deal.

SInce then I've noticed a SIGNIFICANT improvement with Vegas and a couple of other apps. The render times are MUCH better.

I did actually put the "old" memory (G.Skill) onto a test system, and lo, memtest86 (4.0a) found a TON of problems. The new memory has no such problems and the entire system is perceptibly snappier, and clearly more stable.

I'm just putting it out there - but because there are MYRIAD system configs with nearly as many types of memory out there, I'm wondering if Vegas (and other memory intensive aps) are more sensitive to errors in memory location etc (since the information is being moved around VERY quickly.

Anyway, I just rendered a 4 min clip (straight out of the EX1 in XDCAM) into a reference clip with a timeline (MP4). It took 1:05 to render. I put the old memory back in...4:49 to render.

Mere coincidence?

ps: I'm still expecting occasional instability, since the change is very new, but I'm only reporting what I can see, which is a vast improvement in performance.

pps: the new memory is Corsair Vengeance-LP, the old memory is G.Kill RipjawsX

Comments

farss wrote on 1/14/2012, 4:18 PM
"I'm wondering if Vegas (and other memory intensive aps) are more sensitive to errors in memory location etc (since the information is being moved around VERY quickly."

NLE's use more RAM than just about any other kind of app therefore it makes sense that they are going to be the most likely to "find" faulty memory. It's not so much how quickly the data is being moved, it's the amount of data being stored.

Bob.


john_dennis wrote on 1/14/2012, 4:25 PM
The telling thing about your experience is that you were apparently experiencing a hardware failure (memory error) and the only indication you saw was an application failure or poor performance except for an occasional reboot. In the bad old days before everyone on the earth had a computer, systems had hardware failure detection built into the hardware. Aside from making the repair of systems easier or possible, it made it possible to separate hardware failures from OS or problem program failures.

With the personal computer much of that has disappeared. "What's a reboot among friends?" seems to be the operative way of thinking. You have likely discovered the fallacy of that philosophy. Hardware is quite reliable today but it's not perfect and it's all likely to fail over time.

Until a few systems ago I ran ECC memory on every desktop system that I built. Bob (farss) has advocated for ECC memory for multi-socket systems. I'm not sure if the economics of the extra cost of error checking makes sense for the larger population of personal computer users, but I supect that many of the software complaints that people have may actually be from hardware failures. No one knows since the ability to detect hardware failures is limited to none for most of the systems people are running. I once said that if your money or your life depends on it you should be running ECC. For critical applications, these days, servers have it and they are clustered so if one server fails another one takes over the task.
ritsmer wrote on 1/14/2012, 4:31 PM
Right - and yet another reason for using a motherboard with ECC (Error Correction Codes) memory when building a new computer for editing...

Former user wrote on 1/15/2012, 2:24 PM
ritsmer...I actually asked at my local geek shop for ECC memory...they don't even carry any. ?!?! I've been building my own systems since my second computer (the only pre-built system I own (save for laptops) was my very first. I always used to use ECC memory, and now it just doesn't seem as necessary.

The spontaneous and random power-downs were definitely an indication that something wasn't quite right, but what I find really interesting is now, now having good memory (the G.Skill made it through the first pass of memtest86, but more probing tests revealed the failure of TWO sticks of memory as it turns out). The Corsair RAM is apparently flawless (a full night of testing - 16 passes, and it's still all good).

I guess what's most surprising is how Vegas very clearly is working much more smoothly with the new RAM. Renders are obviously so much faster (I actually opened a support ticket about slow render times - I've now pulled that ticket to say it's definitely been resolved). With the "other" RAM I was seeing render times of 1:1 (a 4 minute video was taking 4 minutes to render). I'm now seeing 4:1 (4 minutes of the same video takes 1 minute to render).

I guess there's something to be said for a spec box (one recommended by SCS). I'm going to check to see if the real-time renders are behaving any differently now (since I had client meetings all weekend - fun).

Next mobo and RAM I spec will definitely be thoroughly tested for RAM integrity (and ideally ECC if I can swing it).

The problem these days seems to be that all the high end hardware is for twitchy gamers who just care how hard they can over-clock heir system.
cybercom wrote on 1/15/2012, 9:23 PM
I had maddening reboots on my Intel MB until I replaced the Corsair RAM with a prohibitively expensive memory that was "Intel lab approved." Since then, the system has been rock solid. Cost me $500, but I'm not crashing anymore so it was worth it.

< ")%%%><(
SamAt wrote on 1/16/2012, 10:39 AM
There are actually options that don't require buying the most expensive memory. I buy memory based on the spec of the motherboard and the RAM (and ratings at New Egg which, given sufficient votes, convey some information about quality control on that unit) and how I plan to overclock. I buy matched and with heat spreaders, but not necessarily top of the line. Overclocking requires that you be careful about settings to avoid instability so it's common in the community to run stress tests to ensure the system is stable. All these tools are available for free, just search overclocking forums.

But you're on it - if your system spontaneously dies there are some top suspects: power and RAM as you found, and I would also include overheating - if the cpu core or, for example, a video card, overheats it can cause a reboot in self-defense. I always run Hardware Monitor and watch my core temperatures during long renders just to be sure.

Not trying to pontificate here, just help.

Sam
JJKizak wrote on 1/16/2012, 11:32 AM
Memory is a very sneeky thing that may or may not be compatible with the motherboard. Just a few tenths of a volt one way or another will mean working or not working.
JJK
Former user wrote on 1/22/2012, 8:04 PM
Here's the mind-blender (will it blend...I bet it would):

Second set of RAM...same problems. Two kits of 16GB of premium RAM with all the sticks bad? Hardly.

I've tested each 4GB stick of RAM individually...in each slot. EVERY time the system booted and ran.

This is the kicker: errors persisted in memtest86. This indicated one of two things EVERY stick of memory was bad OR EVERY slot on the motherboard was bad.

After consulting with a couple of net admin friends who are far geekier than me they came to a consensus: new CPUs have a RAM controller on board...they thought I probably had a CPU with a faulty RAM controller.

After a quick trip to the shop, I picked up an identical i7 2600k CPU. After more than 20 consecutive hours of memtest86, another 5 hours of stress tests, and 4 hours of overclocking the system...the result: zero errors in all tests. The old CPU is now back at the shop to verify that it's a dud, and then I'll get a refund and exchange for the new CPU.

Dead RAM controller. A new one on me (to be fair, there was also some talk of faulty cache on the CPU, but they thought that was somewhat more unlikely, but possible).

I'm not sure this is going to have any impact on Vegas 11 since I'm still seeing occasional crashes on render. That said, any improvement is good. I'm also heartened to read in one of the threads that another patch is coming soonish (on the heels of build 521).
rjwerth wrote on 1/24/2012, 2:43 PM
A bad CPU is possible, but I'd bet it is something even more simple than that. These new sockets Intel is using seem to be way more tricky than the "good ol' " pin sockets of old. VERY little surface area between the connection points makes a bad connection way more likely.

As far as memory these days, don't get me started. I have sent back SO MANY sets of memory in the last year, I'm sick of it. Some you could even test for hours w/no errors but run a memory intensive program or try and get the system to sleep/wake properly and you end up with the strangest symptoms of instability you've ever seen.
Former user wrote on 1/24/2012, 6:23 PM
I ran the new CPU and had 1-bit...ONE MEASLY BIT...come up with an error after 20 hours. What a laugh. It was more likely static or something.

But yeah, the new non-pin CPUs with low space tolerences, and mass produced boards make for lots of "outside the line" connections.

The entire thing drives me crazy. Not as crazy as VP11 mind you, but what can you do. I haven't migrated my project work (it's all tests of old old new VP10 projects so far), but I'm hoping this stuff gets sorted out.

In fact, before I looked into the PSU/memory/CPU issue, I was loading up PPro occasionally. I really REALLY do hate it. It's fine, but just so clunky. Why can't dragging two clips over each other just make for a crossfade? Three or four clicks later (more if you want to get it actually right), and you're done roughing out your transition. It's insane. If I used PPro I'd have to charge more because I'd be spending WAY more time editing.

As Roseanne Roseannadanna's mother used to say, "If it's not one thing..."

cybercom wrote on 1/26/2012, 9:11 AM
But at least you'd be editing, not crashing, rebooting, reloading and constantly saving and feeling nervous because you didn't trust the software...

< *)###><
zmobile wrote on 1/28/2012, 2:01 AM
Well this thread was sure worth reading.

I had done a major upgrade from XP Pro 32bit to Win 7 64bit. Everything seemed stable, so I went ahead and upgraded from Pro 8 to 11. And that's when the random crashes started all the way to BSOD. I though build 525 would fix it, nope. Went back to 411 and got by for a while.

Then I read this thread.

So, I got memtest86, found a bad 1 GB module in 20 minutes and replaced it. In fact, I replaced them all so, from 3 GB to 8 GB. So far, the machine is stable and appears to be a little faster, but too soon to tell and to late to take measurements. I still have to test 411 more, finish the current project, then try the new just released build.

In the old days, the "new" programs were the best memory and CPU testers. They "found" all the weak silicon, so to speak. I guess this law is still true.

So for those that have upgraded in a big jump especially, and you are scratching your collective heads... make sure your hardware really is passing all the tests and replace all the bits that are not. I never thought that would happen to me.

Total cost to fix crashing $130 USD, but could have done it for $39.

chuck z
zman Productions
Jeff9329 wrote on 2/6/2012, 4:36 PM
Chuck Z:

Are you still not crashing?

Im thinking about uprgading from VP8 to VP11 but I dont have time for any shenanigans from a bug laden program.

What is your current opinion of upgrading?

Thanks,
Jeff
VanLazarus wrote on 2/6/2012, 8:22 PM
I recently upgraded to high quality Corsair Dominator RAM (16gb), which the tech guy at the computer store said is some of the highest quality RAM you can buy. My system got more stable, but, unfortunately, not Vegas 11.