>>> Is it generic or name brand? What CAS latency is the RAM? <<<
Dell RAM might be lower quality RAM that is more borderline than name brand RAM like crucial or whatever. However, RAM rarely fails and that's why most RAM manufacturers have lifetime warranties on them (Dell is 1 year warranty). RAM does fail sometimes... but really not often enough to care too much about it.

>>>Does it have a heat spreader? How many layers on the PCB? Is it dual channel or single channel? <<<
Heat spreaders are unnecessary. I don't know about PCB. You will have dual channel if you have two sticks of the same model RAM (any DDR PC3200 RAM can be dual channel). CAS latencies make no measurable difference in Vegas rendering speeds (I tried...).

>>> Example #2: 80GB? Who is the manufacturer? Is it one of the one-year warranty drives with 2MB cache or a 3-5 year warranty drive with 8MB cache? Are they smart enough to include an 80 conductor IDE cable so the drive can perform well? <<<
It's Dell so the warranty is 1 year. They don't actually say what the hard drive is... but they are mostly all good. 8MB cache would be nice. On WDs system/appications performance goes up 30% (mainly loading times I think). Dell I'm pretty sure uses a 80-conductor IDE ribbon.

>>> Example #3: Power supply? Who makes it? What is the maximum power draw on it? Does it have good voltage regulation? Is it an Intel- or AMD-approved power supply? How does it perform in an overcurrent situation? Is it the only fan blowing air out the back of the case? <<<
I haven't heard of Dell having problems with bad power supplies. Dell's case design is pretty good with good airflow.

>>> When you start to break this down and examine matters, you're not getting the super deal that you might think from Dell. Sure, it's a "2.8GHz" computer, but it's probably an i848 chipset vs. i865/875, 533MHz FSB versus 800MHz, single channel generic high latency RAM versus dual channel low latency with a heat slug, etc. etc. etc.. That's not the way you build a good stable editing workstation. <<<
The 4600 uses the 865 chipset, 533mhz FSB. The 8300 uses the 875, 800mhz FSB with a hyperthreading Pentium. The 8300 is better but the 4600 can also be very fast. The Dells run dual channel RAM to that's not a problem. I don't forsee any stability problems with a Dell (all the parts should work together) but then again I don't own one. It should be a solid editing station. It's cheap and it should work. I don't see substandard parts.

It's really Dell's fault for pricing their systems loooooooow and trying to bait customers into their overpriced add-ons.

>>> Cheap, fast, good: Pick Two. <<<
A Dell system might actually be cheaper, less time-consuming, and better than buying your own parts and putting together your own system. It's definitely cheaper, which you can use towards buying better parts elsewhere. If you have a large budget then you can't compensate some of the flaws a barebones Dell would have (not the fastest processor). It's less time-consuming since it's half-built, but then you may have to wait around for the right deal. It might be better quality on a low budget (more money to spend on other parts) but it could be lower quality on high budgets.

One thing Dell does very well in its cases is cooling and noise.

They use a heat shroud that keeps the noise level down while moving the hot air from the chip right to the fan.

One thing Dell does not do very well is their new desktop cases are almost all mini-tower which means only only four drive slots. System drive + video drive + DVD burner leaves only one left.

I think you can get a perfectly acceptable Vegas machine from Dell. It will not be top of the line and not perfectly upgradeable, but it will run and it will be stable. It depends on your needs, your budget and your expertise. If your budget is not large and you aren't a system builder, then you shouldn't worry about grabbing a good deal on a Dell.

People, we're working too hard at this.

I have two PCs. One is a homebuilt PC with 800GB of storage that recently became an AMD 2000-based machine (before that, it was an AMD 1200, and before that, an AMD Duron 750). The other is a super-cheap ($599) Dell Inspiron 2650, bought last year.

Both machines run Vegas great, and deliver RenderTest times of about 3 minutes. If I bought state-of-the-art CPU and memory I could get times of under 2 minutes, but for my purposes, both machines are plenty fast.

I went the homebuilt route because I enjoy getting my hands a bit dirty, and I was incensed that my very first PC (a 1998 Compaq Presario) needed a $100 Compaq power supply, when a generic power supply from Fry's had twice the output and cost $29. The nice thing about doing it yourself is that most of the parts of your current machine become parts for your next machine. But if you use a PC for getting work done, and don't care about what "DDR" means, then a Dell or HP or whatever would do just fine.

I guess it's fun to talk feeds and speeds, but for the purposes of creating great video I think it takes us off the track. It does for me, anyway.

Movinpix: Sometimes renders can take a long time. I've discovered (as JL has mentioned) that one can open multiple instances of Vegas and let one of them do the rendering in the background, at low priority, while working on the foreground instance of Vegas on an entirely different part of the project.

Dell RAM might be lower quality RAM that is more borderline than name brand RAM like crucial or whatever. However, RAM rarely fails and that's why most RAM manufacturers have lifetime warranties on them (Dell is 1 year warranty). RAM does fail sometimes... but really not often enough to care too much about it. Heat spreaders are unnecessary. I don't know about PCB.

I was doing applications engineering on memory modules at my last job a year ago, and I can tell you that generic RAM really does have problems. They're usually made with poorly designed PCBs and use parts from the low end of the speed bin. If you take a SPICE model of the cheap RAM module and attach it to a model of the connector, motherboard traces, and north bridge, you'll find there's more ground bounce and simultaneous switching output noise than on a decent Corsair or Kingston. Memory module PCBs on generic RAM typically have fewer layers than the name brand, which compromises signal integrity and crosstalk. Heat also plays a factor in terms of pushing the timing of the module and RAM itself towards the worst-case timing corner, so that's a compounding factor even for a few degrees of cooling that a heat slug will give. If you're already marginal on your signal integrity, you may latch false values and ultimately cause instability because the RAM was given unstable voltage under heat duress and with poor signal integrity. If that happens to be a program counter or pointer value being passed, you're stewed. RAM failure doesn't just take the form of catastrophic failure (which I've seen with generic modules more often than good quality moduels) but it can cause mysterious computer crashes that happen with a particularly hot ambient temperature under high load because of poor signal integrity (and which is another reason to use ECC on a professional workstation). Then the customer goes and screams at Dell as to why their computer is crashing. No wonder they outsourced their consumer support to India - it's cheaper to have a customer deal with instability and call a tech support person getting paid $7/hour than it is to give them the right product the first time.

You will have dual channel if you have two sticks of the same model RAM (any DDR PC3200 RAM can be dual channel). CAS latencies make no measurable difference in Vegas rendering speeds (I tried...).

The question is: does Dell give you two sticks of the same model of RAM module when you order the box? If you're not asking for it, you might not get it. CAS latencies assist in random access of data - a problem that Rambus RDRAM was notorious for. Rendering has a pretty consistent flow of data and often uses burst mode transfer, but for other uses of the computer including multitasking it's good to have.

It's Dell so the warranty is 1 year. They don't actually say what the hard drive is... but they are mostly all good. 8MB cache would be nice. On WDs system/appications performance goes up 30% (mainly loading times I think). Dell I'm pretty sure uses a 80-conductor IDE ribbon.

A professional using a workstation can't afford a high failure rate on a mass storage drive, particularly if their production is trapped inside a failed drive. Why can't Dell put a better hard drive in? The vast majority of people calling Dell won't look to ask for an 80 conductor IDE cable, and it may not make a difference to them if all they're doing is word processing and e-mail. I find the biggest bottleneck in terms of the feeling of system speed is the storage subsystem. When I spec out a system, I make sure that I understand the general dataflow of the customer and plan accordingly. I don't think Dell does or cares about it relative to even their average business customer.

I haven't heard of Dell having problems with bad power supplies. Dell's case design is pretty good with good airflow.

The problem is not that the power supply will suffer a catastrophic failure. The problem is that Dell will spec out the lowest power rating they can with the components they include. Low-quality power supplies provide poorer voltage regulation to the components in the computer. As I mentioned above with the RAM, it is a contributory factor to a compound signal integrity problem that results in instability and crashes. If you want to add more hard drives or other sources, you're pushing the limits of the power supply. The further you are away from the maximum power rating of the power supply, the better the regulation and the more stable the system will be. As for the airflow, is it good, or is it good enough? Again, unless I've actually specified the system I can't be sure. I'm positive that Dell wouldn't post a white paper or schematic of their cooling strategy.

The 4600 uses the 865 chipset, 533mhz FSB. The 8300 uses the 875, 800mhz FSB with a hyperthreading Pentium. The 8300 is better but the 4600 can also be very fast. The Dells run dual channel RAM to that's not a problem. I don't forsee any stability problems with a Dell (all the parts should work together) but then again I don't own one. It should be a solid editing station. It's cheap and it should work. I don't see substandard parts.

Again, what kind of RAM is it? Who is the OEM motherboard manufacturer? Is system stability their primary concern by design, or are they more interested in the bottom line?

(Hint: companies exist first and foremost for profit)

It's really Dell's fault for pricing their systems loooooooow and trying to bait customers into their overpriced add-ons.

And it's really the customer's fault for getting a machine that isn't the best performer out there.

A Dell system might actually be cheaper, less time-consuming, and better than buying your own parts and putting together your own system.

Not if you're having to add your own parts or replace a power supply because the hard drives you added make the voltage regulation marginal. Furthermore, if you have a specification in hand most white box local shops can turn around a machine as fast if not faster than Dell, and you can save money on the shipping charge. I don't typically build systems in my current consulting job, though I do upgrade them on occasion when necessary. Most of the time I contract it out and specify terms and conditions. Most shops seem to balk at the prospect, but I have one or two places that I know and use to build them because they are consistent and deliver quality product that I'm sure works.

It's definitely cheaper, which you can use towards buying better parts elsewhere. If you have a large budget then you can't compensate some of the flaws a barebones Dell would have (not the fastest processor). It's less time-consuming since it's half-built, but then you may have to wait around for the right deal. It might be better quality on a low budget (more money to spend on other parts) but it could be lower quality on high budgets.

It's not cheaper. My clients don't have time to muck around with upgrading because they're under the gun for getting projects out by the specified contract deadlines. Again, a trusted white-box manufacturer will be superior to any Dell because you control every aspect of the design of the box. And, if you order everything from the start, you can forego hunting for parts. Heck, even places like NCIX offer assembly and a 1 year warranty for $50, but you get to spec out everything you want.

>>> poor signal integrity (and which is another reason to use ECC on a professional workstation). Then the customer goes and screams at Dell as to why their computer is crashing. <<<
You have to consider the cost versus benefit here. How likely is it for off-brand RAM to break down versus brand name RAM or ECC RAM? I don't hear about RAM breaking down very often, so I don't see better RAM being worth it. If crappy RAM has a 3% chance of breaking down and you lose, say, $300 if it does, then you should only pay no more than $9 for better RAM. I chose what I think are "worst-case" numbers since I don't know what the real numbers are. Even in the worst-case scenario, paying a few extra dollars isn't worth it. That calculation assumes ECC or brand name RAM breaks significantly less than off-brand RAM. In this the price difference between putting your own RAM or switching from Dell to a DIY solution isn't justified.

>>> No wonder they outsourced their consumer support to India - it's cheaper to have a customer deal with instability and call a tech support person getting paid $7/hour than it is to give them the right product the first time. <<<
I think they get paid much less than $7/hour. Tech support requires only a few weeks of training. It's not even computer training, you have to learn how to speak like an American. For such a low-qualification job the hourly rate should be a lot less.

The argument for a Dell applies to people who can build their own computers and troubleshoot their own problems. Getting a barebones Dell is cheaper than doing everything I believe.

>>> The question is: does Dell give you two sticks of the same model of RAM module when you order the box? If you're not asking for it, you might not get it. <<<
They advertise it, so it should be dual channel. I don't see why Dell would have computers using single channel RAM.

>>> CAS latencies assist in random access of data - a problem that Rambus RDRAM was notorious for. Rendering has a pretty consistent flow of data and often uses burst mode transfer, but for other uses of the computer including multitasking it's good to have. <<<
I don't find that it would be worth paying for lower latency RAM. Some applications will be a few percent faster (maybe MPEG2 encoding???), but the extra cost isn't really justified.

>>> A professional using a workstation can't afford a high failure rate on a mass storage drive, particularly if their production is trapped inside a failed drive. Why can't Dell put a better hard drive in? The vast majority of people calling Dell won't look to ask for an 80 conductor IDE cable, and it may not make a difference to them if all they're doing is word processing and e-mail. I find the biggest bottleneck in terms of the feeling of system speed is the storage subsystem. When I spec out a system, I make sure that I understand the general dataflow of the customer and plan accordingly. I don't think Dell does or cares about it relative to even their average business customer. <<<
For DV work a 7200rpm ATA hard drive is fine. Getting a second 7200rpm ATA for video storage is a very good idea. You should definitely protect data with a back-up scheme and not rely on RAID 0 or higher reliability of a certain hard drive. Project files are small and can be easily backed up onto a CD-R. Video files you can re-capture if your hard drive fails. As for the better hard drives... are certain hard drives definitely better? You may have a favorite brand, but overall the differences in 7200rpm ATA drives are not that great and you can make a case for all of the hard drive manufacturers out there (seagate, hitachi, Western Digital, maxtor) maybe except for Samsung, which I know little about.

>>> The problem is not that the power supply will suffer a catastrophic failure. The problem is that Dell will spec out the lowest power rating they can with the components they include. Low-quality power supplies provide poorer voltage regulation to the components in the computer. As I mentioned above with the RAM, it is a contributory factor to a compound signal integrity problem that results in instability and crashes. If you want to add more hard drives or other sources, you're pushing the limits of the power supply. The further you are away from the maximum power rating of the power supply, the better the regulation and the more stable the system will be. <<<
Is there any evidence that Dell uses low quality/wattage power supplies? They have an interest in avoiding power supply problems (which may be difficult to trouble shoot) so they should be using a reasonable quality power supplies. It might be that they give you a lesser power supply if you don't have a power-hungry video card and a second hard drive. In that scenario the power supply would hamper upgradeability of the computer. But I don't think Dell does that since economically they don't have too much of an incentive to do that. They would have to keep 2 different inventories, the builder has to pick which power supply to put in, customers satisfaction would be lower, higher wattage only costs a little bit more, and there would be more tech support calls.

>>> Again, what kind of RAM is it? Who is the OEM motherboard manufacturer? Is system stability their primary concern by design, or are they more interested in the bottom line? <<<
RAM = DDR400, brand not advertised. Motherboards are custom-built by IBM, they are not standard shape. I don't see why Dells would be unstable. Stability is in their interest since most customers have 3/4-year serivce plans and warranties and customer satisfaction is in their interest. They do seem to be cutting corners on tech support, but that's not an issue unless one of your parts break down. In which case you have to grind though the tech support person's script.

>>> It's not cheaper. My clients don't have time to muck around with upgrading because they're under the gun for getting projects out by the specified contract deadlines. Again, a trusted white-box manufacturer will be superior to any Dell because you control every aspect of the design of the box. And, if you order everything from the start, you can forego hunting for parts. Heck, even places like NCIX offer assembly and a 1 year warranty for $50, but you get to spec out everything you want. <<<
I think Dell is cheaper. In some deals the computer costs less than the parts in it (street prices, not wholesale prices). A white box computer may not be as good since they builders might mess up the installation or run into problems because the combination of parts don't work.

Glenn,

I just want to clarify my statement. You are not going to "save hundreds" on a $500 PC but I did save hundreds on my $1800 PC which Dell sells for $2400! And mine has better parts than Dell’s. (i.e., high performance memory (Kingston HyperX CAS 2), more features on my Gigabyte 8KNXP motherboard than Dell's, etc.) When you get into the high-end systems you can save quite a lot by building it yourself.

~jr

Glenn, let me just say at the outset that it is not my job to prove affirmatively that Dell comptuers are not good editing machines, but it is Dell's responsibility to prove why their computers are good for production editing.

You have to consider the cost versus benefit here. How likely is it for off-brand RAM to break down versus brand name RAM or ECC RAM? I don't hear about RAM breaking down very often, so I don't see better RAM being worth it. If crappy RAM has a 3% chance of breaking down and you lose, say, $300 if it does, then you should only pay no more than $9 for better RAM. I chose what I think are "worst-case" numbers since I don't know what the real numbers are. Even in the worst-case scenario, paying a few extra dollars isn't worth it. That calculation assumes ECC or brand name RAM breaks significantly less than off-brand RAM. In this the price difference between putting your own RAM or switching from Dell to a DIY solution isn't justified.

You skimmed over the core of my RAM stability comment. You appear to have focused entirely on catastrophic failure while I'm talking primarily about issues that cause data corruption and crashes that are next to impossible to determine. You're also talking about a "DIY" solution. This is not Do-It-Yourself, this is spec-it-yourself and get a reputable system integrator to build it. Honestly, I think you missed my main point entirely here.

The argument for a Dell applies to people who can build their own computers and troubleshoot their own problems. Getting a barebones Dell is cheaper than doing everything I believe.

Then you believe incorrectly. A bare-bones Dell is not going to be cheaper when you factor in the total cost of ownership. That cost includes the opportunity cost of losing production time, sending your machine back and possibly breaking contractual obligations. You're also overfocused on this DIY thing again. Again, I'm going to tell you - you specify computers, but you get a reputable system integrator to build them for you.

They advertise it, so it should be dual channel. I don't see why Dell would have computers using single channel RAM.

But what kind? If the sticks aren't matched you'll get dual channel but non-optimized performance. It's still technically dual-channel, but that's what marketing is about. Go read the motherboard manual for Intel's i875P. They tell you what happens. I don't think the sales people really understand what's exactly going into these boxes, nor would they be compelled to tell you.

I don't find that it would be worth paying for lower latency RAM. Some applications will be a few percent faster (maybe MPEG2 encoding???), but the extra cost isn't really justified.

Why skimp? Really? If I'm specing out a top performer, I'm going to do it. It's not like I'm going to overclock with the RAM, but every bit of performance counts. In a production, time is money, even for a few percent more performance. You seem too focused on the front-end cost here.

Is there any evidence that Dell uses low quality/wattage power supplies? They have an interest in avoiding power supply problems (which may be difficult to trouble shoot) so they should be using a reasonable quality power supplies. It might be that they give you a lesser power supply if you don't have a power-hungry video card and a second hard drive. In that scenario the power supply would hamper upgradeability of the computer. But I don't think Dell does that since economically they don't have too much of an incentive to do that. They would have to keep 2 different inventories, the builder has to pick which power supply to put in, customers satisfaction would be lower, higher wattage only costs a little bit more, and there would be more tech support calls.

It's Dell's job to affirmatively prove this, not mine or yours. That's an unacceptable unknown for a production-worthy machine. Even you, however, make mention of how cheap support is. The aggregate cost of support issues due to poor voltage regulation or a cheap power supply may not outweigh the up-front cost savings. Most of Dell's computers are used for consumers who don't upgrade their boxes, or offices that run office applications. That's it. That's different than using them for a four-hour DV-to-MPEG-2 encode. But even this is immaterial compared to Dell affirmatively proving what they use and the MTTF/MTBF under full load.

As for the better hard drives... are certain hard drives definitely better? You may have a favorite brand, but overall the differences in 7200rpm ATA drives are not that great and you can make a case for all of the hard drive manufacturers out there (seagate, hitachi, Western Digital, maxtor) maybe except for Samsung, which I know little about.

The issue is the warranties that are given on these hard drives. Again, in a production environment, you need to have good reliability. Failures translate to lost profits and an expense of dealing with replacement. It's not confidence-inspiring when the hard drive I get has a 1 year instead of a 3 year warranty. I still have a Quantum Viking 4.5GB SCSI drive in my old P2-400 box, and it runs 24/7 for the last nearly six years. It came with a five year warranty out of the gate. It's run for longer than they said it would, but much longer than a 1 year warranty would garner. It's all about reducing risk, not eliminating risk.

RAM = DDR400, brand not advertised. Motherboards are custom-built by IBM, they are not standard shape. I don't see why Dells would be unstable. Stability is in their interest since most customers have 3/4-year serivce plans and warranties and customer satisfaction is in their interest. They do seem to be cutting corners on tech support, but that's not an issue unless one of your parts break down. In which case you have to grind though the tech support person's script.

The cost of extended service plans is not cheap. As the plan gets longer in length, the cost goes up as well. But I say again - how do you track down a marginal stability problem with RAM with any reliability? And how can you risk a production machine on an unknown. You can't. Dell has to prove this affirmatively, not me. And that's the point.

I think Dell is cheaper. In some deals the computer costs less than the parts in it (street prices, not wholesale prices). A white box computer may not be as good since they builders might mess up the installation or run into problems because the combination of parts don't work.

A white box will be better than a Dell if one of my clients asks me to specify it for them and it comes from a reputable builder. The variable of unknown components is eliminated. My clients are editors and videographers, not computer engineers and technicians. Just as they know how to handle a camera and set correct lighting and so forth, it's my job to give sound technical advice. They won't be building it, but they'll ask me to manage that project for them. And I'll guarantee you - all of the machines I've specified are stable and not built by me because I affirmatively know what went into them.

>>> You skimmed over the core of my RAM stability comment. You appear to have focused entirely on catastrophic failure while I'm talking primarily about issues that cause data corruption and crashes that are next to impossible to determine. You're also talking about a "DIY" solution. This is not Do-It-Yourself, this is spec-it-yourself and get a reputable system integrator to build it. Honestly, I think you missed my main point entirely here.
<<<
Ok I think we're talking about two different things here:
A- Buying *all* your parts (probably from newegg.com) and putting everything together yourself
B- buying a *barebones* Dell and putting in your own parts.
C- The above versus getting a computer shop / systems integrator to put together the computer.
I'm talking about A vs B, you're talking about B vs C.

And also...
D- Most bang for your buck video editing system versus...
E- a top end machine for people who edit professionally (where extra performance is justified)
I'm talking more about E, you seem to be talking about E. Now if you're building a top end machine, then a barebones Dell is not that good a choice anymore (overpriced upgrades, no support).

>>> You skimmed over the core of my RAM stability comment. You appear to have focused entirely on catastrophic failure while I'm talking primarily about issues that cause data corruption and crashes that are next to impossible to determine. <<<
Ok that would be a bad scenario too, but *how often* would that occur (versus better brand name RAM)? And then you'd have to figure out how much money you lose if instability occurs to see if the benefits outweigh the costs. Most RAM has a lifetime warranty and I rarely hear about RAM failing... so my wild guess is that the chance of RAM failing is very low. i could easily be wrong since I don't have any good information on how much non-brand name RAM fails.

next to impossible to determine? You can test RAM with memtest86, although it may take you some time to figure out it's a hardware problem and not Premiere's fault or whatever. By deduction you can pinpoint random crashes on hardware failure (CPU or RAM) so it's not hard to figure out.

>>> But what kind? If the sticks aren't matched you'll get dual channel but non-optimized performance. It's still technically dual-channel, but that's what marketing is about. Go read the motherboard manual for Intel's i875P. They tell you what happens. I don't think the sales people really understand what's exactly going into these boxes, nor would they be compelled to tell you. <<<
What do you mean not matched? Not the same model?
From reviews of Dells around the internet, there doesn't seem to be anything wrong with them performance-wise. They hold their own pretty well in benchmarks.

>>> I don't find that it would be worth paying for lower latency RAM. Some applications will be a few percent faster (maybe MPEG2 encoding???), but the extra cost isn't really justified.

Why skimp? Really? If I'm specing out a top performer, I'm going to do it. It's not like I'm going to overclock with the RAM, but every bit of performance counts. In a production, time is money, even for a few percent more performance. You seem too focused on the front-end cost here.
<<<
In the case of Vegas, lower latency RAM just doesn't seem to make a measurable difference at all. Perhaps 5% of the time a video editor would do a task that benefits from lower latency RAM (???MPEG2 encoding???), but the benefits in that case don't outweigh the costs. You have to take into account how little time low latency RAM saves. You won't even feel the few percentage difference.

>>> It's Dell's job to affirmatively prove this, not mine or yours. <<<
Well if they did you'd still be able to call them biased and accuse them of fudging specs, or making subjective claims ("best support!") that they couldn't prove. I think the practical thing to do here is to find good solid information on whether or not a barebones Dell can be turned into a decent video editing machine. I must admit I don't have very solid information since I've only briefly used one for one project. Goof information would be things like people who have actually turned a barebones Dell into a working productive machine. I suppose this leaves it open to the FUD factor (fear, uncertainty, doubt).

>>> [power supplies] That's an unacceptable unknown for a production-worthy machine. Even you, however, make mention of how cheap support is. The aggregate cost of support issues due to poor voltage regulation or a cheap power supply may not outweigh the up-front cost savings.

>>> That's different than using them for a four-hour DV-to-MPEG-2 encode. But even this is immaterial compared to Dell affirmatively proving what they use and the MTTF/MTBF under full load. <<<
MTBF numbers may not be that useful, because:
A- They can be fudged. A lot of hard drive manufacturers just put a high number on without thorough testing and with dubious time-acceleration measures (increasing heat).
B- By the time reliable MTBF numbers are out, the product is already obsolete. You need time and a large sample to get a good measurement.

Now a lot of people use Dells and don't have products with their power supplies delivering too low voltage. Some people did get burned with non-standard power supplies, but Dell doesn't do that anymore.

>>> The issue is the warranties that are given on these hard drives. Again, in a production environment, you need to have good reliability. Failures translate to lost profits and an expense of dealing with replacement. It's not confidence-inspiring when the hard drive I get has a 1 year instead of a 3 year warranty. [...] It's all about reducing risk, not eliminating risk. <<<
Warranties do not affect how reliable a drive is. Shorter warranties may let manufacturers design less reliable drives, but Dells are going to have the same hard drives as all the other manufacturers (there's only 5 manufacturers and they all make decent drives). There's not much difference unless you can pinpoint the more reliable hard drive brands. However, you can definitely go wrong with those predictions. Promax now recommends IBM drives, but IBM had significantly high failure rates with its deathstar/deskstar 75GXP line. Is there a big difference in the hard drives you spec out? Is it probable that they are more reliable? If you have a good backup plan in place then it doesn't cost that much if you lose data (many reasons other than hard drive failure). Video can be re-captured and project files can be copied back onto the computer.

>>> I still have a Quantum Viking 4.5GB SCSI drive in my old P2-400 box, and it runs 24/7 for the last nearly six years. It came with a five year warranty out of the gate. It's run for longer than they said it would, but much longer than a 1 year warranty would garner. <<<
One good experience doesn't say much. Even the most unreliable products will have a lot of happy customers.

I think for people on a budget and want to assemble their own computer, they might want to look into getting a barebones Dell. I used to think that assembling your own computer was always a few hundred cheaper than a Dell, but that is not the case when you get a barebones Dell and put in your own parts. When I price it out the Dell usually comes a hundred or so ahead, and you don't have to assemble as much. I could definitely be wrong. There may be some gotchas that I don't know about, or I may have miscalculated the cost difference between a Dell and a DIY solution.

Dell/DIY solution VS getting a system builder to build it for you. I have no idea about this. It seems like vendors like ABS computers cost a few hundred more than DIY/Dell, but maybe there are builders who will do it for a very low fee and do a quality job too. Again, I have no idea.

BTW, I found a great resource when I was researching building my first PC at PCMech.com. They sell a CD with “how to” videos but they also give you enough free “how to” video that are downloadable so that you can feel comfortable with the steps needed to assemble your own PC.

~jr

I say go with a two-processor system. Even if you don't have top-of-the-line speed in your processors, you'll see significant benefits in rendering time.

I'll tell you, once you've used a Windows XP system with two processors, you'll never want to go back to a single-processor system. You'll notice the lag time, right off the bat.

When you use two processors, even if the rendering software isn't optimized for two processors (and Vegas appears to be, BTW, which is commonly referred to as multi-threaded), XP will offload the OS stuff on one processor, and commit the second processor to the task at hand, which would be rendering.

I'm talking significant speed benefits here. Since Vegas is multi-threaded, it can utilize two processors in a single computer to really get some work done.

I have a dual-Xeon box running at only (ONLY?) 2.2 GHz. I plan to upgrade my processors soon. Still, I can tell that my dual 2.2 configuration can outrun a single 3.0 GHz box, and perhaps a single 3.2 box.

I'll have to give Microsoft some credit here, as XP really does a good job of using two processors. Almost as good as Unix. Vegas shines on my dual-processor machine.

Dual processors will not really help Vegas. If you render a real world project, the second processor is going to be used very little. A P3.2C is going to be faster than 2X2.2ghz most of the time.

I don't have a Xeon box to play with, but I do have a processor with hyperthreading and I can see that the second processor doesn't do much work.

OS stuff doesn't seem to take many clock cycles.

I respectfully disagree, here. I'm talking about two processors, not hyperthreading, which I do not enable on my computer.

My computer, with two processors, makes Vegas render extremely quickly. Way faster than an associate's 2.8 GHz P4. He loves to use my box, when he's here, 'cause the interface is "real time."

Hyperthreading makes for a "virtual" second processor, and that is not the same as having two real processors working for you. If you don't have two processors, then you really don't know what I'm talking about.

Well I’m curious about this. It seems the CPU usage during renders to NTSC DV on my dual xeon averages 60% for both procs; for renders to MPEG2 the usage goes to near 100% for both procs; hyperthreading is not enabled. Can someone explain what’s going on?

For me, one of the advantages of having dual processors is the ability to run multiple instances of Vegas (or other apps) without slowdowns.

JL

Here's what I understand about hyperthreading, SMP:
Each program has a thread that processes data. Because of programming limitations the second thread may not be doing the same thing as the first.

A processor can run 1 thread at a time. If you have multiple threads then each thread takes turns (OS handles that).

Dual processors can run 2 threads.

Hyperthreading is Intel's party trick that tries to get 1 processor to run 2 threads at once. 1 thread doesn't take up all the CPU's capacity so it can actually run 2 threads. However, there are drawbacks to hyperthreading so it doesn't give huge performance gains. Real world gains range from -5% to 50% when both logical/fake processors are equally loaded. In Vegas the gain is about 15-20% when both processors are equally loaded.

Vegas has 2 (or more???) threads. The first thread handles rendering filters, and the second DV encoding and audio processing. In long intensive renders the first thread has a lot to do and the second not very much.

So... the second processor in Vegas doesn't do much and *depends on your render*. The Main Concept encoder seems to be having both threads do the same kind of work, which should make it go nearly twice as fast (real world difference might be 70-80? I've never measured so I don't know).

>>>My computer, with two processors, makes Vegas render extremely quickly. Way faster than an associate's 2.8 GHz P4. He loves to use my box, when he's here, 'cause the interface is "real time." <<<
Maybe he has some sort of bottleneck in his system because his system should be comparable and maybe even faster. Bottlenecks include hard drive fragmentation and using PIO mode instead of DMA. He may have something taking up a lot of CPU cycles like a bad RAID controller (VIA RAID controller on the P4P800deluxe for example) or programs running in the background (malware, distributed computing clients, etc.). He could try Spot's rendertest.veg to see how he stacks up. A P2.8C should come somewhere around 1:30 (give or take 20 seconds).

Ok that would be a bad scenario too, but *how often* would that occur (versus better brand name RAM)? And then you'd have to figure out how much money you lose if instability occurs to see if the benefits outweigh the costs. Most RAM has a lifetime warranty and I rarely hear about RAM failing... so my wild guess is that the chance of RAM failing is very low. i could easily be wrong since I don't have any good information on how much non-brand name RAM fails.

When I bill my clients out at $55/hour, and they ask me why things are going haywire on them, good quality RAM has already paid for itself.

next to impossible to determine? You can test RAM with memtest86, although it may take you some time to figure out it's a hardware problem and not Premiere's fault or whatever. By deduction you can pinpoint random crashes on hardware failure (CPU or RAM) so it's not hard to figure out.

I beg to differ. The only way to fully understand this problem is to get a SPICE model of the memory, module PCB , connector, motherboard PCB and northbridge, and buffers at both ends, and run it through a worst-case corner analysis at the design stage, or get a storage scope and trace out the signal lines and hope and pray that you don't create impedance problems in the interim. I don't think memtest is going to load down the processor to 100%, increase system temperature and affect rail voltages. Trust me, SSTL-2 Type 1 signals are not easy to deal with at 400MHz DDR speeds. This is why some motherboards can't use all of their memory slots - the tx line reflections and ringing make the signal look like crap on toast.

I'm not trying to pull some intellectual superiority trick over you. I'm simply telling you the way it is in the real world. One of my customers at my previous place of employment was evaluating memory modules from several manufacturers for their storage network interface. Without going into the details, we found a direct correlation between relative module quality and signal integrity. In fact, some module manufacturers are awful in their ability to supply proper memory models. After we saw all sorts of noise and ringing, we excluded them fairly quickly.

You have to take into account how little time low latency RAM saves. You won't even feel the few percentage difference.

A few percent multiplied by a hundred productions over the life of the machine is time spent where they couldn't spend it elsewhere.

Goof information would be things like people who have actually turned a barebones Dell into a working productive machine. I suppose this leaves it open to the FUD factor (fear, uncertainty, doubt).

FUD or not, I can't personally afford to take the chance.

MTBF numbers may not be that useful, because:

In my original post, I mentioned MTTF and MTBF. The difference is that one considers the operational time failure of the unit, while the other considers operational time failure PLUS the time that it will take to replace the unit and get it up and running again. I don't think Dell could match that turn-around time unless they had a pretty thorough support contract. And - believe me - I've saved a client who was on the brink of a lawsuit by a few hours because his machine went down at a critical time.

But even considering the "fudged" numbers, they're a better indicator than having nothing.

Now a lot of people use Dells and don't have products with their power supplies delivering too low voltage. Some people did get burned with non-standard power supplies, but Dell doesn't do that anymore.

A lot of people use Dells professionally, but few are using these computers in such a severe way. Video editing, unlike word processing and e-mail, is very taxing on a system. That's the corner case we're addressing here.

Warranties do not affect how reliable a drive is. Shorter warranties may let manufacturers design less reliable drives, but Dells are going to have the same hard drives as all the other manufacturers (there's only 5 manufacturers and they all make decent drives). There's not much difference unless you can pinpoint the more reliable hard drive brands. However, you can definitely go wrong with those predictions. Promax now recommends IBM drives, but IBM had significantly high failure rates with its deathstar/deskstar 75GXP line. Is there a big difference in the hard drives you spec out? Is it probable that they are more reliable? If you have a good backup plan in place then it doesn't cost that much if you lose data (many reasons other than hard drive failure). Video can be re-captured and project files can be copied back onto the computer.

If a warranty doesn't affect how reliable a drive is, then the manufacturer should simply honor a longer warranty! The manufacturers themselves admitted when they dropped the standard 3 year warranty that they were trying to save on RMA costs. It's their own damned fault, to be sure, but there's a direct correlation between drive lifetime and warranty. I'm not going to go into the actuarial analysis of it because I don't need to - long warranty implies a higher probabilty of longer life and that the manufacturer is willing to back that up at their own financial peril.

Now let's focus on Dell. Who supplies Dell's hard drives and of what use are they? I only personally recommend Western Digital or Seagate because of my experience with them (e.g. failure rate, RMA handling). But that's at the moment. I'm not a fanboy, and I don't really care who wins or loses in market share terms. You are correct in asserting that you need a good fallback strategy regardless of inherent hard drive reliability. That's why one of my clients saves his batch capture file on separate media and does backups often. But why introduce an aggravating factor in terms of warranty?

I think for people on a budget and want to assemble their own computer, they might want to look into getting a barebones Dell. I used to think that assembling your own computer was always a few hundred cheaper than a Dell, but that is not the case when you get a barebones Dell and put in your own parts. When I price it out the Dell usually comes a hundred or so ahead, and you don't have to assemble as much. I could definitely be wrong. There may be some gotchas that I don't know about, or I may have miscalculated the cost difference between a Dell and a DIY solution.

The shipping costs of a Dell alone should outweigh the cost savings versus getting a white box built at a reputable dealer. But if you're on a budget to begin with, the seriousness of your projects is probably questionable. Either you're in business to make money, or you're not. Skimping means you're too focused on the short term and immediate cash flow in terms of business. If you're trying to fund a big production with your own money, you're better off using someone else's money (e.g. loan, outside private investment) first in the interest of doing it right.

The last thing about this is that, unless I had the technical know-how, I wouldn't blindly go to anyone and just get the white box built out of a list of specs. That's why I'm in business consulting. I give my clients advice, and if there's a screw-up they know they can fall back on me for support.

>>> I beg to differ. The only way to fully understand this problem is to get a SPICE model of the memory, module PCB , connector, motherboard PCB and northbridge, and buffers at both ends, and run it through a worst-case corner analysis at the design stage, or get a storage scope and trace out the signal lines and hope and pray that you don't create impedance problems in the interim. I don't think memtest is going to load down the processor to 100%, increase system temperature and affect rail voltages. Trust me, SSTL-2 Type 1 signals are not easy to deal with at 400MHz DDR speeds. This is why some motherboards can't use all of their memory slots - the tx line reflections and ringing make the signal look like crap on toast. <<<
So, theoretically lower quality RAM has a higher chance of failing. If the theory is sound, then practically speaking this should translate into the real world where we see bad RAM fail a lot more than good RAM. That I accept. However, you have to take into account real world considerations:
A- How likely is low quality RAM going to fail?
B- Do you have to test RAM to see if it isn't crap on arrival? If you should test low quality RAM, do you also have to test good quality RAM? This could cost you like 20 minutes of your time and tie up your computer for a day. I think a better testing program is prime95, which runs calculations and checks against known results. It loads the CPU well and tests RAM too. memtest86 just writes and reads from the RAM, it doesn't load CPU much.
C- How much would failing RAM (BSODs, subtle data corruption, outright failure) cost you? For people with clients, I didn't consider the possibility that you can lose future projects if one project goes really bad.
D- How does the RAM that comes with your computer stack up?

I think the crucial thing that should be known is A- how likely is low quality RAM going to fail compared to brand name RAM? In some cases if you have a vague idea about that then you can figure out if brand name RAM is worth it. If you run a server and bad RAM would cost the company $100,000, while better RAM costs $100, then better RAM is worth it if it fails 1 in a 1000 times less than lower quality RAM. You'd also have to consider your job security if you were the IT guy. Even if you make the right decision to get slightly lower quality RAM, personal considerations may outweigh the company's.

Another thing you have to consider is that humans aren't very good at dealing with extremely low or large numbers. Many people buy lottery tickets and are overly afraid of sharks, SARS, etc. Winning the lottery, dying from SARS, and plane crashes are all extremely rare events. You can avoid this trap by doing something along the lines of the analysis above, where you try to quantify things and see if your decisions are reasonable before you purchase that lottery ticket, make yourself look like an idiot by wearing a SARS mask, or buy shark repellent. When your data is unreliable, you can check to see if it makes sense even in the worst/best case scenario.

Another thing to consider is that a barebones Dell is so cheap that you can replace the memory with whatever you want. It can still be cheaper than a DIY computer, even if you throw the old sticks into the trash instead of eBaying them (or put them into another computer).

>>> A lot of people use Dells professionally, but few are using these computers in such a severe way. Video editing, unlike word processing and e-mail, is very taxing on a system. That's the corner case we're addressing here. <<<
I figure there must be a significant number of people who play games on their computers, which can stress the CPU and video card a lot. Some of the new video cards use as much energy as the CPU.

>>> If a warranty doesn't affect how reliable a drive is, then the manufacturer should simply honor a longer warranty! The manufacturers themselves admitted when they dropped the standard 3 year warranty that they were trying to save on RMA costs. It's their own damned fault, to be sure, but there's a direct correlation between drive lifetime and warranty. I'm not going to go into the actuarial analysis of it because I don't need to - long warranty implies a higher probabilty of longer life and that the manufacturer is willing to back that up at their own financial peril.
<<<
Dells doesn't make hard drives, so their 1 year warranty doesn't really affect things. And their business counts on customers buying 2-3 year warranties (which I assume a large portion of their customers do), so they have a bit of incentive to buy drives that are more reliable. Most hard drives now have just 1-year warranty, so Dell might be slightly inclined to go wtih more reliable drives. But that doesn't necessarily mean drives with longer warranties. Dimension 2400s come with Seagate drives (because some people with that computer have Seagates in them). At least for now...

>>> The shipping costs of a Dell alone should outweigh the cost savings versus getting a white box built at a reputable dealer. <<<
Dell usually has free shipping.

>>> But if you're on a budget to begin with, the seriousness of your projects is probably questionable. Either you're in business to make money, or you're not. Skimping means you're too focused on the short term and immediate cash flow in terms of business. If you're trying to fund a big production with your own money, you're better off using someone else's money (e.g. loan, outside private investment) first in the interest of doing it right. <<<
Well we've gone through this before. A barebones Dell would be good for people who can build their own computer and don't need top notch performance. You might also have to wait around for the best deals on a Dell. For your line of work getting a barebones Dell may not be very relevant.

I can assure you that Vegas uses two processors quite well.

I can do real-time previews that would choke other systems.

If you are afraid of two-processor systems, then don't get one. I'm here to tell you that they work, and they work really well. I know what I'm talking about, 'cause I use at least one every day. I run Vegas on a dual-Xeon platform, and I run Final Cut Pro on a dual-proc Mac.

Two heads are better than one.

So, theoretically lower quality RAM has a higher chance of failing. If the theory is sound, then practically speaking this should translate into the real world where we see bad RAM fail a lot more than good RAM. That I accept. However, you have to take into account real world considerations:

Believe me, I have taken real-world considerations into account. A marginal stability problem will continue to manifest itself time and again. I'm not talking about alpha particle errors or flip-flop metastability issues, but even these issues can cause problems that should be trapped out both in hardware and in software where possible. Why do you think servers and workstations use ECC RAM? That kind of reliability is necessary in a production environment, and it's the reason why it needs to be spec'd out. Cheap memory uses cheap PCBs for the modules. Fewer layers means less noise immunity, more crosstalk, more ground bounce, etc.. I did the SPICE simulations myself to prove what happens in marginal conditions. It wouldn't surprise me to see weekly lock-ups on this.

I'd really like you to try to understand what I'm telling you, but you'd have to be a semiconductor engineer to understand (which I am). Please take my word for it, it's not worth the risk on a production machine.

B- Do you have to test RAM to see if it isn't crap on arrival? If you should test low quality RAM, do you also have to test good quality RAM? This could cost you like 20 minutes of your time and tie up your computer for a day. I think a better testing program is prime95, which runs calculations and checks against known results. It loads the CPU well and tests RAM too. memtest86 just writes and reads from the RAM, it doesn't load CPU much.

20 minutes of my time? Not if there's data corruption. I've seen it happen, and I saved my one client's bacon because I went in there, did some rudimentary data recovery, and cobbled a solution together to meet a deadline because his customer's lawyer gave him a deadline for getting a version of the production in the customer's hands.

This is also what a good burn-in period is for. I doubt that Dell runs their computers for any significant burn-in, or their overhead costs would go through the roof. Burn-in does two things; namely, it gets rid of infant mortality component failures (which is what you're talking about), and it allows the builder to ferret out stability issues up front. Now, if you've selected the right components, you'll have taken care of the latter. The former will come only with a rigorous burn-in period. If it were up to me, I'd be running a DivX encode through a script that looped, or maybe even timedemo.

C- How much would failing RAM (BSODs, subtle data corruption, outright failure) cost you? For people with clients, I didn't consider the possibility that you can lose future projects if one project goes really bad.

Lock-ups can be costly. See, unlike Vegas, Premiere Pro has no real auto-recovery facility. My client did get burnt on this because of a marginal power supply issue. When his computer crashed, and he went back in, his data was corrupted! That is the problem with a stability problem like this - you never know what will happen. It could be a clean crash, it could hurt data severely. When I'm called on to try and clean up the mess, it isn't worth it for me.

Even losing a day's worth of work could've meant the difference between a lawsuit and not. My client was under the gun because his customer's lawyer sent a nasty letter to have a preview of the production by a certain time and date, or that they would file a lawsuit for breach of contract. What's the cost of that? Maybe you don't deal with those kinds of issues so it's not that important to you. But it sure as hell is important to me and to my clients to do things right.

D- How does the RAM that comes with your computer stack up?

In light of what was mentioned above, I think this is already covered. I always go for Corsair, and it has never failed me one time. Not once in six years since I've been using it. It's more expensive, but you get what you pay for. It's also why overclockers love using it, though I'm not an overclocker myself.

I think the crucial thing that should be known is A- how likely is low quality RAM going to fail compared to brand name RAM? In some cases if you have a vague idea about that then you can figure out if brand name RAM is worth it. If you run a server and bad RAM would cost the company $100,000, while better RAM costs $100, then better RAM is worth it if it fails 1 in a 1000 times less than lower quality RAM. You'd also have to consider your job security if you were the IT guy. Even if you make the right decision to get slightly lower quality RAM, personal considerations may outweigh the company's.

Servers run ECC. The reason they run ECC is because of alpha particle corruption and metastability. These usually result in temporary single-bit failures which the parity bit on ECC allows for automatic recovery without the intervention of software. Even ECC is pretty useless, however, if you don't have a good quality memory PCB that has been designed and tested at the corner cases.

And I already described the issue with failures. I'm not talking about the module crapping out, I'm talking about data corruption or crosstalk. Those types of failures will be far more frequent on a cheaper module, cause more frequent crashes, and result in a higher TCO for the user. A home user might not care, but in a post-production environment, it's not worth it.

Another thing you have to consider is that humans aren't very good at dealing with extremely low or large numbers. Many people buy lottery tickets and are overly afraid of sharks, SARS, etc. Winning the lottery, dying from SARS, and plane crashes are all extremely rare events. You can avoid this trap by doing something along the lines of the analysis above, where you try to quantify things and see if your decisions are reasonable before you purchase that lottery ticket, make yourself look like an idiot by wearing a SARS mask, or buy shark repellent. When your data is unreliable, you can check to see if it makes sense even in the worst/best case scenario.

Again, I don't think you understand the issue here. The types of failures that ECC is designed to accommodate happen pretty rarely. But those are failures that are not inherent to the design. They come from things like alpha particles, which are induced from lead solder and from environmental radiation. The types of failures I'm talking about are ones such as inadequate ground planes or poorly shielded data lines. These will cause failures to happen far more frequently. To draw an analogy with cars: any car can get into a crash because of something someone else did, even if everything else was perfect. Buying a car with skinny tires and a weak engine and expecting it to handle around turns like a sports car or trying to make a short merge onto a 70mph freeway is going to get you in trouble again and again and again.

Another thing to consider is that a barebones Dell is so cheap that you can replace the memory with whatever you want. It can still be cheaper than a DIY computer, even if you throw the old sticks into the trash instead of eBaying them (or put them into another computer).

Why screw around with this in the first place? If it's built right the first time, I don't have to worry about things like this again, or Ebaying my old memory, or getting into the computer to fix it. Most people who are buying Dell are never going to open their computer up, which means that a paid professional will likely have to do it.

I figure there must be a significant number of people who play games on their computers, which can stress the CPU and video card a lot. Some of the new video cards use as much energy as the CPU.

Yes, and a lot of these people experience lock-ups too. Heck, it happened to me a lot while playing Team Fortress, though that had to do with the software and not the hardware. Games are not a critical environment. Post-production is.

Dells doesn't make hard drives, so their 1 year warranty doesn't really affect things. And their business counts on customers buying 2-3 year warranties (which I assume a large portion of their customers do), so they have a bit of incentive to buy drives that are more reliable. Most hard drives now have just 1-year warranty, so Dell might be slightly inclined to go wtih more reliable drives. But that doesn't necessarily mean drives with longer warranties. Dimension 2400s come with Seagate drives (because some people with that computer have Seagates in them). At least for now...

Again, I will say that a warranty is an actuarial risk tradeoff mechanism. If what you say is true, then SCSI hard drive makers shouldn't bother to put more than a year's warranty on their drives destined for high-end workstations and servers. And yet, they do put these five year warranties on their drives because they know they're built to higher standards than a consumer drive. But even the consumer drives that come with a three-year versus a one-year warranty must have something built into them. One could attribute the higher price of an 8MB cache drive with three years of warranty to a 2MB cache drive with one year of warranty to the allocated number of returns due to failures. That's just dealing with the cost of the drive.

What about the data on the drive? How much is that worth? What if it's unrecoverable? If people really sit down and think about things, they'll understand that this is not something to skimp on even if they're a home user.

Dell usually has free shipping.

Yes, but not all the time.

Well we've gone through this before. A barebones Dell would be good for people who can build their own computer and don't need top notch performance. You might also have to wait around for the best deals on a Dell. For your line of work getting a barebones Dell may not be very relevant.

Sorry, but again I have to disagree with you here. A white box built by a reputable local dealer is better than a computer which I have to deal with RMAs, shipping, and other hassles. But the white boxes do have one big disadvantage with respect to Dell - they aren't anywhere near as well marketed. And I'm not biting into the marketing.

Wow! Some of these lengthy yet thoughtful responses would qualify as term papers at many colleges!

Look, the buy vs. build argument has good points on both sides. Let's just leave it at that.

At risk of further flanning the flames, let me mention that prudent business management would dictate that new PCs should be burned in for a certain amount of time, in order to catch the early failures. From my Apple Computer days (back when the Mac first came out), I can recall that management had a very clear handle on the costs involved with warranty returns. So what you do is balance that cost with the added costs of a burn-in period, and you pick the point where the overall costs are lowest.

As mentioned in an earlier post, I have a homebuilt PC that has gone through numerous iterations, and a cheap but bulletproof Dell laptop. I love 'em both, and both run great.

As an aside, one of the posts mentioned how people usually have an irrational fear of rare events. I remember reading a fascinating article where researchers took reasonable scientific data regarding various risks and reformulated it into how much of that activity would be needed to raise your chances of dying by 1 in a million. For example, they discovered that riding in a canoe for 10 miles would do it; riding a bicycle for 300 miles would also. Flying in a jet at 40,000 feet for 6 hours would qualify (gamma radiation), as would living in a granite building for, I think, 5 years (again, gamma radiation from the breakdown of some of the elements in granite). These numbers are to the best of my recollection.

The most common risk factor raising your chances by 1 in a million?--smoking just 1.4 cigarettes!