Tapeless workflows and Sandy Bridge or other PC's: KISS or LOVE?
Life used to be so simple when shooting video on a tape based camera. You shot your material, captured it for editing and stored your precious original footage on tape in a safe and dry place. Sure, it took time to capture, but the big advantage was that if you had a computer or drive failure, you would still have the original tape so everything could be recreated.
Now with tapeless workflows we have the significant advantage of much faster import of the original footage. Connect the flash card or disk drive to the computer over USB and copy the data to a HDD on the computer, ready for editing. The data on the flash card or disk drive can then be erased, so you can reuse it for more shots. But, like Johan Cruyff has said repeatedly, every advantage has its drawback. In this case it simply means that you no longer have the original material to fall back on, in case of computer or drive failures. That is a very unpleasant and insecure feeling.
The easy anwser to that problem is backups. Backup of the original media, backup of projects and backup of exports. This often means a bundle of externals for backup or NAS configurations. One thing is clear, it requires discipline to make regular backups and it costs time, as well as a number of disks. Four as a minimum: 1 for media, 1 for exports and at least 2 for projects. Note: This is excluding a backup drive for OS & programs.
There are different backup strategies in use. Some say backup daily and use one disk for monday, one for tuesday, and so on. Others say one disk for the first backup, the second for the second backup, then the first again for an incremental backup, etc. and once weekly a complete backup on a third disk. Whatever you choose, be aware that shelf live of a disk is far less than tape. There are horror stories everywhere about ball-bearings getting stuck after some time and without original tapes, you better be safe than sorry, so don't skimp on backups.
What is the relevancy of all this? I thought this was about Sandy Bridge and other PC's.
It is and let me try to explain.
Card based cameras are for the most part DSLR and AVCHD type cameras, and we all know how much muscle is required to edit that in a convenient way. Adobe suggests in the system requirements to use raid configurations for HD editing and practice has shown that raid arrays do give a significant performance boost and improve responsiveness, making for a nicer editing experience. The larger the project and the longer the time-line, the more a raid array will help maintain the responsiveness.
One thing you would not do is using a raid0 for projects, media and exports, even if you have backups. The simple reason is that the chance of disk failure multiplies by the number of disks in the raid0. Two disks double the chance of disk failure, three disks triple the chance, four disks quadruples the chance, etc.
Remember: Disaster always strikes when it is most inconvenient.
Imagine you have been working all day on a project, you decide to call it a day and to make your daily backup, but then the raid fails, before you made your backup. Gone is all of today's work. Then take into consideration the time and effort it takes to restore your backups to the state it was in yesterday. That does not make you happy.
Another thing to avoid is using a software or mobo based parity raid, for the simple reason that it is slooowww and puts a burden on the CPU, that you want to use for editing, not house keeping.
For temporary or easily recreated files, like the page-file, media cache, media cache database and preview files, it is very much advised to use a raid0. It makes everything a lot snappier and if disaster strikes, so what? These are easily recreated in a short time.
This was a general overview of what is required with tapeless workflows. Now let's get down to what this means in terms of system design.
Two approaches or train of thoughts
KISS: Keep it stupidly simple or LOVE: Laughing over video editing
The first one, the most economic one, is to use a system with 3 or 4 disks internally and 4 or more backup disks.
A typical disk setup can look like this:
This is a perfectly sensible approach if one does not have large or complex projects, long time-lines and is willing to take the risk of occasionally losing a whole days work, between backups. Many hobbyists and consumers fall in this category.
The KISS approach keeps it stupidly simple. The drawback is that there is no logical way to add more disks or storage. The discipline, diligence and effort required for regular backups make it far from a laughing matter. In fact it can quickly become a bore. Add to that the fact that the disk setup is simple but not very fast, so less suited for situations where lots of clips are involved, multi-cam is a regularly recurring situation or lots of video tracks are involved.
A number of video editors want more from their system than the occasional platonic KISS, they want to really LOVE their system, which lead to the other train of thought.
This is more costly than the KISS approach, but you all know a fiancée or wife is more costly and dear than the occasional kiss on the cheek by an old friend.
Let's start with a typical disk setup. It may look like this:
Two striking differences in comparison to the KISS approach:
1. Much easier disk organization and more disks and thus more space.
2. It requires a hardware raid controller, causing a higher investment cost. It is like an engagement ring. You don't get LOVE for free, one of the guiding principles of the oldest trade in the world.
These are easy statements to make, but what are the benefits or advantages, that you would fall in LOVE with such a system, and what are the drawbacks? Think back to Johan Cruyff's adage.
The only drawback is cost. The advantages are multiple, easier organization, more speed, more storage, snappier editing, no jerkiness, lesser requirements for regular backups and - this is the major benefit - hardly a chance of losing a day's work in case of a drive failure. Keep in mind that a parity raid keeps all your data intact in case of a drive failure, so lessens the need for up-to-date backups.
We all know, we get what we pay for: "If you pay peanuts, you get monkeys. OTOH, if you pay money to monkeys, you get rich monkeys". But in this case you get what you pay for, a much better editing experience with a much easier workflow.
Using a parity raid (be it raid 3/5/6/30/50/60) you get security, ease of mind that you are protected against losing precious media, that you need not worry about the last time you made a backup, that the editing you did today may be lost and you save valuable time editing and a lot of aggravation because of a much more responsive system.
How does this all relate to Sandy Bridge and other PC's?
First of all, the price difference between a Sandy Bridge / P67 platform and an i7-950+ / X58 platform is very small. Of course the new architecture is slightly more expensive than the older one, but the differences are small, almost not worth talking about.
So what are the differences? Look below:
The first thing to keep in mind is that the Sandy Bridge is the successor of the i7-8xx CPU and as such it is much more evolutionary than revolutionary. The CPU power has increased significantly over the i7-8xx due to new architecture and a smaller production process (32 nm), but in essence all the capabilities have remained unchanged. Same memory, same PCI-e lanes, same version, same L3 cache and no support for dedicated raid controllers.
It is great that the processor performs much better than the older i7-8xx CPU's, almost achieving the level of the i7-9xx range of processors, but is still limited:
The Sandy Bridge is unsuitable for anything more than a KISS system.
Why? Because it lacks the required PCI-e lanes to accomodate more than a 16 x PCI-e nVidia card with CUDA support to enable hardware MPE acceleration and the integrated graphics are not supported by CS5.
You may wonder if that is a bad thing. The plain and simple anser is NO. It is a great processor, it delivers great value for money, is a solid performer, but it has its limitations. Intel had a reason to position this CPU as a mid-level CPU, because that is what it is, a mid-level performer in comparison to what is to come.
The term mid-level performer may seem strange when compared to the old generation of i7-9xx CPU's, because they perform almost equally well, but keep in mind that there is a generation difference between them.
So what about the i7-9xx and X58 platform?
It still is going strong. About the same performance as a Sandy Bridge, with only the much more expensive hexa-cores clearly in the lead, both performance and price wise. The quad cores deliver about the same value for money. The main difference however is the platform that allows a dedicated raid controller to be installed, thus making it the platform of choice for those who want to go from a passing KISS to true LOVE.
And what lies ahead?
Sandy Bridge E on the Waimea platform (X68). Now that is revolutionary. More than double almost everything a processor can offer: double the cores, double the PCI-e lanes, triple the memory, more than double the L3 cache, increase the PCI-e support from 2.0 to 3.0, etc...
This is why Intel calls this a high-end CPU / platform.
So what now?
If you prefer a KISS approach, choose either a Sandy Bridge/P67 or an i7-950+/X58 platform.
If you wonder whether in the future you may need multi-cam more frequently, edit more complex projects and longer timelines or even progress to RED, look at KISS/LOVE solutions, meaning the i7-950+/X58.
If you can't have downtime, time pressure is high, delivery dates to clients are critical or you edit highly complex projects, lots of multi-cam situations or lengthy time-lines, choose a LOVE solution, an i7-950+/X58 platform.
If you have the time to wait till Q4/2011, Sandy Bridge E/Waimea looks to be worth the wait.
Hope this gives you some more insight into recent and future developments and helps you make wise investment decisions.