Starting from scratch!!!

[Miz Ducky]

Thank you all for your suggestions. I've just purchased a 2.8 P4, and I'll be buying the rest (mobo, memory, drives, case) very soon. I'll reread the video card info when I'm shopping for it (that will be soon too). I've actually been putting this off for a very long time, so I'll probably be going all out (reasonably speaking of course).

[Miz Ducky]

BTW
Sorry it took me so long to get back. Hurricane Frances had me living without cable (first - no internet) then no power (nothing but candles Damn it got hot!) for a few days.

[madmag9999]

heh that must have been a bad storm i got rain from it up here in pa.

heh intel meh to each his own. and as for video cards just take our suggestions and buy what u think is best, if ur short on funds id suggest one of the lower ati cards like a 9000 maybe. also when buying video cards u dont have to buy an ati or nvidia card just buy a card with the chipset u want. its much cheeper and works just as good.

[dwchang]

about cpu's, your best bet now would be to get a socket 939 amd64 . thats the new socket for the amd64 and should last alot longer than the previos socket, making it good for futer upgrades. its probably the best consumer cpu for vid editing now, unless you wana sell your house and buy a a fx53 . anywaythe new socket amd64's should be out real soon, if they arent allready.

They're out:

<a href="http://www.pricewatch.com/h/prc.aspx?i= ... ">Athlon64 3800+</a>
<a href="http://www.pricewatch.com/h/prc.aspx?i= ... ">Athlon64 3500+</a>
(both Pricewatch pages)

Heh, I actually worked on these products and wow...a lot cheaper than I thought they'd be.

Anyway, sjcks they got a P4 2.8 b/c well....no 64-bits and that is the future. Sooner than you think

[DrngdKreationz]

At Jacon 04, people in the AMV panel mentioned that some video cards come with premier. Is that true? I haven't seen it yet. Which ones come with the software that can be used for making AMV's? if any.....

OK as for you video card question about what was mentioned at Jacon. was that they were talking about video capture cards, some come with either ulead, or premiere for example thiscapture card here comes with premiere 6. now thats just an example, but that is what they were talking about. I personally don't think its worth is unless you're going to be doing actual video capture (if you're getting it off a DVD then you don't need a capture card)

[LovEnPeaCE]

[quote="Otohiko"]
Speaking of the 4200, it's a great card, to sort of address madmag's point. ATI was indeed ways ahead with the 9XXX series over GeForce FX, but nvidia seems to have held the crown with the GeForce up to 4 Ti series and now with 6

yes, the geforce 4 ti series were very good, high quality cards for a lower price, but they are harder to find nowadays because they really aren't being made anymore

[Miz Ducky]

heh intel meh to each his own.

Fine be that way!

Actually, the reason for me is that I've been running a P3 for 5 years and have NEVER had a problem. (well I caused a software problem once, but it's not the computer's fault) Like I said in my first post, it's been very good, but I'm afraid that if I put it under any stress it will have a seizure and die. Sooo, I guess that's what made up my mind. I did read all the reviews, but excluding the 64, the AMD and P4 seamed comparable.

So THERE! humph. j/k

Well my current quandary comes from RAID.
I want a SATA HD. A lot (not all) of the mobos I've been looking at have Raid. The mobos also seem to have issues starting from an SATA device.
I'm wondering. What reasons would I have for using RAID? Do you all use it? What specifications have to be met to use it? What do all those 0's and 1's mean?
Would it be best to start with an IDE HD and then incorporate a SATA later?

[Zero1]

I'm wondering. What reasons would I have for using RAID? Do you all use it? What specifications have to be met to use it? What do all those 0's and 1's mean?
Would it be best to start with an IDE HD and then incorporate a SATA later?

Reasons for using RAID:

Why use RAID you say?

1) Data redundancy (Backup if you like)
2) Increased data transfer rates
3) Use all drives as a single unit

1) Data redundancy:
Data redundancy is a feature offered by all RAID levels except RAID 0. It requires 2 drives minimum (depending on the RAID level) of equal size. Basically what happens is whatever is written to the Primary drive is copied (more commonly reffered to as Mirrored) to the second, mirrored drive.
In short it creates an exact copy of the hard drive you are using onto another harddrive. If the drive in use fails or plain dies, the other drive takes over.
RAID 1 can improve read transfer rates, but write rates suffer.

2) Increased data transfer rates
The sole reason I went for RAID. Harddrives are good these days, and the Western Digital Raptor is one of the best, however, there is only so fast a drive can spin and read. The head can't be in 2 places at once... With RAID it can be. As in the case of RAID 0 (used for maximum performance) you can have 2 - 8 drives on a single controller, and when you write data to it, the data is split over the drives, so you can have 8 drives writing or reading a file at once as opposed to a single drive. Buying multiple small drives usually costs around the same as buying a single drive with the same capacity, the only additional cost is the RAID controller.

3) Use all drives as a single unit
Another great thing about RAID is that you can connect loads of small drives, and have the Operating System treat all the drives as one large drive. The advantage here is obvious. Say you happen to have 2 x 40GB harddrives, but when you rip a DVD and transcode it to huffyuv, the file would have been around 60GB... Problem. The first thing some might do is go and buy a larger drive. With RAID you could connect these drives and it would look like you have a single 80GB harddrive, problem solved.

It's ideal for fileservers too. I was speaking with derobert a while ago, and I believe nago (the server you get your local videos from) is circa 960GB with over 500GB used if memory serves correctly. This would have been very difficult and/or impossible without RAID. Bear in mind the largest HD at the moment is 400GB, a Hitatchi Deskstar (dubbed Deathstar due to it's reliability)


Criteria to be met:
As for specifications, as you mentioned some motherboards come with onboard RAID. Don't let this be a deciding factor in which motherboard you get because you can get PCI RAID cards (which is probably better performance wise anyway). Some manufacturers I would recommend are Highpoint and 3ware.

RAID cards come in a number of different flavours. Some are ATA, some are SCSI, and some are SATA. ATA and SATA are the most common and they are what you should concentrate on. I reccomend SATA strongly because it has a higher throughput than conventional ATA, and is also more scalable. ATA has reached 133MB/s. The current SATA controllers/drives are 150MB/s standard, and is said to reach 300MB/s for second generation SATA and 3rd gen. SATA is said to hit 600MB/s. Also take note that motherboard manufacturers are now including SATA ports; they will be trying to phase out ATA, but not just yet.

From a connection point of view, the SATA drives use a much smaller interface cable (I'd get shielded cables if possible) thus reducing the mess inside the case and increasing airflow that little bit extra. One thing to beware of is the power connector. Most new power supplies have SATA power connectors, I got a Vantec Stealth 520W, but it only has 2 SATA power connectors. Fortunately for me, I have Western Digital drives which can also use the standard molex connector. Failing that, it's possible to find converters in most electronics retailers such as Maplin, and I'd expect PC stores to carry them as a matter of course.

If you plan on building youself something new, don't forget to get a new power supply, It's unlikely that the current one will be adequate. One reason being that all new motherboards (Intel and AMD) have an extra connector on the motherboard (a 4pin +12v). While maybe not essential, it's supposed to help maintain stability.

But the basic components for a simple RAID setup (depending on intended usage of course) would comprise of 2 Hard drives and a RAID controller (either on board or PCI). Never use software RAID, it's slow and IMO is a ticking timebomb.


I've been using RAID for 9 months or there about.

To explain a bit further, here a description I wrote up covering the main RAID levels. The most common levels you will hear about are RAID 0, 1, and 5

RAID Levels:

Although not necessary, it's generally best to use identical drives of the same age size and wear. This will help minimise problems and keep the drives in sync.

RAID 0:
Requirements: 2 drives minimum
Available capacity = 100%

RAID 0 offers improved read and write transfer rates. It does this by "splitting" the files into blocks (of a specified size in the controller's BIOS, I'll talk about this later) over multiple drives, so that the file can be read from multiple drives at the same time, as opposed to a single drive.
It also allows you to combine the capacity of multiple drives, and the OS will detect it as a single drive. It requires 2 drives minimum and can be expanded to 8 (there may be cards supporting more, it may even be possible to RAID arrays over multiple controllers, I'd have to check that though). RAID 0 is strictly a performance mode and offers no data redundancy (backup), that is if one drive dies and/or the array is broken, you lose your data. There is a utility called Active@UNDELETE that may be able to recover broken RAID arrays.

RAID 0 is commonly used for video editing since it allows you to have a much larger storage than any single drive available whilst appearing to the Operating System as one large drive. Currently I have access to a 400GB single drive (though it's a Hitatchi "Deathstar"...) and I believe it's the largest commercially availble drive as I'm typing this. Now 400GB is mountains of storage. Highpoint do an 8 port SATA RAID card, so imagine 400GBx8 = 3200GB or 3.2TB! Thats more storage than the .org's server!, and all this would show as a single 3.2TB hard drive. Nice!

I currently run a modest 120GBx2 in RAID 0, it's ideal for editing Huffyuv files since they require a large throughput.

As for setup, arrays (a name given to a set of hard drives operatining in RAID) are created in the BIOS. It's usually a plug in BIOS that appears after the POST screen. RAID 0 arrays are easy to create, first enter the RAID card's BIOS, then you select the drives you want to have as an array and specify the stripe size the stripe size defines how often files will be split. My particular card allows 16, 32, 64, 128, 256, 512, 1024, 2048KB stripes. This basically means that if you have a 16KB stripe size and you write a 32KB file, 16KB of that file will be stored on one drive and the other 16KB on the other drive. The operation alternates.

There is a general rule of thumb for setting stripe size, if you want to optimise transfer rates of small files, set a large stripe size, if you want to optimise the transfer of large files (such as videos with high sequential reads) set a small stripe size.


RAID 1:
Requirements: 2 drives minimum
Available capacity = 50%

RAID 1 offers data redundancy, which is essentially an identical copy of your harddrive. You also get improved read transfer rates, but write transfer rates suffer. You will get roughly double the transfer rate of a single drive for read transfers but the write transfer speed is the same as a single drive, so this mode only increases the read speed.
As with RAID 0, it requires at least 2 drives, one as your Primary drive, and one as an identical copy. The drawback to RAID 1 is that you need double the amount of storage required for it to work. This means that if you want 120GB available for use, (for arguements sake let's say it's 1 x 120GB drives) you will need another 120GB for the array to be mirrored. So in short, you only get use of half the available storage, so in this instance on a 240GB RAID 1 array, 120GB would be available, and the other 120GB would be an identical copy of the other drive.

RAID 1 is suitable for fileservers and mission critical systems where near to 100% uptime is required. Dependant on which drives die, it can sometimes survive multiple failures.

RAID 1 is an expensive, yet very reliable type of array, it's exactly the opposite to RAID 0. Another rule of thumb is RAID 0 is used for performance and tempoary storage, whereas RAID 1 is used for permanent secure storage. Both cases offer the ability to have arrays much larger than any commercially available drives.

This probably isn't a good choice for your average user since it's very cost ineffective. You basically have to think of a number and double it, so this is why it's generally used by large companies such as banks and places that hold valuable data. For end users that require data redundancy, they usually choose RAID 5 as an alternative.


RAID 2:
Has been omitted

I've decided to leave RAID 2 out of this explanation because the entry level cost very high and requires very high transfer rate requirement to justify. I'm also led to believe that no commercial implementations exist and/or are not commercially viable. In other words, it sucks.


RAID 3:
Requirements: 3 drives minimum
Available capacity: Sum of all drives less the capacity of 1 drive

This is my favourite RAID level, it offers the best (or most of) of two worlds, in this case it gives you data striping like RAID 0, allowing high read and write throughputs and also employs data redundancy with the need of only 1 extra drive. So how does it offer data redundancy with only 1 extra drive, whereas RAID 1 requires as many extra drives as there are user available drives? It uses a simple, yet absolutely genius formula to write data to the spare (or Parity as I will refer to it as) drive. I'll start with a 3 drive setup to keep things simple.

Say for instance you have a RAID 3 array that consists (3 in total, 120GB each) of 2 striped drives and 1 parity drive. You will have 240GB available for your use, and a single 120GB drive set aside for Parity information. As you may know, data is stored as bits (binary digits), which can be represented as on or off, or 0 and 1. Now if you don't understand binary, don't worry, it's relatively easy. The way a parity works is it adds the data of drive 1 and drive 2 together, and stores the answer on drive 3. It's just a basic equation like a + b = c. Where a is Drive 1, b is Drive 2, and c is Drive 3 (the parity drive). Now from this equation (providing you have at two of the values) you can work out the missing value. This is exactly how the RAID controller works if a drive would happen to fail. A binary value can only be a 0 or a 1, so how does this work?, the number rolls over.

So if drive 1 has a value of 1, and drive 2 has a value of 1, unless you are bad at maths, that makes 2! But as I said, a drive can store a 0 or a 1, so since it passes 0, then 1, and there is no number 2, it cycles back to 0. Here is a quick table of the combinations for the array

Code: Select all

Drive 1		|Drive 2		|Drive 3 (Parity)
-------------+-------------+-------------+
		0		|		0		|		0
		1		|		0		|		1
		0		|		1		|		1
		1		|		1		|		0

Using the formula Drive 1 + Drive 2 = Drive 3, if one of the values were missing, you can find the missing value by reversing the operation.
So if Drive 1 has a 1 written to it, Drive 2 is dead, and the Parity Drive has a 1 on it, well 1 - 1 = 0 (using the formula Parity - Drive 1 = Drive 2)

Well that's a pretty crude explanation of it, no doubt someone will pick me up on something (seeing as I'm not a techie lol)

The great thing about RAID 3 is that you can have as many user available drives as you want, and you only need one drive for the Parity since the parity is the sum of all drives. I believe the method is called XOR, though I'm not 100% certain.


RAID 4:
Has been omitted

RAID 4 is pretty much the same as RAID 3, the only notable difference being that the data is stored across the drives in blocks as opposed to striped.


RAID 5:
Requirements: 3 drives minimum
Available capacity: Sum of all drives less the capacity of 1 drive

RAID 5 is where it starts getting messy. RAID 5 stores data as blocks, as opposed to being striped. It also offers the same level of redundancy as RAID 3 and 4, but doesn't have a dedicated Parity drive. Instead it stores the Parity in a distributed location. So instead of storing all the Parity information on one Drive, it stores the parity information across the whole array in blocks.

As with RAID 3 and 4, RAID 5 still uses the simple but clever XOR technique.

It has a high read transfer rate, and a medium write transfer rate. Block data will only transfer as fast as 1 drive since it's not striped. Also rebuilding a RAID 5 array is said to be difficult. In the event of a drive failure, it has a medium effect on transfer rates.

RAID 5 is probably the most easilly available RAID level with good efficiency and cost effectiveness to end users. It's one of the more common RAID levels compared to RAID 3 and 4.


As for 0's and 1's, I assume we are talking binary digits, theres an explanation in the bir abour RAID somewhere


Well until next time

[Miz Ducky]

well, well, well! That is way more than I could get out of the guys at work! I'm lovin' this! Thank you guys all so much for the info! If you were here I'd probably want to hug you!