Best 30gb+ IDE HD for video capture? (1 Viewer)

[JasenP]

I need a big and fast IDE Hard Drive. I have upgraded my system RAM and I am still getting a high amount of frame loss, I am going to try a faster HD.
Any suggestions or online resources?
------------------
"Strange women lyin' in ponds distributin' swords is no basis for a system of government!"

300+ Reasons why I drive A 1991 Cavalier.

 

[Rob Gillespie]

IBM GXP range. The newer Maxtors are really fast too but the IBM track record has shown greater reliability.
Make sure you're running on ATA/66 with optimum IDE channel arrangement, with a fast processor on a good motherboard.
If this is still too slow, be prepared to spend $$$ and go to SCSI.
------------------
Link Removed
"One does not simply walk into Mordor."
Link Removed

 

[Abdul Jalib]

What is your configuration? I would suspect something other than the disk drive, though it's possible that overengineering the disk drive would compensate for your real problem.
I use Western Digital Caviar WD600 60 GB 5400 RPM drives and Maxtor 30 GB 5400 RPM drives for HDTV recording without a problem - quiet and cool. I have a WD 30 GB 7200 RPM drive for my main system drive, and it's overall quiet but not silent. I believe it has gotten good reviews for being a quiet 7200 RPM drive. Most of them are pretty quiet nowadays, though. TiVos are Linux PC's that use 5400 RPM (or even 4500 RPM) drives without a problem, usually.
The recent IBM drives have had a high failure rate. There is some speculation that the problem is actually a bug in Windows, not the drives, and that IBM drives are just more susceptible due to having bigger caches.
------------------
http://groups.yahoo.com/group/Philips_HDTV/joinClick to subscribe to Philips_HDTV discussion group

 

[Rob Gillespie]

The recent IBM drives have had a high failure rate.

I think that was restricted to a large batch of 46gb drives in the 75GXP range, all from the factory in Hungary. The same drive from other parts of the world (my two were made in Taiwan) have no such issues.
------------------
Link Removed
"One does not simply walk into Mordor."
Link Removed

 

[JasenP]

I have a 500mhz K6 with 380mb of RAM and I am using an ATI All-In-Wonder 128 PCI (No AGP slot, thanks again Compaq!)
Is there a better software/capture codec combination that might help?
------------------
"Strange women lyin' in ponds distributin' swords is no basis for a system of government!"

300+ Reasons why I drive A 1991 Cavalier.

 

[Ken Chan]

Capturing is a combination of three things: frame size, compression time, and disk transfer rate (for writing of course).
For example, you could capture full screen (704x480 or 720x480 with NTSC) with no compression. Then CPU speed doesn't matter and all you need is a really fast hard drive subsystem (in the neighborhood of 15 to 20? MB/sec).
You can get away with slower drives (like a TiVo) if you have effective real-time compression, probably done with hardware. An HDTV transport stream is already compressed at -- what -- 19.4 Mbits/sec, not even 3MB/sec. Any modern drive can handle that.
Note that it is possible to drop frames if the codec can't keep up. The worst situation would be a CPU-intensive codec writing to an IDE drive using PIO instead of DMA, with the CPU having to oversee the data transfers, all with a low-end CPU.
Looking at the specs for the All-In-Wonder, it looks like the compression is CPU-dependent (although it may be hardware-assisted to some degree). There's a small matrix listing quality for certain frame size and CPU class combinations.
I would first try capturing using the smallest supported frame size. If you can't capture, say, 160x120 without frame loss, something is fishy. You might also try turning on DMA for the drive (under the Device Manager properties).
Any new IDE drive you buy today would probably be fast enough for moderate captures; it's everything else that could be the bottleneck: drivers, bus, CPU, codec.
Have you run a benchmark on your current drive? You can download SANDRA for free from download.com. It has drive benchmarking.
//Ken

 

[Shayne Lebrun]

SCSI RAID chain. It'll cost. Lord, will it cost.
In this case, RAID 0 (striping without parity) across, say, three or four 18 gig, 10K RPM drives.

 

[JasenP]

Ken, thanks for the Sandra Tip, what a great piece of software!
DMA is on and I am running Windows ME
Here are the results of the benchmark:
Buffered Read: 2847 kbs
Sequential Read: 2862 kbs
Random Read: 1884 kbs
Buffered Write: 2886 kbs
Sequential Write: 2889 kbs
Random Write: 2884 kbs
Average Access Time: 12ms (estimated)
The only warning was a Low Disk index?
Can anyone interpret these results for me?
------------------
"Strange women lyin' in ponds distributin' swords is no basis for a system of government!"

300+ Reasons why I drive A 1991 Cavalier.

 

[Abdul Jalib]

I'm not familiar with the K6 motherboards, but with Duron and Athlon VIA-based motherboards, you have to get the 4-in-1 drivers for just about anything to work right. Anyway, you should definitely go to your motherboard manufacturer's website and download and install the latest BIOS and drivers for your motherboard.
------------------
http://groups.yahoo.com/group/Philips_HDTV/joinClick to subscribe to Philips_HDTV discussion group

 

[Rob Gillespie]

Jasen, I've just downloaded Sandra and done the same tests.

This on the first partition of my first drive, an IBM 75GXP 46gb. The drive runs on UDMA66.
[*]Buffered Read: 58mb ps[*]Sequential Read: 36mb ps[*]Random Read: 7mb ps[*]Buffered Write: 35mb ps[*]Sequential Write: 31mb ps[*]Random Write: 9mb ps[*]Average access time 7ms

These results are from the last 10gb partition on the second hard drive, same model and UDMA66 again:
[*]Buffered Read: 57mb ps[*]Sequential Read: 25mb ps[*]Random Read: 6mb ps[*]Buffered Write: 34mb ps[*]Sequential Write: 24mb ps[*]Random Write: 8mb ps[*]Average access time 7ms

As you can see, there is a little bit of a performance hit by using the outer most parts of the disc.

This is the first time I've used Sandra so perhaps someone else can put the results in context.

------------------
Link Removed

"One does not simply walk into Mordor."

Link Removed

 

[Ken Chan]

Rob, tracks are numbered from the outside in, with the outer tracks -- and earlier partitions -- being faster, as your scores indicate. (The outer tracks are longer, and have more sectors; discs are CAV, so the outer tracks have more sectors passing under the drive heads per second.)
Jasen, I believe the low disk index just means your scores are really low, which they are 3MB/s sequential write just won't cut it; 10MB/s is a good minimum to shoot for.
The fact that your buffered and sequential read scores are practically identical may be because the interface is the limiting factor. Unfortunately I could not find info in Sandra that reports which interface your drive is using. (Some BIOSes report drive modes on summary screens during bootup, but that's probably the modes the drive supports, not the mode that is actually being used by the OS, especially since the OS hasn't loaded yet.)
Note how Rob's buffered read in the high 50s is near the theoretical maximum 66 for a UDMA66 interface, and considerably higher than the sequential read. (Just how old is your computer anyway? The only multi-word DMA mode that would match is the oldest, mode 0, with a max of 4.2MB/s. Do you know the brand/model of your drive? What's the size?)
If that is the case, simply buying a big new drive would not help. You would also have to buy a new PCI IDE controller card. UDMA100 is the current standard, but it's wasted bandwidth, because there are no drives remotely close to 100MB/s sustained transfer rate. A UDMA66 controller would be fine. Then you'd want to disable the motherboard controllers.
Or maybe there's some magic BIOS setting you're missing that would unleash your drive's potential (seems unlikely, though). Since you're using ME, it should have reasonably good DMA drivers, so that's probably not the problem.
I recommend checking out storagereview.com. They review lots of drives. If you check out the sustained transfer rates (for current reviews, under "Low-Level Measurements") you'll see that current drives have read speeds in the 20s and 30s, even 40s. Writes would be a bit lower than that (unfortunately not listed), but still within the desirable range. Also check out the reference section -- it has more than you want to know about hard disks (actually reprinted from pcguide.com another good site).
//Ken

 

[JasenP]

Would it help if I created a partition that was used exclusively for video capture? If so, How large should it be? I have a 12.5 gb HD right now, it's small but if I can use a partition to test and see if this eliminates my dropped frames problem..I may have to try it.
------------------
"Strange women lyin' in ponds distributin' swords is no basis for a system of government!"

350+ Reasons why I drive A 1991 Cavalier.

 

[Larry Carlson]

For what it's worth, the week's Best Buy flyer has a Maxtor 60gb 7200 rpm hard drive for $170 - fast drive and the best price I've seen for it.
Larry

 

[Ken Chan]

Would it help if I created a partition that was used exclusively for video capture?

Not unless you solve the interface problem first (if that is indeed the problem as I suspect). You can get a new Ultra DMA 66 PCI IDE [gee, what a string of mumbo-jumbo] controller for around $20. Put that in, attach your existing drive to it, turn off the IDE controllers in the motherboard and run the benchmarks again. See if that improves matters.
Then yes, you could optimize the drive by making your data partition first. If you get decent partitioning software like Partition Magic, I believe it's perfectly valid to have your primary boot partition (with the OS and your apps) at the end of the drive and your data partition inside an extended partition at the beginning (the faster end).
//Ken