I just purchases 4 BenQ 1650 drives and it took 20 mins to burn 2 different data dvd-r on the fly simultaneously which involves all 4 drives working. Previously I can achieve 10 mins using Lite-on drives. All 4 drives are attached to my primary and secondary ide channel. It doesn't matter if it is burning from master to slave or vice versa, or from the same IDE channel or different IDE channel. As long as all 4 drives are working, burning at "8x" takes 20 mins. However if I burn only 1 data dvd-r which only involves 2 drives working, it can complete on the fly burning at 10 mins which is what I expected. I'm returning these BenQ drives for Lite-on and will never look back. Plus it only uses green light, it's hard to know if it's burning or reading or idle. BTW, I'm using Nero.

Liteons are UDMA4 whereas BenQs are UDMA2. But there are other things to take into consideration too. I did some tests a while ago (with three burners burning simultaneously) and realized that if I fed images from 3 different hard drives to 3 different burners there were no bottlenecks (2 BenQs as sole devices on IDE1 and IDE2 and my Samsung via IDE to SATA adaptor). I then tried to feed the same image (residing in my fastest Raptor drive) simultaneously to all three drives and the burns took about 20 minutes to complete. The software/hardware buffers were erratic, which means that even the Raptor, (one of the fastest hard drives in the SATA market), cannot keep up with feeding 3 burners at once. And my rig is fast, look at the specs below.

The last test I did was to feed 3 different images from the Raptor to the three burners. Again there were bottlenecks with the drive buzzing like a demented wasp, trying to read and feed from different locations on its surface. I also found that there was a slight increase in PIEs in the last two scenarios. I don't know how realistic these tests are, but since then I only feed 1 image per hard drive per burner, and had no bottlenecks. All tests were done under minimal system load conditions with no other programs running in the background...

You are so right, I have a raptor too. I just burn a DVD video using Nero to 2 BenQ 1650 and it took 15 mins at "8x". I've burnt to 4 lite-ons at 16x in 6 mins with no problems. All the good reviews about BenQ are misleading when it comes to speed for reading and burning.

Is there a DVD burner higher than UDMA4?

As far as I know there are no UDMA5 burners, but there are SATA burners. I suspect that such writers work the same way my ATA to SATA adapter works on my Samsung burner. In Device Manager it shows as "Serial ATA Generation 1 - 1.5G", but I'm pretty sure that the real standard is really UDMA2 since the drive itself is not capable of more. SATA writers provide the connectivity of SATA without the need of a standalone adaptor, but I seriously doubt that they are nearly as fast as the SATA standard allows. More input on this matter from more experienced users would be welcome.

Unless you can show us some proof, I found this a false statement. I have many UDMA4 drives and burning 16x (or 16x on-the-fly) from the same IDE channel?? There's a difference between reviewing a single drive performance and multi-drive performance.

Leave it to Zevia to get and keep those drives humming!

I was about to say... I've yet to read a review about reading and burning on any drive that mentions its performance "on the fly" with 2 drives or more. I've only ever seen tests for "on the fly" with one drive and burn/read/ripping for one drive. One drive might have been measured against other drive's performance, but never all together like you mention, kamen

I meant burning to 4 lite-ons from a raptor hdd. On-the-fly usually takes me 9 mins to 3 lite-ons since one is being used as a reader.

kamen555, it is still hard to believe that you can burn a single image to 4 burners at 16x sucessfully at 6 minutes, even from a raptor hdd and four UDMA4 drives. A raptor hdd has a very fast seek times but there are bottlenecks in your system that made such task impossible.

Also on-the-fly from 1 reader to 3 burners are even worse especially with your setup (2 IDE channels for 4 burners).

From my past experience, I can only get max 12x on-the-fly copy on two UDMA4 drives (Lite-On to Plextor) in the same IDE channel, burn time about 8 minutes. When I tried 16x the buffer and write speed will go crazy and thus it will void the writing strategy used, plus the burn time was actually longer -- 10 minutes.

Using two UDMA2 drives in the same IDE channel I can get max 6x though.

Anyway, what software did you use to copy on-the-fly? If you want to make lots of duplicates in short time, I would suggest trying ImgBurn with Queue. You can open multiple instance of ImgBurn at once. In your case I'd suggest opening two ImgBurn, add the same image multiple times in the Queue, and burn them to two burners simultaneously at 8x or 12x to get successful burn quality. Also don't forget to set "Insert Next Disc - Auto OK" to 1 second so it will take the next burn almost immediately and you can insert another blank media to the idle drive.

Hi,My Samsung SH-W163A is recognized as UDMA-2 device by the controller. Nero CD/DVD Speed reports a burst rate of about 50MB/s.

Michael

zevia, I use Nero, from my experience it's the most versatile software. I can only specify the time it took to burn and not the speed because Nero doesn't display the actual speed of the burn. When I get my new Lite-ons in a couple of days, I'll do a burn from raptor HDD to 4 lite-ons, maybe it's not 6 mins but 7 mins at 16x. I usually only need to make 1 copy of each source of data DVD-R, so I avoid making an image on HDD at all costs since it'll take longer, so I don't think Imgburn will help in my case. By your comments 8 mins will be the fastest I can achieve for on the fly burning which sounds correct to me, I usually get 9-10 mins.

Is there a review/tutorial on how to built a fast duplication system on your home PC somewhere on the net? I would like to know how to build a fast burning on the fly duplication system that consists of 6 or more drives.

To show the actual speed in Nero 6 or 7, you can grab a registry edit here: ShowExactSpeed. It's in my sig for a while btw.

copying on the fly on the same ide cable is like two way traffic on a one lane road,
forget dma's, irrelevant.

how long does a 2 meg cache hold data?

Hi Zevia: I believe they have enabled this in Nero 7
http://club.cdfreaks.com/attachment....id=82806&stc=1

One thing puzzles me though, why would one prefer a UDMA2 over a UDMA4 dvd burner when prices are the same?

Because U2 or U4 is in most cases not relevant for what the hardware is made for: producing (good) burns.

michael/mciahel: Same here, the 163 bursts like hoyada.

Speed matters though. Shouldn't it be possible to achieve high speed and good burns at the same time? It seems this can only be achieved at UDMA4.

I think the reason is one of safety. It's safer for BenQ to go at a lower speed than a higher one, since electrical noise problems will always be magnified at higher speeds. This is important since the drive will be mated to all kinds of motherboards - old and new. And when you're selling a drive for $30.00, each support call instantly negates all profits for that drive (and a few others), so safety matters.

BenQ is louder than Lite-ons, at least 1650 vs 1693, I think it's because it spins at a higher rate.

zevia is right, i'm wrong. Just tested my setup, the time is so way off from 6mins.

I get variable speeds using Nero 6 and my BenQ 1650. The best I've had for a full SL burn is slighly less than 6:30, but I've had up to 8 mins on others.

I think it depends on buffer underruns (drive slows down so buffer can keep up) and caching of files. I've found burning single file ISO's is a lot faster than burning files and folders

deleted

This is interesting.

I only have 2 burners installed in my system - both on the motherboard's secondary IDE channel, one set as master and the other as a slave. Currently, one of the burners is a BenQ 1640 crossflashed to the external version of firmware so that I can do DMA mode-4 (instead of the standard firmware's mode-2 limitation) high speed transfers. The other is an LG 4163 (only supports DMA mode 2). As long as one of these drives is running DMA mode-4, I can use Nero to write the same data simultaneously to both of them at 8X successfully. However, if I attempt to go any faster than 8X, the transfer rate typically collapses - often to perhaps 1X or maybe 2X burn rate. Sometimes I have to reboot before I can again write to either of the burners at full speed. While I am able to burn to both drives successfully at 8X, I'm a bit paranoid when it comes to burning DVD's, so I always configure Nero to verify the burned data after a write. Since, unlike the write, I don't have control as to how fast the data reads back, Nero will attempt to error-check data from both drives simultaneously - and at full speed. At this point, the transfer rate will again collapse to 2X - 3X (note this is during read-back so the drive's write buffer level has nothing to do with it).

I researched this issue briefly and I came across information which suggested that Windows itself checks the IDE bus for errors and will automatically lower the DMA mode if any errors are found. I concluded that my problem related to the fact that I was attempting to overload the secondary IDE controller with too much data and was causing these IDE errors to occur.

I've since debated about adding a secondary IDE controller to try and work around these issues - but have not yet done so. So I'm a bit surprised to see that you're running into something similar, but with only one drive per IDE channel.

If I were troubleshooting this issue, I'd consider some of the following:

1. I'd make sure that all of my IDE interconnect cables were 80-conductor ribbon cables (and not rounded ones, since I've heard of too many problems reported about some of these). You don't want the system to be resending your data because of a checksum error on one of these cables.

2. I'd worry that my power supply was good enough to power all of these drives (especially since a drive tends to burn a lot of power while it's writing) and that it was clean enough to keep electrical noise to a minimum. Again, you don't want electrical noise induced into your IDE cable causing checksum errors and retransmits. It's not clear to me how I'd go about checking this, though...

3. Perhaps the motherboard itself is having problems moving all of that data? Again, I don't know how to check this, other than perhaps temporarily replacing some of the IDE burners with IDE hard drives, and then retrying the transfers to see if data moves properly. This is a pretty extreme measure, though...

DW1650's aren't UDMA4? (so they're slower than say a Pioneer 111D ?)

Is there any point using a 80pin IDE cable with a DW1650 then?

The most likely problems are:

1) More then one drive on an IDE channel. This doesn't work, as there is too much latency switching between devices on an IDE channel, so don't do it.

2) You are pushing a single hard drive too hard by forcing it to feed too many burners with data. Consider yourself lucky if you can feed 2 burners from one hard drive.

3) You are saturating the PCI bus on your motherboard. This will occur if you use PCI IDE/SATA cards and often if your motherboard contains an extra SATA or IDE controller (like the promise ones)

4) Copying DVDs on the fly. Readers are not even fast enough to generally even handle copy-on-the-fly between one reader and one burner.

Things it isn't:

1) CPU load. The CPU should have no problems moving the data around, as it is mostly DMA between devices.

2) DMA modes. Ultra DMA mode 2 is fine for 16 speed burning. Mode 4 is better, but if set up correctly, it won't make a difference.

Things it probably isn't:

3) The power supply. Hard drives and Optical devices are not huge power drains (unlike CPUs and GPU's) .... if you have a half decent PSU, you should have no problems.


Suggestions I can think of:

1) If you have a really powerful system, or speccing a new system, put 8gig of memory in it and set aside about 5gig as a ramdrive. Use this to cache DVD images to burn to multiple burners, rather then hoping a hard drive is able to handle the load. A ramdrive in system memory will have awesome multiple access performance as its transfer rate is infinite (doesn't need to transfer data anywhere) and it has very quick access rates (way less then 1ms)

2) If you can't do that, you should have 1 drive on each IDE channel and use SATA hard drives. Have 1 hard drive for each burner. For example, if you want to use 4 burners, you can do the following:

a) Install a PCI ide card. Put 2 optical drives on it (1 to each channel) and 2 optical drives onto the motherboard.
b) Install 4 SATA hard drives (whatever you can find, cheap is good) and install them onto the 4 motherboard SATA ports.
c) Burn images from each separate drive to each burner.

4) If you need more SATA ports, get a PCI-E card rather then a plain PCI card. That way, you won't overload the PCI bus.

rather expensive experiment, please post results, screen shots etc