HN10 - Very low speed burning disc to disc

vbimport

#1

Hi, I am newbie to this forum, and I’ll tell you my problem:

I am regularly copying Verbatim DVD+RW 4x to Verbatim DVD+R 16x with Recordnow on-the-fly, and although everything runs fine, it takes about 30 min to write a 4.7 GB disc. It begins at 16x but after having 100 MB writen, it falls down to about 2x and ends, let’s say below 4x. However, if I copy the DVD to HDD and then to the H10N, it varies between 16x, 10x, 12x.

So, I began to think that the problem is on the IDE bus. I have only one, with the LG H10N as master, a Liteon XJ-HD166 dvd reader as slave, the rest are WD 160 GB disks on SATA, two of them on RAID1. The processor is a Dell Dimension 8400, P4 3.2 Ghz, 3 GB ram. WinXP Home SP2 to the last update.

And Nero CD/DVD Speed gave me some reason as the reader and the writer, each one alone, took more than 50% of cpu utilization at its higher speeds.

I use 80-wire cable, but the Device Manager shows UDMA2 for both devices, that could be the reason for high utilization and low speed. Specifications for the Liteon reader says it is only UDMA2 capable, but I have seen in this forum that I should force H10N to UDMA4 by flashing some firmware from TDB.

There. at TDB site, I have seen three possible firmwares: JL11 and JJ11 for H10N, and UJ12 for H12N, that could be valid too. But I have also read about people that had problems with those fw and were not able to go back. By the way, my firmwares are the latest from Liteon (DD06) and LG (JL12).

I have also thought that riplock in the dvd reader could be part of the problem, but I don’t know if its possible to avoid it.

So, questions are: 1) Any other possible reasons for this performance? riplock? 2) Should I risk my writer box by trying those TDB firmwares? 3) Which of them, if it doesn’t matter too much 16x or 18x?, and 4) Any solution if I would decide to come back to the JL12 LG firmware?

Thanks in advance


#2

Hi, If you are using stock LG firmware for the H10 then it will show up as UDMA2. The buffalo JJ firmware changes it to UDMA4 and so does the UJ12 H12N firmware but its 18x so it needs to be. :bigsmile:

If your reading from a RW disc and recording on the fly the times will be slower. DVD drives do not read data from a RW the same way the do from a normal DVD , its speed is restricted to 4x pr 8x max. Even though the burner and media are more than capable of doing this. I would sugesst ripping to the harddrive instead of a RW then burning to your Verbatim media. Will be quicker. :bigsmile:


#3

Totally agree with this. And it’s much safer as any read error - quite possible with RW media - will ruin a blank DVD for you.

Additionally, it’s never going to be quick when both devices are on the same IDE channel.


#4

But the point is that I have a number of DVD+rw that I would like to copy to +r and leave my “+rw pool” free again. And if I copy the original to hdd, and hdd to the target, time for both parts is about 25 minutes too. And by the way, I have never ruined any dvd in that way, is just a matter of time to copy.

Anyway, I have just made the test of copying one Verbatim +r MCC 004 to another, on the fly, and the behaviour is exactly the same. It takes about 25 minutes for a 4.29 GB dvd and after copying 40 MB, write speed slow down to 2.3x - 2.7x until the end. So, no difference if the source is +rw or +r.

However, if I copy +rw to +rw at 4x maximum speed, it keeps the pace of 4x all along the copy time, taking about 15 min. But copying from +rw to +r, if I limit the write speed to 4x, it goes down to 2.4x and takes again 25 min. I cannot understand this. I have to check this again.

On the other hand, even with just one IDE channel, if I’d achieve a lower utilization, the performance should improve, mainly if the bottleneck is the IDE channel. Just queuing theory, but I could be wrong, of course.

And, on the contrary, if the cpu is heavely loaded in this environment, adding a second IDE by means of a PCI controller card, should not solve the problem, I think, if the cpu goes to 100% during the process (50% due to each channel as I have seen in Nero CD/DVD speed). So udma4 for one of both devices could alleviate the problem.

Do you agree with this reasoning?
Should I use the Buffalo JJ firmware better than the UJ12 to check udma4?
Does someone know if there would be a possibility to go back or not?

Thanks a lot


#5

4x will be about the max safe speed for an on-the-fly copy with both drives on one channel. Results will very regardless of the type of source disc used, as the reading drive may elect to slow down at any time. Optical drives are simply not a reliable source for sustained transfers, too many things can make them slow down. Once an under-run does occur, the writing drive will down-shift the burn speed anyhow. The UDMA status matters not, 2 or 4 is the same in this instance.

There is a firmware mod available for the LiteOn 166 that will give you 16x read speeds on all discs, I think. But that still does guarantee a smooth speed curve.

Reading a disc at 16x, then writing one at 16x will be faster than an on-the-fly copy at 4x.

CPU utilization of 50% due to optical drive reading or writing is certainly not normal, but CDSpeed is not a reliable indicator of CPU use, use task manager instead.


#6

If this takes 25 minutes, something is wrong. Post the result of the Nero CD-DVD Speed burst transfer test, you need a disk with some data in the drive to run the test. It should display about 23 MB/sec or more. Check the DMA settings of your source and target drive.
Also post a cdspeed transfer rate scan reading a RW from your DVD-Rom drive


#7

I mostly agree with this.

I’ve just ripped a DVD+R disc on my Samsung burner in 6:40. I then burned it to a 4x +RW disc & ripped again.
This time the rip time was 10:00. Add to this a burn time of , say , 8mins for 12x gives a total time of 18mins. In my case the Verb 4x RW disc was ripped at a max of 8x but if the reader was unable to reach 8x on the RW disc then we start to see the time get pushed out to closer to 25mins.

I’m ripping to a SATA HDD which is nicely defragged , but if ripped to a badly fragmented HDD we again could see the time get pushed out.

To go back to the on-the-fly situation on a single IDE channel. The speed fluctuation of the burner is being caused by the burner continually stalling as the data cannot be fed to it quick enough which is why a burn speed of 4x to an RW disc works best as the RW disc cannot be written to faster than 4x & the burner’s firmware is optimised for that speed and the speeds match.
When you burn to the DVD-R disc , presumably a 16x media, at 4x the burner’s firmware not being optimised for that speed will again keep stalling the burn. This last bit is more supposition than provable fact.

However, what is fact is that this continual stalling will dramatically affect the overall time taking as both devices will keep adjusting their read or write speeds.


#8

Just be doing a bit of testing.

All these tests are on the same IDE channel.

  1. DVD+R to DVD-R at 8x took 10mins.
  2. DVD+RW to DVD-R at 4x took 15mins
  3. DVD+RW to DVD-R at 8x took 20mins with an average write speed of 2.7x & the buffers fluctuated all over the place.

I couldn’t be bothered to waste another -R at 12x or 16x as I this this test substantiates what I was saying.

It seems it’s primarily the read speed of a DVD+RW that’s the real issue & once the reader can’t keep up with the writer then burn speed takes a nose dive.


#9

I agree with all of you. Thanks for your comments. I was wrong about the cpu load in the tests, that in fact is very low, below 10% in all the cases. I initially took the data from cd/dvd speed, that seems not to be meaningful for this case.

I have just done three tests, for a typical 4.3 GB DVD video :

  1. Rip from Liteon +RW 4x to HDD SATA RAID1 (mostly unfragmented). Result 8% cpu, 9 min, 8 MB/sec eq. to around 6x
  2. HDD to LG H10N dvd +R 16x. Result 5% cpu, 6 min, 12 MB/s eq. to 9x
  3. Disc to disc Liteon +RW to LG +R. 3% cpu load, 25 min, 3 MB/s, 2x
    In the first two tests, the speed kept more or less constant. In the third, one, it began at 16x and 1 min later went to 2.3 - 2.7 oscillating for about 5 min to go slowly to almost 4x till the end.

Is interesting to point out that the disc to disc load is much lower than the sum of the two first, so the copy time gets higher. This indicates, I think, that cpu load is not the bottleneck, so the discussion about udma2 or udma4 is nonsense. Btw, both devices appear as UDMA2 in Windows Device Manager.

But 8 + 12 MB/s could give 20 MB/s average, that could lead to maybe more than 33 MB/s in bursts. So, if the bus were limited to atapi/33=udma2 (??) this could be the limiting factor. But I assume that the bus should be ata/66 or even ata/100 because the ribbon cable is 80-wire. And this made me think that maybe the failure could be at BIOS level. In fact I have level A01 (Dell 8400) and there is an A09 available. I tried to find a specification for the speed of bus or devices in BIOS and I could not find it. So I will try to flash the latest BIOS in the next days.

When I was thinking that udma4 could be good, I saw a DVD reader Asus DVD-E616A3 supporting udma/100 (udma5?) for just 17 € that could serve as a replacement for my old Liteon. And I still keep on thinking that it could support a higher transfer rate (assumed I would flash the new Dell BIOS and there were no restrictions from the IDE bus side, if this would be the problem). Any idea on this?

Here are the results from cd/dvd speed reading a 4.38 GB dvd video:

Nero CD Speed Test results

General Information

Operating System Windows XP Home Edition (5.01.2600 Service Pack 2)
Drive JLMS DVD-ROM XJ-HD166
Firmware Version DD06
Serial Number
Disc DVD+RW
MID MKM A02
Capacity 4.38 GB

Transfer Rate

Start 0.86x
End 8.73x
Average 6.43x
Type CAV

Seek Times

Random 108 ms
1/3 121 ms
Full 212 ms

CPU Usage

1X n/a
2X n/a
4X n/a
8X n/a

Interface

Burst Rate 26243 KB/sec


I don’t know why the cpu usage shows as n/a. Yesterday I got
1x 11%
2x 18%
4x 29%
8x 51%

in the same test case.

Tell me, please, what do you think about this. BIOS upgrade? New ATA/100 DVD reader? A second IDE controller card? Or just “leave it as it is”?..

Thank you.


#10

Nonsense. Your reader (8MB av) is slower that the writer (12MB av), it can not supply the data rate needed. So it is quite normal that the poor burner does not know what to burn if the incomming data rate is so slow. Also the packets collide when using both drives on the same IDE bus, further reducing the data rate. If you insist on burning on the fly, try 8x burn speed.


#11

So, do you suggest that changing the dvd reader or adding a second IDE controller would solve the problem?

I would say to do the first and then write to 8x or sometimes do copys through hdd.

In that case, I have the doubt to buy the reader I mentioned before Asus DVD-E616A3 or Sony DDU-1615B2 or Teac DV-516G. Do you know some web site where I could compare characteristics or read some reviews about them?

Thank you very much


#12

The Asus would be a decent choice but even with a PCI card you’ll struggle because of the relatively low read speeds of 4X RW media.

And my tests show that trying to burn at 8x in your configuration is a no-no.

A problem is that many PCI IDE cards don’t support optical drives very well so unless you get the right one you’ll be no better off. There is a thread dedicated to PCI IDE cards that’ll be worth reading.

Personally I’d probably just copy to the HDD & burn from that.


#13

There’s nothing wrong with the LiteOn 166 as a reader. You may want to use the firmware patching tools available to unlock it’s read speeds. You can get the fastest reader in the world and it won’t change your situation at all. The problem is the IDE controller, not the drives being used. It’s also good to note that UDMA2 is WAY faster than the data rates required. UDMA4 adds nothing to the mix.

Also, referring to average speeds or data rates is misleading. Running a CDSpeed transfer rate test with the 166, using the different disc types, will show you the current speed curves on the different types. If the read speed starts out below 6x, (and most will), then your burn speed curve must also start out at or below that speed. The LG H10 will also ramp up the burn speed very quickly to 8x. So you have to compare the 2 speed curves. 8x read speed curves do not keep up with 8x burn speed curves.

Bottom line is that it’ll be much easier, and probably quicker, to copy to HD and burn at high speed.


#14

Ok. Now I am convinced that the best to do in my case is doing copies to hdd and then burn from there. I’ll do it that way. Is much easier than any other adventure.

Thanks a lot to everyone. I don’t have to tell you how much I’ve learned just in one day.


#15

According to everything discussed here, I am making dvd copies through hdd.

I also discovered that a recent upgrade to Recordnow was responsible for some performance loss. With this solved and an important change in my Dell 8400 BIOS (from A01 to A09), that seems to get also a slight better performance, I reduced the time for a 4.7 GB on the flight copy (+RW 4x to +RW 8x) from 18 to 12 min (just for testing). But is clear that without the IDE bottleneck, times could be better.

So, I still keep on looking for a solution for the future, and a possible one could be to move the writer from IDE to SATA (buying a new SATA writer). But I read a lot of oppinions telling that this field is not mature yet, and that there is a lot of compatibility issues between SATA optical devices and motherboards, mainly if you use raid as in my case (2 sata raid1 hdds, 1 sata non-raid hdd, 1 sata port free; the new bios seems to accept optical on sata, raid controller is Intel Application Accelerator).

Any experience on this area? People seems to prefer NEC vs Samsung, LG, Liteon or Asus SATA writers.

Thanks


#16

Intel Application Decellerator is infamous for causing issues with optical drives.

You really should avoid doing burns where frequent buffer under-runs are occurring. Yes, the burner is designed to handle them, but burn quality can suffer a lot.


#17

Once again: Your reader is too slow. Reading 8x CAV is too slow for burning 8x PCLV.


#18

I also have a problem with my H10n. My burst rate is only 11 MB/s, and it used to be 26MB/s. I’m using a LG GSA-H10N with the JL12 firmware and i have Windows XP with SP2. The thing is that my burst rate on Vista is 26 MB/s for one reason. I tried reinstalling the OS again and i still have 11 MB/s on Windows XP. I also changed the ide cables but it’s still the same speed. I noticed one thing though, right when i copy something from it to hard drive i can hear the disc just stopping from spinning so fast as it uses to at high speeds. It works on UDMA2, as it used to work all the time.