SATA burning slow

vbimport

#1

Hi,

I just purchased an SATA PCI controller card - a Promise TX4302 - and two internal SATA DVD burners - a Lite-On LH20-A1S 20x burner and an Asus DRW-1814BLT 18x burner.

My problem is that I can’t ever get the speed to hit above 8.8x with Imgburn , and the average writing speed is about 6x, with both SATA burners.

I’m dumbfounded because I can burn at the same speed or faster on the other 2 burners I have connected to the IDE controller on my motherboard - an old Sony DRU-700A, and a Plextor PX-750A.

I tried both Verbatim RG20 16x DVD-R media and TDK 16x DVD+R from CMC, and the results are similar.

What’s interesting however is that the SATA burners are able to read very fast for the verification in IMGBURN, near 16x, but not to burn fast.

Here is the raw data I collected, burning the same full 4.7GB image from my SCSI 15k rpm hard drive onto each burner :

  1. with Verbatim 16x DVD-R media MCC 03RG20

a) Sony DRU-700A, primary EIDE master : write speed average 6.5x, peak 8.1x . Read speed 5.7x, peak 8.3x.

b) Plextor PX-750A, secondary EIDE master, write speed average 10.8x, peak 15.9x. Read speed 6.9x average, peak 9.9x.

c) Lite-on LH20-A1S, SATA channel 3, write speed average 6.0x, peak 8.5x . Read speed 11.0x average, 16.2x peak.

d) Asus DRW-1814BLT, SATA channel 4, write speed average 4.8x peak speed 8.8x; read speed 8.1x average, 12.0 peak.

  1. TDK 16x DVD+R media CMC MAG M01

a) Lite-on LH20-A1S SATA #3, write speed 6.1x average, 8.7x peak; read speed 11.0 average, 16.3x peak.

b) Asus DRW-1814-BLT SATA #4, write speed 6.1x average, 8.7x peak; read speed 11.0x average, 16.3x peak.

I have nothing on SATA channels #1 or #2 on the SATA controllers - these are external eSATA ports.

In all cases I am burning the ISO image from a freshly formatted Seagate ST336752LC 36 GB 15,000 rpm Ultra160 SCSI hard drive . The ISO image is the only file on the NTFS partition which spans the entire disk, so it’s not fragmented for sure. I did all burnings one at a time, without running any other programs.

The SCSI controller for the hard drive is an LSI21040 3.3V PCI 66 MHz / 64 bits based on a Symbios 53C1010 chip.

I am doing this on WinXP which is booting from another hard drive, also Ultra160 SCSI. All latest OS updates are installed. I have the latest drivers for both the LSI SCSI controller and the Promise SATA controller.

The motherboard is an Asus A7M266-D, dual Athlon MP, with two Athlon MP 2800+ CPUs, and 768 MB of PC2100 ECC DDR.

The Sony and Lite-ON burners are connected to the EIDE controller on the motherboard - both as masters on primary and secondary channels, and there are no slaves on each channel.

The system has the following boards installed, filling all of the slots :

  • Gainward USB 2.0 and Firewire PCI controller
  • Promise SATA 300 TX4302 controller
  • SB Live value sound card
  • Intel EtherExpress Pro/1000 gigabit ethernet card (64 bit/66 MHz 3.3V PCI)
  • LSI 21040 Ultra160 controller (64 bit/66 MHz 3.3V PCI)
  • ATI 9200 AGP video card

Can anyone please shed some light on how I can improve SATA burning performance on my system ? Right now I can only use the SATA drives as readers since the burning performance on them tops at 6x average, and I have to keep burning to the EIDE drives which are faster.

Thanks !


#2

I think you’ll find that SATA burners need to be run only on chipset-based SATA controllers, no PCI cards or 3rd-party motherboard controllers, (SIL, etc).

That said, your PCI bus is hopelessly overloaded.


#3

PROMISE controllers are made for HARDDISKS ONLY!


#4

yup thats 100% correct,the sata controller is the issue


#5

CDan, the motherboard actually has 3 separate PCI buses, so I don’t think it’s that overloaded. But I will try taking the USB/Firewire, soundcard and network card out to see if it makes any difference. Not that I could really live without any of them.

The dual athlon motherboard does not have USB due to a chipset bug, so a separate PCI card is required. I need the NIC for internet access - not NIC is built-in. I like the SB Live because it has SPDIF out - and the CMedia built-in to the motherboard does not, and has horrendous sound distortion (from hard drives, etc) on the analog out like most built-in audio. Being an amateur musician I’m really sensitive to that sort of thing.

Chipset-based SATA is not an option as they no longer make new dual Athlon motherboards, and none ever had built-in SATA. Even though SATA peaks at 150MB/s, PCI is still capable of up to 133 MB/s which is way more than needed for DVD burning.

Also, I am getting 16x speed on SATA for reading, which would seem to indicate the bus bandwidth is sufficient, at least in one direction. Why doesn’t it ever get close to that speed for writing ?


#6

My PCI to SATA card works well. I have one with the Silicon Image 3114 chip.

I avoid Promise like the plague. Is your card designed for optical drives? Are there any BIOS updates for it?


#7

Steve,

The card specs are at http://www.promise.com/product/product_detail_eng.asp?segment=Non-RAID%20HBAs&product_id=168

It is not a RAID card. The specs say it supports serial ATAPI, so I think it’s fine for optical drives.
It has the latest BIOS on it, and the latest drivers.

This is the 2nd card I try. The first one was a Silicon Image 3112. I had complete failure with that card due to a BIOS bug. Even with the latest BIOS flashed into it (flashed with a floppy), the Silicon Image BIOS conflicted with my LSI SCSI controller BIOS. I couldn’t boot to the SCSI hard drive as a result while the SIL-3112 controller was connected. The Promise card on the other hand does not have this conflict problem - I can boot with it just fine. And it seems very stable. Everything is fine with it, except writing performance.


#8

Hi,

Would you mind providing a source for that information ?
My Promise card is not a RAID card. The specs at
http://www.promise.com/product/product_detail_eng.asp?segment=Non-RAID%20HBAs&product_id=168 say it supports serial ATAPI.


#9

Are there any other good SATA PCI cards I could try ? I’ll head to my local Fry’s before it closes and buy a bunch of different ones. Hopefully one will boot and actually perform and I will return the others …


#10

you might find this thread helpfull
http://club.cdfreaks.com/showthread.php?t=214312


#11

Thanks. These seem to be recommended the SIL-3112 chipset. I already tried one card with that chipset, but it was unusable to the BIOS bug. Maybe Adaptec customizes its BIOS and doesn’t have this bug. I’ll try it.

For now, I was able to achieve slightly better performance with the Promise. I adjusted some motherboards BIOS settings, removed the USB/Firewire card and the SB card, disabled the EIDE and the motherboard.

I did a DVD-R burn on the Lite-On at 8.4x average, 11.4.x peak; and DVD+R on the Asus at 8.5x average, 11.6x peak. After about 11.5, Imgburn showed the device buffer not being filled, so the drive pulled back in speed down to 8x … I never have this problem with my EIDE Plextor which burns up to 16x towards the end. Maybe the buffer on those SATA drives is too small.


#12

You can believe it or not, but no Promise or Highpoint card has ever been shown to properly support optical drives, in spite of generalized assurances from the makers that they do. The same can be said of the majority of PCI cards, and this has never been more true than with SATA optical drives.

A simple data disc burn test with CDSpeed will produce a graph that answers a lot of questions, and I’ll guess that it won’t be pretty. It has nothing to do with the drive, or drive buffer, it’s the controller.


#13

Well, I think I have just proven otherwise - it’s possible to get good results out of the Promise. I’ll grant you that it wasn’t the drive or drive buffer. But it seems that it wasn’t the controller either, but the bus rather, surprisingly.

I moved my TX4302 card to a 3.3V 66 MHz PCI slot on my motherboard . This is actually a 64-bit 66 MHz 3.3V PCI slot. I only have two such slots on my motherboard. One was occupied by the LSI SCSI controller, which is 64-bit 3.3V only, and thus it has to be there - it can’t be inserted into any other slot due to voltage. The second high-speed slot was previously occupied by a 32-bit 66 MHz Intel Pro/1000 Gigabit ethernet card. But this is a mixed voltage card, so I can also put in one of the three slower 32-bit 33 MHz 5V slots, at the price of a little performance loss on the LAN, which I guess I can live with, I’m not running a big server at home, though I like the gigabit speed for file transfers / backups.

Right now, I booted the system with only the AGP video card, the LSI SCSI in one of the 3.3V slots, and the Promise 4302 in the other 3.3V slot.

And, using the SATA Lite-on LH20-A1S drive, I was just able to achieve a burn on DVD-R that averaged 11.2x and peaked at 16.1x, in other words, as it should be. Read was 10.9x average and 15.9x peak. The device buffer remained full at 100% the whole time in Imgburn. I am doing another burn on the Asus drive which is giving similar results right now.

Next, I will try concurrent burns on both SATA burners, from 2 different physical SCSI hard disks. I will also try to put back all my other PCI boards and re-enable the IDE on the motherboard to see if performance declines.

And if I’m really ambitious later, I’ll play with the Lite-on drive firmware and see if I can burn the Verbatim media at 20x instead of 16x …

Before I discovered this, I bought one more card tonight at Fry’s , an SIIG : http://www.siig.com/product.asp?catid=7&pid=466 . I was unable to find any Adaptec. The SIIG is based on the Silicon Image SIL-3114 . As with the other SIL-3112 I had on sunday, I was unable to boot the system to my SCSI disk when the SIIG card was in the machine. It’s a very nasty conflict. I bet everything would be OK if only the BIOS could be turned off. So, this SIIG card is going back to Fry’s. And hopefully I will be happy with the TX4302 now that it’s in the proper slot. I will be sure to stay away from Silicon Image. No worry though, my PCI slots are all full. I guess my next upgrade will be a new motherboard and new CPUs, but I like to at least double performance when I do that, and for 32-bit apps, the current CPUs don’t do that affordably. I would need to have at least a quad-core to double the performance from the Dual Athlon MP 2800 which is 2.133 GHz.


#14

Hi,

:disagree: sorry to disagree. The PDC20579 from my sig is the onboard version of SATAII150TX2+. The drives do fine.

A simple data disc burn test with CDSpeed will produce a graph that answers a lot of questions, and I’ll guess that it won’t be pretty.
It is pretty with my SH-S182D, and it was pretty with my SH-W163A (SATA).
It has nothing to do with the drive, or drive buffer, it’s the controller.
It has very much to do with the drive (or even with its firmware): Having connected my Philips DVDR1660, it failed severely when writing the Lead-In, my Benq DW1650 (only difference to the Philips is the faceplate and the firmware) worked without problems. Go figure.

Michael


#15

No. I suggest doing some reading.

Michael


#16

FYI, my concurrent burns with the 2 SATA burners from 2 SCSI hard disks went fine, and the burning speed was identical to the separate burns. I’m a happy camper now, with the Promise TX4302.


#17

After having done some additional tests with fw BCIC, I’ll have to admit that the Benq fails in a similar way as the Philips. Maybe fw BCDC was better then, but I don’t like to flash back and rearrange the drives now.

Michael