USB CD Rip Speed issues

[OzPeter]

Hi all,
I have been experimenting with dBPowerAmp 11.5 (trial) using an external USB 2.0 CD/DVD drive (LaCie) d2, on both my laptop (Dell Inspiron 8500) and a work desktop (another Dell, can't think of the model, but 3.2G with hyper threading enabled) both systems running XP SP2. (Although I tested the desktop, I expect to be ripping/encoding from the laptop)

From my tests I would say that the laptop is CPU bound when it comes to encoding as I see the CPU usage maxing out at 95% to 100%. The desktop seems to be USB transfer rate limited, as the max rip speed is around 40x and the CPU usage is around 65%.

Now comes the strange part. On both setups I am seeing the same behaviour regarding rip and encode speeds (albeit the desktop has the higher actual speed). When a rip/encode (to mp3 192kb) starts, I see a lowish rip speed and encode speed. Over the 1st half of the CD, these numbers slowly increase in a linear manner. The CD drive is running quietly at this time. Then around track 9 (of 15) there is a step jump in disk speed, and the rip and encode speeds shoot up accordingly. In the case of the desktop, the rip speed went from 14x to 40x, and the encode speed increased from 12x to 16x - and there was a 50% step increase in CPU usage. Of course at the same time the noise from the CD becomes much louder. In the laptop case, the rip speeds went from 16x to 25x and the encode speeds jumped from 10x to 12x (and the CPU was pegging out near 100%)

Obviously both systems can handle the speeds in the latter parts of the rips. But I would prefer it if they ripped at the high speed for the entire CD.

I have read everything in these forums to do with ripping/encoding speeds, and in general all the solutions point to DMA being enabled on the drives. However USB drives do not have this configuration option.

I am not sure why the CD speeds up around track 9. I have tried different encode rates (192 and 128) and got the same behaviour around track 9, which suggest it might be the physical track location that is triggering the switch.

I have also seen the same behaviour on the laptops internal CD drive (which is USB based), so it is not drive dependent.

So can anyone help me out with suggestions to get these USB drives working at their full potential? Even if it is to tell me that this isn't the right forum :-)

Regards
Peter

[LtData]

Re: USB CD Rip Speed issues

First, CD drives use a read method called "Zone-CAV" in which the speeds on various areas of the discs are different, that is the rotation speed changes. See here for more information: http://www.pcguide.com/ref/cd/constSpindle-c.html. This explains the change in sounds you hear and the increase in transfer speeds.

Second, I assume both of your devices have what is called "USB 2.0"?

As for the difference in CPU usage, that is probably due to a few factors. First, the rotation speed of the hard drives most likely differs. Most desktop hard drives spin at 7200RPM where as laptop hard drives spin at 5400RPM or even 4800RPM. Second, CPU usage depends on your USB controller which would also explain the higher CPU usage and slower transfer speeds. Also, of course the CPU speed itself plays a factor.

There isn't really a good way to speed up USB CD drives, as they are limited by the USB bus itself, which I seem to remember topping out at 20MB/s where CD drives can deliver about 30-40MB/s in transfer speeds. The better external connection is Firewire or IEEE1394, which is faster (30-40MB/s) but not on as many computers.

[OzPeter]

Re: USB CD Rip Speed issues

Hi there,
Thanks for the speedy reply. But in reviewing what I wrote, and then what you replied, I think I misled you into not answering what I wanted.

I think I can put better as this:

I have a external LaCie USB2.0 CD drive on a USB 2.0 (high speed port). In doing trial rips I have seen the CD run at both a low speed and a high speed, and the speed change from low to hi kicks in around track 9 of 15 every time.

When going from low to high speeds there is a marked improvement in the rip and encode speeds (25% for my slow system and 50% for my fast system).

I want to do a significant amount of ripping and encoding so overall performance is important for me. It looks like if I can get the drive to run at its higher speed for the entire rip, then I should get shorter rip and encode times and hence a faster throughput.

After reading the link you posted in your reply, I am thinking that the speed change may be being tripped by the drive itself. So my question really is:

Is there any known way to force or encourage the drive to start off at its higher speed?

Regards
Peter

[LtData]

Re: USB CD Rip Speed issues

No, the drive speed is controlled internally. Every modern CD drive changes the speed of the drive, though most aren't as audible or noticeable as your drive is.