Does Having A Bigger Cache Buffer Make A difference?

[craiger]

Just wondering if having a 2mb cache buffer or 8mb cache buffer makes that much of a difference? Does it just prevent more buffer underuns? Should you get a CD burner or combo drive based on price or have the bigger buffer? Thanks, Craig.

[CDRecorder]

A lot depends on the computer in which you are going to install the burner, craiger. What type of system are you using?

My Lite-On LTR-32123S with 2-MB buffer runs great (buffer rarely if ever under 98%) on my 1.2-GHz Athlon with 512 MB of RAM and 60-GB hard drive.

For most reasonably fast computers, a 2-MB buffer seems to be quite sufficient. IMHO, the buffer size isn't the most important feature.

[craiger]

It's a Gateway P2 350 with 256mb of ram.

[Turkeyscore.com]

I'm not sure about with cd drives, but when I bought a Maxtor 120 GB 7200 RPM Hard drive i noticed that it is quite fast so I imagine that a cd drive with a bigger cache would perform better. I havent had any buffer underrun problems with the hard drive i have now but then again, my cdrw drive is not fast enough to get buffer underrun.

[Inertia]

Just wondering if having a 2mb cache buffer or 8mb cache buffer makes that much of a difference?

Not really, since the advent of buffer underrun prevention technology. Theoretically it may be better to not have a burn interrupted, but even if buffer underrun prevention is invoked, I have seen no evidence that it has any detrimental effects on the recording. The newer forms of this technology are better than ever, resuming at the exact location that the burn was interrupted.

Does it just prevent more buffer underuns?

No. This would have been true before buffer underrun prevention was available, but in new burners buffer underruns are a thing of the past.

It is important, however, to the "my buffer is bigger than your buffer" crowd.

There are no buffer enlargement creams available, so they go for the devices that have the biggest buffers.

[CDRecorder]

With a machine that old, you probably will want a larger buffer. I'm not sure you'll be able to burn faster than 24x or 32x without buffer underruns because the machine is so slow. I installed a 48x12x48x Lite-On on a Celeron 366 with 288 MB of RAM, and it can't burn faster than 24x without buffer underruns. DMA is enabled for all drives.

[Inertia]

With a machine that old, you probably will want a larger buffer. I'm not sure you'll be able to burn faster than 24x or 32x without buffer underruns because the machine is so slow. I installed a 48x12x48x Lite-On on a Celeron 366 with 288 MB of RAM, and it can't burn faster than 24x without buffer underruns. DMA is enabled for all drives.

The System Requirements for the LiteOn LTR-52246S as specified by LiteOn are "CPU Pentium 166MHz or faster, 128MB DRAM required". Craiger's PC should be able to support high speed burning

The term "buffer underrun" is an error that causes a burn to be aborted (and creates a coaster), usually during an audio burn when the buffer empties and there is no way to recover the location of the last written information.

If a buffer empties when buffer underrun prevention is enabled, the burn pauses while the buffer refills. After refilling the buffer, the recording then resumes. This is not a buffer underrun, although some are starting to use this term incorrectly.

At a 52x burn speed, data throughput (cooked) is about 7.8 MB/sec. A 8 MB buffer will allow a burn to continue about 769 milliseconds longer than a 2 MB buffer, so the practical difference when burning at high speed is negligible.

[CDRecorder]

I know about Lite-On's system requirements. The problem is that this type of system won't be able to burn at 52x without pausing frequently. Interestingly, CenDyne's 48x drive (this particular one was a re-packaged Lite-On) says that it requires a 400-MHz PII. I also know from experience that a Celeron 366 with 288 MB of RAM (very similar to Craiger's system) simply can't burn faster than 24x without pausing. Really, a more powerful system is needed to get maximum performance while burning.

[Inertia]

Interestingly, CenDyne's 48x drive (this particular one was a re-packaged Lite-On) says that it requires a 400-MHz PII.

Speaking of the CenDyne System Requirements 52x24x52 Drive, they set a very high requirement indeed. In addition to the "400Mhz MMX Processor or Higher", it doesn't support Win98SE or Win98. Strange, since I have two computers running LiteOn LTR-52246S drives on Win98SE systems.

But CenDynes's most stringent requirement is for a hard drive with an access time of of 15ns (nanoseconds) or less, or 15 billionths (1.5e-008) of a second. Since this about a million times faster than an average hard drive access, no available drives can match this specification.

Therefore, using CenDyne's System Requirements, we can see that their drive won't work with anything.

In case someone misses it, this is irony.

[CDRecorder]

CenDyne probably meant "ms", not "ns".

[dodecahedron]

possibly Lite-On's system requirements are such that they mean that a low-spec drive can burn at 52x but with many "buffer underruns" (pauses), i.e. the "buffer underrun protection technology" activated many times when the buffer empties to prevent coasters.

the most important info in this discussion is IMO:

At a 52x burn speed, data throughput (cooked) is about 7.8 MB/sec. A 8 MB buffer will allow a burn to continue about 769 milliseconds longer than a 2 MB buffer, so the practical difference when burning at high speed is negligible.

basically, craiger, if your system is low spec, then it's possible, or even probable, that no matter the size of the buffer, there will be "buffer underruns". maybe there will be less of them with 8MB than 2MB buffer.
so the question is: do you care if there are "buffer underruns" in your burns or not?
if you do, you'll probably have to burn at slower than 52x anyway.
if you don't then the buffer size doesn't matter anyway.

[Inertia]

dodecahedron, a "buffer underrun" is not a pause and this is an incorrect use of the term. I know that you like to communicate with precision, so in case there is some question about my explanation, see:

http://www.cdrfaq.org/faq04.html#S4-1
http://www.cnet.com/Resources/Info/Glos ... under.html
http://www.webopedia.com/TERM/B/buffer_underrun.html

This is why the protection technology is called "buffer underrun prevention". It doesn't prevent pauses, on the contrary it uses a pause to refill the buffer and prevent coasters:

http://h40067.www4.hp.com/press_cdwriter.htm
http://www.teac.com/DSPD/support/dt_ide ... _drive.htm
http://support.gateway.com/s/SOFTWARE/m ... faq1.shtml

This reminds me of the "I could care less" discussion.

Inertia

[dodecahedron]

agreed.
that is why everywhere i said "buffer underrun" with double-quotes!

[cfitz]

CenDyne probably meant "ms", not "ns".

Inertia is quite aware of this, LiteOnGuy. That was the basis of his joke.

cfitz

[CDRecorder]

I'm sorry. I didn't mean to ruin your joke, Inertia.

[QQ]

Well, I can burn @ 48x with my C300A CPU, 256mb of ram without any buffer drops whatsoever. Most it ever falls is like 95%. Well, I have to add that I have a fast drive (WD800JB, Western Digital 80gb with 8mb cache), but even CPU-wise I have no problems.

[Inertia]

agreed.
that is why everywhere i said "buffer underrun" with double-quotes!

Sorry if I overreacted, dodecahedron.

I didn't realize that you were using the quotation marks as a device to denote irony.

I was afraid that some dummy like me would read your comments and interpret them literally.

Well, I can burn @ 48x with my C300A CPU, 256mb of ram without any buffer drops whatsoever. Most it ever falls is like 95%. Well, I have to add that I have a fast drive (WD800JB, Western Digital 80gb with 8mb cache), but even CPU-wise I have no problems.

Thanks, QQ, I was hoping that someone with a system in your power range would respond with their high speed burning experience.

You have saved me the effort of cobbling together an old Gateway Pentium Pro 200 with a 140 watt power supply to run some tests. It's in pieces on the floor now, and I'll just put it away again.

In any event, it proves the point that burning is not particulary CPU intensive. DVD playback and MP3 encoding are more CPU intensive because of the respective decompression and compression involved. For instance burning a MP3 CD-R on the fly from WAV files on the hard drive is CPU intensive and could bog down an underpowered system, but this is because of the MP3 compression demands and not the burning requirements.

I'm sorry. I didn't mean to ruin your joke, Inertia.

Don't give it a thought, LiteOnGuy, it didn't make any difference.

I can't help wondering if your previous high speed burning difficulty with the Celeron 366 couldn't be DMA related. The symptoms seem that way. I know that you stated that DMA was enabled for all drives, but I have seen instances where the system shows that DMA is enabled but in fact it is not properly implemented. Frequently a change or update to the busmaster drivers or BIOS can fix this.

If you still have this system or have access to it, it would be interesting to run some transfer speed tests with CD Speed. Abnormally high CPU utilization is usually a giveaway that DMA is not working, whether or not the system says it is enabled.

[CDRecorder]

Don't give it a thought, LiteOnGuy, it didn't make any difference.

Thank you, Inertia! That makes me feel better!

Well, I can burn @ 48x with my C300A CPU, 256mb of ram without any buffer drops whatsoever. Most it ever falls is like 95%. Well, I have to add that I have a fast drive (WD800JB, Western Digital 80gb with 8mb cache), but even CPU-wise I have no problems.

I can't help wondering if your previous high speed burning difficulty with the Celeron 366 couldn't be DMA related. The symptoms seem that way. I know that you stated that DMA was enabled for all drives, but I have seen instances where the system shows that DMA is enabled but in fact it is not properly implemented. Frequently a change or update to the busmaster drivers or BIOS can fix this.

If you still have this system or have access to it, it would be interesting to run some transfer speed tests with CD Speed. Abnormally high CPU utilization is usually a giveaway that DMA is not working, whether or not the system says it is enabled.

This machine has the latest BIOS (July 1999, so not very new). The IDE drivers for this motherboard recommend you don't use them with Win98 (that is what the machine has) because the Win98 ones are better. Also, DMA does seem to be working; this type of computer can only transfer data at about 10x from the CD when DMA is disabled. The computer does have an ancient 4-GB ATA33 HDD on it which is very slow. This could be the problem.

I'll ask the owner (it isn't my machine, but I have a very similar one with an ATA133 controller card with a 60-GB ATA133 Maxtor drive on it) to let me run some CDSpeed checks for reading and writing, and I'll write back with my results. I'll pay attention to CPU usage and HDD activity during the test.

I can't try the tests on mine because I have what I think is a driver issue, which is that when the HDD is connected to the ATA133 controller, CD transfer speeds are limited to 20x (I've tried to diagnose this, but I haven't been successful yet).

Thank you all for your kindness and suggestions!

[CDRecorder]

Oh, and I'd also like to apologize for doubting your system requirements, Inertia. I thought the Celeron 366's CPU was the limiting factor, but my theory was disproven by QQ.

[Inertia]

No need to apologize to me for doubting the system requirements. They aren't mine, just posted information and I take no ownership of them. I would hate to stake my life on some posted system requirements, i.e., CenDyne's.

But thanks anyway, LiteOnGuy, for your consideration.

[CDRecorder]

You're welcome. Thank you for being so forgiving, Inertia.

BTW, this is post #250!