On Sun, May 06, 2001 at 01:22:35PM -0400, abrahams@acm.org wrote:
Curtis Rey wrote:
Paul Abrahams wrote:
My hard drive claims to be ATA100; the motherboard claims UDMA 33/66 IDE. I'll admit that I know nothing of the relationship between ATA and UDMA, but I had assumed that the numbers were comparable, especially since most hard drives seem to be ATA66. Can you enlighten me? From my experience the differences between ATA and UDMA are a matter of semantics. I do beleive that the marketing departments of the various HD makers use this term ad-hoc (though I have read that there is a difference from a technical standpoint) For example mobo makers refer to their IDE channels as UDMA and the HD makers will use both terms (ATA and UDMA). So am I likely to gain anything by hooking my ATA100 hard drive to an ATA100 controller rather than hooking it directly to an UDMA 33/66 mobo?
AFAIK, ATA (AT Attachment) refers to the basic protocol by which the controller talks to the disk, whereas UDMA (Ultra-DMA) refers to the ability of the drive/controller combination to transfer data directly from the disk to the processor memory. I believe that the numbers (33/66/100) refers to the speed of the controller-drive bus. Therefore, if your drive is (for example) marked as ATA100 with UDMA, then it is UDMA100 compatible (i.e. the numbers can be placed after ATA or UDMA, and they mean the same thing). I believe that the correct way to say it is to put the speed after 'ATA', and put the UDMA separately, but, as usual, PC manufacturers seem to use the terms interchangeably. Anyway, as for whether it'll be faster, your PCI bus is limited to 33 MHz, 32 bit, and ATA is 16-bits wide, so an ATA66 card can (theoretically) saturate the PCI bus. There will be a slight performance advantage from ATA100, as the drive will be able to get data to the card faster, so will be able to ensure that the controller card never runs out of data to put on the PCI bus, but I don't think that the performance increase will be as much as you might hope for. Basically, latency for new accesses is determined more by the seek time of the drive (even the whizzy fast ones), and for large accesses, the data transfer rate will be limited by the PCI bus. Going from ATA33 to ATA66 will probably make a significant difference; from ATA66 to ATA100 won't. Sorry. Of course, I could be totally wrong... -- David Smith Tel: +44 (0)1454 462380 (direct) STMicroelectronics Fax: +44 (0)1454 617910 1000 Aztec West TINA (ST only): (065) 2380 Almondsbury Home: 01454 616963 BRISTOL Mobile: 07932 642724 BS32 4SQ Work Email: Dave.Smith@st.com Home Email: David.Smith@ds-electronics.co.uk