UDMA Mode 5

[scorch]

How can I determine if my system will support UMDA Mode 5?

[araaraara]

UDMA5 is ATA100. Support depends on your motherboard chipset and your harddrive. All new harddrives support it, but older motherboards may only support upto UDMA4 (ATA66) or UDMA3 (ATA33). In most cases, the motherboard and harddrive will run at the highest speed that the 2 support. If you can tell us your motherboard, then we can tell you more what it supports.

[scorch]

Heres the specs:
OS: Windows 98 Second Edition
VIA i2K-KT133s/686A; VIA 8363 chipset, sockett A
BIOS CHIP: AWARD 1998; PCI/PNP 686; 269230556
4\1 Driver: ver.432\VIA sound ver. 1.09
Mobo BVK1M Ver. 1.0 (Boot screen shows: K1M1006 )
800MHz AMD Duron w\fan
AC97 Onboard Sound
MT AC-Link MR Modem (Si3024-XS8)
ELSA Gladiac 511 GeForce 2 MX-400; 64MB SDR; AGP4X w/TV Out w\cpu fan
NVidia Driver: Detonator XP; Win9x-ME_23.11.exe
640Mb P100 SDRAM (1-128Mb P100 SDRAM; 2-256Mb P133 SDRAM)
Seagate ST340823A 40GB HDD
52x CD-ROM
12x8x36 CD-RW (Philips)
TM FLCS
BIOS Manufacturer : Award Modular BIOS v6.00PG
BIOS ID : 10/07/2000-8363-686A-6A6LMB09C-00
BIOS Date : 10/07/00
BIOS OEM Signon : K1M1006
BIOS ROM Size : 256K
Chipset : VIA 82C305 rev 3
Super I/O Chip : VIA 686 found at 0h:7h:0h

[scorch]

P.S. the mobo is apparently a Procomp, as they list boards with model #'s leading to BVK1M, but not including. O f course site is under construction and email addresses return as invalid.

[otheos]

UDMA3 (ATA33)

That would be UDMA2 or ATA3 or UDMA33 or ATA33.

UDMA3 was never used (48MB/s). :)

Now your mobo has the 686A VIA southbridge that supports up to ATA5 or UDMA4 or ATA66 or UDMA66, essentially 66MB/s max.

Let noone fool you, ATA66 will run each and every modern ATA66/100/133 (or UDMA4/5/6 or ATA5/6/7) hard disk at its full speed without any performance issues.

[scorch]

It can run UDMA Mode 5?? The St340823A I'm using's specs say it can support UDMA Mode 5 in Ultra ATA\100 mODE if the UATA utilityb is run. I did and it shows the drive in mode 4, the highest setting allowed, apparently by mobo. Startup screen shows it as ATA 66. I just ordered a ST360021A 60Gb and it lists the same potential.So it will run to full potential but won't be listed as such?
Cause it says to use the 80 combo cable, 18" max. Well, I'M a paraplegic, and could use a longer cable for easier reach. If I can't get UDMA mode 5 ultra ATA/100, would a longer cable interfere with trhe mode 4 that it seems to be running now?
Also, you know more about a Procomp board then I've been able to find. Can you hook me up with a site that covers it? TIA

[otheos]

Your motherboard supports "only" ATA66 and hence the drives operate at ATA66 (UDMA4) mode as they are backwards compatible.

However, there is no performance loss due to that and don't let anyone convience you otherwise :)

Your new hard drve will work perfectly and at its max capabilities with ATA66 instead of ATA100 as the latter offers only marketing hype.

Calbes? Well I wouldn't suggest longer than 45cm as they tend to cause problems. You can try (they're not that expensive).

As long as the cable is 80wire 40pin (ATA66/100/133 compatible) the mode (UDMA4) will not be affected. If it is longer than the specification you may have some data corruption.

EDIT: About your motherboard: do you speak hungarian? if you do: look here (http://plusabit.tvnet.hu/html/lap/bvk2a/index_800.php)
Other than that it looks like a standard KT133 board, that should support all 200Mhz FSB Athlon/Durons and PC133 RAM. IT should be fine :)

[bassman]

Scorch, about the RAM config, mixing frequencies is often a cause of crashes and malfunction of the system...hope it didn't happen to you...

[Rugor]

Bassman,

He's running all his RAM at the same frequency. I don't believe there is or ever has been a motherboard that can support running different sticks of RAM at different speeds.

An Ultra 66 controller can burst data from the IDE controller to the Hard Drive at 66mb/second. An Ultra 100 controller can burst data at 100mb/second. However, those rates apply to the controller, not the hard drive.

The real transfer limit on a Hard Drive has nothing to do with the controller, but depends on the speed at which the platters pass under the heads. No drive can read data from the platters any faster than that. A fast 7200 RPM drive is usually limited to 35-40mb/second because it can't get the phyical data bits into position to be read any faster than that. Even that rate can only be sustained when the data is sequentially located on the Hard Drive, if the heads have to move to seek it, the transfer slows down again.

That rate can be exceeded in short bursts when reading to and from the HD's buffer, which normally ranges from 512kb on budget drives to 2mb on most performance models. But anything not in that buffer is limited by the physical mechanics of the data transfer.

That's why you don't see any real difference between Ultra 66 and Ultra 100, because they're both faster than the drive can physically transfer data.

The only time Ultra 100 may provide a slight improvement is when you have two drives on one controller, because it takes more than one drive to saturate the Ultra 66 bus.

I just figured someone should explain WHY none of the newer UDMA modes provide any real world advantages in normal use.

[otheos]

Great post Rugor.

One correction:

The only time Ultra 100 may provide a slight improvement is when you have two drives on one controller, because it takes more than one drive to saturate the Ultra 66 bus

Well, IDE is a very limited interface and as such even with 4 drives on two channels the 66MB/s per channel are more than enough. Why? because not two drives on the same channel can use the bus simultaneously. They have to take turns, and thus each drive has the whole bus on its own when it uses it. This is why you wouldn't want two hard drives on the same channel as copying from one to the other will be a dog (read the first, cache it, switch to the other, write it etc).

So if you say have two drives each capable of 40MB/s then your total required bandwidth is NOT 40+40 but still 40MB/s. SCSI on the other hand allows devices to disconnect from the bus and share the total bandwidth with other devices. In SCSI you do add each drives ability to obtain the total bandwidth necessary.

So, 66MB/s is more than enough for all of today's IDE drives (and this includes RAID0 arrays as they are bound to the same limitations).

[Rugor]

Thanks otheos, that part slipped my mind because I was so focusing on the physical limitations of drive transfers.

Actually I avoid the problem when setting up a RAID 0 array because I won't put two drives on the same channel because that would defeat the purpose of improved speed. Of course then you run into your 133mb/sec saturation issues with the PCI bus.

[shadow]

ATA/100 - Real Performance or Marketing Hype?
To add to Otheos's posts:

http://www.realworldtech.com/page.cfm?ArticleID=RWT011701000000