Announcement

**LvR** · 11 February 2001, 20:05

SCompRacer

Interesting

Your scores on those GXP disks - are they on single disks or on your RAID 0.?

Personally - no ATA 100 patch applied or required, and in W2K, 30G and 45G GXPs return Sisoft scores of around 25.

WD200BB single disks return about the same.

Raid 0 of 2 WD200BB disks return score of 39 on an Iwill Sideraid 100.

All of this on ACPI W2K,NTFS only,Intel 815EEAAL PIII 800 with only Intel drivers and no ATA 100 patches applied. Intel utility also reports disks all running EIDE mode 5 (ATA100).
------------------
Lawrence

[This message has been edited by LvR (edited 12 February 2001).]

**SCompRacer** · 12 February 2001, 04:38

RAID 0, FAT32, Win2000 VIA drivers as the later 4 in 1's seem to lower scores even more from Post's I read.

I have just recently got Win2000 up and running on second partition, Win98SE on first. My SiSoft on 98 is in the 39,000 range. I have been told the partitioning slows it down but I don't know if thats an absolute.

Many report disabling ACPI doubles their benches in Win2000. I have seen many posts in the 14,000 range with SiSoft in RAID 0, ACPI enabled and then it doubles with ACPI disabled. I was getting high numbers with ACPI enabled compared to the posts I saw. I was even told you can even use Win2000 to set up a smoking RAID setup but it cannot be on a partitioned hard drive.

[This message has been edited by SCompRacer (edited 12 February 2001).]

**LvR** · 12 February 2001, 07:17

SCompRacer

Thanks for the feedback, glad you got something out of it.

I think my rigs will most likely be only 5-10% faster on W98 - I prefer NTFS on any/all disks when disaster strikes - I have found it to be much more tollerant of user/driver stuff-ups.

I was asking about performance because I am sure this is not a valid "shortcoming"/"bug" in W2K - There is no way for Microsoft to have foreseen the finalization of future technologies (ata100 for now but ata160 etc soon), and the fact that some mobo manufacturers actually manage to supply drivers running at the correct settings, support that line of thinking.

I think W2K does "support" ATA100 - it is just a question of mobo manufacturers having to start using it in their drivers AFTER THEY HAVE FAMILIARIZED THEMSELVES WITH THE OPERATING SYSTEM ABILITIES.

The patch is for those mobo drivers that have not properly taken ATA100 into account, and I suspect may cause some fun-and-games somewhere because of exactly that!

------------------
Lawrence

**Gurm** · 12 February 2001, 07:19

I am not understanding this patch. It has no effect on my system, since my U100 drives have always and will always be enumerated as SCSI devices on the Promise controller, and have always run at U100.

I've been confused on this topic ever since people started releasing "patches" for Ultra66 drives. Which chipset does this "patch"? Not the VIA, not the Promise, and the previous ones didn't patch the Highpoint either. Apparently it patches Intel controllers. Woo! Umm... who has one? Not I.

Does this do something mysterious for Promise controllers or am I just missing the boat?

- Gurm

------------------
Listen up, you primitive screwheads! See this? This is my BOOMSTICK! Etc. etc.

**MultimediaMan** · 12 February 2001, 08:00

Gurm has it right:

The UDMA-100 issue under Win2K only pops up on the "South Bridge" (Controller chipset) IDE controllers.

On Onboard or Regular IDE RAID controllers, the device is enumerated differently, and the OS goes with what the device BIOS and Drivers dictate. Onboard OEM IDE controllers drivers that Win2K uses are outdated, an update will be made in SP2: There are other issues here as well. There was a well-known fault with Fast Win98SE/WinME machines and fast HDDs with large onboard caches: similar problems will need to be addressed for Win2K. Win2K is much more complex than Win9x will ever be, so testing and development could and should take longer to get done right.

M^2

**LvR** · 12 February 2001, 08:12

Gurm

Running Intel 815 controllers here - they simply love the sound of ATA100 with no W2K patch required. Also running Iwill Sideraid 100 (Hghpoint 370) at 100 with no "known" hassles.

------------------
Lawrence

**SCompRacer** · 12 February 2001, 11:54

Well this has sure been enlightening, but why do I still feel confused? Have to try a Promise controller and see what happens there.

**LvR** · 12 February 2001, 12:17

Maybe Microsoft knows something about the re-introduction of the Intel 440LX chipset

**Gurm** · 12 February 2001, 13:45

I'm trying to figure out which U100 controllers actually enumerate their drives as IDE and use the Microsoft ATAPI.SYS or PCIIDE.SYS files.

Let's see...

Promise: Nope. Uses their own drivers, and appears as a SCSI controller. Drives appear as SCSI disks.

Highpoint: Nope. Uses their own drivers, and appears as a SCSI controller. Drives appear as SCSI disks.

VIA 133/133A: Nope. Uses their own drivers, although controller appears as an IDE controller. ATAPI and PCIIDE aren't used. Drives appear as SCSI disks.

Intel: I would have guessed yet, but apparently (from what I've heard) 810/820/815/850 uses the same approach as all the guys above. Although they may appear as IDE, they have their own drivers.

So whose controller uses IntelIDE, PCIIDE, or ATAPI and is also U100 capable? None that I know of.

- Gurm

------------------
Listen up, you primitive screwheads! See this? This is my BOOMSTICK! Etc. etc.

**paulcs** · 12 February 2001, 23:17

There's a lot of commotion about the ATA-100 Hotfix and the recently annointed official VIA 4in1 4.28 set on both Viahardware.com and HardOCP. From what I can gleen, W2K is recognizing ATA-100 devices as ATA-66. The VIA 686B Southbridge has native ATA-100 support, and some guy on the HardOCP forums (and now the front page) claims that the 4.28 busmastering drivers attempt to force the issue, which causes a conflict with Windows 2000. The ATA-100 hotfix addresses this conflict, and ATA-100 devices work the way they're suppose to.

Viahardware is recommending that only people using motherboards with the 686B Southbridge install the Hotfix.

I don't know how this effects Intel chipsets or third party onboard controllers. It does seem to help people in SCompRacer's position, as his board uses the 686B Southbridge.

Paul
paulcs@flashcom.net

**paulcs** · 12 February 2001, 23:29

Correction: Kyle is corresponding with some guy about the 4in1's and the Hotfix. Here's what Kyle and the guy had to say. Kyle's comments are in bold text; Cody's are in italics.

Cody Frisch MAY be onto something here. I will pass it along for you guys if you want to try it out. I have not tried this due to the fact I am not done with unscrewing my box from the initial v4.28 4n1 install on my Win2K box. I will supply you with a transcript of his mails.

I am using it in win2k - no problems whatsoever so far. Have been using the betas for weeks without problems. I may suggest installing the 3.11 standalone driver from VIAHardware it has a utility with it. But I have not had ANY problems at all. My guess is that people are not updating the driver for the actual controllers. Once it installs the ATA driver before you reboot you need to go to the other two devices and update their drivers to the VIA BM driver or whatever its called. Its .inf should be in \winnt\system32\drivers. ALSO make DAMN sure you install the ATA/100 fix from Microsoft FIRST, regardless of if you have ATA/100 drives.

Now I am a tad bit lost on what he means by, "Once it installs the ATA driver before you reboot you need to go to the other two devices and update their drivers to the VIA BM driver or whatever its called." My 4.28s installed the new drivers fully throughout the "IDE ATA/ATAPI controller" tree but it only showed it after reboot of course. He also went on to ad this about this about the MS fix for ATA100 posted over at VIAHardware.

What's happening is that Win2k defaults all ATA/100 drives to ATA/66, this patch corrects that. The problem is that the new VIA driver conflicts with that and tries to force ATA/100. Speculation on my part I admit - but the problems don't exist this way for me, they do without it for me.

Paul
paulcs@flashcom.net

[This message has been edited by paulcs (edited 13 February 2001).]

**LvR** · 13 February 2001, 04:42

http://support.microsoft.com/support.../Q247/9/51.ASP

Just wondering - how does this little bit of muck tie in with the rest?

------------------
Lawrence

Announcement

Win2000 ATA100 Patch

Win2000 ATA100 Patch

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment