|
Comment #41
Dear Sir,
For your readers who like a bit more credibility, this will be useful. The information
in regard to NT not supporting Block Mode access is here.
Article ID: Q152307
http://support.microsoft.com/support/kb/articles/Q152/3/07.asp
That details NT and its ability to address IDE.
Sincerely Yours,
Ryu Connor
Hello Ryu,
Thanks for that great link! It definitely clears up all the confusion about whether to
enable IDE HDD Block Mode or not. It's going into the BIOS guide immediately.
Adrian Wong
Adrian's Rojak Pot
http://www.rojakpot.com/
http://www.adriansrojakpot.com/
Dear Sir,
Back at it again. For your readers who like a bit more hard facts here is a Microsoft
support article that details all types of BIOS Shadowing and caching. It would seem 9x
*might* benefit, but not NT. Article ID: Q92766 http://support.microsoft.com/support/kb/articles/q92/7/66.asp
Sincerely yours,
Ryu Connor
Hello Ryu,
Thanks for another great link! It definitely clears up all the confusion about whether
to enable BIOS shadowing or not. Just like WinNT, Win9x will not benefit from BIOS
shadowing as it accesses the hardware directly via drivers. The BIOS is only used when DOS
mode is engaged. This is going into the BIOS guide immediately.
Thanks again! :)
Adrian Wong
Adrian's Rojak Pot
http://www.rojakpot.com/
http://www.adriansrojakpot.com/
Comment #42
The BIOS guide rules. It's one of the best things I've ever found on the Internet.
Enix
Thanks, Enix! :)
Adrian Wong
Adrian's Rojak Pot
http://www.rojakpot.com/
http://www.adriansrojakpot.com/
Comment #43
In reference to your BIOS guide, there are a few things I would point out in reference
to questions some other people raised...as far as video or BIOS shadowing, this feature is
mostly useless, and even more so under Linux, which someone inquired about. Linux can
manage these things much more efficiently on its own.
As far as the PNP issue goes, you can turn off every feature and Win 95/98 will work
fine, it initializes and assigns resources to the PNP devices itself irrespective of the
BIOS. I run multiple OS's that do not use PNP and have always had these features disabled
and never had a problem with Windows. Another person said that LBA is not compatible with
Win NT. This is bizarre, since I have run NT on two computers with LBA enabled.
Earthman
Hello Earthman,
I agree with you over your point about the shadowing of the video and system BIOS.
Normally, if you don't use that many IRQs, enabling or disabling PnP won't have much of
an effect. It's basically meant for systems that need to share IRQs. Enabling the PnP OS
allows Win9x to shuffle and allocate the IRQs as it sees fit.
I don't think he meant LBA. I think you are referring to Chris Bope's statement about
the IDE HDD Block Mode feature being incompatible with WinNT. It has been confimed by a
Microsoft document discussing about Enhanced IDE operation with WinNT 4.0, thanks to Ryu Connor for the link.
Adrian Wong
Adrian's Rojak Pot
http://www.rojakpot.com/
http://www.adriansrojakpot.com/
Comment #44
Thanks for the site, it's very helpful and much appreciated! I've got it bookmarked for
future reference. Looking forward to the Integrated Peripherals page and any further
refinements of the existing pages.
Thanks again!
Harold Nobles
Hello Harold,
Glad you liked it! :) I'll be continuously working to
expand and improve this guide. You can count on that!
Adrian Wong
Adrian's Rojak Pot
http://www.rojakpot.com/
http://www.adriansrojakpot.com/
Comment #45
Hey, great site, although I could use a little more handholding on the PCI IRQ
information, as in how interrupt A, B, etc. become IRQ 9, 10, etc.
Anyway, I'm writing to let you know that users of Windows 2000 should set PNP OS to NO,
not yes. This isn't because Windows 2000 isn't PnP, it's because Windows 2000 wants to
handle all aspects of PnP, and not have the BIOS trying to muck with it. This is part of
the distinction between APM/PnP, which are old standards, and ACPI which is the new PnP
power management standard. With APM/PnP, most of the pnp/power management work was done by
the BIOS, and each BIOS was written differently. With ACPI, each BIOS should behave very
similarly, and allow the OS (Windows 2000) to handle all of the PnP and power management
tasks. Or so I understand from speaking with Microsoft (I might be off a little on the
reasons, but according to MS, turn that PNP OS Installed switch to No).
It's also imperative to have the latest BIOS revision installed, some motherboards such
as the Abit BX6 rev 2 (which I have) won't work unless you manually disable APCI during
Windows 2000 Setup, or you install a BIOS rev from June or newer.
Keep up the good work, I have your page bookmarked and have tweaked my BIOS
accordingly. :)
Sincerely,
Jason Gerend
freelance technical writer/author
Hello Jason,
As soon as time permits, I'll expand on the manual assignment of the IRQs.
Thanks for that neat piece of information! :) I was only
aware that in order for ACPI to work with Windows 2000, APM needed to be disabled. I
didn't know that PnP OS was also to be disabled as well. So, thanks for that tip. I'm sure
that will help many Windows 2000 users.
Adrian Wong
Adrian's Rojak Pot
http://www.rojakpot.com/
http://www.adriansrojakpot.com/
Wow, I just wanted to say I'm impressed with how fast you updated your site after
receiving my comment. Now only if I updated my own site that fast... :) (Address withheld
due to lack of updates :) ).
Keep up the good work, I'm going to put a link to your site up on my site... as soon as
I get around to updating it :)
Jason Gerend
freelance technical writer/author
LOL! I'm only able to do so because I was using my parent's personal Internet connection
in Malaysia. Now that I'm in India and without a personal connection (yet), updates will
be much slower. :(
Adrian Wong
Adrian's Rojak Pot
http://www.rojakpot.com/
http://www.adriansrojakpot.com/
Comment #46
Well, might as well add more fuel to the fire about Shadowing/Caching....
The only time that caching video RAM would help would be if your video driver was
directly manipulating the data that is stored there. I have heard that this was used to
generate some effects way back when video cards were little more than 2D accelerators with
some 3D toys. I would agree that caching video RAM is pointless because most of the time
any modifications that are done to an image that is on the screen is done in normal RAM
and then sent off to Video RAM, where the 3D chip does any processing on it (such as Z
buffer I believe) Nowadays, in my opinion, anyone who uses a recent video card should not
have to enable Video RAM Caching.
Enabling shadow ROM, video or system, has a very unlikely chance of causing system
crashes or corruption. Back when ISA cards were the norm, a lot of them needed a memory
address to work. Some of the drivers that were written for these cards were
"beta" to say the least. These drivers would sometimes look for their respective
card and crash if it encountered a shadowed area. Now, PCI cards can still use memory
addresses, but they are put way up in the 3 to 4GB area, which is almost impossible to
achieve with real RAM unless you have a huge server. Anyway, I guess that I'm saying that
shadowing does not hurt your stability if you are using recent operating systems and
hardware. If it helps your performance by turning it on or off, go for it.
Oh yes. IDE MultiSector Mode and 32 bit transfers are fine for any hard drive that you
buy now. It used to be an issue with older hard drives when these options first came into
the market. Some drives (notably the Western Digitals) had problems with this under any
operating system when you were pushing a lot of data across the IDE bus. I should also
mention that IDE MultiSector Mode is only effective if you have a decent amount of cache
on the hard drive. Cheap drives, such as Fujitsu, have no cache on the board. Since there
is no cache on the board, it can't properly queue up data to be sent along the IDE bus in
MultiSector mode. 32 bit transfer rate only applies to how wide the data bus is between
the IDE controller chip to the rest of the system. The IDE bus is still only 16 bits wide.
Some older chipsets couldn't support 32 bit transfers, but any recent chipset (Later
Pentium and Pentium II) supports this option properly.
I think that this is enough for one email. The fire is high enough as it is. If you
have any questions about what I said, please let me know. Also, if you would like a second
opinion about any topic related to your optimization guide, please let me know.
Curtis Kobelsky
http://stick.dyndns.com
Hello Curtis,
Thanks for your e-mail. It's great!
I added the bit about the possibility of crashes with shadowed video and system BIOS
because quite a few manuals mentioned that possibility. Even some graphics card manuals
state that Video BIOS Shadowing should be disabled. But I guess you are right. This
problem has been licked and there's no need to hark on it anymore. The guide will be
amended to reflect this.
Yes, I agree with you. But I have just received confirmation via a Microsoft document
about EIDE operation with WinNT. According to it, there has been some cases of data
corruption when IDE HDD Block Mode (or 32-bit Disk Access) was enabled. So, Microsoft is
recommending that WinNT users disable IDE HDD Block Mode. It's more of a WinNT problem,
rather than a problem with the hardware itself.
Please feel free to write in about this article or any other article on this site. Your
views are most welcomed! :)
Adrian Wong
Adrian's Rojak Pot
http://www.rojakpot.com/
http://www.adriansrojakpot.com/
Comment #47
Hi Adrian :-)
I must tell u that I am a complete amateur when it comes to computers, if u can't point
and click I can't use it.
That being said, I must congratulate u on your excellent site and the easy to follow
tweak instructions (the things I don' t understand I don't do). This is the first page
that I actually got information that I could apply without crashing my computer and it
seems to have increased performance. I don't have much of a system, ( a no longer in
production Matsonic 5010 motherboard, 233MHz Pentium MMX, 32MB of RAM, 1.2GB hard disk) so
any performance boost is appreciated. Thanks for the tips.
Also, it seems that many of the other informational sites are in many ways dependent on
the revenues of the very manufacturers that they have to review, that may be why they
don't want unbiased reports linked on their page.
Keep up the good work!
Regards,
Joseph Colangelo
Thanks, Joseph! :D
Adrian Wong
Adrian's Rojak Pot
http://www.rojakpot.com/
http://www.adriansrojakpot.com/
Comment #48
Hi,
Very nice guide, good work...
Question:
In the Chipset Features Setup guide you say that using ECC memory does affect
performance, but in the BIOS Features Setup guide you say nothing about a performance hit
for enabling ECC checking of the L2 cache. I always thought that disabling the L2 ECC
would give a small performance boost. Is the L2 ECC checking "free"?
I read this at http://www.glideunderground.com
"Michael Ma has just sent me news via email on a research he as done on Enabling
L2 Cache ECC for Celerons which doesn't actually support it. I've recently done some
research on this issue. Like thousands out there, I have an overclocked P2 Celeron 300A
running at 450. If you look on Intel's site, none of the Celerons have "ECC L2
Cache", but all the regular non Celerons P2s and P3s do. I had the option enabled for
a couple of months before I found this out. I'm pretty sure that if your motherboard is
good enough (I have an ABIT BX6), it knows how to automatically shut off ECC in L2 cache,
like it does with ECC RAM. So for at least good motherboards, having the option enabled
won't hurt anything, but having the option disabled when you have ECC L2 cache would be
loss of some potential in your CPU. Michael Ma
Thanks Michael for the news! "
Is that true? That the Celeron does not have ECC cache?
Krister Bergman
Hello Krister,
Sorry about that, Krister. I overlooked that point. I have corrected it. L2 ECC
checking isn't free but the performance hit is negligible.
If I'm not mistaken, the Celeron has ECC L2 cache onboard. My own Celeron 300A won't
even overclock to 450MHz stably unless ECC L2 Cache Checking has been enabled.
If the Celeron does not have ECC L2 cache onboard, the feature would be meaningless.
Enabling or disabling it won't affect performance - not even a tiny bit. If you can detect
a slight performance drop, that shows that the ECC function is being used. However, most
of the time, the performance difference is negligible.
I doubt that Michael's motherboard automatically shut-off ECC checking for the
Celeron's L2 cache. The default setting for both ECC functions for the RAM and L2 cache
must have been Disabled for his motherboard. If the BIOS can detect that the Celeron has
no L2 ECC cache, it would have removed the option, just like the Processor Number
Feature option which will only appear if you install a Pentium III processor.
Adrian Wong
Adrian's Rojak Pot
http://www.rojakpot.com/
http://www.adriansrojakpot.com/
Comment #49
Hi Adrian,
I just read what you said about the BIOS cacheable and Video BIOS cacheable BIOS
settings. I am no expert on computers, but I always thought that the EEPROM containing the
BIOS is slower to access than the SDRAM, so if the BIOS is copied to RAM (when you enable
this setting) it speeds it up. I don't think enabling these two settings uses up any L2
cache space. Can you clarify this?
Thanks,
Dr. Michael Feldman
Hello Dr. Feldman,
Definitely. ROM BIOS is very slow, compared to SDRAM. The BIOS will be accessible much
quicker if it's shadowed. However, that doesn't mean it will improve the system's
performance. Since it will not be used, if you are using the proper drivers and all, it's
just taking up RAM and doing nothing at all.
That was for system and video BIOS shadowing. But I see that you are talking about the
System and Video BIOS Cacheable settings. In those cases, the BIOS takes the process one
step further and caches the BIOS in the L2 cache. That further improves the BIOS'
performance but again, if the BIOS isn't being used, what's the point of wasting precious
L2 cache on it?
There are a few users who propose that enabling System and Video BIOS Cacheable won't
use up the L2 cache until the BIOS is accessed. However, the benefit of L2 caching will
primarily only be evident for subsequent BIOS accesses. As it goes, the BIOS is rarely
accessed, if ever. Again, the L2 cache's bandwidth will be wasted caching data that's not
constantly accessed.
IMHO, it would really be better to reserve the L2 cache for the system RAM, especially
if you own a Celeron which already has such a limited amount of L2 cache.
Adrian Wong
Adrian's Rojak Pot
http://www.rojakpot.com/
http://www.adriansrojakpot.com/
Comment #50
Nice little guide..
However I am not sure that the A20 Gate explanation is correct. I could be wrong, but I
think I saw at some point that the A20 line was used during the V86 to real mode switch
which can only be performed by a processor reset (by holding the RST line down) and that
in order to do that, the A20 line and the KBD controler were solicited to perform this
operation (this is vague, but it's as far as I can remember)...
Windows 3.1 made heavy use of this technique (remember the C:\WINA20.386 file?), but I
don't think it is used for protected mode to real mode switches (a la win95) and even less
in Windows NT (protected mode only). So my feeling here is that this BIOS option is fairly
useless/harmless today unless you're still a Win3.1 user and you are having timing
concerns...
Another word about shadowing... My opinion on this is :
A) Video ROM BIOS shadowing should definitely be enabled if you are still using DOS
type applications that are using the VBE extensions (original Quake for example)...
B) About shadowing address ranges
1) If you are using an add-on card with a driver which is not using ROM code in this area,
the shadowing is useless
2) If you are using an add-on card with a driver that is making use of ROM code in this
area, it may be useful
3) If you are using an add-on card which is using some CXXX-EFFF area for I/O, then
shadowing would probably prevent the card from working because the memory R/W requests
might not be passed to the ISA bus.
All this needs to be checked/verified...
Thank you,
Ivan Warren
Hello Ivan,
What you said about Gate A20 is correct but these days, the chipset has taken over that
task from the keyboard controller. The keyboard controller was just too slow.
Yes, I can see how shadowing the video BIOS would be useful in the case of a DOS game.
Thanks for reminding me about it. I'll be adding this to the guide.
The addon card's driver should be allowing the system to bypass the BIOS and access the
hardware directly. Can anyone verify this? BTW, thanks for that tip about the CXXX-EFFF
area.
Adrian Wong
Adrian's Rojak Pot
http://www.rojakpot.com/
http://www.adriansrojakpot.com/ |
