Question for CAT MAKER - Concerning LargeSystemCache

[Monolyth]

Hello,
I first just wanna say what an awesome job you guys @ ATI have done w/the driversets, I am extremely glad I purchased my 9800 Pro.

One thing has bothered me though, ever since the early CAT 3.x driversets setting a certain registry value has resulted in "Delayed Write Failures" on bootup of XP for many people. The setting would be:

[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Contro l\Session Manager\Memory Management]
"LargeSystemCache"=dword:00000000

This setting can push a little more performance from a persons computer, it is a rather small amount, but we all know how "tweakers, o/c's, and enthusiasts" can be about performance. I was wondering if the CATALYST Team was going to be looking into this when they got a chance?

No real rush, only reason I ask is that some "Tweak" guides don't fully reveal that there is a conflict, I just don't want people going down the same path as I did (reformatting twice, hehe).

Thank you for your time, and for these awesome drivers :) Definately sticking with ATI for awhile!

Peace!

[GamblerFEXonlin]

I want Large system cach back too, its very nice with games with big save files like neverwinter nights and baldurs gate.

moving around files, unpacking files is also very pleasant, but you need lots of ram. Taking screenshots in games is also nice then.

Here's an old thread about it on Rage3d:

http://www.rage3d.com/board/showthread.php?threadid=33683515&perpage=20&pagenumber=1

possible workaround:
http://groups.google.com/groups?selm=_lcLb.8134%24uF6.2528978%40news1.news. adelphia.net&oe=UTF-8&output=gplain

(i havent tried it i cant understand what he say)

Might be an hardware error with the radeon since it so hard to fix? It's obviously a bug not a feature, or else a strict warning should be when you install your Radeon+Catalyst drivers as HD corruption might happen.

Powercolor Radeon 9600
Catalyst 4.2
WinXP pro

[Monolyth]

Wow that guy has no idea how to write a legible guide lol, but anyways, I may try it this weekend when I have time, and try to legibly explain what he did if in fact it works. He could of course be bs'ing. But oh wells.

I'll let ya know if I get workin :)

Tweakers of the world unite! All those that say this registry "tweak" is useless go do something else!

[NickSpolec]

Unforturnately, ATI's driver team doesn't add features (no matter how useful) or fix bugs unless it's something major, great demand for it, or they see it as relevant.

[Monolyth]

Ye of little faith.
:)

[ChrisW]

http://mirror.ati.com/support/infobase/4217.html

[Razor04]

Originally posted by GamblerFEXonlin
Might be an hardware error with the radeon since it so hard to fix? It's obviously a bug not a feature, or else a strict warning should be when you install your Radeon+Catalyst drivers as HD corruption might happen.
It isn't ATI's problem...it can occur with any piece of software. The driving factor in whether or not it is a problem seems to be Windows mood on any given day. I know I wrote a thread about this in the past (not here though) and I will try to dig it up as I really don't want to type all this again. Give me a few hours and I will check after my test is out of the way. Oh and why should there be a warning? It is random and occurs with a feature not intended for home systems and isn't specific to ATI.

[Monolyth]

Naw naw I wasn't blaming ATI for anything, I'm a Web Developer, and trust me I play the blame game enuff at work w/network people.

But anyways, I just think that giving light to things like this is better than just leaving them be forever. It's not like everyone posts everyday on this topic, it's just something I like bringing up whenever a new X. set of CAT's come out, just to kinda check things.

BTW 2 reboots, no delayed failure w/Cache enabled and no ATI "workaround". But like you said it probably is just a matter of time till Windows decides to crap out, but most of the time in the past I've had it error out the moment I installed ATI drivers. We'll see, I really need me a test system, cuz it would be a major pain to rebuild this sucker again hehe.

[Razor04]

Originally posted by Monolyth
BTW 2 reboots, no delayed failure w/Cache enabled and no ATI "workaround". But like you said it probably is just a matter of time till Windows decides to crap out, but most of the time in the past I've had it error out the moment I installed ATI drivers. We'll see, I really need me a test system, cuz it would be a major pain to rebuild this sucker again hehe.
I found the posting from Beyond3D:

http://www.beyond3d.com/forum/viewtopic.php?t=9227&start=20

Part about Large System Cache's starts a few posts down...if you really want to know everything about that feature then read that whole post.

[Monolyth]

20 reboots and no failure...by now I SHOULD have croaked according to the "general system specs" straight from ATI:

System Memory greater than 512 Meg. (1 gigabyte of RAM is common) -
1GB, and HUGE FRIGGIN (3GB) swapfile, on its own partition.


Large NTFS disk volumes. And multiple large volumes. (60-100 gigabyte hard drives possibly in RAID arrays) -
2 x 74GB WD Raptor's in SATA RAID 0
Maxtor 250GB SATA


AGP graphics with large AGP resource requirements (AGP aperture greater than default) -
About the only setting that's "pseudo default" - 128MB in BIOS


Large file transfers -
I do lots of video editing, so I have multigb movies on my storage drive and are copied and moved all the time.


So with all this do you really think I could have escaped this "bug" for 20 reboots? I seriously doubt not...
With that said, I would recommend anyone with a spare system running an ATI Radeon w/latest CAT's should follow the next few steps and see if it actually works on "You're" setup. I take no responsibility if your system goes kaputt (that's why it's a test setup anyways :-). This is an "english" understandable guide adapted from the steps taken in the guild here: http://groups.google.com/groups?selm=_lcLb.8134%24uF6.2528978%40news1.news. adelphia.net&oe=UTF-8&output=gplain

-Go to SMARTGART
-Turn FastWrites Off and AGP OFF
-Reboot
-Check SMARTGART to make sure Fast Writes and AGP are OFF
-Reboot
-Do the Registry Edit
-Reboot
-Re-Enable the Settings in SMARTGART
-Reboot
-Pray to God.

Good luck, I know it looks entirely retarded that doing it like that might work, but what I'm thinking is that it has to do with some form of memory "clash" that occurs between SmartGART and System Caching. They attempt to write to the same memory location and end up overflowing and crashing the lsass.exe service (Storage Service). So it could be that turning off the SmartGART enhancements and "THEN" setting and then enabling caching, and re-enabling smartgart after clean boots from each change allows the two to access the lsass function and choose their own memory location without causing an overflow. I don't know if it does croak on me I will edit this post ASAP and remove the steps and make big ole letters saying we're fux0red, and apologize, but until that happens I will stick to this.

Peace!

[Razor04]

Originally posted by Monolyth
2 x 74GB WD Raptor's in SATA RAID 0
I just want to say you suck and I am quite jealous... :)

[Monolyth]

Yeah i had the 37's before too, bought the 37's @ 250 each and now their like 100 bux a piece >:|
but I needed more space on my Array *sigh* dern customers all want uber HQ video's now.

[Quitch]

Originally posted by Razor04
I just want to say you suck and I am quite jealous... :)

(S)He'd be better off without the RAID, and using one as the OS drive and the other for programs. Patented StorageReview wisdom that :)

[Razor04]

Originally posted by Quitch
(S)He'd be better off without the RAID, and using one as the OS drive and the other for programs. Patented StorageReview wisdom that :)
It isn't that...it is the fact that I am a poor college student and could never afford something like that. Hell I sold off a couple of old drives to pay off a student loan interest payment a couple months back. Now I am stuck with carefully managing my disk space. :(

[Monolyth]

Do you have a link to the benchies for that? I did a brief look on SR.com but didn't find anything. I could see how that would be true if a person was using a PCI33 card for their RAID, considering the limit for PCI's 33 spec is 133mhz aka 133MB/s, and you would never be able to reach that IRL. I use onboard SATA RAID right now, which does not suffer near the amount of speed issues that a PCI33 card would have.

This is just a guess as to why they would recommend using it on std. SATA settings. I don't know what they revealed as well. I also use the machine for gaming so I enjoy faster load times. ;)

[Quitch]

Can't link off-hand, but I've read it when I built my machine and was toying with RAID0 (search for posts by Quitch)

Post on their fourms about it. They will link the article AND tell you why it's not a good idea... however, if load times are the only thing you worry about, then RAID0 is a good idea.

Overall though, you're better without.

[Monolyth]

Wilst I found this:
http://faq.storagereview.com/tiki-index.php?page=SingleDriveVsRaid0

I still did not see where this guy had much proof at all to support his "hypothesis" especially concerning 2 threads hitting the array at once. Even with a single drive setup, 1 for OS, 1 for applications, you would still see that if you ran 2 applications at once the head would still have to hit 2 places on the same drive in the order that they requested, which doesn't give you anything except common IDE performance. And SCSI is frankly not recommended in my opinion for a board that does not contain a PCI66 slot. The PCI bus total can only handle up to 133MB/s, and having a U160 or U320 on there would saturate the bus, leaving Windows crawling.

I was not impressed by that article at all, the only part that is useful is concerning the pagefile but I already knew this. Thx for pointing me to the link Quitch, but with something like that you have to provide quantitive data in order for such "hypothesis" to be accepted. That was just someone trying to explain away their point of view, as is their right, but IMHO it doesn't belong in a FAQ.

Cheers!

[Quitch]

I'd rather you posted that on their forums so as to give them a chance to defend their position. I'd be interested to see that.

[budd_wm]

So did anyone else ever try the fix. I'd do it but for the moment can't back up the stuff on my RAID array. I'll do it once I can, but that may be awhile....

BTW, your system still running Monolyth?

[Quitch]

I don't see why this shouldn't work and why it isn't support, even Microsoft say it is for servers OR programs that need it. It is of benefit (IIRC) to any program that accesses a lot of files, and ones like Baldur's Gate II which access large numbers of small files will gain from this setting.

The fact that it isn't for desktops is a nonsense. All this setting does is control how much memory is set aside for the system cache, Application Performance being the smaller value. The reason this is the default is because desktops tend to have less memory than servers, and Microsoft don't want you to run out.

What people fail to grasp is that Windows, as it ships, is configured to work for everyone, in every circumstance. It comes shipped as a jack of all trades, but a master of none. If you know you are never going to run low on memory then there is NO down side to changing this setting to System Cache, since you have more than enough RAM to take the extra hit.

Just because 50 services are on by default, doesn't mean I need them. They're on because I MIGHT need them, and I MIGHT not know how to configure services. This is why Windows is the desktop system of choice, it just works. But if you know what does what and why, then OF COURSE you're in a better position to set the options than Microsoft, they had to configure for every PC out there, while you need to only worry about yours.

[GamblerFEXonlin]

It seems the problem is indeed with Windows XP

With a lot of RAM and a lot of I/O operations (for instance read/write to harddisk) windows "get dry" of Page Table Entries (PTEs). This causes blue screen or other funny errors.

A high AGP aperture size make the problem worse (they mention "higher then default", maybe smaller than default doesnt help).

Increasing the Page Table Entries might "fix it" according to
http://mirror.ati.com/support/infobase/4217.html

I dont know if its a real fix or a workaround that will fail if you go extreme with RAM and IO operations and AGP aperture size.


BUTT my GEForce 3 64MB, 1GB ram and windowsxp SP1 it worked completely.

Then I dare claim it is either
64MB graphics ram vs 256MB
AGP4x vs 8x
NVIDIA have workaround in their driver

NVIDIAs workaround could be
- hack registry and increase page table entries
- less IO operations and conservative use of Page Table entries provided by OS
(driver is more clean or doesnt have long FP32 and lacks PS 2.0 performance :D

[Razor04]

Originally posted by GamblerFEXonlin
It seems the problem is indeed with Windows XP

With a lot of RAM and a lot of I/O operations (for instance read/write to harddisk) windows "get dry" of Page Table Entries (PTEs). This causes blue screen or other funny errors.

A high AGP aperture size make the problem worse (they mention "higher then default", maybe smaller than default doesnt help).

Increasing the Page Table Entries might "fix it" according to
http://mirror.ati.com/support/infobase/4217.html

I dont know if its a real fix or a workaround that will fail if you go extreme with RAM and IO operations and AGP aperture size.

BUTT my GEForce 3 64MB, 1GB ram and windowsxp SP1 it worked completely.

Then I dare claim it is either
64MB graphics ram vs 256MB
AGP4x vs 8x
NVIDIA have workaround in their driver

NVIDIAs workaround could be
- hack registry and increase page table entries
- less IO operations and conservative use of Page Table entries provided by OS
(driver is more clean or doesnt have long FP32 and lacks PS 2.0 performance :D
For the last time...

* It is a Windows XP problem...I have never experienced it under Windows 2000
* It can occur with any piece of software...not just ATI's

The real solution to this problem would be for MS to find the problem and fix it...but if you think that will happen anytime soon you are sorely mistaken. You know what the funniest thing about this is...it is that this problem can occur with an NV setup too (and has for some people) but ATI is always mentioned because they have a page detailing the problem. In my eyes that is unfair...just because they documented it doesn't mean it is their problem (as the fix they suggest shows) or present only on systems with ATI cards.

[budd_wm]

Well he never responded so I guess his system crashed...;)

[AngelGraves13]

I had this problem too, but the Cat 4.2 fixed it for me. I don't get it anymore. I hope it wasn't just luck and that its been fixed.

[Quitch]

I had this problem whenever I installed the latest Intel chipset drivers :)