Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

TX chipset Limitation?

0 views
Skip to first unread message

Paladin1

unread,
Apr 7, 1998, 3:00:00 AM4/7/98
to

Could anyone explain more about TX ChipSet Limitation?
I mean how come they design a MB that Limit only 64MB memory will
cached?
and how could I avoid the problem or is there any solution that would
increase the performance if memory is execeed 64MB???

Thank You,

François Compagne

unread,
Apr 7, 1998, 3:00:00 AM4/7/98
to

The first Intel Pentium chipset of triton series was the FX.
Then intel derived it in two version , HX geared for servers
(Multiprocessiong, ECC and so) and the VX for stations.
The VX was limited : 64 M cacheable, no multi CPU.
Then the VX evolved to TX (SDRAM, UDMA...)
The HX was stopped because for intel high end Pentium are a no way : Intel
promoted then Pentium PRO and Pentium II for workstatuion and server
(thoses who needed more than 64 Meg).

Also, look at the memory price one or two years ago.

If you want to avoid that problem of 64M cacheable, avoid MB with TX
chipset.

Caveat : Other chipsets can cache more than 64M provided they have enough
Level 2 cache and enough tag ram.
Some chipset are able to cache up to 512 M, but the way some MB manufacters
use it, (one chip for tag ram, 512K cache not extensible), the real
cacheable mem is only 64 M, and spec written by marketing dpt are very hard
to decode.

Look at http://www.anandtech.com/p5xv3.html for an example of bad things.
(Informationn about 2 months old, perhaps fixed).


Paladin1 <ky...@lightspeed.bc.ca> a écrit dans l'article
<352a2d66...@news.lightspeed.bc.ca>...

KALLE

unread,
Apr 7, 1998, 3:00:00 AM4/7/98
to

On Tue, 07 Apr 1998 13:45:08 GMT, ky...@lightspeed.bc.ca (Paladin1)
wrote:

>Could anyone explain more about TX ChipSet Limitation?
>I mean how come they design a MB that Limit only 64MB memory will
>cached?

It's very simple: AMD and other has also made good socket 7 Chips. So
Intel decided to kastrade the cachable area so power users must buy
Intel P II !

>and how could I avoid the problem or is there any solution that would
>increase the performance if memory is execeed 64MB???

I realy doubt!

>Thank You,

KALLE

see http://www.jump.net/~lcs/kalle.htm for ASUS P55T2P4 information.

Coffee or Tea

unread,
Apr 7, 1998, 3:00:00 AM4/7/98
to

Unfortunately, I have purchased one piece of NEC 32 Mb SDRam in order to
increase the memory to 96 Mb SDRam. Before I read this message, I don't
know there is limitation of TX chipset. But right now, I face a problems on
the combination of ram.

I have two old ram 2 x 32 Mb -12ns and they work properly. After I
installed the new ram, error occur.

combination of ram :
1) new ram at socket 1, old ram at socket 2&3..............when windows 95
started, fatal error occurred as a result system halted.
2) any combination of ram except item 1, for instant, new ram at socket 2,
old ram at socket 1 &3. When himem.sys loaded, unreliable memory detect at
3029f??? address as a result system halted.

Could I still solve this problem? I need to throw the new ram away or treat
it as a rubbish? BTW, how much the system will be slow down when the memory
is higher than 64 Mb?

Best Regards,
Coffee

Nathaniel Eccs

unread,
Apr 7, 1998, 3:00:00 AM4/7/98
to

On Tue, 7 Apr 1998 23:20:47 +0800, "Coffee or Tea"
<coffe...@cheerful.com> wrote:


I would take back the one that is giving a himem.sys error

Coffee or Tea

unread,
Apr 8, 1998, 3:00:00 AM4/8/98
to

any suggested remedy?

Coffee

Nathaniel Eccs 撰寫於文章 <352a4af7...@news.erols.com>...

TomMiami

unread,
Apr 8, 1998, 3:00:00 AM4/8/98
to

In response to your original message listed below:
I too have increased to 96mb on the TX board. At first I encountered many
problems, then began to look throught the bios settings. I began slowing
things down on the memory. Leave 60ns ok, but increase other items from 111
and 222 to 333 or 444 and upward. Until all works ok now... Checked
performance and it pretty much matches when I had 64MB ram....
Tom Graham
DataTron Computer Services, Inc.
e-mail: TomMiami+AEA-datatron.net
url: http://www.datatron.net
Nathaniel Eccs wrote in message +ADw-352a4af7.9202452+AEA-news.erols.com+AD4-...
+AD4-On Tue, 7 Apr 1998 23:20:47 +080-, +ACI-Coffee or Tea+ACI-
+AD4APA-coffee+AF8-tea+AEA-cheerful.com+AD4- wrote:
+AD4-
+AD4-
+AD4-I would take back the one that is giving a himem.sys error
+AD4-
+AD4APg-Unfortunately, I have purchased one piece of NEC 32 Mb SDRam in order to
+AD4APg-increase the memory to 96 Mb SDRam. Before I read this message, I don't
+AD4APg-know there is limitation of TX chipset. But right now, I face a problems
on
+AD4APg-the combination of ram.
+AD4APg-
+AD4APg-I have two old ram 2 x 32 Mb -12ns and they work properly. After I
+AD4APg-installed the new ram, error occur.
+AD4APg-
+AD4APg-combination of ram :
+AD4APg-1) new ram at socket 1, old ram at socket 2+ACY-3..............when windows 95
+AD4APg-started, fatal error occurred as a result system halted.
+AD4APg-2) any combination of ram except item 1, for instant, new ram at socket 2,
+AD4APg-old ram at socket 1 +ACY-3. When himem.sys loaded, unreliable memory detect
at
+AD4APg-3029f??? address as a result system halted.
+AD4APg-
+AD4APg-Could I still solve this problem? I need to throw the new ram away or
treat
+AD4APg-it as a rubbish? BTW, how much the system will be slow down when the
memory
+AD4APg-is higher than 64 Mb?
+AD4APg-
+AD4APg-Best Regards,
+AD4APg-Coffee
+AD4APg-
+AD4APg-
+AD4-

Dave Tatosian

unread,
Apr 9, 1998, 3:00:00 AM4/9/98
to

In article <352a2d66...@news.lightspeed.bc.ca>,

ky...@lightspeed.bc.ca (Paladin1) wrote:
>Could anyone explain more about TX ChipSet Limitation?
>I mean how come they design a MB that Limit only 64MB memory will
>cached?
>and how could I avoid the problem or is there any solution that would
>increase the performance if memory is execeed 64MB???

First, it's not a "motherboard design" issue - it's an "Intel would rather
have you buy a P2 system" problem. Ie: blame the marketeers, not the mainboard
designers.

Second, the only way to avoid the TX chipset limitation is to not buy a TX
chipset board.

/dave

Idi Amin

unread,
Apr 11, 1998, 3:00:00 AM4/11/98
to

In article <6ggvqr$1g56$1...@news.gate.net>, TomM...@datatron.net says...

> In response to your original message listed below:
> I too have increased to 96mb on the TX board. At first I encountered many
> problems, then began to look throught the bios settings. I began slowing
> things down on the memory. Leave 60ns ok, but increase other items from 111
> and 222 to 333 or 444 and upward. Until all works ok now... Checked
> performance and it pretty much matches when I had 64MB ram....
> Tom Graham

And those settings are?
--
Remove the "*" to reply...

TomMiami

unread,
Apr 13, 1998, 3:00:00 AM4/13/98
to

Actually, I played around with the Chipset Features in the bios untill I got
all working without fail. Since leaving prior message posted, I've tweaked a
little more with success. I found that the '16 bit I/O Recovery Time' set
faster than 4 BUSCLK (lower number) is what was giving me trouble. I only
have 1 16-bit card (sound blaster 64). I now am working with the following
in order:
x222
x222
10T/6T
3T
4T
Enabled
Enabled
3T/4T/7T
3T
Enabled
Enabled
Enabled
4BUSCLK
5BUSCLK
--

Tom Graham
DataTron Computer Services, Inc.
Miami, Florida
remove .NS (NoSpam) from e-mail address
e-mail: TomMiami.NS+AEA-datatron.net
Idi Amin wrote in message ...
+AD4-In article +ADw-6ggvqr+ACQ-1g56+ACQ-1+AEA-news.gate.net+AD4-, TomMiami+AEA-datatron.net says...
+AD4APg- In response to your original message listed below:
+AD4APg- I too have increased to 96mb on the TX board. At first I encountered many
+AD4APg- problems, then began to look throught the bios settings. I began slowing
+AD4APg- things down on the memory. Leave 60ns ok, but increase other items from
111
+AD4APg- and 222 to 333 or 444 and upward. Until all works ok now... Checked
+AD4APg- performance and it pretty much matches when I had 64MB ram....
+AD4APg- Tom Graham
+AD4-
+AD4-And those settings are?
+AD4---
+AD4-Remove the +ACIAKgAi- to reply...

0 new messages