Very Computer

by **chiej** » Fri, 27 Oct 1995 04:00:00

Hi there,

Will Linux run on Intel Zappa or Endeavor board? Is there any known
problem with these two boards?

Thanks,

Chie-Jin
--
------------------------------------------------------------------------------
Chie-Jin Cheng nFX Corporation
A new type of Special Effects........
nth dimension * Effects........

(408) 748-9200 (voice) 4800 Great America Parkway
(408) 748-9272 (fax) Santa Clara, CA 95054

by **Kai Leibrand** » Sat, 28 Oct 1995 04:00:00

Linux is fine on the Endeavour, and also on the Zappa. I can recommend
getting the Endeavour - it gives you the better (Triton) chipset.

Kai.
--
____________________________________________________________________

Interaction Design Research
____________________________________________________________________

Design Researcher

Brunel University fax: +44 (0)1784 472879
Department of Design voice: +44 (0)1784 431341 Ext. 244

by **Witold Robert Jachimcz** » Sun, 29 Oct 1995 04:00:00

: Linux is fine on the Endeavour, and also on the Zappa. I can recommend
: getting the Endeavour - it gives you the better (Triton) chipset.

: Kai.
: --

Endeavour gives better : Triton chipset? What? Well I don't know
anything about Endeavour but Zappa has Triton Chipset.

I have Zappa, it's great (especially with MR BIOS)!

Witek

by **roo** » Thu, 02 Nov 1995 04:00:00

It is true. Zappa had some problems with their design and the Triton
chipset. Good chipsets don't always make good motherboards. I know
that MR BIOS fixes some of this problems, but I never tried it.
Where did you get this MR BIOS anyway?

According to Intel, problems with Zappa are fixed on Eneavour.

|> : Linux is fine on the Endeavour, and also on the Zappa. I can recommend
|> : getting the Endeavour - it gives you the better (Triton) chipset.
|>
|> : Kai.
|> : --
|>
|> Endeavour gives better : Triton chipset? What? Well I don't know
|> anything about Endeavour but Zappa has Triton Chipset.
|>
|> I have Zappa, it's great (especially with MR BIOS)!
|>
|> Witek

--
Jovan Ilic

by **Chris Eilbe** » Thu, 02 Nov 1995 04:00:00

: Sorry about that!
: I thought the Zappa still had the Mercury chipset...

No! I have both Zappa and Endeavour and both have the Triton chipset.
Both work fine with Linux and are considerably faster (33%) than an Opti
board I used to have with the same chip.

: Kai.

Chris

--

Room PC126, DRA Malvern. Tel +44 1684 896083. Fax +44 1684 896121

by **Xavier Bo** » Sat, 04 Nov 1995 04:00:00

By the way...

Has someone tried the new ide drivers (with Bus Master DMA
support for the Triton chipset) with the Advanced/EV (aka
Intel Endeavor)???

I cannot make the driver work... It doesn't recognize the
Bus - Master capabilities of the motherboard...

Thanks in advance!

--
-------------------------------------------------------------
o o o Xavier Boix i Marcos. Becari de Sistemes Operatius
o o o Departament d'Arquitectura de Computadors - UPC

U P C WWW home page: http://www.ac.upc.es/~xavier
-------------------------------------------------------------

by **Kai Leibrand** » Tue, 07 Nov 1995 04:00:00

Well, that's because Intel for some obscure reason decided to _not_ use
Triton internal PCI-IDE interface, but stick a seperate I/O chip on the
Endeavour that also has a PCI-IDE port on it. It disables the Triton
PCI-IDE, and I don't think you can re-enable it.
Sorry! I have the same problem!

Kai.
--
____________________________________________________________________

Interaction Design Research
____________________________________________________________________

Design Researcher

Brunel University fax: +44 (0)1784 472879
Department of Design voice: +44 (0)1784 431341 Ext. 244

by **Kim G. Fr** » Wed, 08 Nov 1995 04:00:00

|>: Well, that's because Intel for some obscure reason decided to _not_ use
|>: Triton internal PCI-IDE interface, but stick a seperate I/O chip on the
|>: Endeavour that also has a PCI-IDE port on it. It disables the Triton
|>: PCI-IDE, and I don't think you can re-enable it.
|>: Sorry! I have the same problem!
[snip]
|>Can anyone else confirm this? (That Intel's own Triton-based motherboards
|>have a Linux-incompatible IDE channel?) It seems amazing. If you are
|>running Linux on a Zappa or Endeavour, please post your experiences, good
|>or bad.

I'm running on a Zappa board (p5 133, 32 Mb ram, std. bios). I'm only
using the primary IDE channel for a CD-ROM drive (Mitsumi FX100) and I
don't have any problems. I do get an unexpected interrupt from the IDE
driver at boot, but it doesn't seem to harm. So at least the internal
interface on the Zappa motherboard is detected and works.

-kim
----------------------------------------------------------------
Kim Frei Tel: +45 45 99 95 99 AVS/UNIRAS A/S

DK-3460 Birkerod
Opinions are mine(TM) Denmark

by **Kai Leibrand** » Wed, 08 Nov 1995 04:00:00

> Can anyone else confirm this? (That Intel's own Triton-based motherboards
> have a Linux-incompatible IDE channel?)

^^^^^^^^^^^^^^^^^^
That's not what I said. What I said was that the IDE port supplied by
the Triton chipset isn't used, but rather a seperate one provided by
the I/O componentry on the Motherboard. It just means you can't use
Triton's BusMaster DMA.
Comparing normal Triton (non-busmaster) transfer speeds with those
achieved by the seperate IDE I have on my Endeavour yields _no_
performance difference, as stated before.
It is _not_ incompatible with linux!

Kai.

--
____________________________________________________________________

Interaction Design Research
____________________________________________________________________

Design Researcher

Brunel University fax: +44 (0)1784 472879
Department of Design voice: +44 (0)1784 431341 Ext. 244

by **roo** » Thu, 09 Nov 1995 04:00:00

>Can anyone else confirm this? (That Intel's own Triton-based motherboards
>have a Linux-incompatible IDE channel?) It seems amazing. If you are
>running Linux on a Zappa or Endeavour, please post your experiences, good
>or bad.

Linux 1.3.37, Intel Zappa 100MHz, *old* 70ns simms on 30->72 adapters, Maxtor
7540A IDE HD:

Runs like a dream. So easy to install and operate, I was depressed.
I wanted to hack more to get my hot new MB to work. But it does work....
(although I feel I should note that my Pro Audio Spectrum doesn't, now.
It's manual says it wants an 8 MHz bus, and I can only slow mine down to 8.3
MHz. Seems to make enough of a difference....)

by **Mark Lo** » Thu, 09 Nov 1995 04:00:00

>By the way...

>Has someone tried the new ide drivers (with Bus Master DMA
>support for the Triton chipset) with the Advanced/EV (aka
>Intel Endeavor)???

>I cannot make the driver work... It doesn't recognize the
>Bus - Master capabilities of the motherboard...

Try again with linux 1.3.37, and email me the results (if you want my help).
--

For Linux IDE (big/many) help, see: /usr/src/linux/drivers/block/README.ide
For latest Linux kernels: ftp.cs.helsinki.fi:/pub/Software/Linux/Kernel/v1.[23]

by **Kai Leibrand** » Fri, 10 Nov 1995 04:00:00

> >Has someone tried the new ide drivers (with Bus Master DMA
> >support for the Triton chipset) with the Advanced/EV (aka
> >Intel Endeavor)???

> >I cannot make the driver work... It doesn't recognize the
> >Bus - Master capabilities of the motherboard...

Well, again: the Andeavour uses a seperate I/O component, which replaces
the Triton-internal PCI-IDE with another, non-DMA-bustering non-PCI EIDE
port. Why Intel decided to do this is unclear. The component used is
from NSC, and is the same as on the Zappa. The Zappa is said to have
more problems with this chip, so that some people have reported to have
to use the compatability mode in the Win95 IDE drivers - not good for
performance.
However, performance penalties are unlikely to be big, unless you have a
fast spinning IDE drive. In fact, in C't 10/95 test results suggest that
the interfaces on the Asus P/I-P55TP4Xe and the Endeavour have virtually
identical transfer rates...

Kai.
--
____________________________________________________________________

Interaction Design Research
____________________________________________________________________

Design Researcher

Brunel University fax: +44 (0)1784 472879
Department of Design voice: +44 (0)1784 431341 Ext. 244

by **Nat Makarevit** » Fri, 17 Nov 1995 04:00:00

KL> Well, again: the Andeavour uses a seperate I/O component, which
KL> replaces the Triton-internal PCI-IDE with another, non-DMA-bustering
<...>
KL> performance penalties are unlikely to be big, unless you have a fast
KL> spinning IDE drive. In fact, in C't 10/95 test results suggest that
KL> the interfaces on the Asus P/I-P55TP4Xe and the Endeavour have
KL> virtually identical transfer rates...

Indeed, because the transfer rate is directly related to the density and
rotation speed of the drive used.

But DMA busmaster eats much less CPU power, meaning that processes waiting
for an I/O to complete do not hang other processes.

Is there any benchmark around, showing Bonnie and IOZone results of a given
EIDE box and disk with and without DMA ?

--
Linux

by **Mark Lo** » Sat, 18 Nov 1995 04:00:00

> KL> Well, again: the Andeavour uses a seperate I/O component, which
> KL> replaces the Triton-internal PCI-IDE with another, non-DMA-bustering
><...>
> KL> performance penalties are unlikely to be big, unless you have a fast
> KL> spinning IDE drive. In fact, in C't 10/95 test results suggest that
> KL> the interfaces on the Asus P/I-P55TP4Xe and the Endeavour have
> KL> virtually identical transfer rates...

>Indeed, because the transfer rate is directly related to the density and
>rotation speed of the drive used.

Well, yes and no. Yes, the *sustainable* transfer rate is directly related
to the speed of the media, but the shorter burst interval is a direct function
of the interface timings. And the timings should be identical with the two
chipsets.

Quote:>But DMA busmaster eats much less CPU power, meaning that processes waiting
>for an I/O to complete do not hang other processes.

>Is there any benchmark around, showing Bonnie and IOZone results of a given
>EIDE box and disk with and without DMA ?

Nope. Here is what my P120 Triton system gives, with and without DMA.

This is with the drive's built-in write cache feature turned on (optimal
state for either test), and a command line of "time iozone 40 8192":

(1) With DMA:
IOZONE performance measurements:
5121250 bytes/second for writing the file
3560529 bytes/second for reading the file
kernel CPU: 5.43sec

(2) No DMA:
IOZONE performance measurements:
5090174 bytes/second for writing the file
3396197 bytes/second for reading the file
kernel CPU: 10.11sec
--

For Linux IDE (big/many) help, see: /usr/src/linux/drivers/block/README.ide
For latest Linux kernels: ftp.cs.helsinki.fi:/pub/Software/Linux/Kernel/v1.[23]