Linux-Hardware Digest #716, Volume #10 Fri, 9 Jul 99 16:13:33 EDT
Contents:
Diamond Fire GL 1000 PRO ("Alex Korshunov")
Re: CD ROM mount problems... (toby)
Re: Linux/KDE; KDat backup on dat tape proggy (Marc SCHAEFER)
Re: Internal Modem (Michael McConnell)
Re: Celeron, what's the catch? (Anthony Hill)
Re: Help! My awe32 suddenly stopped working (Martin)
Re: Mounting probelm extended VFAT 32 (Michael McConnell)
Re: Making MPEG movies from AVI? (Walter B Kulecz, PhD)
Re: Windows easy to install? BULLSHIT! (Hobbyistİ)
Re: Mounting probelm extended VFAT 32 (toby)
Re: Video Card Recommendation? (Rahul Karnik)
Re: SoundBlaster Live! (Rahul Karnik)
Re: SCSI CD-RW: yamaha 4416s: HowTo REWrite CD-RW's? (Nektarios Lathiotakis)
Win95 -> Linux via null-modem .... help! (Pankaj Pant)
Re: CPU Question (Johan Kullstam)
Re: Celeron, what's the catch? (Marc Mutz)
Re: Celeron, what's the catch? (Marc Mutz)
CDROM changer recommendation? (Alan Jaffray)
Cirrus Logic CL5446 rev 69 ("Michael Swanson")
----------------------------------------------------------------------------
From: "Alex Korshunov" <[EMAIL PROTECTED]>
Subject: Diamond Fire GL 1000 PRO
Date: Fri, 9 Jul 1999 01:06:01 +0400
Linux RedHat 5.0
Windows X not support my video cards. Help me, plz!
------------------------------
From: toby <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup
Subject: Re: CD ROM mount problems...
Date: Fri, 09 Jul 1999 13:09:18 -0400
Reply-To: [EMAIL PROTECTED]
try:
shell>mount /mnt/cdrom
>
>
> I used to mount my cdrom using "mount /cdrom" ... now it always tells
>
> me:
>
> "mount: can´t find /cdrom in /etc/fstab or /etc/mtab"
>
> http://www.searchlinux.com
------------------------------
From: Marc SCHAEFER <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup,comp.arch.storage
Subject: Re: Linux/KDE; KDat backup on dat tape proggy
Date: 9 Jul 1999 09:16:12 GMT
In comp.arch.storage Gene Heskett <[EMAIL PROTECTED]> wrote:
: Has anyone managed to get the KDat proggy to run their dat drive yet?
Yes.
But that was long time ago (maybe 1.1pre1). It was working mostly.
However, from the impression I got, I would recommend using tar and/or
cpio directly from the shell. Maybe KDat has improved. But. do you
really want your data's integrity to depend on a graphical interface ?
: device as mt is (a freshly made /dev/nst0) when I tested that, choosing
: instead to cover all its sins with a simple "no tape in drive" message.
: Messages to the author have gone un-acknowledged for about 20 hrs now.
what does:
mt -f /dev/nst0 status
say ?
------------------------------
From: Michael McConnell <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup
Subject: Re: Internal Modem
Date: Fri, 9 Jul 1999 12:48:40 +0100
Wouldn't be so sure about that. IIRC USR actually labelled their winmodems =
as
such. I wish other manufacturers would do the same.
Also... (but I may be wrong here) the USR Sportster PnP cards could be
jumpered too.
-- Michael
(still using a non-PnP USR 28k8 card that's a little long in the tooth)
On Thu, 8 Jul 1999, James wrote:
> I think this internal modem is a Winmodem, and, as it,
> it's not a real modem. In fact, those pseudo-modems
> work with software and this softs are not developped
> for Linux by the manufacturer. Winmodem emulate a
> real modem thanks to CPU (the software). So, I'm
> really sad to tell you that your "modem" will never work
> under Linux, except if the manufacturer provide the soft !
> Sorry.
>=20
>=20
> Asim Shankar a =E9crit dans le message <7m1fpt$jup$[EMAIL PROTECTED]>...
> >Hi!
> >
> >I have an internal ISA Plug-N-Play USRobotics 33.6Kbps FAX modem.
> >I just installed Caldera OpenLinux 2.2
> >
> >How do I get the modem working in Linux??
> >
> >I retrieved some info from the Windows' Device Manager, namely the IO
> >addresses and IRQ for the modem, if that helps. It's also assigned COM3 =
in
> >Windows.
> >
> >IO - 03E8-03EF
> >IRQ - 5
> >
> >Please tell me how to get this modem functional for Kppp etc.
> >
> >Thanks,
> >
> >-- Asim Shankar
> >
> >
> >
>=20
>=20
>=20
>=20
-- Michael "Soruk" McConnell=09=09=09 [Red Hat 6.0 Available!]
Eridani Star System -- The Most Up-to-Date Red Hat Linux CDROMs Available
Email: [EMAIL PROTECTED] http://www.amush.cx/linux/ Fax: +44-8701-600807
------------------------------
From: [EMAIL PROTECTED] (Anthony Hill)
Crossposted-To: comp.sys.ibm.pc.hardware.chips,comp.sys.intel
Subject: Re: Celeron, what's the catch?
Date: Thu, 08 Jul 1999 22:39:20 GMT
On 6 Jul 1999 05:38:01 GMT, [EMAIL PROTECTED] (Chris Robato
Yao) wrote:
>>not MY benchmark results. Where are YOURS? Couldn't find any eh? That
>>poor look up ability of yours I suppose...
>
>You are the one claiming a Dual Celeron setup (using 333s) can be faster
>than a K6-III 450 on LInux. The Burden of Proof is on you.
Agreed, the dual Celeron route is more expansive (the dual
slot 1 motherboard adds an absolute minimum of $50 over the price of
most super7 boards, plus another $30 for the two slotkets, plus the
cost of two Celerons), and in most situations on a P6 architechture,
you'll be lucky to see more then 50% increase in performance (there
are some obvious exceptions to this rule of course). At best you
could expect a dual-Celeron 333 to match the performance of a K6-III
450 under most Linux applications, and that's only when applications
are able to take advantage of the two processors. As some others have
pointed out, compiling the kernel under Linux is one such situation.
Now, of course if you're comparing a non-overclocked K6-III to
a pair of overclocked Celerons, obviously things aren't going to be
the same. As long as Celerons are very overclockable, it'll be tough
for AMD to provide better bang for their buck to the tiny percentage
of tweakers out there. Mind you, they don't really have to, as sales
to this small market aren't a very significant portion of the market.
Celerons are also becoming somewhat harder to overclock. The 300MHz
Celerons have largely disappeared, and even the 333MHz ones are
starting to go away. 366MHz Celerons might make it to 550MHz, but
you're probably looking at about a 50/50 chance of being able to
manage to overclock that high with a reasonable degree of stability,
and in a dual setup things are only likely to get worse.
>But you have not shown anything to back your claims.
>
>Have you seen http://www.cpureview.com?
>
>Article shows a K6-III 400 is faster than a Cel @ 450 in compiling.
The K6-III is indeed a very good processor for compiling
(among other things). Clock for clock it's probably the fastest x86
processor out there when it comes to compiling (except maybe Cyrix
MIIs, but they're running at less then 300MHz for the most part, so
they don't really count).
>>>You have not yet answered my other question on your ratios of K6 vs. PII
>>>FPUs.
>>
>>benchmarks, hardware & programming sites. That & I own a dog of a cpu(k6-2
>
>I have not see any benchmark or website that suggests a .55 ratio for
>the P2 FPU vs. K6. You made this up. Back it up.
In raw FPU number crunching which has is produced with at
least semi-optimized code, the PII could manage to be close to twice
the speed of the K6. I'd imagine that if AMD ever actually released
SPEC numbers for the K6-2 it would probably be only just slightly
better then 1/2 the score of a PII at the same clock rate. Now, mind
you, you won't see that big of a difference in real world
applications, but the FPU of the K6 line is a touch weak. With the
K7, AMD appears to have really concentrated on this (as seen by the
dramatic increase in FPU performance as compared to very little
improvement in integer).
>On the contrary, there is plenty that suggests otherwise. Read the
>K6-III review of http://www.combatsim.com. According to them, a K6-III
>450 is barely equivalent to a PII 400 on non 3DNow games.
This is due primarily to the fact that games don't use all
that many floating point calculations. The K6-III slows down a fair
bit for the floating point calculations in these games, but makes up
for it with excellent integer performance, bringing it's overall
performance close to being on-par with that of a PII or Celeron in
many non-3DNow! games, and close still in games which do support
3DNow!.
Anthony Hill
[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED] (Martin)
Crossposted-To: comp.windows.x.kde
Subject: Re: Help! My awe32 suddenly stopped working
Date: Fri, 09 Jul 1999 18:29:14 GMT
Wed, 07 Jul 1999 18:45:23 GMT / =?iso-8859-1?Q?Torbj=F8rn?= Grindhaug
<[EMAIL PROTECTED]>
>I was listening to an mp3-file in xmms, then I closed this app and
>started 'krabber' to rip an audio-track.
>When I shut down krabber, and I started xmms again I was told to check
>if my soundcard was set-up correctly. I didn't even restart KDE after
>this.
>A reboot didn't seem to help either. (Btw. the sound card currently
>works fine in win95.)
>When starting/exiting KDE again I noticed it couldn't load 'maudio' or
>whatever it's called. I also tried using 'window maker', but the sound
>was not working there either.
> After using Linux for more than one year I haven't had a single
>problem with my soundcard.
> Here are (some of) my configuration settings:
>
>RedHat 5.1, Kernel 2.2.10 (Also tried with 2.0.36)
>KDE 1.1.1
>SBawe32 (not PnP) at irq 7, dma 1,5
>SB16 driver compiled into kernel.
The default IRQ for most soundcards is 5, maybe something is trying to
put it there ?
------------------------------
From: Michael McConnell <[EMAIL PROTECTED]>
Crossposted-To: alt.comp.linux.xxx,alt.linux,alt.os.linux,comp.os.linux.setup
Subject: Re: Mounting probelm extended VFAT 32
Date: Fri, 9 Jul 1999 13:02:13 +0100
On Thu, 8 Jul 1999, Henk Toorman wrote:
> I can mount my C-drive which is a 16bits FAT primary partition.
> But I have problem to mount the drives D and E which are part of the
> extended partition which is 32 bit FAT..
> No command seems to work.
>
> Who can help me to solve this problem?
Which kernel version are you using?
-- Michael "Soruk" McConnell [Red Hat 6.0 Available!]
Eridani Star System -- The Most Up-to-Date Red Hat Linux CDROMs Available
Email: [EMAIL PROTECTED] http://www.amush.cx/linux/ Fax: +44-8701-600807
------------------------------
From: killspam@wkulecz$pam$uck$.bigfoot.com (Walter B Kulecz, PhD)
Crossposted-To: comp.os.linux.misc
Subject: Re: Making MPEG movies from AVI?
Date: Fri, 09 Jul 1999 18:43:24 GMT
In article <7m2pnp$[EMAIL PROTECTED]>, "Jeff Volckaert"
<[EMAIL PROTECTED]> wrote:
>
>I'm not married to MPEG, just small video clips. We'll see if the Linux
>version works with the Hauppauge card. It recommends the 2.2 kernel so it
>must be pretty new. I wonder how compression compares between MPEG and
>Realvideo?
Enormous quality tradeoff, factor of 7-8 in bandwidth.
352x240 MPEG-1 runs about 10 MByte/min. RealG2 "LAN" runs about 1.5
MByte/min for 230x240 size.
A high quality video source captured at "high" resolution 704x480 and
resized to 352x240 in a good encoder like Xing will give about the
same quality as VHS tape when played back full screen.
RealG2 is so blocky and blurry that its degradation is noticible in
even small 320x240 window and generally useless for anything other
than "talking heads".
--wally.
------------------------------
From: [EMAIL PROTECTED] (Hobbyistİ)
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.linux.setup
Subject: Re: Windows easy to install? BULLSHIT!
Date: Fri, 9 Jul 1999 13:41:05 -0500
Reply-To: [EMAIL PROTECTED]
On Fri, 9 Jul 1999 10:53:45 -0700, [EMAIL PROTECTED] scrawled these
sagacious words ...
: On Fri, 09 Jul 1999 11:08:35 -0400, Brian Hartman <[EMAIL PROTECTED]> wrote:
: >I can't and won't defend M$ product support. However, the sources of support are
:far more numerous for M$ than for Linux. Besides the
: >Usenet newsgroups, there are countless articles in mainstream computer magazines
:catered to solving problems in both Windows and its apps.
: >My experience of Linux is that you're limited essentially to Usenet and chat
:groups. (Again, there are notable exceptions to this rule, but
: >not many). My argument continues to be that if you're not paying for the product,
:the programmer has no reason to support it, other than the
: >kindness of his/her own heart.
:
: What is a magazine but an overpriced Web Page commited to paper
: with far too many banner ads? Most of the popular computer press
: is a joke, especially Ziff Davis.
:
: [detetia]
It's really amazing how these linux shortcomings are rationalised in
these ridiculous ways.
Hey linux is hard to install and configure => Well you shouldn't be doing
that by yourself. A trained professional should be doing it so your
difficulties are irrelevant.
Hey, Linux lacks some spiffiness in it's appearance. It looks so dusky
and dry.... => You microsoft weenies seem to love the glitzy appearance
that gives you that warm and fuzzy feeling .... blah, blah, blah. {I
don't know about you, but I like when my car looks good and also
functions well. The fact that it functions well, doesn't mean that it's
looks are unimportant. A car dealer with that viewpoint will soon be out
of business. The same goes for a proposed desktop OS.}
Now the latest rationalisation is on linux helpers not being in
magazines. => Magazines are no more than an overpriced .... garbage,
garbage....
So magazines are a useless medium? We can dispense of them now, right?
I could go on but ....
<final hand wave> Plunk.
--
A Hobbyist.
I use what works best for me and not what works best
for others.
>>>Down with the irrational OS zealot!!!<<<
------------------------------
From: toby <[EMAIL PROTECTED]>
Crossposted-To: alt.comp.linux.xxx,alt.linux,alt.os.linux,comp.os.linux.setup
Subject: Re: Mounting probelm extended VFAT 32
Date: Thu, 08 Jul 1999 16:34:36 -0400
Reply-To: [EMAIL PROTECTED]
Do you have support for vfat built into the kernel or the module
installed?
Henk Toorman wrote:
> I can mount my C-drive which is a 16bits FAT primary partition.
> But I have problem to mount the drives D and E which are part of the
> extended partition which is 32 bit FAT..
> No command seems to work.
>
> Who can help me to solve this problem?
>
> e-mail: [EMAIL PROTECTED]
------------------------------
From: Rahul Karnik <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.x
Subject: Re: Video Card Recommendation?
Date: Fri, 09 Jul 1999 14:51:36 -0400
Personally I recommend a Matrox G200 PCI (under $100 these days, if you look
hard 'nuff) -- 2D quality is the best around. It is supported by the latest
version of XFree86 and runs very well.
I myself have a G200 AGP and a Viewsonic PS790 and run in 1280x1024, 24 bit
color.
Rahul
------------------------------
From: Rahul Karnik <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup
Subject: Re: SoundBlaster Live!
Date: Fri, 09 Jul 1999 13:50:04 -0400
Ok, with the drivers from Creative, I am able to play both audio CDs with xplaycd
and MP3 files with kmp3. In Windows MP3 files are played thru' the wave interface
and if they are in Linux too, then there is no reason why your SB LIve! should not
play wav files. Of course, there may be some other reason, not connected to the SB
Live! like mixer settings...
Richard Wilkinson wrote:
> > Got the CD player working after forcing the package but i'll be damned > if I
> > can make any wave audio devices function.
> > If anyone has any ideas...
>
> On the OSS page it says that their driver only has CD playing
> capabilities and that the other sound stuff isn't supported yet.
> Maybe Creative have only got this far as well (unless of course it's
> the same driver).
>
> Cheers,
>
> Rich
Rahul
------------------------------
From: Nektarios Lathiotakis <[EMAIL PROTECTED]>
Subject: Re: SCSI CD-RW: yamaha 4416s: HowTo REWrite CD-RW's?
Date: Fri, 9 Jul 1999 18:34:57 GMT
Carsten Cimander <[EMAIL PROTECTED]> wrote:
: Hi,
: I have a SCSI-CD-RW drive yamaha 4416s in my linux box
: (4x write, 4x re-write, 16x read)
: along with a SCSI CD-ROM TEAC CD-532S (/dev/scd0)
: 1) How to get it work with CD-RW's?
: 2) How to erase CD-RW's?
: 3) How to (re-)write?
: I have cdrecord and xcdroast installed.
: Many questions. I know
: Any help is welcome.
: Carsten
Hi,
If it works with Recordables it ll work with rewritables too.
I supose it does. I use :
cdrecord dev=0,3,0 blank=fast speed=4
to erase a disk and Rewrite it with xcdroast in iso9660. Unfortunately erase
is not supported by xcdroast. See man page of cdrecord for details for
dev= numbers.
There is also an other filesystem (udf) designed for rewritables and
DVDs but although the filsesystem support is in good stage, the device
can be mounted read only and cant be used with rewritables (yet).
see http://trylinux.com/projects/udf/ for details about udf filesystem.
It is the filesystem that adaptec's directCD program uses in windows for
Rewritables.
Anyway, you can (re)write CD-RW in iso9660 format as if they were
recordables if you erase them each time.
good luck
Nektarios
------------------------------
From: Pankaj Pant <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup
Subject: Win95 -> Linux via null-modem .... help!
Date: 09 Jul 1999 15:02:53 -0400
Hi,
I'm trying to hook up my 486 (win95) to a K6-II running Linux via a null-
modem. I've followed the steps in an article in the Linux Gazette issue 41.
http://www.linuxgazette.com/issue41/smyth.html
I won't put in all the detail here ... but here's my problem. I can connect
through Hyperterm to the Linux m/c ... so the link is working.
However, the only reason for doing all this is that I want to be able
to run netscape and other x-apps on the 486. So as mentioned in the
above article I created a user called ppp95 and set it's shell to
point to pppd. Finally I put a .ppprc in /home/ppp95 with the
following lines:
connect '/usr/sbin/chat -v ATH OK AT OK ATE0V1 OK ATX3 OK ATDT CONNECT'
-detach
modem
crtscts
lock
:192.168.100.4
After this I start Dial-up Networking on the 486 and log in as ppp95.
However after saying "dialing" for a few seconds it quits saying that
the modem is not responding. I'm using "Standard 28800 Modem" to connect.
On the linux side, I can see (in /var/log/messages) that it receives
ATH, AT and ATE0V1 and then dies waiting for ATX3 ....
So, my question is: what should I try next? I'm completely lost. I've tried
reading all the HOWTO's, but I either can't find anything relevant or don't
understand the jargon (I'm pretty new to all this).
I'm hoping that some kind soul can bail me out of this one.
Thanks in advance. - Pankaj.
PS: If you need more details regarding the setup, please see the link
above. I've tried to follow it as closely as possible.
------------------------------
From: Johan Kullstam <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup,redhat.general
Subject: Re: CPU Question
Date: 09 Jul 1999 13:15:13 -0400
[EMAIL PROTECTED] (The 2-Belo) writes:
> Joceli Mayer wrote:
>
> >[EMAIL PROTECTED] wrote:
> >
> >> I am in the market for a new computer and my first priority is to get Linux
> >> running on it, but I need to know some things about a CPU. I have recently
> >> been looking at an Intel Celeron processor, and I was wondering if Linux
> >> is compatible with this processor, and some of the pitfalls, if any, I may
> >> come across if I go with the Celeron.
> >>
> >> ------------------ Posted via SearchLinux ------------------
> >> http://www.searchlinux.com
> >
> >It works. Also check http://www.amd.com for an AMD K6-3 or K6-2, cheaper and
> >faster than celeron. Also there is the new K7 chip, they rule all intel chips
> >in a performance-price tradeoff, so far ...
>
> A Celeron is essentially a Pentium without the on-board cache memory. The
> structure of the chip itself is no different.
even the old celeron without cache uses the pentium-ii/pentium-iii core.
the celerons today are pentium-ii with a smallish 128KB l2 cache.
however this cache runs at clock speed as opposed to p-ii which run
l2 cache at half-speed. it's basically a toss-up as to which is
faster (for same clock rate) the celeron or pentium-ii.
all indications are that the amd-k7 will be a performance winner.
however, it may be a few months yet before you can get a k7 and mobo.
--
johan kullstam
------------------------------
Date: Fri, 09 Jul 1999 18:50:01 +0200
From: Marc Mutz <[EMAIL PROTECTED]>
Crossposted-To: comp.sys.ibm.pc.hardware.chips,comp.sys.intel
Subject: Re: Celeron, what's the catch?
Charismo wrote:
>
> What is this? The might professor unless I miss my guess coming down from
> above to warn us about our quoting style? I get enough of that at my
> University...
> Anton Smith
>
> Undergraduate BSc Compsci and Electronics
> Massey University
>
Well, thanks for the professor (btw: did you see any title in my sig?)
Recommended reading for those of you who do not understand what I am
talking about:
in newsgroup news.announce.newusers
read article "A Primer on How to Work With the Usenet Community"
Marc
--
Marc Mutz <[EMAIL PROTECTED]> http://marc.mutz.com/
University of Bielefeld, Dep. of Mathematics / Dep. of Physics
PGP-keyID's: 0xd46ce9ab (RSA), 0x7ae55b9e (DSS/DH)
------------------------------
Date: Fri, 09 Jul 1999 20:08:05 +0200
From: Marc Mutz <[EMAIL PROTECTED]>
Crossposted-To: comp.sys.ibm.pc.hardware.chips,comp.sys.intel
Subject: Re: Celeron, what's the catch?
Michael wrote:
>
<snipped what Mike agreed with>
> Agreed.
>
> So you are saying that a 128K second level cache holds windows
> operating system code, outlook express, Lotus 123, Netscape,
> numerous background tray programs and never goes to main memory when
> you task switch....? <interrupted>
>
I think you underestimate the speed of modern processors. Just because
you have to wait for your new Office package doesn't mean the processor
is *that* slow.
Let me explain:
Typical L1 bandwidth is up to and exceeding 1GB/sec (!). Calutating how
often the L1 cahe can be refilled within a single second s left as an
exercise to the reader.
Typical L2 bandwidth ... let me have a look ... have no data by hand.
What happens when the context is switched (which can happen up to 100
times/sec in Linux)? The cache holds the data of the previously running
application. If the next time slice goes to the same application, then
most of the cache content is still there, so the application runs mostly
as if nothing happened.
If, however, the context switches, i.e. another task gets the next time
slice, the cache lines are almost suddenly replaced by different ones
(again: calculating how long that takes is left as an exercise), fitting
the needs of the new CPU owner. Now the task has the CPU for itself for
1/100 sec (except when an interrupt occurs). Almost all applications are
running almost all of the time in loops that fit entirely in the L1
(well, 32K is _very_very_ much code for a loop, typical loops are a
hundred byte, perhaps a kB or two). So L2 accesses are relatively
seldom. The L2 cache mostly serves as a kind of prefetch buffer for the
L1, enabling fast filling of L1 cache lines.
To summarize: After a _very_ short time the L2 cache does _not_ contain
a significant amount of code/data belonging to an application other than
the one that is currently owning the CPU.
This implies that a context switch (and you will agree that raising
windows is one) leads almost always to a complete reload of the caches
(L1,2+3).
Agreed?
OK.
> I admit the discussion is a bit blurred, I am
> not maintaining that a larger 2nd level cache gets your better
> application performance (beyond the limits you outline below). What I
> am saying is that the larger caches hold more code so that task
>Michael wrote:
>
> Marc Mutz <[EMAIL PROTECTED]> wrote:
>
> >Michael wrote:
> >>
> >> [EMAIL PROTECTED] (chrisv) wrote:
> >>
> >> >Nonsense. The extra cache of the k6-3 does you no good whatsover for
> >> >having "numerous multiple windows open." Do you really think that
> >> >the extra peasly 128k helps with this? Not.
> >> >
> >> You really don't know what your talking about. How fast is your
> >> celeron without that peasly 128 cache that Intel grafed onto the
> >> cacheless one? Is your Celeron a hotrod without that cache?
> >>
> >Sorry to interrupt your little dispute, but I want to clarify on the
> >subject of cache size from the point of view of someone who spends much
> >of his time in the linux-kernel mailing list et al. Also, you might take
> >a look at my signature.
> >
> >Mike, you cannot compare the performance ratio of pre-A celeron (with no
> >L2 cache) to post-A celeron (with L2 cache) with the speed ratio of two
> >cache-equipped chips. Here's why:
> >On a chip w/o L2 cache and only a few kB of L1 cache, almost every data
> >fetch will result in DRAM access, hence will be slow if the dataset
> >needed will exceed the size of the L1. That is something everyone agrees
> >with, I guess.
>
> Agreed.
> >Now it shows when you do measurements of *real world* applications
> >(BAPCo comes to mind), the size of the L2 does almost nothing. The
> >difference is too small to be readily distunguished from the statistial
> >blur of measurements.
> >Read my lips: It does _not_ matter if you have L2's of 256K@66MHz,
> >512K@100MHz, 1M@100MHz, 2M@133MHz. The performance difference will be in
> >all these cases within 5% and don't tell me 5% will make any difference
> >that you can see
>
> So you are saying that a 128K second level cache holds windows
> operating system code, outlook express, Lotus 123, Netscape,
> numerous background tray programs and never goes to main memory when
> you task switch....? I admit the discussion is a bit blurred, I am
> not maintaining that a larger 2nd level cache gets your better
> application performance (beyond the limits you outline below). What I
> am saying is that the larger caches hold more code so that task
> switching is more readily "instant". <interrupted>
see above
> I don't think any standard of
> measurement is available to measure how a business user actually uses
> windows. I am not certain that beyond anecdotal experience you can
> measure what I am talking about.
Between your actions with mouse and keyboard lie _ages_ from a CPU's
point of view. So it does not matter how you actually handle your
applications: Almost all the time they are waiting for you to say them
what you want :-)
> >The same applies to 128K@400MHz or 512K@200MHz. The only difference you
> >will find is when measuring L2 bandwidth.
> >_What_ will make a difference, though, is whether you are running with
> >512K@100MHz (like the AMD K6-2) or 256K@400MHz with the added benefit of
> >having decoupled L2 and DRAM/L3 paths (like the K6-3).
> >
> This will make a difference in the actual speed of the specific
> appication in the foreground, or possibly its speed while running in
> the background. On this we agree.
There is no such thing as 'foreground' and 'background' on the level of
the processor. All tasks are more or less equal in the face of the CPU
(except wrt their priority).
<snipped passage Mike agreed with>
> >You may _tell_ what you want, I _know_ better. Prove what you say :-0
>
> Not certain I follow here, and if you would please explain. (may be
> related to my comment immediately above).
Sorry, English is not my native tounge...
I meant that you can tell me what you want about cache speed's impact on
raising windows. I can give you counter arguments proving (or at least
convincing most of the people) that cache has nothing to do with that.
It's almost surely the chipset. And I asked you to prove what you stated
about cache influence on BitBlt performance.
> So based on your qualifications how much code can fit into these
> caches when multitasking and task context switching? Does everything
> fit into the first 128K second level cache. <interrupted>
see above.
> By the way, the speed of
> the celeron cache is the same as the processor speed and can be as
> fast as the K6 III's cache or faster. This alone doesn't account for
> the quickness of the k6 III over the celeron. Any elucidation you can
> give us is appreciated.
The quickness of the AMD over the celeron stems from its L1 cache being
twice as big.
Marc
--
Marc Mutz <[EMAIL PROTECTED]> http://marc.mutz.com/
University of Bielefeld, Dep. of Mathematics / Dep. of Physics
PGP-keyID's: 0xd46ce9ab (RSA), 0x7ae55b9e (DSS/DH)
------------------------------
From: [EMAIL PROTECTED] (Alan Jaffray)
Subject: CDROM changer recommendation?
Date: Fri, 9 Jul 1999 18:58:45 GMT
I'm ripping several hundred audio CDs, and doing this by changing
discs every hour for a few weeks is getting tiresome. So I'd like
to get a cheap CDROM changer, preferably IDE, that works well under
Linux. (Then write a quick script to run cdparanoia, change discs,
run cdparanoia, change discs, repeat N times for N-disc changer.)
Any recommendations on brands and models to look for, or to avoid,
or where to buy them?
Thanks,
Alan
--
Alan Jaffray <[EMAIL PROTECTED]>
there's nothing you can do that can't be done
------------------------------
From: "Michael Swanson" <[EMAIL PROTECTED]>
Subject: Cirrus Logic CL5446 rev 69
Date: Fri, 9 Jul 1999 11:46:48 -0700
Anyone know where I can find the drivers for this?
please reply to email
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.hardware) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Hardware Digest
******************************
