🚀 go-pugleaf

siberia@nospam.demon.co.uk (Claire) wrote:
>       8Mb AGP ATI 3D RagePro Exp 98

I have an ATI @Work (@ home :-) which works fine and I can get
1600x1200 at 24-bit colour. (Not sure if this info. helps you
but the cards probably use the same X server binary).


> Finally, I'm having difficulty choosing an ISDN card.  I want
> something cheap, which works well under both Linux and NT4.
> Some ppl suggested that a BT Speedway is reliable and good value:
> any other suggestions?  (It's for use with Home Highway,
> connecting to Demon).

I've had no problems with the Speedway card.

Paul

On 10-Nov-99 00:16:39, Claire spoke thus:

<snip> video card considerations

>Also, I'm getting a CD-writer, and the most attractive option
>seems to be an IDE Mitsumi 4804TE (24X4X4 speed). 

Get a SCSI one. No, really. You can get IDE writers working via
a kernel recompile, but SCSI is best for an OS like linux.

>Finally, I'm having difficulty choosing an ISDN card.  I want
>something cheap, which works well under both Linux and NT4.

Least painful method ? Get an external one -you won't need to alter 
anything on linux (or need any drivers/kernel support) since it 
appears as a fast modem, and you don't use up a slot.

The disadvantages of this is an extra box on the desk (but with more 
flashy lights ;) an extra  mains plug used up, and longer ping times 
(only relevent if you're an online quake-head).

-- 
Mike

> >Also, I'm getting a CD-writer, and the most attractive option
> >seems to be an IDE Mitsumi 4804TE (24X4X4 speed).

> Get a SCSI one. No, really. You can get IDE writers working via
> a kernel recompile, but SCSI is best for an OS like linux.

You don't need to recompile any more.  I just installed Redhat 6.1, 
and I've discovered that I can drive my IDE CD writer through the 
IDE-SCSI emulation just with the appropriate boot-time option and 
module installation.  I can even now leave my standard CD drive as an 
IDE device, just selecting the writer to pretend to be SCSI.

(The reason for needing this is that cdrecord, which I use to write 
CDs, only supports SCSI interfaces.  However, since ATAPI is 
essentially SCSI commands over an IDE interface, a very lightweight 
ide-scsi emulator allows it to use ATAPI writers as well.)



Jon Moore

In uk.comp.os.linux Claire <siberia@nospam.demon.co.uk> wrote:
[snip]
:       4Mb AGP S3 Trio

DON'T. Getting drivers is a pig.

[SNIP]
-- 
-- 
               Politics: The conduct of public affairs for private advantage
                        Ambrose Bierce
Raphael Mankin
E-Mail: raph@panache.demon.co.uk
----------------------------------

Is USB supported by Linux yet?

-- 
Oz

In article <5o1f08.m04.ln@linux.a-taylor>, Andrew Taylor <news@hot-
shot.demon.co.uk> writes

>You can get a pci adaptec 2904 scsi card for like �25. Well worth it.

Can you boot from these?  I suspect not but it's worth asking.

I want a cheap bootable Linux supported narrow [1] SCSI card[2].

[1] Well, I won't complain if it's wide as well...

[2] I got the nice 18Gb UW disk and the 2Gb narrow one is still in there
but there is little point in having it there with 18Gb above it.  It is
faster (and obviously bigger) than the 1Gb IDE in the main Linux box.

TB       Busby the cat 1984 - 22.9.1999  :(

> I was planning on using SUSE 6.2.  Does anyone know if it has the
> IDE-SCSI emulator too?

It's part of the kernel.  If it's Linux, it's there.

>	  Is this a reliable tool?

IME yes - the only failures I've ever had were from one batch of cheapo
unbranded CDRs, and I'm often doing quite demanding work while burning a CD.
For CD writing, I think the critical elements are the quality of your
hardware and media.

-- 
Nick Kew

> Hmm, it looks a bit more expensive, but not much.  The only thing
> is that I am not sure about is what sort of SCSI card I need: I
> am confused by the difft variations of SCSI -- Ultra, Ultra-Wide,

Get a cheapo for CDs, tapes, etc.  You only need something with "ultra"
or "wide" in its name if you're using a SCSI HD.

> but it is SCSI-".  Does that mean I must have a SCSI-2 card, or 
> can I use an older one I have in a box?

A SCSI-2 card is about 30 pounds.  And I very much doubt you have a SCSI
card that is older than SCSI-2.

> In general, I prefer external modems, so I guess the same logic
> goes for ISDN.

If you say so ...

>	  But if I have channel-bonding and/or compresion
> working, won't I be in danger of getting overruns on the serial
> port, as well as hammering the system with interrupts?

Are you sure you need channel-bonding (which will of course double your
phone charges)?  As for compression, it's already 'discounted' in
the line speeds you see quoted for connectivity.

> The most elegant solution would be USB,
           ^^^^^^^
You mis-spelt "trendy and hyped".


-- 
Nick Kew

Hi Mike,

thanks for your help (and to the others who replied too)

On 10 Nov, in article
     <829.983T600T6492860@redrobe.demon.co.uk>
     mike@redrobe.demon.co.uk "Mike Redrobe" wrote:

MR> On 10-Nov-99 00:16:39, Claire spoke thus:
MR> 
MR> >Also, I'm getting a CD-writer, and the most attractive option
MR> >seems to be an IDE Mitsumi 4804TE (24X4X4 speed). 
MR> 
MR> Get a SCSI one. No, really. You can get IDE writers working via
MR> a kernel recompile, but SCSI is best for an OS like linux.

Hmm, it looks a bit more expensive, but not much.  The only thing
is that I am not sure about is what sort of SCSI card I need: I
am confused by the difft variations of SCSI -- Ultra, Ultra-Wide,
SCSI-2, etc.  I see a Yamaha SCSI CD-RW at a reasonable price, 
but it is SCSI-".  Does that mean I must have a SCSI-2 card, or 
can I use an older one I have in a box? New SCSI controllers seem
more expensive than the drive!

MR> >Finally, I'm having difficulty choosing an ISDN card.  I want
MR> >something cheap, which works well under both Linux and NT4.
MR> 
MR> Least painful method ? Get an external one -you won't need to alter 
MR> anything on linux (or need any drivers/kernel support) since it 
MR> appears as a fast modem, and you don't use up a slot.
MR> 
MR> The disadvantages of this is an extra box on the desk (but with more 
MR> flashy lights ;) an extra  mains plug used up, and longer ping times 
MR> (only relevent if you're an online quake-head).

In general, I prefer external modems, so I guess the same logic
goes for ISDN.  But if I have channel-bonding and/or compresion
working, won't I be in danger of getting overruns on the serial
port, as well as hammering the system with interrupts?

The most elegant solution would be USB, but Linux doesn't support
hat yet -- or a router, but they are a lot more expensive :(

Thanks again,
Claire

-- 
Anti-spam address in use. The "Reply-to:" address given will be valid
for a few weeks; thereafter, change everything left of the @ to "c-mac"
when replying.  NB - some mail systems do not properly handle the
given reply-to address: if your msg is bounced, please use "c-mac"

On Wednesday, in article
     <19991110.12522000@cgovia.lut.ac.uk> J.P.Moore@lboro.ac.uk
     "Jonathan Moore" wrote:

JM> > >Also, I'm getting a CD-writer, and the most attractive option
JM> > >seems to be an IDE Mitsumi 4804TE (24X4X4 speed).
JM> 
JM> > Get a SCSI one. No, really. You can get IDE writers working via
JM> > a kernel recompile, but SCSI is best for an OS like linux.
JM> 
JM> You don't need to recompile any more.  I just installed Redhat 6.1,=20
JM> and I've discovered that I can drive my IDE CD writer through the=20
JM> IDE-SCSI emulation just with the appropriate boot-time option and=20
JM> module installation.  I can even now leave my standard CD drive as an=20
JM> IDE device, just selecting the writer to pretend to be SCSI.

I was planning on using SUSE 6.2.  Does anyone know if it has the
IDE-SCSI emulator too?  Is this a reliable tool?

Claire

-- 
Anti-spam address in use. The "Reply-to:" address given will be valid
for a few weeks; thereafter, change everything left of the @ to "c-mac"
when replying.  NB - some mail systems do not properly handle the
given reply-to address: if your msg is bounced, please use "c-mac"

Claire wrote:

> I was planning on using SUSE 6.2.  Does anyone know if it has the
> IDE-SCSI emulator too?  Is this a reliable tool?

I know it has one but I have never had cause to use it.

-- 
_____________________________________________________________________
  ____                              mailto:jamie@rmcs.cranfield.ac.uk
   /        All comments expressed should be assumed to be my opinion
\_/ amie    only and you should get your own opinion.

Jamie <billy.gotes@usa.net> wrote in message
news:382AF69B.5B9FE1C7@usa.net...
> Claire wrote:
>
> > I was planning on using SUSE 6.2.  Does anyone know if it has the
> > IDE-SCSI emulator too?  Is this a reliable tool?
>
> I know it has one but I have never had cause to use it.
>
> --
> _____________________________________________________________________
>   ____                              mailto:jamie@rmcs.cranfield.ac.uk
>    /        All comments expressed should be assumed to be my opinion
> \_/ amie    only and you should get your own opinion.


You can get a pci adaptec 2904 scsi card for like �25. Well worth it.

Andy

On 10 Nov 1999, Raphael Mankin wrote:

> :       4Mb AGP S3 Trio
> 
> DON'T. Getting drivers is a pig.

?? Drivers? The AGP S3 Trio cards are supported by XFree86 already. Even
the S3 Trio 3D is now supported as of 3.3.5

Jason Clifford                                  http://www.definitelinux.com/
      Definite Linux - The UK's leading distribution with crypto support
               **** Linux Workstation and Server Systems ****

{R}@sunshine.tm wrote:
: In article <6HB5$DA3EyK4Ewv$@dbqa.demon.co.uk>, 
: hamster@nospam.demon.co.uk says...
: }I want a cheap bootable Linux supported narrow [1] SCSI card[2].
: }
: }[1] Well, I won't complain if it's wide as well...
: }
: }[2] I got the nice 18Gb UW disk and the 2Gb narrow one is still in there
: }but there is little point in having it there with 18Gb above it.  It is
: }faster (and obviously bigger) than the 1Gb IDE in the main Linux box.

: What you want is a 2940U2W in an overclocked Celery 300A and twin 9Gb U2W 
: drives, and 256K ram.

: My little 166MMX 80Mb Thinkpad takes 15 mins to compile the linux kernel
: Kane does the same in 27 seconds. I think its processor bound :((

Richard is winding you up - and trying to start a dick-size contest.
NEVER overclock, particularly a Celeron. It is the fast track to grief.

The big memory is more relevant, anyway.

{R}@sunshine.tm <{R}@sunshine.tm> doodled in uk.comp.os.linux:

> What you want is a 2940U2W in an overclocked Celery 300A and twin 9Gb U2W 
> drives, and 256K ram.

Good heavens. I knew it was expensive, but... ;)

> My little 166MMX 80Mb Thinkpad takes 15 mins to compile the linux kernel
> Kane does the same in 27 seconds. I think its processor bound :((

WTH is Kane? And did your make include 'clean'?!

~Tim
--
| Geek Code: GCS dpu s-:+ a-- C++++ UBLUAVHSC++++ P+++ L++ E--- W+++(--) N++ 
| w--- O- M-- V-- PS PGP++ t--- X+(-) b D+ G e++(*) h++(*) r--- y-	     
| The sun is melting over the hills,         | http://www.glutinous.custard.org
| All our roads are waiting / To be revealed | piglet@glutinous.custard.org

> What you want is a 2940U2W in an overclocked Celery 300A and twin 9Gb U2W
> drives, and 256K ram.

To be honest I don't think 1/4 of a MB of RAM is going to go very far. I
know it's expensive but this is being ridiculous.

Andy

In article <80jdh2$8um$1@hek.netix.org.uk> C. Newport
<crn@NOSPAM.netunix.com> shared the following with us in demon.tech.pc:
[snip]
>Richard is winding you up - and trying to start a dick-size contest.
>NEVER overclock, particularly a Celeron. It is the fast track to grief.
[snip]

Sorry, I do believe that you are talking out of your backside. This
machine has been running a Celeron 466 at 582 for 3 months. It's fast,
stable and more reliable than my previous K6 300 ABIT setup.

-- 
Marcus Durham
The UMTSDW Homepage. News, Reviews, Features and Locations. 
http://www.zenn.demon.co.uk/drwho/drwho.htm

Marcus Durham (Marcus@zenn.demon.co.uk) wrote:
: In article <80jdh2$8um$1@hek.netix.org.uk> C. Newport
: <crn@NOSPAM.netunix.com> shared the following with us in demon.tech.pc:
: [snip]
: >Richard is winding you up - and trying to start a dick-size contest.
: >NEVER overclock, particularly a Celeron. It is the fast track to grief.
: [snip]

: Sorry, I do believe that you are talking out of your backside. This
: machine has been running a Celeron 466 at 582 for 3 months. It's fast,
: stable and more reliable than my previous K6 300 ABIT setup.

So how come about 50% of the people who I try to help with obscure
crashes are overclocking celerons. The problem always goes away
when they put it back to standard.

The fact that you have got away with it so far does not indicate that this 
is a reliable practice. If a 466 could be reliably run at 582 then
Intel would label them as 550 and charge more money for them.

In article <80kieh$bu7$1@hek.netix.org.uk>,
C. Newport <crn@NOSPAM.netunix.com> wrote:
[snip]
>The fact that you have got away with it so far does not indicate that this 
>is a reliable practice. If a 466 could be reliably run at 582 then
>Intel would label them as 550 and charge more money for them.

Not necessarily - this isn't how the clock rate markings work.

There *are* sometimes differences between CPUs (and other components)
rated at different speeds, but often the components are made using
exactly the same methods and are simply rated after the event.  The ones
which run happily at 600 are rated 550, and the ones which will only
run happily at 500 are rated 450.  However, it can happen that the
demand for 450s outstrips the supply of chips which have failed the
600 test.  In this case chips which work perfectly at the higher speed
will be stamped with the lower one and sold at the lower price.

The price a CPU chip is sold at has nothing to do with the cost of
manufacturing that individual chip.  Chips can only be sold at the
higher price if there are buyers there wanting to buy them.  If all
the buyers want the lower specced chips then that's how they'll be
labelled and priced.  This is why Intel have been so keen of late
to build in mechanisms to ensure you only run chips at the speed
you paid for.

John
-- 
John Winters.  Wallingford, Oxon, England.

The Linux Emporium - the source for Linux CDs in the UK
See http://www.linuxemporium.co.uk/

In uk.comp.os.linux, john@polo.demon.co.uk (John Winters) writes:

> If all the buyers want the lower specced chips then that's how
> they'll be labelled and priced.  This is why Intel have been so keen
> of late to build in mechanisms to ensure you only run chips at the
> speed you paid for.

If all of the chips come from the same fabrication process with the
same die, how can the "chip" know at what speed it is allowed (has
been rated) to run? Or is this something put into the carrier (which
the silicon can read) rather than in the silicon itself?

In article <m2d7tdjyml.fsf@barnowl.demon.co.uk>,
Graham Murray  <graham@barnowl.demon.co.uk> wrote:
>In uk.comp.os.linux, john@polo.demon.co.uk (John Winters) writes:
>
>> If all the buyers want the lower specced chips then that's how
>> they'll be labelled and priced.  This is why Intel have been so keen
>> of late to build in mechanisms to ensure you only run chips at the
>> speed you paid for.
>
>If all of the chips come from the same fabrication process with the
>same die, how can the "chip" know at what speed it is allowed (has
>been rated) to run? Or is this something put into the carrier (which
>the silicon can read) rather than in the silicon itself?

I believe the latter, although I don't know the fine detail.

John
-- 
John Winters.  Wallingford, Oxon, England.

The Linux Emporium - the source for Linux CDs in the UK
See http://www.linuxemporium.co.uk/

John Winters wrote:
> 
> I believe the latter, although I don't know the fine detail.

The clock multiplier always used to be set by the motherboard 
[by applying, or not, a voltage to a number of processor pins] 
but naughty people used to play with the settings to overclock 
lower speed processors and Intel now set the multiplier in the 
carrier directly. 

The motherboard clock multiplier settings are now irrelevant 
to Intel CPUs.
-- 
                 "IR35: New Labour, Old Habits!"
I still remember the last Labour government; electricity strikes,
bread strikes, the "Closed Shop", Grunwick, rubbish piled high in 
the streets, dead bodies waiting months to be buried...

In article <80lufd$aru$1@polo.demon.co.uk>, John Winters
<john@polo.demon.co.uk> writes
>In article <m2d7tdjyml.fsf@barnowl.demon.co.uk>,
>Graham Murray  <graham@barnowl.demon.co.uk> wrote:

>>If all of the chips come from the same fabrication process with the
>>same die, how can the "chip" know at what speed it is allowed (has
>>been rated) to run? Or is this something put into the carrier (which
>>the silicon can read) rather than in the silicon itself?
>
>I believe the latter, although I don't know the fine detail.

It could be laser trimmed or fuses blown after testing and selection.
This is not hard and is done for a few pence commercially and in large
number for other devices.

-- 
Oz

In article <80m04m$cen$1@reef.news.technocom.net>, Philip Bragg
<philb@vossnet.co.uk> writes

>then I used my Black and Decker paint stripping
>heat gun to remove the EEPROM from an old 3com ethernet card

Be warned, now you have two blowlamp experts ....  :-)

-- 
Oz

In article <80kieh$bu7$1@hek.netix.org.uk> C. Newport
<crn@NOSPAM.netunix.com> shared the following with us in demon.tech.pc:
[snip]
>The fact that you have got away with it so far does not indicate that this 
>is a reliable practice. If a 466 could be reliably run at 582 then
>Intel would label them as 550 and charge more money for them.

Yeah right. Do you know anything about how these chips are made? If the
yield is good then it's quite possible you have a faster chip that you
thought. The marked clock speed isn't the be all and end all of it. This
chip is running a few degrees hotter and it doesn't crash. In fact, to
the 50%(!) of people you know who are having problems, I'd say it wasn't
the overclocked chip but an overclocked PCI bus due to cheap peripheral
devices. If you are going to up the bus speed from 66MHz then make sure
your motherboard and other peripherals are reliable at that speed.

BTW anyone ever see that rant on a newsgroup by an Intel employee saying
that overclocking was a criminal offence and people who did it were
evil? The only criminal thing is Intels pricing.

-- 
Marcus Durham
The UMTSDW Homepage. News, Reviews, Features and Locations. 
http://www.zenn.demon.co.uk/drwho/drwho.htm

On Wed, 10 Nov 99 00:16:39 GMT, siberia@nospam.demon.co.uk (Claire)
wrote:

>
>      4Mb AGP S3 Trio
>      8Mb AGP Intel I740
>      8Mb AGP ATI 3D RagePro Exp 98
>      8Mb AGP Voodoo III 1000

The ATI cards work well with Linux.

Get that one.

_________________________________________________________________
Digital Networks United Kingdom                      www.dnuk.com
Linux and Windows NT workstations and servers.      info@dnuk.com

In article <942332540snz@siberia.demon.co.uk>, Claire
<siberia@nospam.demon.co.uk> writes

>Hmm, it looks a bit more expensive, but not much.  The only thing
>is that I am not sure about is what sort of SCSI card I need: I
>am confused by the difft variations of SCSI -- Ultra, Ultra-Wide,
>SCSI-2, etc.  I see a Yamaha SCSI CD-RW at a reasonable price, 
>but it is SCSI-".  Does that mean I must have a SCSI-2 card, or 
>can I use an older one I have in a box? New SCSI controllers seem
>more expensive than the drive!

A cheap one can be had for 30 quid.  A real star purchase before Diamond
was bought out was the FirePort 40.  This was ultra wide and had a
booting BIOS.  I have one and it really gets the most out of this
UltraStar UW hard drive.  Sustained transfer can be as high as 20Mb/s.
Burst transfers will be in the region of 40Mb/s with minimal CPU usage.

SCSI cards have a lot more brain than IDE will ever have.  The SCSI card
manages the transfers to and from the devices taking that load off the
processor not to mention queued commands and the ability to attach 15
devices to one bus, with dual bus cards this is 30 devices on one IRQ.
Unless these dual bus cards use 2 IRQ.  Still better than 2 per channel
with no guarantee of compatibility.

TB       Busby the cat 1984 - 22.9.1999  :(

In article <MPG.1296dacfe40858659896b6@horizon.co.uk>, {R}@sunshine.tm
writes
>In article <6HB5$DA3EyK4Ewv$@dbqa.demon.co.uk>, 
>hamster@nospam.demon.co.uk says...
>}I want a cheap bootable Linux supported narrow [1] SCSI card[2].
>}
>}[1] Well, I won't complain if it's wide as well...
>}
>}[2] I got the nice 18Gb UW disk and the 2Gb narrow one is still in there
>}but there is little point in having it there with 18Gb above it.  It is
>}faster (and obviously bigger) than the 1Gb IDE in the main Linux box.
>
>What you want is a 2940U2W in an overclocked Celery 300A and twin 9Gb U2W 
                                                              ^^^^^^^^
Disk size war?  eh?

18Gb UW 7200RPM UltraStar, 2Gb Conner, 1.5Gb Digital.  All SCSI.

Total: 21.5Gb.

>drives, and 256K ram.
                ^ my 286 had more than that... :)
>My little 166MMX 80Mb Thinkpad takes 15 mins to compile the linux kernel
>Kane does the same in 27 seconds. I think its processor bound :((

Get a 366 from ukgamers and do it at 550MHz.  As soon as I get a job,
I'll ditch the old slow 417MHz I have now to get a 550.  417 is so last
week.

TB       Busby the cat 1984 - 22.9.1999  :(

In article <80kieh$bu7$1@hek.netix.org.uk>, C. Newport
<crn@NOSPAM.netunix.com> writes

>The fact that you have got away with it so far does not indicate that this 
>is a reliable practice. If a 466 could be reliably run at 582 then
>Intel would label them as 550 and charge more money for them.

What and kill their chances of selling the (shit for the money) P3?

TB       Busby the cat 1984 - 22.9.1999  :(

In <80kieh$bu7$1@hek.netix.org.uk> crn@NOSPAM.netunix.com (C. Newport)
wrote:

>So how come about 50% of the people who I try to help with obscure
>crashes are overclocking celerons. The problem always goes away
>when they put it back to standard.

Because they aren't cooling their systems properly. Overclocking is
fine provided you accomodate the extra heat generated and use proper
cooling.

Rich
-- 
Rich Wood, Systems Manager, Royal United Hospital, Bath, UK
Work: rich@ruh-bath.swest.nhs.uk
Play: rich@dynamite.org

I put it to you that in article <80jdh2$8um$1@hek.netix.org.uk>, C.
Newport <crn@NOSPAM.netunix.com> did almost intelligently write
>Richard is winding you up - and trying to start a dick-size contest.
>NEVER overclock, particularly a Celeron. It is the fast track to grief.

I don't think he is winding anyone up, although he can be prone to willy
waving at times. Your point about Celerons is way off the mark, what are
your sources? If you are such a scaredy cat to try overclocking just say
so but don't try and lecture everybody else. I know of dozens of happy
overclockers without any problems. The celeron is _the_ most
overclockable of processors and you can virtually guarantee that it will
go faster than advertised. What's the matter, do you resent people
getiing better performance than you but paying less cash for it? Got an
inadequate dick as well?
-- 
Regards, Petier
ICQ# 28188540
Lossiemouth RUFC Fixtures Sec. Check us out at http://surf.to/lossiemouth_RUFC
Please sign the guestbook so it looks like we've got some mates.

In article <382E894F.241DE57@ftp.porn.org>,
NoSpam  <NoSpam@ftp.porn.org> wrote:
>
>The clock multiplier always used to be set by the motherboard 
>[by applying, or not, a voltage to a number of processor pins] 
>but naughty people used to play with the settings to overclock 
>lower speed processors and Intel now set the multiplier in the 
>carrier directly. 
>
>The motherboard clock multiplier settings are now irrelevant 
>to Intel CPUs.

What would happen if you put a faster clock crystal in? I suppose your
modem or anything else that relies on signals generated by the motherboard
clock (video card or does that have its own?) would get a bit upset when
what it thinks is one frequency is in fact something completely different.

 - Richard.

-- 
   _/_/_/  _/_/_/  _/_/_/ Richard Corfield <Richard@Littondale.freeserve.co.uk>
  _/  _/    _/    _/      Web Page, CV:   http://www.littondale.freeserve.co.uk
 _/_/      _/    _/       Dance (Ballroom, RnR), Hiking, SJA, Linux, ... [ENfP]
_/  _/  _/_/    _/_/_/    PGP2.6 Key ID: 0x0FB084B1     PGP5 Key ID: 0xFA139DA7

In article <80kieh$bu7$1@hek.netix.org.uk>, at 20:41:53 on Sat, 13 Nov
1999, C. Newport <crn@NOSPAM.netunix.com> spewed forth the following:
>Marcus Durham (Marcus@zenn.demon.co.uk) wrote:
>: In article <80jdh2$8um$1@hek.netix.org.uk> C. Newport
>: <crn@NOSPAM.netunix.com> shared the following with us in demon.tech.pc:
>: [snip]
>: >Richard is winding you up - and trying to start a dick-size contest.
>: >NEVER overclock, particularly a Celeron. It is the fast track to grief.
>: [snip]
>
>: Sorry, I do believe that you are talking out of your backside. This
>: machine has been running a Celeron 466 at 582 for 3 months. It's fast,
>: stable and more reliable than my previous K6 300 ABIT setup.
>
>So how come about 50% of the people who I try to help with obscure
>crashes are overclocking celerons. The problem always goes away
>when they put it back to standard.
>
>The fact that you have got away with it so far does not indicate that this 
>is a reliable practice. If a 466 could be reliably run at 582 then
>Intel would label them as 550 and charge more money for them.
>
>
The problem is adequate cooling I should think... that's the only reason
it'd crash overclocked but not normally clocked =>  Buy a better
heatsink/fan =)
-- 
Michael Bungey

Michael Bungey <spam@fuelwise.clara.co.uk> writes:

> >The fact that you have got away with it so far does not indicate that this 
> >is a reliable practice. If a 466 could be reliably run at 582 then
> >Intel would label them as 550 and charge more money for them.
> >
> >
> The problem is adequate cooling I should think... that's the only reason
> it'd crash overclocked but not normally clocked =>  Buy a better
> heatsink/fan =)

I suspect that it is more than cooling, there are variations in
manufacturing chips. Should the metal layer be slightly under-etched
then the tracks would be slightly wider, these would have a higher
capacitance and would then only run reliably at lower speeds.

I am certain that INTEL et al will measure each chip to see what speed
they will reliably run at and then sell them at that speed/price to maximise
their profits.

Personally I would never over-clock any components as the risk of
loosing data would be much more costly than the benefit of that little
extra speed of the processor. The processor is only one component that
makes up the total speed of an application.

Robin

In article <x2k8nisnem.fsf@zdbaora.nat.bt.com>, Robin Smith
<smithrc@zdbaora.nat.bt.com> writes

>I am certain that INTEL et al will measure each chip to see what speed
>they will reliably run at 

correct

>and then sell them at that speed/price to maximise
>their profits.

Which means meeting demand. They will label a chip as a lower spec than
it *could* be run at if that's what they are short of. This is
particularly common as a plant gets into full swing and produces very
few of the lower spec devices because it yields large amounts of hi-
spec. This sort of thing has been going on in the semiconductor industry
for decades.

-- 
Oz

On Tue, 16 Nov 1999 11:51:26 +0000, Oz <Oz@upthorpe.demon.co.uk> spouted:
>Which means meeting demand. They will label a chip as a lower spec than
>it *could* be run at if that's what they are short of. 

But that's the risk you take, isn't it? No one is suggesting that
overclock is risk free - you simply move the trade-off point slightly
further along the line from 'Total reliability' towards 'Ultimate
speed'. You don't know whether you've got a chip that's faster than its
label and is now running at its proper rated speed, or one that is going
to turn into a small but expensive firework.

I would regard, say, a home system as a much more suitable candidate
for overclocking than a commercial server. Especially since, in the
commercial server case, it's someone else's money you're spending ;>

JM "Or you could just pop the whole box in the fridge and cut a small
grommeted hole in the door for the power and ethernet cables" Y

John Winters wrote:
> However, it can happen that the
> demand for 450s outstrips the supply of chips which have failed the
> 600 test.  In this case chips which work perfectly at the higher speed
> will be stamped with the lower one and sold at the lower price.

Absolutely correct.
Some time back, in the great and glorious days of the former
Amiga empire, many accelerator cards used 68EC030 chips, which
are 68030's which failed the MMU tests but passed the CPU tests.

Not that many failed the MMU tests, and the market for them far
outstripped supply, to the extent that after a while, the majority
of 68EC030's were only 'EC' because Motorola had stamped it on the
chip.  I quite happily ran an EC @ 25MHz as a non-EC for some years,

Tim Haynes <piglet@glutinous.custard.org> wrote:
>Robin Smith <smithrc@zdbaora.nat.bt.com> doodled in uk.comp.os.linux:
>>Bollocks, the phrase "Device Speed is always a function of temperature" is
>>wrong. "temperature is a function of device speed" is correct. 
>An inverse function is also a function, as long as ranges, domains and non-
>smooth bits are sorted out.

Eh, it depends. To a mathematician, perhaps; but I think in this context
'X is a function of Y' implies a relationship where Y causes X in some
way.
-- 
David/Kirsty Damerell.                       damerell@chiark.greenend.org.uk
http://www.chiark.greenend.org.uk/~damerell/ w.sp.lic.#pi<largestprime>.2106
|___| "Sic gorgiamus allos subjectatos nunc." Confessed Mercrediphile. |___|
| | |  Or, in Klingon:    "nucharghqangbogh chaH DISopchu' 'e' wItIv." | | |

Dave Smith <dsmith@contract15.inmos.co.uk> doodled in uk.comp.os.linux:

> Correct.  What generates this heat?  Resistance.  The capacitance of a
> device is not affected by temperature, but the resistance *is*.  Ironically,
> the hotter a device is, the higher the resistance, so the more heat it
> generates.

Hmmm. Given that it's quite a few years since I did any physics, let alone
electronics, why shouldn't the capacitance vary with temperature? (The idea of
surface area comes rapidly to mind here.)

~Tim
--
| Geek Code: GCS dpu s-:+ a-- C++++ UBLUAVHSC++++ P+++ L++ E--- W+++(--) N++ 
| w--- O- M-- V-- PS PGP++ t--- X+(-) b D+ G e++(*) h++(*) r--- y-	     
| The sun is melting over the hills,         | http://www.glutinous.custard.org
| All our roads are waiting / To be revealed | piglet@glutinous.custard.org

David Damerell <damerell@chiark.greenend.org.uk> doodled in uk.comp.os.linux:

> >>Bollocks, the phrase "Device Speed is always a function of temperature" is
> >>wrong. "temperature is a function of device speed" is correct. 
> >An inverse function is also a function, as long as ranges, domains and non-
> >smooth bits are sorted out.
> 
> Eh, it depends. To a mathematician, perhaps; but I think in this context
> 'X is a function of Y' implies a relationship where Y causes X in some
> way.

Yup, I'd admit to being relatively pedantic, but *how* does one determine
whether it's speed that affects temperature *or* the other way round, in a
given case? (Other than, say, it being you wiggling the thermostat, sort of
thing!)

~Tim
--
| Geek Code: GCS dpu s-:+ a-- C++++ UBLUAVHSC++++ P+++ L++ E--- W+++(--) N++ 
| w--- O- M-- V-- PS PGP++ t--- X+(-) b D+ G e++(*) h++(*) r--- y-	     
| The sun is melting over the hills,         | http://www.glutinous.custard.org
| All our roads are waiting / To be revealed | piglet@glutinous.custard.org

In article <MPG.129e09e421382f1e9896d5@horizon.co.uk>, {R}@sunshine.tm writes:
|> In article <3833dc8c.0@nnrp1.news.uk.psi.net>, 
|> andy+news@DELETE.pteron.demon.co.ME.uk says...
|> 
|> }I suggest you look at the address Dave writes from. ST is a 
|> }semiconductor manufacturer. I am a semiconductor designer and
|> }fully agree with Dave.
|> 
|> I suggest you consider and address the original statement that a Celeron 
|> 300A which was, as explained, an Intel marketing ploy.

This may have been the original statement, but, as often happens on this kind of 
thing, this is not the statement that I was referring to.

|> The Celeron 300A was marked as 300MHz but had a choice of bus speeds 
|> 66MHz and 100MHz, it will overclock to 450Mhz simply by setting the Bus 
|> speed to 100 and the internal multiplier to 1.5.

This may be true.  I never claimed it to be true or false.  The statement that
I responded to was the claim that the only reason for extra cooling in an
overclocked machine was to stop it overheating.

|> Then some know it all, without any research into this specific issue, 
|> comes along and says "overclocking is wrong" [or similar].

I never said that overclocking was wrong.  In fact, I said that you *can* 
overclock a device, provided that you keep it cold enough.  However, overclocking
can be risky, as it depends on the quality of the device that you get, as well as
the temperature.  The device that you are overclocking may be one that has
been derated due to commercial reasons (they are making too many of the faster 
devices), or it may be a lower-spec device.  Obviously, a device that has
been commercially derated is better, but you are taking a gamble about what
you get.

Personally, I don't overclock, as I don't need the extra CPU power (I run on a
DX2-66, by the way...).  Overclocking is a personal choice, and is a tradeoff 
between the advantages of more CPU power, and the risk of damaging your CPU and/
or data corruption.  Some people choose to overclock, just as some people choose
to drive fast.  Others don't overclock, like some people drive to speed limits.

|> Please go and find out the facts before making judgements.

Please look at what people wrote before slagging them off.  If these comments
were not aimed at me, then I apologise; I've been working too hard...

|> Do not "silicon designers" dislike overclocking for commercial reasons or 
|> do you just make false tits.

<g>.

I think you're confusing us with "silicone designers".  There are enough tits
around without us having to make false ones... :-)

Intel probably dislike overclocking, because it reduces their revenue.  (or maybe
they like it because people have to buy new chips when they overclock their old
ones too far...)

As far as I am concerned, I don't care.  The devices we make here (in Bristol)
are for embedded systems, and the clock speed is generally defined by the 
customer.  I'd like to see what happens if someone tries to overclock the 
CPU inside their Sky Digibox...

-- 
David Smith                   Tel: +44 (0)1454 462380 (direct)
STMicroelectronics            Fax: +44 (0)1454 617910
1000 Aztec West    TINA (ST only): (065) 2380
Almondsbury                  Home: 01454 616963
BRISTOL                    Mobile: 07932 642724
BS32 4SQ               Work Email: dsmith@bristol.nospam.st.com
                       Home Email: David.Smith@gazpacho.nospam.demon.co.uk