rec.autos.simulators

Here's a thought...

Mike Whit

Here's a thought...

by Mike Whit » Mon, 21 May 2001 15:41:04

Based on what I've been able to obtain from the messages, the problem
rotates around the fact that GPL only uses the lower 32 bits of the 64 bit
value retrieved from the RDTSC instruction.

For those not familiar with the RDTSC instruction it's a 64 bit counter that
increases with every clock cycle.  A 1 ghz processor would increase the
value 1,000,000,000 times per second. Likewise a 500 mhz processor would
increase the value 500,000,000 times per second.

A patch COULD be applied to right shift the 64 bit value 1 digit then only
use the lower 32 bits which should allow processor speeds up to 3 ghz.  By
shifting right 3 digits a processor speed of approximately 12 ghz should
work.

This still requires access to the source code to "squeeze" the shifting into
the proper RDTSC references.





> >> No, why would I waste money on a possibly flawed Intel chip?

> > IT turns out it's a *speed* bug, *not* a P4 bug. *All* chips that fast
> > will have problems.

> In that case it does interest me. I have yet to hear or read of anyone
> actually experiencing this problem w/ an AMD chip, do you have a link or
> any other reference to the statement of somebody with firsthand knowledge?

> It's not that I don't believe you, but I'd like to read what this person
> wrote himself. Hearsay is inadmissible in r.a.s. :-) (*Poof* where did
> half the posts go?)

> Cheers,
> Jens

ymenar

Here's a thought...

by ymenar » Tue, 22 May 2001 07:20:04


> Based on what I've been able to obtain from the messages, the problem
> rotates around the fact that GPL only uses the lower 32 bits of the 64 bit
> value retrieved from the RDTSC instruction.

> For those not familiar with the RDTSC instruction it's a 64 bit counter
that
> increases with every clock cycle.  A 1 ghz processor would increase the
> value 1,000,000,000 times per second. Likewise a 500 mhz processor would
> increase the value 500,000,000 times per second.

> A patch COULD be applied to right shift the 64 bit value 1 digit then only
> use the lower 32 bits which should allow processor speeds up to 3 ghz.  By
> shifting right 3 digits a processor speed of approximately 12 ghz should
> work.

> This still requires access to the source code to "squeeze" the shifting
into
> the proper RDTSC references.

Wow Mykey I didn't know you were that much of a nerd ;-D

You are right, it would totally make sense.

--
-- Fran?ois Mnard <ymenard>
-- May the Downforce be with you...
-- http://www.ymenard.com/
-- This announcement is brought to you by the Shimato Dominguez
Corporation - helping America into the New World...

Thom j

Here's a thought...

by Thom j » Tue, 22 May 2001 09:10:17

I know nothing about code, OS language etc.. but Mike if it is correct
then it *Will* be & *Easy* to patch GPL if you have the code...Yes?
Thanx for the info
Cheers Thom_j.


| > Based on what I've been able to obtain from the messages, the problem
| > rotates around the fact that GPL only uses the lower 32 bits of the 64
bit
| > value retrieved from the RDTSC instruction.
| >
| > For those not familiar with the RDTSC instruction it's a 64 bit counter
| that
| > increases with every clock cycle.  A 1 ghz processor would increase the
| > value 1,000,000,000 times per second. Likewise a 500 mhz processor would
| > increase the value 500,000,000 times per second.
| >
| > A patch COULD be applied to right shift the 64 bit value 1 digit then
only
| > use the lower 32 bits which should allow processor speeds up to 3 ghz.
By
| > shifting right 3 digits a processor speed of approximately 12 ghz should
| > work.
| >
| > This still requires access to the source code to "squeeze" the shifting
| into
| > the proper RDTSC references.

Mike Whit

Here's a thought...

by Mike Whit » Tue, 22 May 2001 10:32:20



> > Based on what I've been able to obtain from the messages, the problem
> > rotates around the fact that GPL only uses the lower 32 bits of the 64
bit
> > value retrieved from the RDTSC instruction.

> > For those not familiar with the RDTSC instruction it's a 64 bit counter
> that
> > increases with every clock cycle.  A 1 ghz processor would increase the
> > value 1,000,000,000 times per second. Likewise a 500 mhz processor would
> > increase the value 500,000,000 times per second.

> > A patch COULD be applied to right shift the 64 bit value 1 digit then
only
> > use the lower 32 bits which should allow processor speeds up to 3 ghz.
By
> > shifting right 3 digits a processor speed of approximately 12 ghz should
> > work.

> > This still requires access to the source code to "squeeze" the shifting
> into
> > the proper RDTSC references.

> Wow Mykey I didn't know you were that much of a nerd ;-D

> You are right, it would totally make sense.

Sonny boy, I was bit twiddling before you were thumb twiddling :)
Ruud van Ga

Here's a thought...

by Ruud van Ga » Tue, 22 May 2001 18:49:58

On Sun, 20 May 2001 06:41:04 GMT, "Mike White"


>Based on what I've been able to obtain from the messages, the problem
>rotates around the fact that GPL only uses the lower 32 bits of the 64 bit
>value retrieved from the RDTSC instruction.

>For those not familiar with the RDTSC instruction it's a 64 bit counter that
>increases with every clock cycle.  A 1 ghz processor would increase the
>value 1,000,000,000 times per second. Likewise a 500 mhz processor would
>increase the value 500,000,000 times per second.

>A patch COULD be applied to right shift the 64 bit value 1 digit then only
>use the lower 32 bits which should allow processor speeds up to 3 ghz.  By
>shifting right 3 digits a processor speed of approximately 12 ghz should
>work.

Sounds like a 15-minute upgrade to me. :)

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Free car sim  : http://www.marketgraph.nl/gallery/racer/

Uncle Feste

Here's a thought...

by Uncle Feste » Tue, 22 May 2001 21:39:32


> > > A patch COULD be applied to right shift the 64 bit value 1 digit then
> only
> > > use the lower 32 bits which should allow processor speeds up to 3 ghz.
> By
> > > shifting right 3 digits a processor speed of approximately 12 ghz should
> > > work.

> > > This still requires access to the source code to "squeeze" the shifting
> > into
> > > the proper RDTSC references.

One word.  Sourcer.  :-))

--

Fester

Ian

Here's a thought...

by Ian » Wed, 23 May 2001 01:13:23

Got a spare 15 minutes to do an upgrade ? ;)

--
Ian P
<email invalid due to spam>



J

Here's a thought...

by J » Thu, 24 May 2001 17:09:49

Nothing can happen to me, since I wisely put aside my trusty ol' P200 (No-MMX)
instead of selling off that rocket.

JensSchumi



--
LWFF Ball Bearing conversion at:
http://members.nbci.com/JensSchu/

GPLRank: -10.1
Monsters of GPL: sub 200


rec.autos.simulators is a usenet newsgroup formed in December, 1993. As this group was always unmoderated there may be some spam or off topic articles included. Some links do point back to racesimcentral.net as we could not validate the original address. Please report any pages that you believe warrant deletion from this archive (include the link in your email). RaceSimCentral.net is in no way responsible and does not endorse any of the content herein.