Re: Program Computes incorrectly under VPC.
user_at_domain.invalid
Date: 02/12/05
- Next message: Ian Robinson: "Re: Really bad performace on a G5"
- Previous message: Steve Jain: "Re: VPC getting too big is Norton Ghost the answer?"
- In reply to: John McGhie [MVP - Word and Word Macintosh]: "Re: Program Computes incorrectly under VPC."
- Next in thread: Michael Paine: "Re: Program Computes incorrectly under VPC."
- Reply: Michael Paine: "Re: Program Computes incorrectly under VPC."
- Reply: Steve Jain: "Re: Program Computes incorrectly under VPC."
- Reply: John McGhie [MVP - Word and Word Macintosh]: "Re: Program Computes incorrectly under VPC."
- Messages sorted by: [ date ] [ thread ]
Date: Sat, 12 Feb 2005 17:52:47 -0500
John McGhie [MVP - Word and Word Macintosh] wrote:
> Hi Rob:
>
> I can tell you that the people who made VPC have already read your messages:
> they are lurking in this newsgroup all the time.
>
> I can also tell you that rounding or truncation errors in computing rarely
> produce "small" errors. Often, a number will go negative or otherwise be
> absurd and you will spot the error very quickly.
>
> However, always remember the saying that "anyone can make a mistake, but to
> really stuff things up requires a computer."
>
> I should also mention, in the interests of setting realistic expectations,
> that this problem is unlikely to be fixed anytime soon. Of course the
> impact of this is having a large and possibly dangerous impact on your work:
> I do understand that.
>
> But I should also point out that Microsoft is going to be busy for quite a
> while just fixing problems that are having a much wider impact on a greater
> number of users. They do warn us somewhere in the fine print to expect
> problems of this nature.
Microsoft has taken on a huge burden of liability by knowing this
problem exists but not fixing it or putting a huge warning on their
software that it should not be used by the scientific community (hmm,
memories of the first pentium come to mind). If I design a building that
fails due to incorrect output caused by a known problem with VPC and
heaven forbid someone dies, not only will I get sued , but so will
Microsoft with their deep pockets.
Microsoft needs to make a public statement concerning this immediately
before disaters happen. Additionally, they should refund my money for
this program (and maybe throw in one of those OS's that work with the
remote desktop :>)
They can contact me by email if they would like to investigate this further.
xlengr@att.net
>
> I suppose that the only thing we could suggest is that you spend $500.00 on
> a cheap PC. Stick it in the corner of the room somewhere, and use it from
> your Mac with Remote Desktop Connection (the Mac client for OS X is a free
> download from Microsoft) -- you don't even have to buy a keyboard or monitor
> for the PC, it's only there to run Windows for you.
>
> I have that set up here so I can use heavyweight PC graphics and
> text-handling programs at full speed on my Mac. Post back if you need
> details of how to set this up -- it will take you all of half an hour to get
> it working :-)
Sounds like a plan.
Thanks for your help.
>
> Cheers
>
> On 12/2/05 7:06 PM, in article 420DB904.5040601@domain.invalid,
> "user@domain.invalid" <user@domain.invalid> wrote:
>
>
>>Steve Jain wrote:
>>
>>
>>>On Fri, 11 Feb 2005 14:45:04 -0500, user@domain.invalid wrote:
>>>
>>>
>>>
>>>>I am running an engineering program under VPC6.1 called AISIWIN
>>>>(www.clarksteel.com - free download) The program give invalids numerical
>>>>results under certain input condtions. I had a friend download the same
>>>>program and run it on a real PC and it gives correct results. This is a
>>>>serious problem as I use this program to design parts of structures.
>>>>Does anyone know why this could happen?
>>>>
>>>>I have a G4DP w/ 768 ram running 10.3.6
>>>>
>>>>Thanks,
>>>>
>>>>Rob Munach, PE
>>>
>>>
>>>The floating point processor on the PowerPC isn't as precise as that
>>>on an Intel FPU, 63 to 80, I think. This will can cause rounding
>>>errors in some applications.
>>>
>>>
>>
>>Steve,
>>
>>
>>These results are waaay off. I should be see output of about 2500 lbs
>>instead of 5 lbs or 6 lbs like I am seeing. I have compared the same
>>program on a real PC and am getting significantly different results. I
>>can't imagine the small differences in the FPU are causing this. There
>>must be a an underlying problem that MS should know about and address.
>>Can you get me in touch with someone at MS who can investigate this? If
>>you would like to run the program side by side on a Mac and PC, I can
>>let you know how to download it and what cases I am having problems
>>with. This is a VERY serious problem.
>>
>>Thanks for your help.
>>
>>Rob Munach, PE
>
>
- Next message: Ian Robinson: "Re: Really bad performace on a G5"
- Previous message: Steve Jain: "Re: VPC getting too big is Norton Ghost the answer?"
- In reply to: John McGhie [MVP - Word and Word Macintosh]: "Re: Program Computes incorrectly under VPC."
- Next in thread: Michael Paine: "Re: Program Computes incorrectly under VPC."
- Reply: Michael Paine: "Re: Program Computes incorrectly under VPC."
- Reply: Steve Jain: "Re: Program Computes incorrectly under VPC."
- Reply: John McGhie [MVP - Word and Word Macintosh]: "Re: Program Computes incorrectly under VPC."
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|