Numpy, Matplotlib crash Python 3.8 Windows 7, 32-bit - can you help ?

[Thomas Passin]

On 3/22/2023 8:09 AM, a a wrote:
> On Saturday, 18 March 2023 at 20:12:22 UTC+1, Thomas Passin wrote:
>> On 3/17/2023 11:52 AM, a a wrote:
>>> On Friday, 17 March 2023 at 16:32:53 UTC+1, a a wrote:
>>>> On Friday, 17 March 2023 at 16:03:14 UTC+1, Thomas Passin wrote:
>>>>> On 3/16/2023 8:07 PM, a a wrote:
>>>>>> Crash report:
>>>>>>
>>>>>> Problem Caption:
>>>>>> Problem Event Name: APPCRASH
>>>>>> Application name: python.exe
>>>>>> Application version: 3.8.7150.1013
>>>>>> Application time signature: 5fe0df5a
>>>>>> Error module name: _multiarray_umath.cp38-win32.pyd
>>>>>> Version of the module with the error: 0.0.0.0
>>>>>> Time signature of the module with the error: 63dfe4cf
>>>>>> Exception code: c000001d
>>>>>> Exception offset: 000269c9
>>>>>> Operating system version: 6.1.7601.2.1.0.256.48
>>>>>> Regional Settings ID: 1045
>>>>>> Additional information 1: 0a9e
>>>>>> Additional information 2: 0a9e372d3b4ad19135b953a78882e789
>>>>>> Additional information 3: 0a9e
>>>>>> Additional information 4: 0a9e372d3b4ad19135b953a78882e789
>>>>> This exception has been reported to have many causes, but one
>>>>> possibility seems to be that your computer may not support an advanced
>>>>> instruction set that the .pyd was compiled for. I found this one
>>>>> specifically mentioned on the Internet: Advanced Vector Extensions. If
>>>>> that were the case, you would either need to find a different version of
>>>>> the module, or upgrade the computer/OS.
>>>>>
>>>>> It would be worth trying to downgrade the multiarray version to an
>>>>> earlier one and see if that fixes the problem.
>>>> Thank you Thomas
>>>> for your kind reply.
>>>>
>>>> I am fully aware to be living on an old machine, old OS, Windows 7, 32-bit system
>>>> but I have visited every social chat support forum on the Internet: from Python to Matplotlib, Numpy, Twitter, Github.
>>>>
>>>> As a newbie I am not aware how to downgrade "the multiarray version to an
>>>> earlier one
>>>>
>>>> I simply tried to test Python code from
>>>>
>>>>
>>>> https://www.section.io/engineering-education/reading-and-processing-android-sensor-data-using-python-with-csv-read/
>>>>
>>>> ====
>>>> # Python program to read .csv file
>>>>
>>>> import numpy as np
>>>> import matplotlib.pyplot as plt
>>>> import csv
>>>> ----
>>>>
>>>> "After importing the libraries, we now read the .csv file:
>>>>
>>>> with open('accl1.csv', 'r') as f:
>>>> data = list(csv.reader(f, delimiter=',')) #reading csv file
>>>>
>>>> ====
>>>> Just read about AVE from Wikipedia
>>>>
>>>> https://en.wikipedia.org/wiki/Advanced_Vector_Extensions
>>>
>>>
>>> downloaded and run
>>> HWiNFO
>>> and AVE not supported, not greened out
>> That's too bad; you may be out of luck. It's possible that someone has
>> compiled the .pyd library in such a way that it does not need the
>> instruction set extensions. I'm sorry but I don't know how to find out
>> except by trying internet searches - or by downgrading to earlier
>> versions of Numpy hoping to find one that works and also can be used by
>> the other libraries/programs that need to use it.
> 
> 
> Thank you Thomas for youre kind help.
> 
> You are the real Python PRO, you deserve Nobel Prize in Python.
:)

> I operated an old Dell computer with Windows XP preinstalled
> and upgraded XP to Windows 7 to get some web services to work.
> 
> Unfortunately I failed to find and install driver for video controller since none supported by Dell.
> 
> Visited many driver sites (Intel Driver Assistant included and more)
> without any success.
> 
> So life with an old PC is not easy

I reused my 10-year-old Sony VAIO laptop (it had Windows 8, IIRC) to be 
a Linux machine - I got a 1T external solid state drive, set up the BIOS 
to boot from it, and installed Linux Mint.  If you are willing to tackle 
Linux, this might be a good way to go.  I recommend Mint for newcomers 
to Linux.  The computer is much snappier and pleasant to use than it was 
under Windows.

I mostly use it as a backup computer.  I had to to without my main 
computer for a week or so, and the old machine made a fine substitute. 
I even copied all my Thunderbird emails over and used email all the week 
without losing any messages. Actually, the keyboard on that old computer 
is much better than I've got on my new one, although a few keys are 
getting a little flaky.

I was able to compile some version of Python on it, though I forget why 
I needed to do that.  With this setup, you could install a newer version 
of Python, and Numpy would work - it might get compiled during 
installation, but that's not a problem.  It happens automatically.

If fact, I know that it works because I have Numpy working on the 
computer.  Of course, my computer has the instruction set extensions and 
your does not, so who knows if can be compiled for you.  But it would 
probably be your best bet.

Anyway, if you decide to try it out, let us know.  And if you hit any 
problems, I might be able to help you.  I'm not a Linux expert but I've 
installed various distributions maybe 20 times or more as virtual 
machines, and twice using an external drive, including running Tomcat 
and MySQL as services.  Once you get it installed and working, and 
learned some of its quirks (not too bad, mostly about installing 
programs and configuring the desktop to be more to your liking), it's 
not much different from using Windows.  Well, batch files are really 
different...