Questions About Grading of Assignment 0

Questions About Grading of Assignment 0

by Erik Fabrizzi -
Number of replies: 5

Dear PTFS team,

I have some questions about the grades that I recieved for Assignment 0,

I seem to have lost some points for things like "not printing pi" or "not providing the code for float" or "not writing how many cycles per iterations the code computes" wich are actually all there. I wanted to ask you who or how I should get in contact with you to discuss my results as I think some things may have been gone unseen by mistake.

Thanks a lot for your attention! 

In reply to Erik Fabrizzi

Re: Questions About Grading of Assignment 0

by Ayesha Afzal -
Dear Erik,

Since you mentioned these details in the file rather than the main report, they were overlooked. Although the grades have now been adjusted, kindly prepare an organized report for the future.

Best regards,
In reply to Ayesha Afzal

Re: Questions About Grading of Assignment 0

by Zurab Mujirishvili -

Could you kindly also look at the comments for our submission? Additionally, if there is a "proper" way to respond to the assignment feedback, that would also be nice to know.

Thanks in advance.
In reply to Ayesha Afzal

Re: Questions About Grading of Assignment 0

by Erik Fabrizzi -
Thanks a lot! There is still some things that I am not understanding tough,
This are the comments left on task 3 where Iost I 15 points:
  1.  missing pi
  2.  and runtime contribution of the code for single float
  3.  the float code for pi is also missing
To wich I have the following comments:

  1.  It is true that the value of pi for the float version is not mentioned in the report, mostly because the question only asked to discuss what happens in that case, wich I did. But I can understand If you would have liked us to write a value.
  2.  Maybe I am misunderstanding what you mean by "runtime contribution" but there is a plot in the report that displays runtime and cycles per iteration  for both the float version and the double version.
  3. I am also not sure what is meant by that, since I did submit the code for the float version, and the float function is defined right next to the double version in the source file bench.c. I did not copy paste the code into the report because again it is not mentioned that it has to be in the report.

Sorry for bothering you again and thanks for your support!

As a short addend:

I do tend to add build\usage intruction in a in the code folder, I do think that this is a better practice than putting it in the report but if you prefere that please let me know.

In reply to Erik Fabrizzi

Re: Questions About Grading of Assignment 0

by Krishna Sai Lakshmi Gayatri Manda -
Dear Erik,

The reason for loss of credits
1. You did not provide the Pi approximation.
To know whether you ended up at correct result.
Hence you lost 5 credits here.

2. Eventhough, you didnot mention the runtime or time of the benchmark runned.
By checking the plot, I awarded you 5 credits.

3. You provided the base code for double is report and not for float.
So I checked the code still no marks reduced for that.
It is just a suggestion to show the changes in report for float code.

4. The remaining 10 marks are reduced because you did not mention anything about the loss of accuracy due to float.
You discussed why the performance is high but the question is about accuracy.

5. You can add build or usage instructions. It is highly appreciated.
But report does not have any pi approximation number for both double and float.
You added plot for frequency vs time. But the question is simple and just for 2.4 GHz.
All the required values that are needed for grade are not written report rather they are drawn from file.
In reply to Krishna Sai Lakshmi Gayatri Manda

Re: Questions About Grading of Assignment 0

by Erik Fabrizzi -
Thanks for your answer! Just to be clear:
2. I can read the runtime from the plot, is this not sufficient?
3. The question asked to "submit" the code, in the future should we also copy the code to the report?
4. "The float version failed at computing an estimate of pi due to the high number of iterations,
which lead to an overflow of the datatype. I confirmed that a correct result is obtained for
slices 1000000." This is on page 2, I understand if it is not enough but why is this "nothing" about the lost of accuracy?

Again, sorry to bother but I want to get all cleared out to not repeat the same mistakes in the following assignments and I think it will help everybody to understand more how to structure the reports.
I wish you a nice evening!