XLS for java. Image conversion is very slow.

[juneyounmoon]

It takes 5 minutes to convert a 600cell Excel file filled with only text "1" into an image.
The result is a 235KB png image.

It is an occupied resource.
It occupied a certain amount of resources from start to finish.
- cpu usage: 10%
- Memory Usage: 600MB
- Disk Write: 900 MB/s (very strange)


This is a test spec.
- Windows 10
- jvm mem: 1G
- processor: 12th Gen Intel(R) Core(TM) i9-12900, 2400Mhz, 16 cores, 24 logical processors
- Spire.XLS for java: 13.2.4

This is the code used.
Contents containing personal information have been replaced with "-".

Code: Select all

    public static void main(String[] args) {
        com.spire.license.LicenseProvider.setLicenseKey("-");

        Workbook workbook = new Workbook();

        workbook.loadFromFile("-");

        Worksheet worksheet = workbook.getWorksheets().get(0);

        worksheet.saveToImage("-");
    }


I have attached the xls file used for the test and the png file that is the conversion result, so please check it.

[Triste.Dai]

Hi,

Thanks for your feedback.
In your document, there are 162 rows and 119 columns, so the document contains 19278 cells in total. I did a test with your document and reproduced your issue. I have logged this issue into our bug tracking system with the ticket number SPIREXLS-4498, our developers will investigate and fix it. Sorry for the inconvenience caused. Once the issue is fixed, I will inform you asap.

Sincerely,
Triste
E-iceblue support team

[juneyounmoon]

Are you sure it's a bug?

[Triste.Dai]

Hi,

Thanks for your inquiry.
Actually, it is not a bug, but after testing, the convert speed is slow, I communicated with our developers, they will investigate this issue and try to improve the speed of conversion. Once our developers make some progress, I will inform you in time.

Sincerely,
Triste
E-iceblue support team

[juneyounmoon]

I wonder if the developers also agree that it's too slow.

Asking questions in detail to share within my team.
thank you.

[Triste.Dai]

Hi,

Thanks for your inquiry.
I have communicated with our developers, according to their feedback, they converted your document which contains near 20000 cells within 10 seconds with their computer, this issue may be caused by the especial environment. Please spare us a little more time to find the real cause. Thanks for your understanding.

Sincerely,
Triste
E-iceblue support team

[Triste.Dai]

Hi,

Thanks for your patience.
After investigating, our development team has determined that there are no issues with the code logic of our product.
However, our developers have adjusted the Jvm heap space size to 2G, which has resulted in a conversion time of around 30 seconds - faster than what you have experienced. We suggest that you adjust your heap space size accordingly to achieve optimal performance.
If there are any other questions, just feel free to contact us.

Best regards,
Triste
E-iceblue support team

[juneyounmoon]

Hi.

Your words sound a bit odd.

When the developer tested it, he said that he converted a document containing 20000 cells within 10 seconds, but when he tested it with the document I gave him, he said that the problem was reproduced.
The developers did the conversion under the same conditions, but the problem only occurred in the documentation I gave you.

But why the sudden talk of heap space?

Why is the documentation created by the developer different from the documentation created by me?

Why does it require more heap space?

Why is it taking 30 seconds, 3x slower than the 10 seconds the developer tested, even with more heap space?

My English is poor, so it may sound a little aggressive.
Thank you for your understanding.

[Triste.Dai]

Hi,

Thanks for your inquiry.
Apologize for that I did not explain it clearly to you. Your document contains approximately 20,000 cells, our developers tested it on the development computers, and the conversion time was around 10 seconds. They also tested it on their personal computers with a 2GB JVM heap space, the conversion time was around 30 seconds.
We adjusted the JVM heap space to ensure that the objects could be recreated without having to release it first when the heap space was insufficient. This adjustment can help improve the conversion performance. Moreover, our developers have reviewed the source code and found no logical issues.
We hope this explanation clears up any confusion or misunderstandings about the reason for the slower conversion speed. If you have any further questions or concerns, please do not hesitate to contact us.

Best regards,
Triste
E-iceblue support team