IA X2 Claim to further compress files 30%. Not seeing it

[spicymikey]

Marketing literature is often filled with hyperbole. We all know that. But this IA claim was pretty specific:

"64-bit Compression: Shrink apps and data by up to 90%, or enjoy an extra 30% size reduction if you were already on our compression platform(2). Beat Microsoft's own compression on runtimes such as .NET. "

From my testing, I see NO improvement in compression of our installer whatsoever compared to IA18. I did notice two additional compression levels to select in the slidebar, but even sliding over to the far right yielded no difference for me.

What's up with that? Was this complete marketing BS or do I need to do something else to get this to work. I'd certainly like a smaller installer if possible.

[FrancescoT]

Dear Mikey,

about marketing BS ...

did you try to generate a package using the Maxim compression level currently available in IA X2 (requires an x64 processor and significant memory)?

You could run a simple test;
- create a project from scratch and add to it the .Net 4.6 runtimes.
- build the process using the standard Best compression level and then rebuild the project using the latest Maxim compression level.

Finally compare the results.

Regards

[spicymikey]

Not sure really. I went to PROJECT SETTINGS, then BUILD COMPRESSION. There's a slide bar, as there was in 18. I noticed it was no longer at the far right but two notches over now. I slid it back over to "Best". The setting is described as:

"InstallAware will pre-process data from a continuous stream, using a 1gb buffer for LZMA compression with binary trees running on 2 threads. Requires a 64 bit Operating system, 10 GM of memory during compression, and 1GB during extraction."

If that is "Maxim", then yes, I tried that.

I have a PC running Wn 8.1 x64 with 12g of installed RAM.

Thanks

[spicymikey]

Hi Francesco

I never heard back from you after Thursday's post. So I was wondering if you could explain (for everyones benefit) what I might be doing wrong to cause me not to see this 30% improvement in compression. The advertising emails made it a major point in the reason to upgrade. I may be doing something wrong, but I followed the limited instructions and see zero improvement is package sizes.

If anyone else using IA X2 has any suggestions I'd greatly appreciate it.

Thanks in advance

Mike

[FrancescoT]

Dear Mike,

with further verifications, I can only say that results may be different based on the Runtime package used the test the compression.

I just repeated the test I did last week using Net 4.6 only and producing a Single compressed EXE.

- With the standard Best compression the generated package size is 231.068 kb.
- With the current maximum compression level, it's instead 65.669 kb.

Both compression processes required some time to complete and of course, much more time using the current maximum compression level (I missed to measure the effective time spent ...).

Regards

[spicymikey]

So are you saying this is only a benefit for those creating single compressed EXEs? Maybe that's the reason. I generate webdeploy packages.

If that's not it then we have a mystery. I understand there would be differences. But as I said I see ZERO difference while you see over a 200% difference. That is a big variance. Something is wrong unless its true this is only for single compress exe's.

Please advise how to get to the bottom of this. The statement you made; "I can only say that results may be different" does not explain this one.

Thanks
Mike

[FrancescoT]

Dear Mike,

No, I am not saying that is only a benefit for those creating single compressed EXEs ... more simply, I just tried with creating single compressed EXEs.

Let me try with building a Web Exe.

BTW, dis you try to compile a compressed exe (...just to compere our results)?

Regards

[spicymikey]

OK, well that's the only thing I thought could explain it. If it should significantly improve WEb Media Blocks also, then I am unable to explain it.

The literature seems to suggest you have significantly improved your compression algorithms and this should just work. I slid the compression bar all the way to the right for "Best". If there is something else we need to do then please let me know

Thanks

[FrancescoT]

Dear Mike,

I just tried to generate a Web Compressed exe compiling the same project I used previously (it only includes Net 4.6).

The results are all most identical;

- 2.539 kb for the Web Exe
- 63.062 kb for the Net 4.6 web media block

Please note! It's required a x64 processor to use the current maximum compression level.

Hope this helps you.

Regards

[spicymikey]

Yes, that's what I get also. The point is, that's no better than what you got with IA 18. It's the same!

I am using an x64 processor of course. I am also using a 64 bit version of Windows 8.1

Go ahead and try running this same thing with IA18 and best compression. it's the same numbers

Thanks
Mike

[FrancescoT]

Dear Mike,

how can it be possible to replicate the same identical test in IA 18, if the Net 4.6 package is only available in X2?

You can eventually use the lower BEST compression in IA X2 to replicate the same IA 18 condition ... and this is exactly what I did before.

Regards

[spicymikey]

Sorry, I used .net 4.5 for my test. it came out the exact same at 68.8meg. This isn't a test of compression for a particular file. This is a test for compression in general. Don't use a iA runtime, use something third party that can be run through both installers.

I have an entire production ready installer that creates a dozen packages totally about 240meg. It's all exactly the same size in IA X2 as it is in IA 18.

[FrancescoT]

Dear Mike,

as I said earlier, that results may be different based on the file used the test the compression.
This is not a trick or an excuse to hide something, it's just in this way.

For example, the compression rate of a JPG file will be not the same of a plain text file and this because the JPG file, due its nature, it is already compressed.

The above is just an example, but the same occurs with any other kind of file involved with the process.

If you want to replicate the exact IA 18 compression settings and results, just use the same identical best compression level that it's still available in IA x2. Then use the new x64 compression settings to do any comparison.

You can go further with any other possible and additional tests if you want.

hope this clarifies any doubt.

Regards

[spicymikey]

Francesco,

Now I'm getting angry. This conversation is going no where. Not sure the problem or why you are not following me, but I am telling you the following:

**** I took a complete IA 18 installer which contains over a dozen packages and ran it with IA X2, IT MADE NO DIFFERENCE. Zero. Nada. Nothing. The size of the installer was the same.

I understand jpg's are already compressed. In fact EVERTHING IS ALREADY COMPRESSED that we would run through your installer. EVERY third party MSI or EXE installer is already compressed. SO are you saying that this "30% compression improvement" is only for text files and files not already compressed? if so, then the feature is basically useless since the bulk of a projects size is not from the script files. That's tiny already. it's from all the 3rd party msi's etc.

I'm starting to belive i was right from the beginning. The claim made by Jason with his never ending sales emails was BS. its a misleading claim, at best, and a lie, at worst. if I am wrong, then prove it to me and everyone reading this thread. Compress SQL Server 2012 from the Microsoft site using IA18 and IA19 and show me your results. For me it was 152meg with ether installer.

[FrancescoT]

Dear Mikey,

please excuse me, but if there is someone that should get angry, this is me.

I am not saying that this "30% compression improvement" is only for text files or that it can't be used with files already compressed.
I am only saying that this compression as any other possible compression technique available, may be influenced by the files involved with the process. This is not a BS.

If you really want to make test over a compression process,YOU SHOULD NOT USE AN ALREADY COMPRESSED FILE. Because it simply not possible to produce an effective and real test using such kind of files. Instead use a folder that contains some files of any kind and do a test.

Now if you believe that what reported by Jason it is not true or that the feature is basically useless, you are totally free to believe what you prefer.

It seems to me that I had a collaborative conduct with our discussion, but unfortunately, I didn't get the same feeling by your last reply.
Consequently I have to ask you to refrain in continuing the discussion in this way.

For sure I totally agree with your words when you say, this conversation is going no where .... and there is no reason to continue it.

Regards