Android Question Improve B4A speed using the video card?

I think your processor architecture has more than sufficient resources for compiling. Your hardware is definitely not the problem. Adding an additional processor wouldn't change that in the slightest. The graphic clearly shows that there is no lack of available computing power. All cores are evenly utilized except for a few peaks. It's not a hardware bottleneck. The “problem” in this case is quite clearly the compiler itself. The cores process pending compiler tasks too quickly and then go into wait state. The compiler's program flow simply cannot provide enough feed to keep the processors busy. These then “suffer” from underemployment. Who can make the compiler faster? Quite simply: the compiler programmer. Now this one will almost certainly not be a beginner either. But it is in the nature of a compiler that it also uses libraries and other suppliers and has to wait for their results. This is a general problem with high-level languages and object-oriented programming. It is like it is. But no one is going to sit down and write process-oriented code in assembler, this would probably solve the problem to a large extent. Let's be honest (and I hope you understand my irony): Compile your program on an older Pentium with only one core. A few gigabytes less RAM would also be helpful. Then you know what waiting means. I know similar problems with graphics controllers. And I don't mean a graphics card. I mean high-speed PCs with the latest technology, architecturally designed to render large graphic files in digital print and then send this RAW data to a printer (e.g. a Fiery). Many users try to make their Fiery faster with more memory. But that's usually enough and more RAM doesn't help at all. It is the program flow of the rendering itself that simply takes time despite optimal programming. Providing needed resources during the compiler run produces exactly the same effect. The compiler itself has to wait because it has no other choice due to its programming. Compiling the program on another computer might not be a bad idea. You have the direct comparison for this phenomenon. Only if the CPU utilization is permanently at 100% then there is a hardware problem. Even with a slightly weaker computer, you will need similar times for the compiler to run as long as the processor does not reach its limits.

BlueVision said:

Even with a slightly weaker computer, you will need similar times for the compiler to run as long as the processor does not reach its limits.

Click to expand...

Upgrading the RAM from 32GB to 96GB ( for other purposes, not B4X) did not improve compiling time in B4X at all.

yo3ggx said:

Upgrading the RAM from 32GB to 96GB ( for other purposes, not B4X) did not improve compiling time in B4X at all.

Click to expand...

Exactly. Because it is not the bottleneck...

yo3ggx said:

All 12 logical CPUs are used during compilation.

Click to expand...

This is only what the operating system shows you, in reality only one CPU is used.

But you can ask Erel if he has programmed the compiler to be multithreaded.

Filippo said:

This is only what the operating system shows you, in reality only one CPU is used.

But you can ask Erel if he has programmed the compiler to be multithreaded.

Click to expand...

Not only does it use all cores (8), but also all threads (16). Are you saying that what Windows Task Manager shows is fake?

Filippo said:

But you can ask Erel if he has programmed the compiler to be multithreaded.

Click to expand...

The IDE uses many threads for all kinds of tasks.
The compiler is managed by a single background thread but some of the toolchain components are multithreaded.

Erel said:

The IDE uses many threads for all kinds of tasks.
The compiler is managed by a single background thread but some of the toolchain components are multithreaded.

Click to expand...

This is what I observed, all processor threads are used.