The problem with sploop

Hi,

Thanks for your post.

Please read the below E2E post:

https://e2e.ti.com/support/dsp/tms320c6000_high_performance_dsps/f/112/p/11376/99894#pi317286=2

To know more on SPLOOP, please refer section 3.4 in the c6000 optimizing compiler user guide as below:

http://www.ti.com/lit/ug/spru187v/spru187v.pdf

There are c6000 optimization workshop and materials with c64+ optimization techniques involved. please refer the below wiki's:

http://processors.wiki.ti.com/index.php/Optimization_Techniques_for_the_TI_C6000_Compiler

http://processors.wiki.ti.com/index.php/TMS320C6000_DSP_Optimization_Workshop

http://processors.wiki.ti.com/index.php/Optimized_Sort_Algorithms_For_DSP#Sort_Algorithms

There are also techniques involved in optimizing c64x+ compiler to produce efficient code which includes:

Restrict keyword & usage with structure fields

Impact of supplying trip count info

Balancing resources

Optimizing “if” statements

Handling function calls

Understanding optimizer output etc.

For more details, please refer the below wiki presentation:

https://processors.wiki.ti.com/images/e/e8/C64plus_cgt_overview.pdf

Also, there are c64/c64+ compiler optimization tricks involved. To know more on this, please refer the below wiki presentation:

http://processors.wiki.ti.com/images/6/6e/C64p_cgt_optimization.pdf

Thanks & regards,

Sivaraj K

-------------------------------------------------------------------------------------------------------

Please click the Verify Answer button on this post if it answers your question.

-------------------------------------------------------------------------------------------------------

Stanislav Gritsutenko said:

Is there a document that describes the resource conflict in loop buffer?

Basically instruction set reference guide is sufficient, but you have to use imagination. VLIW is about expressing parallelism on processor's terms. In the context "processor's terms" mean that you can't schedule multiple instructions that are executed on same execution unit/port in the same cycle. Commonly you express parallelism explicitly with ||, right? And you have no problem grasping what does resource conflict mean in that case, right? In SPLOOP on the other hand parallelism is expressed implicitly, but you still have to lay down instructions in manner that doesn't cause conflicts. SPLOOP 1 means that execution of the loop body will start every cycle. So at 1st cycle add.l2 will execute. But in the next cycle you want processor to execute second add.l2 and first add.l2. No go, you can't execute two instructions on same port in same cycle. You have to imagine "vertical" SPLOOP "horizontally" in order to see the conflict. And the reference guide provides examples of such visualization, so study it harder...

Just in case. Is explicitly-expressed parallelism incompatible with implicit SPLOOP? No. You can use || even in SPLOOP, but you still can't schedule more than one instruction on same execution unit/port in the same cycle. So that if there is ||-group in SPLOOP, whatever that simultaneously flies by from adjacent iterations may not cause conflict.