Quantcast
Channel: Forums - Recent Threads
Viewing all articles
Browse latest Browse all 262198

Compiler/AM3517: Float Point Calculation Unit(VFP) Performance Question

$
0
0

Part Number:AM3517

Tool/software: TI C/C++ Compiler

We are using AM3517(ARM Cotex-A8) chip and running software compiled  by Sourcery GNU GCC Toolchain [ gcc version 4.4.1 (Sourcery G++ Lite 2010q1-202), arm-none-linux-gnueabi-gcc] .

We do some comparison tests to to check VFP module performance, in our test doing some calculation(such as sin,cos, tan, sqrt, matrix multiple etc), I see the calculation perormance of using compiling option( -mfpu=vfp, -mfloat-abi=softfp,  enabled VFP module) is about twice times than compiling option(-mfloat-abi=soft,  disabled VFP module) ,  So my questiion is :

1)  Is the above test result  reasonable?  OR How many times should be reasonable?

2)  Besides we also want to know the performance difference between using compling option( -mfpu=vfp, -mfloat-abi=softfp) and option( -mfpu=vfp, -mfloat-abi=hard).  But we can't running software compiled using option (-mfpu=vfp, -mfloat-abi=hard),  whether option(-mfloat-abi=hard) is much better than option(-mfloat-abi=softfp) ?

3) Where can i get a suitable toolchain to compile software and compiling option (-mfloat-abi=hard)  can be enabled?  we also tried to use CCS V7 toolchain to build software, but I can't running software on our board :(

 

Could someone give me some support or suggestion for abvoe questions?  Thanks a lot in advance.

 

 


Viewing all articles
Browse latest Browse all 262198

Trending Articles



<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>