optimized build targets

[Vladimir Kozlov]

Chuck,

I think nobody care about performance of optimized build since it is still much faster then fastdebug VM and we don't 
expect it to perform as product VM.
Several years ago I did some cleanup of debug vs non-product flags and corresponding code. But I would not surprise we 
still have some debug code under #ifndef PRODUCT. If you can point such code we will appreciate it.

Thanks,
Vladimir

On 7/1/10 2:53 PM, Chuck Rasbold wrote:
> Coleen, Vladimir,  thanks for the answers and for jogging my memory.
>
> As a follow-up, with optimized, do either of you see more performance degradation problems vs. product than is warranted?
>
> I seem to recall some diagnostic JVM code that is under #ifndef PRODUCT which really should be protected by #ifdef
> DEBUG.  That is, are there runtime sanitty checks going on that were really only intended for debug/fastdebug, but are
> inadvertently being included in the optimized build because of a flipped #ifdef?
>
> -- Chuck
>
> On Thu, Jul 1, 2010 at 12:19 PM, Vladimir Kozlov <vladimir.kozlov at oracle.com <mailto:vladimir.kozlov at oracle.com>> wrote:
>
>     I am using it from time to time. Some logging code (compiler statistic, for example) is under #ifndef PRODUCT so you
>     don't need debug VM to get it and generated code is the same as product.
>
>     Vladimir
>
>
>     On 7/1/10 12:06 PM, Chuck Rasbold wrote:
>
>         Does anyone still utilize the "optimized" build target for HotSpot?
>
>         I used to believe that optimized was nearly identical to product,
>         but with the symbols not stripped out.  I see from optimized.make though,
>         that -DPRODUCT is not defined, and might have guessed that the G_SUFFIX
>         would be set, which it is not.
>
>         What's the current rationale for these settings?
>
>         -- Chuck
>
>