On TechRepublic: Weirdest error messages of all time
BNET Business Network:
BNET
TechRepublic
ZDNet
TalkBack 14 of 27:
Next »
« Previous
In a perfect world...
Yes, those points may be true in a perfect world.

> The JIT compiler can optimize for a specific CPU, rather than for the lowest denominator in a CPU family

If an application is targeting peak performance then it is reasonable to assume that the end-user has the same goals. It would not be unreasonable to expect that such clients would have top of the line hardware so if you have a product that runs on the PPC you compile it for the top of the line PPC processor; you would not compile it for the least common denominator. You don't gain something from JIT in that case. I have yet to see a JIT compiler that outperforms the top C/C++ compilers, if you have then I would be interested in reading the specifics. So far this claim has been nothing more than blue sky optimism on the part of JIT technology fans.

>A JITd VM can adjust for changing branch prediction results

How may cycles are required to make that judgment as to which is the best branch? Unless this is a very simple decision you would probably spend more cycles than you would gain.

3 is a good point.

Garbage collection is not ideal for a real time application. You still need to deallocate memory using traditional methods so you are not gaining anything with C/C++ there however, long pauses can be unacceptable in cases even with multi-generation garbage collectors. Garbage collect's greatest contribution is it's ability to cover for sloppy coders.
Posted by: balsover   Posted on: 12/07/05 You are currently: a Guest | Members login | Terms of Use

Alert moderator to an offensive message

Subscribe to this discussion via Email or RSS

poor performance per dollar  ~doolittle~ | 12/07/05
Not really  Roger Ramjet | 12/07/05
and you had a fantastic compiler  balsover | 12/07/05
The idealized one  Roger Ramjet | 12/07/05
Then it is pointless, isn't it? Not such a great idea after all  balsover | 12/07/05
History lesson  Roger Ramjet | 12/07/05
Great Comments!  DemonX | 12/07/05
Well...  techboy_z | 12/07/05
The InHell mac  Roger Ramjet | 12/07/05
Mactel  jheine | 12/07/05
Actually...  techboy_z | 12/07/05
Partly true  Roger Ramjet | 12/07/05
Re. compile for speed  techboy_z | 12/07/05
In a perfect world...  balsover | 12/07/05
Remember the InHell iAPX432 ?  Steven J. Ackerman | 12/07/05
From the article you linked to  balsover | 12/07/05
A few comments on what is "good" architecture for the future  michael_t | 12/08/05
Always starts. . . . . . .  a.techno.geek | 12/07/05
What people don't get  Roger Ramjet | 12/07/05
Perhaps you are the one that is wrong  balsover | 12/07/05
Did they really blow it?  Roger Ramjet | 12/08/05
Yes they did.  balsover | 12/09/05
Thumbs up!  sys_engineer | 12/07/05
A better question  Roger Ramjet | 12/08/05
cost?  balsover | 12/09/05
Message has been deleted.  myfevertoy | 10/22/06
How a big company MISSES a good chance and conversely  michael_t | 12/08/05

What do you think?

SponsoredWhite Papers, Webcasts, and Downloads

advertisement

SmartPlanet

Click Here