java-processor at Yahoo! Groups

Re: SUPERCOMPUTING MULTI-CORE JOP PROJECT

Martin Schoeberl — Tue, 10 Nov 2009 06:18:50 GMT

Why would you need to switch contexts? That would imply that you have more parallel code than actual execution units. With a highly parallel hardware (either

Re: SUPERCOMPUTING MULTI-CORE JOP PROJECT

Tobias Gogolin — Tue, 10 Nov 2009 06:04:14 GMT

Would it be an efficient approach though? I represent the thesis that to achieve breakthrough performance the program with all its objects and methods should

Re: Instantiated Memory

Dan Clemmensen — Tue, 10 Nov 2009 02:31:43 GMT

My bootloader and a tool called ExtractBoot are included in the tarball. ExtractBoot post-processes the bytecode to convert constructs that need to access main

Re: Instantiated Memory

Martin Schoeberl — Tue, 10 Nov 2009 01:30:39 GMT

Ok, understand. Putting a boot loader in Java into the M$ instead of booting via microcode was also on my list of further ideas. It would make the generation

Re: Instantiated Memory

Dan Clemmensen — Tue, 10 Nov 2009 00:45:51 GMT

Here is XBlockGen, together with some related files. These may be useful for review, but they are by no meas a complete project.

Re: Instantiated Memory

Dan Clemmensen — Tue, 10 Nov 2009 00:42:51 GMT

I had forgotten that. My project initializes the bytecode cache with a bootloader and uses two BRAMs for the microcode, so I have four initialized BRAMs. I use

Re: SUPERCOMPUTING MULTI-CORE JOP PROJECT

David Blubaugh — Mon, 09 Nov 2009 23:46:59 GMT

Sir, Would you be interested in helping me to initiate such a project ?? David ... From: Martin Schoeberl Subject: Re:

Re: JDWP Implementation to debug in JOP.

Martin Schoeberl — Mon, 09 Nov 2009 23:33:00 GMT

Looks like there is a make target missing to build JopDebugger.jar The debugger project is a little sad story: almost done, the development on it was stopped.

Re: mem_main.dat file question

Martin Schoeberl — Mon, 09 Nov 2009 23:29:14 GMT

What do you mean by 'implement the memory as Java class files'? I don't understand. If you need the main memory content in a Java based simulation, just read

Re: SUPERCOMPUTING MULTI-CORE JOP PROJECT

Martin Schoeberl — Mon, 09 Nov 2009 23:25:51 GMT

I'm not aware of such a project. I would be skeptic to use JOP for high performance computing. It is designed for time-predictable computing, which is a

Re: Instantiated Memory

Martin Schoeberl — Mon, 09 Nov 2009 23:22:21 GMT

In the actual configuration of JOP only the stack memory uses initialized BRAMs with VHDL sources generated by BlockGen. The Java bytecode cache (jbc) uses an

Re: Instantiated Memory

Dan Clemmensen — Mon, 09 Nov 2009 18:43:08 GMT

Well, first, I must publish it somewhere. Furthermore, you would need to re-structure your project to use my BRAMs instead of the ones in the official project.

Re: Instantiated Memory

c.p.bridges@... — Mon, 09 Nov 2009 18:29:38 GMT

Also, Martin, would this error in BRAM generation for the stack give an incorrect initial address read? Cheers, Chris ________________________________ From:

Re: Instantiated Memory

c.p.bridges@... — Mon, 09 Nov 2009 18:19:06 GMT

Hey Dan, how do I use Xblockgen? Cheers, Chris ________________________________ From: java-processor@yahoogroups.com [mailto:java-processor@yahoogroups.com] On

Instantiated Memory

Dan Clemmensen — Mon, 09 Nov 2009 18:05:30 GMT

BRAM initializers use INIT_00 through INIT_3f. I seems that blockgen is emitting too many initializers. I guess that this could occur if the input .dat file