purchasing for free...
...
...
yup, you gotta love modern technology.
| OSDev.org https://forum.osdev.org/ |
|
| SIMICS https://forum.osdev.org/viewtopic.php?f=1&t=8508 |
Page 1 of 1 |
| Author: | be [ Wed Jul 30, 2003 11:31 pm ] |
| Post subject: | SIMICS |
Does anyone kno about such emulator? I wanna know can I purchase it for free some way? |
|
| Author: | bkilgore [ Thu Jul 31, 2003 12:07 am ] |
| Post subject: | Re:SIMICS |
purchasing for free... ... ... yup, you gotta love modern technology.
|
|
| Author: | bkilgore [ Thu Jul 31, 2003 9:38 pm ] |
| Post subject: | Re:SIMICS |
Guess you were right about that free part... I happened to go to one of the schools they list as working with, so I decided to apply for a free 1-year academic license. The next day I got a reply with the FlexLM license file and information on how to download it. I haven't had a chance to try it yet but it sounded interesting. Kinda like a commercial bochs with more architectures supported. Like bochs everything is deterministic, but it's supposed to provide a lot more information, like dumping information about cache misses, etc. Also, it can simulate clusters and even network computers (apparently for a demo they had it simulate two computers, each with two processors, one running linux, the other windows, talking to each other over the network) they say even the networked computers run deterministically, so it sounds good for debugging hard-to-recreate race conditions, etc. Anyways, I'm gonna start trying it out now, I'll let you know how it works. - Brandon |
|
| Author: | mystran [ Fri Aug 01, 2003 12:12 am ] |
| Post subject: | Re:SIMICS |
When one has multi-tasking enabled, couldn't one add some indeterminism to bochs. Like, one could have a kernel task that used some decent PRNG, which could be seeded either automatically, by taking some value like time on boot, or manually with a kernel/module parameter, so you could still reproduce the same sequence. This task could then introduce some "random" delays into the code, so things like interrupts would appear a little less deterministic. Maybe other things too, like allocating/deallocating some random amounts of things (kernel objects, memory, pages). If this was a loadable module, you could add it into your test-setup, take the automatically generated seed, and if something odd happens, you could reproduce it and test it again. ps. this was just a random idea.. no idea if it's really useful thing to do.. |
|
| Author: | bkilgore [ Fri Aug 01, 2003 10:59 am ] |
| Post subject: | Re:SIMICS |
I can hardly imagine it being useful to add indeterminism to an emulator.... Besides, not everything is "deterministic" as long as there is some connection between the simulated and the real, chaotic world. For instance, when it comes to keyboard interrupts, if something about a keyboard interrupt occuring somehow causes a race condition or some other inconsistent behavior in the kernel, it would be very hard to recreate. But if something is triggered not so much by the exact time something started, as by the relative timing between code, determinism can help debug that greatly.
|
|
| Page 1 of 1 | All times are UTC - 6 hours |
| Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |
|