Messages in this thread | | | Date | Wed, 5 May 1999 08:08:55 -0400 | From | Matthew Hunter <> | Subject | Re: Mindcraft III |
| |
On Wed, May 05, 1999 at 04:34:08AM +0000, Andrew Morton wrote: > Mindcraft are not releasing the results of their > recent retest. They'll be running it a third time: > http://www.mindcraft.com/openbenchmark.html
This is an interesting change of position. Perhaps they are really trying to be honest, since this report has been widely publicized and their reputation is at risk. However, reading their proposal, I notice a potential problem: they are limiting the patches that can be applied to the Linux system to those available from a limited list of sites within a limited timeframe.
In short, if you want constant foo to have a known, non-default value, the typical way to do it would be to modify the code to change the constant, then recompile. It's a trivial change for anyone familiar with the system, but under the new rules, it doesn't seem like it would be allowed. Instead, you would have to find a publically-available patch and apply that.
In effect, the result would not be "tuning", but instead something vaguely reminiscent of MS's service packs; "vaguely official updates only".
The restriction is presumably intended to simulate a knowledgeable expert who doesn't want to make any "dangerous" changes, but then, changing certain constant values is not necessarily dangerous for a kernel expert; NT probably allows similar constants to be changed via the registry, thus giving NT an advantage (albeit possibly a minor one).
If anyone is going to take them up on this one, the issue is probably worth clarifying.
-- Matthew Hunter (mhunter@andrew.cmu.edu) Do not meddle in the affairs of sysadmins, for they are root, and merciless.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |