Re: [PATCH] Use x86 SSE instructions for clear_page, copy_page

[Ingo Molnar]

* Ingo Molnar <mingo@xxxxxxx> wrote:

> executed SSE code? You are corrupting those registers.

doh - should read before i write. The code is perfectly fine.

Arjan's cache arguments remain. What i'd suggest to test is the precise
speed of compiling the kernel, fully done in ramfs. (if you dont have
enough RAM for this then compile a portion of the kernel.) For the
testing, add a /proc/sys switch to turn the SSE functions on/off
runtime, hence you can eliminate the effects of page placement. If the
compilation timings are stable enough then you can try the runtime
switch to see whether it has any effect. There should be a small but
visible change (in one direction or the other), as compilation brings in
lots of new pages and copies around stuff too.

	Ingo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/