| To: | Jeff Garzik <jgarzik@xxxxxxxxxxxxxxxx> |
|---|---|
| Subject: | Re: [timers] net/core/* |
| From: | Andi Kleen <ak@xxxxxx> |
| Date: | Wed, 31 May 2000 22:24:34 +0200 |
| Cc: | Andi Kleen <ak@xxxxxx>, kuznet@xxxxxxxxxxxxx, Andrew Morton <andrewm@xxxxxxxxxx>, netdev@xxxxxxxxxxx |
| In-reply-to: | <393573CD.DADE1A55@xxxxxxxxxxxxxxxx>; from Jeff Garzik on Wed, May 31, 2000 at 10:19:37PM +0200 |
| References: | <39352AB3.7C609B8C@xxxxxxxxxx> <200005311903.XAA23547@xxxxxxxxxxxxx> <20000531212143.64448@xxxxxxxxxxxx> <393573CD.DADE1A55@xxxxxxxxxxxxxxxx> |
| Sender: | owner-netdev@xxxxxxxxxxx |
On Wed, May 31, 2000 at 10:19:37PM +0200, Jeff Garzik wrote: > > gcc 3.0 will fix this, with -freorder-blocks (which knows some common > > heuristics like == NULL is usually not taken already) and __builtin_expect() > > > > it'll allow exorcise of about a zillion gotos from the kernel tree. > > I wish... :) We'll still have to support gcc 2.7.2... Not with the highest efficieny though. gcc 2.7.2 kernels will be most likely slower than gcc 3.0 kernels (3.0 has a much better scheduler at least on x86). Adding a few more cycles difference probably do not hurt -- especially when it makes the code easier to read and maintain. -Andi -- This is like TV. I don't like TV. |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: [timers] net/core/*, Jeff Garzik |
|---|---|
| Next by Date: | Re: [PATCH] eepro100 device name <-> pci bus/slot/func mapping, Jeff Garzik |
| Previous by Thread: | Re: [timers] net/core/*, Jeff Garzik |
| Next by Thread: | [timers] net/sched/*, Andrew Morton |
| Indexes: | [Date] [Thread] [Top] [All Lists] |