Messages in this thread | | | Date | Thu, 18 Jun 1998 13:05:56 -0700 (PDT) | From | <> | Subject | Re: SYSENTER based syscalls patch, 2.1.105, RFC |
| |
On Thu, 18 Jun 1998, Andrea Arcangeli wrote:
> > [mingo@hal fastcall]$ ./fastcall > > (INT $0x80 based getpid(), got pid 497) latency:282 cycles > > (SYSENTER based getpid(), got pid 497) latency:138 cycles > > andrea@dragon:~$ ./a.out > (INT $0x80 based getpid(), got pid 949) latency:115 cycles > Illegal Instruction > [...] > > On my 266 Mhz P5MMX I am going faster than your PII?!?! I am using a #SMP > (no-SMP ;-) kernel but this should not change nothing since the test is > done outside the kernel code...
yep. It's probably because on the PII it's more expensive to do segmentation related stuff. And call gates (int $0x80) do _alot_ of that stuff ...
-- mingo
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |