Re: [patch 0/3] kvm tool: Serial emulation overhaul

[Pekka Enberg]

On Sun, Dec 11, 2011 at 5:53 PM, Ingo Molnar <mingo@xxxxxxx> wrote:
> Okay, but look at it from another angle: the top output i
> generate is about 300k characters. 5000 msecs to execute it
> means 16 usecs overhead per character - or about 50k cycles - on
> a top of the class x86 CPU.

I'm seeing 1.5 usecs per character for this little benchmark:

$ cat tick.S
#define DBG_PORT	0xe0
	.code16gcc
	.text
	.globl	_start
	.type	_start, @function
_start:
	mov	$0x64,%cx
	mov	$0xffff,%bx

outer_loop:
inner_loop:
	mov	$0x3f8,%dx
	mov	$0x2e, %al	# .
	out	%al,%dx

	sub	$0x1,%bx
	jne	inner_loop
	
	sub	$0x1,%cx
	jne	outer_loop

	/* not a valid port to force exit */
	outb	%al, $DBG_PORT

$ gcc -nostdinc -c tick.S -o tick.o
$ ld -Ttext=0x00 -nostdlib -static tick.o -o tick.elf
$ objcopy -O binary tick.elf tick.bin
$ perf stat ./vm run ./tests/serial/tick.bin > /dev/null
  Warning: ./tests/serial/tick.bin is not a bzImage. Trying to load it
as a flat binary...

 Performance counter stats for './vm run ./tests/serial/tick.bin':

       9984.468850 task-clock                #    0.995 CPUs utilized
            16,597 context-switches          #    0.002 M/sec
                46 CPU-migrations            #    0.000 M/sec
             1,200 page-faults               #    0.000 M/sec
    32,500,683,485 cycles                    #    3.255 GHz
         [82.17%]
    14,986,643,686 stalled-cycles-frontend   #   46.11% frontend
cycles idle    [83.57%]
     9,400,515,530 stalled-cycles-backend    #   28.92% backend
cycles idle    [67.11%]
    25,213,133,751 instructions              #    0.78  insns per cycle
                                             #    0.59  stalled cycles
per insn [83.57%]
     5,111,389,696 branches                  #  511.934 M/sec
         [83.57%]
        14,565,759 branch-misses             #    0.28% of all
branches         [83.58%]

      10.030175553 seconds time elapsed
--
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/