2.4.20-rc2 strange L1 cache values

2.4.20-rc2 strange L1 cache values

[Margit Schubert-While]

Another peculiar thing - Why is /proc/cpuinfo showing "ht" ?

processor           : 0
vendor_id           : GenuineIntel
cpu family           : 15
model                 : 2
model name       : Intel(R) Pentium(R) 4 CPU 2.40GHz
stepping             : 4
cpu MHz             : 2400.159
cache size          : 512 KB
fdiv_bug             : no
hlt_bug               : no
f00f_bug            : no
coma_bug          : no
fpu                      : yes
fpu_exception   : yes
cpuid level         : 2
wp                      : yes
flags                   : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr 
pge mca cmov pat pse36
                              clflush dts acpi mmx fxsr sse sse2 ss ht tm
bogomips          : 4784.12

Re: 2.4.20-rc2 strange L1 cache values

[Dave Jones]

On Wed, Nov 20, 2002 at 09:40:51PM +0100, Margit Schubert-While wrote:
 > Another peculiar thing - Why is /proc/cpuinfo showing "ht" ?
 > 
 > processor           : 0
 > vendor_id           : GenuineIntel
 > cpu family           : 15
 > model                 : 2
 > model name       : Intel(R) Pentium(R) 4 CPU 2.40GHz
 > flags                   : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr 
 > pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm

All Pentium 4's seem to show it, though not all have them have
the extra sibling.

		Dave

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

2.4.20-rc2 strange L1 cache values

[Margit Schubert-While]

Any ideas anybody ?

<6>CPU: L1 I cache: 0K, L1 D cache: 8K
<6>CPU: L2 cache: 512K
<6>Intel machine check architecture supported.
<6>Intel machine check reporting enabled on CPU#0.
<7>CPU:     After generic, caps: 3febfbff 00000000 00000000 00000000
<7>CPU:             Common caps: 3febfbff 00000000 00000000 00000000
<4>CPU: Intel(R) Pentium(R) 4 CPU 2.40GHz stepping 04

Re: 2.4.20-rc2 strange L1 cache values

[Dave Jones]

On Wed, Nov 20, 2002 at 07:05:06PM +0100, Margit Schubert-While wrote:
 > Any ideas anybody ?
 > <6>CPU: L1 I cache: 0K, L1 D cache: 8K
 > <6>CPU: L2 cache: 512K
 > <6>Intel machine check architecture supported.
 > <6>Intel machine check reporting enabled on CPU#0.
 > <7>CPU:     After generic, caps: 3febfbff 00000000 00000000 00000000
 > <7>CPU:             Common caps: 3febfbff 00000000 00000000 00000000
 > <4>CPU: Intel(R) Pentium(R) 4 CPU 2.40GHz stepping 04

P4 Trace cache isn't recognised.
Apply ftp://ftp.kernel.org/pub/linux/kernel/people/davej/patches/2.4/2.4.20/descriptors.diff

		Dave

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

Re: 2.4.20-rc2 strange L1 cache values

[Steffen Persvold]

On Wed, 20 Nov 2002, Dave Jones wrote:

> On Wed, Nov 20, 2002 at 07:05:06PM +0100, Margit Schubert-While wrote:
>  > Any ideas anybody ?
>  > <6>CPU: L1 I cache: 0K, L1 D cache: 8K
>  > <6>CPU: L2 cache: 512K
>  > <6>Intel machine check architecture supported.
>  > <6>Intel machine check reporting enabled on CPU#0.
>  > <7>CPU:     After generic, caps: 3febfbff 00000000 00000000 00000000
>  > <7>CPU:             Common caps: 3febfbff 00000000 00000000 00000000
>  > <4>CPU: Intel(R) Pentium(R) 4 CPU 2.40GHz stepping 04
> 
> P4 Trace cache isn't recognised.
> Apply ftp://ftp.kernel.org/pub/linux/kernel/people/davej/patches/2.4/2.4.20/descriptors.diff
> 

Yep that works (I have two Xeon boxes with the same issue) :

(2.4.20-rc2 with descriptors.diff)
CPU: Trace cache: 12K uops, L1 D cache: 8K
CPU: L2 cache: 512K
CPU: Physical Processor ID: 0
Intel machine check reporting enabled on CPU#0.
CPU:     After generic, caps: 3febfbff 00000000 00000000 00000000
CPU:             Common caps: 3febfbff 00000000 00000000 00000000
CPU0: Intel(R) XEON(TM) CPU 1.80GHz stepping 04

But why does this P4 Trace cache report as L1 I cache on 2.4.18 ? 

(2.4.18)
CPU: Before vendor init, caps: 3febfbff 00000000 00000000, vendor = 0
CPU: L1 I cache: 12K, L1 D cache: 8K
CPU: L2 cache: 512K
CPU: Physical Processor ID: 0
CPU: After vendor init, caps: 3febfbff 00000000 00000000 00000000
Intel machine check reporting enabled on CPU#0.
CPU:     After generic, caps: 3febfbff 00000000 00000000 00000000
CPU:             Common caps: 3febfbff 00000000 00000000 00000000
CPU0: Intel(R) XEON(TM) CPU 1.80GHz stepping 04

Does this have any implications on the 2.4.18 performance (or the 
2.4.20-rc2 performance without the descriptors.diff) ?

Regards,
-- 
  Steffen Persvold   |       Scali AS      
 mailto:sp@scali.com |  http://www.scali.com
Tel: (+47) 2262 8950 |   Olaf Helsets vei 6
Fax: (+47) 2262 8951 |   N0621 Oslo, NORWAY

Re: 2.4.20-rc2 strange L1 cache values

[Dave Jones]

On Wed, Nov 20, 2002 at 07:48:27PM +0100, Steffen Persvold wrote:

 > >  > <6>CPU: L1 I cache: 0K, L1 D cache: 8K
 > Yep that works (I have two Xeon boxes with the same issue) :
 > But why does this P4 Trace cache report as L1 I cache on 2.4.18 ? 

Look again above, and you'll see .18 said it had 0K L1 (which is
correct, L1 != Trace cache).
 
 > Does this have any implications on the 2.4.18 performance (or the 
 > 2.4.20-rc2 performance without the descriptors.diff) ?

The SMP weighting should be done with L2 cache size, which
was correct on .18 too, so it should be ok.

		Dave

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

Re: 2.4.20-rc2 strange L1 cache values

[Steffen Persvold]

On Wed, 20 Nov 2002, Dave Jones wrote:

> On Wed, Nov 20, 2002 at 07:48:27PM +0100, Steffen Persvold wrote:
> 
>  > >  > <6>CPU: L1 I cache: 0K, L1 D cache: 8K
>  > Yep that works (I have two Xeon boxes with the same issue) :
>  > But why does this P4 Trace cache report as L1 I cache on 2.4.18 ? 
> 
> Look again above, and you'll see .18 said it had 0K L1 (which is
> correct, L1 != Trace cache).

The original poster reported on 2.4.20-rc2 (which reports 0K), not 2.4.18. 
The output I provided was from .18 and that clearly says 'L1 I cache:12K'.
Here are some 'sniplets' of my 2.4.18 dmesg :

Linux version 2.4.18-sca8smp (root@r2) (gcc version 2.96 20000731 (Red Hat Linux 7.3 2.96-112)) #1 SMP Wed Nov 6 20:17:23 CET 2002

[snip]

CPU: L1 I cache: 12K, L1 D cache: 8K
CPU: L2 cache: 512K
CPU: Physical Processor ID: 0
Intel machine check reporting enabled on CPU#0.
CPU0: Intel(R) XEON(TM) CPU 1.80GHz stepping 04

>  
>  > Does this have any implications on the 2.4.18 performance (or the 
>  > 2.4.20-rc2 performance without the descriptors.diff) ?
> 
> The SMP weighting should be done with L2 cache size, which
> was correct on .18 too, so it should be ok.

Ok, since this is now fixed (with your patch), I really don't care about 
earlier kernels anyway (however some of my customer might have if it was 
an issue). Do you know if your patch is going into 2.4.20 release (it's a 
rather small and useful patch) ?

Thanks,
 -- 
  Steffen Persvold   |       Scali AS      
 mailto:sp@scali.com |  http://www.scali.com
Tel: (+47) 2262 8950 |   Olaf Helsets vei 6
Fax: (+47) 2262 8951 |   N0621 Oslo, NORWAY

Re: 2.4.20-rc2 strange L1 cache values

[Dave Jones]

On Wed, Nov 20, 2002 at 08:23:04PM +0100, Steffen Persvold wrote:

 > The original poster reported on 2.4.20-rc2 (which reports 0K), not 2.4.18. 
 > The output I provided was from .18 and that clearly says 'L1 I cache:12K'.
 > Here are some 'sniplets' of my 2.4.18 dmesg :
 > CPU: L1 I cache: 12K, L1 D cache: 8K

My bad. Yes, earlier kernels did confuse the two.
the 12K actually means the trace cache holds 12,000 uops
rather than you have 12KB of cache.

The only bug is the reporting unit.

 > Ok, since this is now fixed (with your patch), I really don't care about 
 > earlier kernels anyway (however some of my customer might have if it was 
 > an issue). Do you know if your patch is going into 2.4.20 release (it's a 
 > rather small and useful patch) ?

I've resubmitted it to Marcelo, but it still didn't show up..

		Dave

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs