ht CPU flag

ht CPU flag

[eugene]

Hello dear developers!

Here is my system configuration.

Linux ns 2.6.19-gentoo-r5 #5 SMP Thu May 3 00:45:12 AMST 2007 i686 
Intel(R) Pentium(R) D CPU 3.00GHz GenuineIntel GNU/Linux

Gnu C                  4.1.1
Gnu make               3.81
binutils               2.16.1
util-linux             2.12r
mount                  2.12r
module-init-tools      3.2.2
e2fsprogs              1.39
reiserfsprogs          3.6.19
Linux C Library        > libc.2.5
Dynamic linker (ldd)   2.5
Procps                 3.2.7
Net-tools              1.60
Kbd                    1.12
Sh-utils               6.4
udev                   104
Modules Loaded         ipt_ULOG iptable_filter iptable_mangle iptable_nat 
ip_nat ip_tables x_tables rtc nls_iso8859_5 ntfs ip_conntrack_ftp 
ip_conntrack nfnetlink e1000 e100 intelfb i2c_algo_bit intel_agp agpgart 
i2c_i801 i2c_core ehci_hcd pcspkr uhci_hcd usbcore sg
---------------------------------------------------------------------------

I have Pentium D CPU, which many Windows utilities like cpuz, wcpuid, 
everest identify as D 930 (Dual Core, 3GHz). From Intel site I find out 
that it has no HT feature, nor Windows XP identify it as HT.

Why do I have "ht" flag in cpuinfo?
-----------------------------------

cat /proc/cpuinfo
-----------------
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 15
model           : 6
model name      : Intel(R) Pentium(R) D CPU 3.00GHz
stepping        : 5
cpu MHz         : 2992.732
cache size      : 2048 KB
physical id     : 0
siblings        : 2
core id         : 0
cpu cores       : 2
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 6
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 pbe nx lm 
constant_tsc pni monitor ds_cpl est cid cx16 xtpr lahf_lm
bogomips        : 5990.31

processor       : 1
vendor_id       : GenuineIntel
cpu family      : 15
model           : 6
model name      : Intel(R) Pentium(R) D CPU 3.00GHz
stepping        : 5
cpu MHz         : 2992.732
cache size      : 2048 KB
physical id     : 0
siblings        : 2
core id         : 1
cpu cores       : 2
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 6
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 pbe nx lm 
constant_tsc pni monitor ds_cpl est cid cx16 xtpr lahf_lm
bogomips        : 5985.19
---------------------------------------------------------------------------

So, how must I choose kernel settings like:

number of CPU's?
HT support?

Best regards, Eugene.

Re: ht CPU flag

[Bernd Eckenfels]

In article <Pine.LNX.4.64.0705182131030.25685@ns.armcci.am> you wrote:
> I have Pentium D CPU, which many Windows utilities like cpuz, wcpuid, 
> everest identify as D 930 (Dual Core, 3GHz). From Intel site I find out 
> that it has no HT feature, nor Windows XP identify it as HT.

the ht flag reported by the CPU and cpuinfo is not a reliable detection if
HT is available on your CPU or your motherboard/bios.

> Why do I have "ht" flag in cpuinfo?

Because your CPU reports it. You will see that also in cpuz output.

However, you can see ht in the sibblings value (for a single core it will be
2 if you have HT, I am not sure if it is 4 for a dual core CPU)

Gruss
Bernd

Re: ht CPU flag

[H. Peter Anvin]

Bernd Eckenfels wrote:
> In article <Pine.LNX.4.64.0705182131030.25685@ns.armcci.am> you wrote:
>> I have Pentium D CPU, which many Windows utilities like cpuz, wcpuid, 
>> everest identify as D 930 (Dual Core, 3GHz). From Intel site I find out 
>> that it has no HT feature, nor Windows XP identify it as HT.
> 
> the ht flag reported by the CPU and cpuinfo is not a reliable detection if
> HT is available on your CPU or your motherboard/bios.
> 
>> Why do I have "ht" flag in cpuinfo?
> 
> Because your CPU reports it. You will see that also in cpuz output.
> 
> However, you can see ht in the sibblings value (for a single core it will be
> 2 if you have HT, I am not sure if it is 4 for a dual core CPU)

IIRC, the HT flag is also reported for multicore CPUs.

	-hpa

Re: ht CPU flag

[Siddha, Suresh B]

On Fri, May 18, 2007 at 11:45:59AM -0700, H. Peter Anvin wrote:
> IIRC, the HT flag is also reported for multicore CPUs.

Yes. Thats correct.

Re: ht CPU flag

[Bernd Eckenfels]

On Fri, May 18, 2007 at 12:07:09PM -0700, Siddha, Suresh B wrote:
> On Fri, May 18, 2007 at 11:45:59AM -0700, H. Peter Anvin wrote:
> > IIRC, the HT flag is also reported for multicore CPUs.
> 
> Yes. Thats correct.

And for some Single-Core Non-HT CPUs.

Gruss
Bernd
-- 
  (OO)     -- Bernd_Eckenfels@Mörscher_Strasse_8.76185Karlsruhe.de --
 ( .. )    ecki@{inka.de,linux.de,debian.org}  http://www.eckes.org/
  o--o   1024D/E383CD7E  eckes@IRCNet  v:+497211603874  f:+49721151516129
(O____O)  When cryptography is outlawed, bayl bhgynjf jvyy unir cevinpl!

Re: ht CPU flag

[Chris Snook]

eugene@ns.armcci.am wrote:
> ---------------------------------------------------------------------------
> 
> I have Pentium D CPU, which many Windows utilities like cpuz, wcpuid, 
> everest identify as D 930 (Dual Core, 3GHz). From Intel site I find out 
> that it has no HT feature, nor Windows XP identify it as HT.
> 
> Why do I have "ht" flag in cpuinfo?
> -----------------------------------

The "ht" flag merely means "I know how to report hyperthreaded logical 
processors if I have them."  My Woodcrest Xeon 5110 and my Athlon64 X2 both have 
the "ht" flag, and correctly report the zero hyperthreaded logical processors 
they each have.

	-- Chris

Re: ht CPU flag

[eugene]

On Fri, 18 May 2007, Chris Snook wrote:

> eugene@ns.armcci.am wrote:
>>  ---------------------------------------------------------------------------
>>
>>  I have Pentium D CPU, which many Windows utilities like cpuz, wcpuid,
>>  everest identify as D 930 (Dual Core, 3GHz). From Intel site I find out
>>  that it has no HT feature, nor Windows XP identify it as HT.
>>
>>  Why do I have "ht" flag in cpuinfo?
>>  -----------------------------------
>
> The "ht" flag merely means "I know how to report hyperthreaded logical 
> processors if I have them."  My Woodcrest Xeon 5110 and my Athlon64 X2 both 
> have the "ht" flag, and correctly report the zero hyperthreaded logical 
> processors they each have.
>
> 	-- Chris
>
>

Thanks, Chris.

Am I right that is chipset on mainboard, who is saying - "I know....", not 
CPU itself? Is it better to switch off HT support in BIOS?
Is it possible to generate CPU name as: "Pentium D 930" in /proc/cpuinfo?
On the other server I have some 2GHz HT Xeons which can't be identified on 
Intel site because of strange naming pattern.
I tried to find any utility for Linux to solve this, but it looks like 
everybody are using /pros/cpuinfo, which is not enough :)

Regards, Eugene.

Re: ht CPU flag

[Chris Snook]

eugene@ns.armcci.am wrote:
> 
> 
> On Fri, 18 May 2007, Chris Snook wrote:
> 
>> eugene@ns.armcci.am wrote:
>>>  --------------------------------------------------------------------------- 
>>>
>>>
>>>  I have Pentium D CPU, which many Windows utilities like cpuz, wcpuid,
>>>  everest identify as D 930 (Dual Core, 3GHz). From Intel site I find out
>>>  that it has no HT feature, nor Windows XP identify it as HT.
>>>
>>>  Why do I have "ht" flag in cpuinfo?
>>>  -----------------------------------
>>
>> The "ht" flag merely means "I know how to report hyperthreaded logical 
>> processors if I have them."  My Woodcrest Xeon 5110 and my Athlon64 X2 
>> both have the "ht" flag, and correctly report the zero hyperthreaded 
>> logical processors they each have.
>>
>>     -- Chris
>>
>>
> 
> Thanks, Chris.
> 
> Am I right that is chipset on mainboard, who is saying - "I know....", 
> not CPU itself? Is it better to switch off HT support in BIOS?
> Is it possible to generate CPU name as: "Pentium D 930" in /proc/cpuinfo?
> On the other server I have some 2GHz HT Xeons which can't be identified 
> on Intel site because of strange naming pattern.
> I tried to find any utility for Linux to solve this, but it looks like 
> everybody are using /pros/cpuinfo, which is not enough :)
> 
> Regards, Eugene.

While the BIOS can disable HT, the ability to *report* HT or the lack thereof is 
within the CPU itself.  I doubt my socket AM2 board has the faintest clue what 
HT is.

As for human-readable names, the model name in /proc/cpuinfo comes from the 
processor name string, and includes however much or little information the 
vendor thinks the lay user wants to know.  If you want more detail, you can 
always interpret the model, stepping, and other related fields to determine the 
model number.

If you don't want to manually look up model info on a CPU version table, x86info 
and dmidecode will give you more information.

	-- Chris

Re: ht CPU flag

[eugene]

On Fri, 18 May 2007, Chris Snook wrote:

> eugene@ns.armcci.am wrote:
>>
>>
>>  On Fri, 18 May 2007, Chris Snook wrote:
>> 
>> >  eugene@ns.armcci.am wrote:
>> > >   --------------------------------------------------------------------------- 
>> > > 
>> > > 
>> > >   I have Pentium D CPU, which many Windows utilities like cpuz, wcpuid,
>> > >   everest identify as D 930 (Dual Core, 3GHz). From Intel site I find 
>> > >   out
>> > >   that it has no HT feature, nor Windows XP identify it as HT.
>> > > 
>> > >   Why do I have "ht" flag in cpuinfo?
>> > >   -----------------------------------
>> > 
>> >  The "ht" flag merely means "I know how to report hyperthreaded logical 
>> >  processors if I have them."  My Woodcrest Xeon 5110 and my Athlon64 X2 
>> >  both have the "ht" flag, and correctly report the zero hyperthreaded 
>> >  logical processors they each have.
>> > 
>> >      -- Chris
>> > 
>> >
>>
>>  Thanks, Chris.
>>
>>  Am I right that is chipset on mainboard, who is saying - "I know....", not
>>  CPU itself? Is it better to switch off HT support in BIOS?
>>  Is it possible to generate CPU name as: "Pentium D 930" in /proc/cpuinfo?
>>  On the other server I have some 2GHz HT Xeons which can't be identified on
>>  Intel site because of strange naming pattern.
>>  I tried to find any utility for Linux to solve this, but it looks like
>>  everybody are using /pros/cpuinfo, which is not enough :)
>>
>>  Regards, Eugene.
>
> While the BIOS can disable HT, the ability to *report* HT or the lack thereof 
> is within the CPU itself.  I doubt my socket AM2 board has the faintest clue 
> what HT is.
>
> As for human-readable names, the model name in /proc/cpuinfo comes from the 
> processor name string, and includes however much or little information the 
> vendor thinks the lay user wants to know.  If you want more detail, you can 
> always interpret the model, stepping, and other related fields to determine 
> the model number.
>
> If you don't want to manually look up model info on a CPU version table, 
> x86info and dmidecode will give you more information.
>
> 	-- Chris
>

As I already told, there is no stepping 4 for Xeons on Intel site! So 
/proc/cpuinfo, dmidecode, x86info are all wrong.
Moreover, x86info is too old and can't work with /sys fs.

Also Linux is poor on giving FSB and Memory frequency, which I need to be 
sure that memory timing is correct (SPD gives slower parameters then are 
on SDRAM labels :( ).

Re: ht CPU flag

[H. Peter Anvin]

eugene@ns.armcci.am wrote:
> As I already told, there is no stepping 4 for Xeons on Intel site! So
> /proc/cpuinfo, dmidecode, x86info are all wrong.
> Moreover, x86info is too old and can't work with /sys fs.
> 
> Also Linux is poor on giving FSB and Memory frequency, which I need to
> be sure that memory timing is correct (SPD gives slower parameters then
> are on SDRAM labels :( ).

Linux doesn't give a hoot.  Setting up the memory is the BIOS' job.

I'm sure the authors of the tools can use a hand in updating them if
they are out of date.

	-hpa

Re: ht CPU flag

[eugene]

On Fri, 18 May 2007, H. Peter Anvin wrote:

> eugene@ns.armcci.am wrote:
>> As I already told, there is no stepping 4 for Xeons on Intel site! So
>> /proc/cpuinfo, dmidecode, x86info are all wrong.
>> Moreover, x86info is too old and can't work with /sys fs.
>>
>> Also Linux is poor on giving FSB and Memory frequency, which I need to
>> be sure that memory timing is correct (SPD gives slower parameters then
>> are on SDRAM labels :( ).
>
> Linux doesn't give a hoot.  Setting up the memory is the BIOS' job.
>
> I'm sure the authors of the tools can use a hand in updating them if
> they are out of date.
>
> 	-hpa
>

To authors of /proc/cpuinfo:

Plz, fix stepping identification:

cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 15
model           : 2
model name      : Intel(R) XEON(TM) CPU 2.00GHz
stepping        : 4
cpu MHz         : 2000.062
cache size      : 512 KB
physical id     : 0
siblings        : 2
core id         : 0
cpu cores       : 1
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        : 4006.39

processor       : 1
vendor_id       : GenuineIntel
cpu family      : 15
model           : 2
model name      : Intel(R) XEON(TM) CPU 2.00GHz
stepping        : 4
cpu MHz         : 2000.062
cache size      : 512 KB
physical id     : 0
siblings        : 2
core id         : 0
cpu cores       : 1
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        : 3999.80

processor       : 2
vendor_id       : GenuineIntel
cpu family      : 15
model           : 2
model name      : Intel(R) XEON(TM) CPU 2.00GHz
stepping        : 4
cpu MHz         : 2000.062
cache size      : 512 KB
physical id     : 3
siblings        : 2
core id         : 0
cpu cores       : 1
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        : 3999.98

processor       : 3
vendor_id       : GenuineIntel
cpu family      : 15
model           : 2
model name      : Intel(R) XEON(TM) CPU 2.00GHz
stepping        : 4
cpu MHz         : 2000.062
cache size      : 512 KB
physical id     : 3
siblings        : 2
core id         : 0
cpu cores       : 1
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        : 4000.08

---------------------------------------------------------------------------
uname -a
Linux rau 2.6.15-gentoo-r1 #5 SMP PREEMPT Sat Jan 20 10:13:08 AMT 2007 
i686 Intel(R) XEON(TM) CPU 2.00GHz GenuineIntel GNU/Linux

---------------------------------------------------------------------------

There is NO "stepping 4" for XEON, as Intel site states!

Thank you in advance.

Re: ht CPU flag

[Bernd Eckenfels]

In article <Pine.LNX.4.64.0705191328090.31481@ns.armcci.am> you wrote:
> To authors of /proc/cpuinfo:
> 
> Plz, fix stepping identification:

It might be a bug in the kernel code, however it is pretty unlikely: It just
reads what the CPU reports. You have an odd chip, nothing what the kernel
can do about. I see "family 15, model 2, stepping 4" usually beeing a P4
2.4GHz CPU. Maybe thats one of those relabling cases.

The information you list is a dual-cpu with HT configuration.

You could give us the output of any of your windows tools, but I am quite
sure they will report the same thing.

Greetings
Bernd

Re: ht CPU flag

[H. Peter Anvin]

eugene@ns.armcci.am wrote:
> 
> There is NO "stepping 4" for XEON, as Intel site states!
> 

Tough.  Reality disagrees.

	-hpa

Re: ht CPU flag

[eugene]

Thanks all for clear answers.
Take my apologies for my ignorance in Intel way to make CPUs.
>From now I'll use Intel recomendation to run their utility to resolve all 
questions.
Let me finish the thread.

Re: ht CPU flag

[Bernd Eckenfels]

In article <Pine.LNX.4.64.0705191704530.32247@ns.armcci.am> you wrote:
> From now I'll use Intel recomendation to run their utility to resolve all 
> questions.

You mean this, right?
 
http://www.intel.com/support/processors/sb/cs-001632.htm

Greetings
Bernd

Re: ht CPU flag

[Bernd Eckenfels]

In article <Pine.LNX.4.64.0705190023420.25685@ns.armcci.am> you wrote:
> Am I right that is chipset on mainboard, who is saying - "I know....", not 
> CPU itself?

It is a feature bitfield read directly from the CPU.

> Is it better to switch off HT support in BIOS?

The CPU will still report that flag. Might speed up the boot, not sure.

> Is it possible to generate CPU name as: "Pentium D 930" in /proc/cpuinfo?

No, cause those are marketing names, not reported by the CPU. You can only
lookup family, model and stepping with the vendors data sheets to get a
first impression of the possible chip.

Some chip cores get reconfigured by the vendor depending on QA (if it is not
able to cleanly process at high speed it gets sold as a slower chip).

> On the other server I have some 2GHz HT Xeons which can't be identified on 
> Intel site because of strange naming pattern.

Google for model and stepping.

Gruss
Bernd

Re: ht CPU flag

[H. Peter Anvin]

Bernd Eckenfels wrote:
> 
>> On the other server I have some 2GHz HT Xeons which can't be identified on 
>> Intel site because of strange naming pattern.
> 
> Google for model and stepping.
> 

... and use x86info.

	-hpa

Re: ht CPU flag

[eugene]

>> To authors of /proc/cpuinfo:
>>
>> Plz, fix stepping identification:

>It might be a bug in the kernel code, however it is pretty unlikely: It just
>reads what the CPU reports. You have an odd chip, nothing what the kernel
>can do about. I see "family 15, model 2, stepping 4" usually beeing a P4
>2.4GHz CPU. Maybe thats one of those relabling cases.

>The information you list is a dual-cpu with HT configuration.

>You could give us the output of any of your windows tools, but I am quite
>sure they will report the same thing.

>Greetings
>Bernd

The problem of getting windows utility output now is that it's the server 
in the datacenter, running 24/7 for many users. That's why I wanted to 
find out everything just in Linux.

This server is on Intel mainboard with Intel chassis, but it's not 
compatible with Win2k3, that's why they gave it to me for Linux :) 
Sometimes it's unstable under Linux too. I feel that the problem is in the 
hardware, may be CPU settings, cache etc... I set "acpi=ht pci=routeirq" 
to make it work at all. It has newest available BIOS installed.

Here is what Intel say about supported CPUs for this mobo:

SL623	2 GHz	N/A	N/A	400 MHz	0.13 micron	B0	512 KB 
Micro-FCPGA 0F24
SL5Z9	2 GHz	N/A	N/A	400 MHz	0.13 micron	B0	512 KB 
Micro-FCPGA 0F24
SL6YT	2 GHz	N/A	N/A	400 MHz	0.13 micron	D1	512 KB 
Micro-FCPGA 0F29h
SL6JY	2 GHz	N/A	N/A	400 MHz	0.13 micron	C1	512 KB 
Micro-FCPGA 0F27h

But none of them has HT! And my configuration - has! It's a puzzle for me.


>> From now I'll use Intel recomendation to run their utility to resolve all
>> questions.

>You mean this, right?

>http://www.intel.com/support/proces...b/cs-001632.htm

>Greetings
>Bernd

Yeah. Definitely Intel has some secret tables to identify CPU :)