Re: Pentium 805D has an interesting surprise

[Bill McGonigle]

On Apr 25, 2007, at 15:16, Ben Scott wrote:


 Who designed this thing, Mel Brooks?  ;-)


Nah, he'd favor the "We Can Do It" approach of reporting actual  
capabilities.


The problem is you mind is a-glow with whirling transient nodes of  
thought, careening through a cosmic vapor of invention.


-Bill

-
Bill McGonigle, Owner   Work: 603.448.4440
BFC Computing, LLC  Home: 603.448.1668
[EMAIL PROTECTED]   Cell: 603.252.2606
http://www.bfccomputing.com/Page: 603.442.1833
Blog: http://blog.bfccomputing.com/
VCard: http://bfccomputing.com/vcard/bill.vcf

___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Bill McGonigle]

On Apr 25, 2007, at 17:01, Bob King wrote:


That seems a bit odd.So you are saying the flag indicates only that it
understands that flag, rather than has that capability?


yes.


Wouldn't it be more
useful to show what capabilities it HAS rather than those it knows  
about?


yes.


-Bill
-
Bill McGonigle, Owner   Work: 603.448.4440
BFC Computing, LLC  Home: 603.448.1668
[EMAIL PROTECTED]   Cell: 603.252.2606
http://www.bfccomputing.com/Page: 603.442.1833
Blog: http://blog.bfccomputing.com/
VCard: http://bfccomputing.com/vcard/bill.vcf

___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Thomas Charron]

On 4/26/07, Thomas Charron <[EMAIL PROTECTED]> wrote:

  Look for siblings : 2 in the cpuinfo.  If it has the HT flag, but
siblings is 1, it answers the question.
  Think of it as a matrix question.  'Just remember, there is no
spoon, err, HT.'


 As an addendum, the lack of a siblings line also means no hyperthreading.

--
-- Thomas
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Thomas Charron]

On 4/25/07, Bob King <[EMAIL PROTECTED]> wrote:

On 4/25/07, Bill McGonigle <[EMAIL PROTECTED]> wrote:
> Right, 'ht' just reports whether the CPU knows it has hyperthreading
> or knows it doesn't have hyperthreading.  This tripped me up for a
> good hour once!
> I have a 2.5GHz P4 here without hyperthreading, but it knows that it
> doesn't:
>$cat /proc/cpuinfo
>flags   : fpu vme de pse tsc msr pae mce cx8 apic mtrr pge
> mca cmov pat
>pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe up cid xtpr
 That seems a bit odd.So you are saying the flag indicates only that it
understands that flag, rather than has that capability? Wouldn't it be more
useful to show what capabilities it HAS rather than those it knows about?


 The flag itself means you can ask the question.  :-)

 Look for siblings : 2 in the cpuinfo.  If it has the HT flag, but
siblings is 1, it answers the question.

 Think of it as a matrix question.  'Just remember, there is no
spoon, err, HT.'

--
-- Thomas
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Ben Scott]

On 4/25/07, Bob King <[EMAIL PROTECTED]> wrote:

So you are saying the flag indicates only that it understands that flag,
rather than has that capability? Wouldn't it be more useful to show
what capabilities it HAS rather than those it knows about?


 Our mistake is in assuming that usefulness was a design criterion.

-- Ben
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Jarod Wilson]

On Apr 25, 2007, at 22:24, Paul Lussier wrote:


<[EMAIL PROTECTED]> writes:


Actually it looks like the newer Xeons (dual and quad cores) don't
support HT.

I googled and found this comment:

"Hyperthreading is only on the NetBurst based 50xx Xeons - the Core2
based 51xx Xeons do not support hyperthreading"

The same is apparently true of the 5300 quad-core Xeon.  So much for
my hope that the new 3GHz eight-core MacPro would hyperthread to
give 16 virtual CPUs :-(


How does one tell wether the Xeon is NetBurst 50xx-based, or Core2
51xx-based, or 5300-based?


Should be able to tell for the most part based on cpu family, model  
and stepping:


http://en.wikipedia.org/wiki/Procfs#.2Fcpuinfo

(as well as core count and common sense)

Based on that data, I was able to verify that my dual dual-core  
system is indeed an older netburst 5000-series, not a 5100 series.  
Oh, and quad-core chips are of the 6/15/x variety as well.


--
Jarod Wilson
[EMAIL PROTECTED]




PGP.sig
Description: This is a digitally signed message part
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Paul Lussier]

<[EMAIL PROTECTED]> writes:

> Actually it looks like the newer Xeons (dual and quad cores) don't
> support HT.
>
> I googled and found this comment: 
>
> "Hyperthreading is only on the NetBurst based 50xx Xeons - the Core2
> based 51xx Xeons do not support hyperthreading"
>
> The same is apparently true of the 5300 quad-core Xeon.  So much for
> my hope that the new 3GHz eight-core MacPro would hyperthread to
> give 16 virtual CPUs :-(

How does one tell wether the Xeon is NetBurst 50xx-based, or Core2
51xx-based, or 5300-based?
-- 
Seeya,
Paul
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Bob King]

On 4/25/07, Bill McGonigle <[EMAIL PROTECTED]> wrote:


Right, 'ht' just reports whether the CPU knows it has hyperthreading
or knows it doesn't have hyperthreading.  This tripped me up for a
good hour once!

I have a 2.5GHz P4 here without hyperthreading, but it knows that it
doesn't:

   $cat /proc/cpuinfo
   flags   : fpu vme de pse tsc msr pae mce cx8 apic mtrr pge
mca cmov pat
   pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe up cid xtpr



That seems a bit odd.So you are saying the flag indicates only that it
understands that flag, rather than has that capability? Wouldn't it be more
useful to show what capabilities it HAS rather than those it knows about?
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Jarod Wilson]

On Wednesday 25 April 2007 15:19:48 [EMAIL PROTECTED] wrote:
> Actually it looks like the newer Xeons (dual and quad cores) don't support
> HT.
>
> I googled and found this comment:
> "Hyperthreading is only on the NetBurst based 50xx Xeons - the Core2 based
> 51xx Xeons do not support hyperthreading"

Hrm. That would indicate that I'm mistaken, and the cpus in my workstation are 
50xx-series, not 51xx. Entirely likely.

> The same is apparently true of the 5300 quad-core Xeon.  So much for my
> hope that the new 3GHz eight-core MacPro would hyperthread to give 16
> virtual CPUs :-(

I can definitely confirm that my core 2 quad system has no HT available to 
turn on in the bios (but does show ht in the cpuflags, which aligns with 
other comments in the thread).

-- 
Jarod Wilson
[EMAIL PROTECTED]
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[mike ledoux]

On Wed, Apr 25, 2007 at 02:15:55PM -0400, Paul Lussier wrote:
> <[EMAIL PROTECTED]> writes:
> 
> > There are also a few dual cores with hyperthreading, Xeon was one
> > product line I noticed had such models.
> 
> Yeah, I just checked one of our new machines which is a dual
> dual-core.  It reports 4 "GenuineIntel" CPUs in /proc/cpuinfo, all of
> which have the 'ht' flag set. 
> 
> So, in theory, if I were to enable HT in my kernel, I might well see 8 CPUs ?

Unless the 'ht' flag is set on dual-core CPUs that don't actually
support it, yes. /proc/cpuinfo on the system I'm working on now (two
3.0GHz Xeon dual-core hyperthreading CPUs) reports 8 CPUs.

-- 
[EMAIL PROTECTED]  OpenPGP KeyID 0x57C3430B
Holder of Past Knowledge   CS, O-
"Fnord is neither a particle nor a wave."

___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[bmcculley]

Actually it looks like the newer Xeons (dual and quad cores) don't support HT.  

I googled and found this comment: 
"Hyperthreading is only on the NetBurst based 50xx Xeons - the Core2 based 51xx 
Xeons do not support hyperthreading" 

The same is apparently true of the 5300 quad-core Xeon.  So much for my hope 
that the new 3GHz eight-core MacPro would hyperthread to give 16 virtual CPUs 
:-(

Guess we're both sol.

-brucem

 Original message 
>Date: Wed, 25 Apr 2007 14:15:55 -0400
>From: Paul Lussier <[EMAIL PROTECTED]>  
>Subject: Re: Pentium 805D has an interesting surprise  
>To: Greater NH Linux User Group 
>
><[EMAIL PROTECTED]> writes:
>
>> There are also a few dual cores with hyperthreading, Xeon was one
>> product line I noticed had such models.
>
>Yeah, I just checked one of our new machines which is a dual
>dual-core.  It reports 4 "GenuineIntel" CPUs in /proc/cpuinfo, all of
>which have the 'ht' flag set. 
>
>So, in theory, if I were to enable HT in my kernel, I might well see 8 CPUs ?
>
>-- 
>Seeya,
>Paul
>___
>gnhlug-discuss mailing list
>gnhlug-discuss@mail.gnhlug.org
>http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Ben Scott]

On 4/25/07, Bill McGonigle <[EMAIL PROTECTED]> wrote:

Right, 'ht' just reports whether the CPU knows it has hyperthreading
or knows it doesn't have hyperthreading.  This tripped me up for a
good hour once!


 So the HT flag does not indicate HyperThreading capability.  Rather,
the HT flag indicates whether the actual mechanism for indicating the
presence of HyperThreading is present.

 Who designed this thing, Mel Brooks?  ;-)

-- Ben
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Jarod Wilson]

On Wednesday 25 April 2007 13:56:35 Ben Scott wrote:
> On 4/25/07, Thomas Charron <[EMAIL PROTECTED]> wrote:
> > Yes, this is exactly why I say the differences from a software
> > perspective are debatable.
>
>   In the sense of what you have to program to make use of it, yes.
> Either way, you have to worry about shared memory concurrency,
> locking, re-entrance, all that stuff.  But the performance
> characteristics are quite different.
>
>   Dual core and quad core is just SMP (symmetric multi-processor)
> implemented in a single IC package.  It used to be this was done at
> the motherboard level -- two sockets, two IC packages, and special SMP
> chipsets.  Dual core just fits everything into a single IC package,
> and makes the SMP chipset standard.  A single dual-core processor is
> "exactly" the same thing as a conventional dual processor system.
>
>   (I put "exactly" in quotes because there are some differences.  For
> example, I understand some cache sharing is possible on some chips.
> But it's really really really close to classic x86 SMP.)

There's another big difference in the AMD64 world that should be noted. Since 
the memory controller is on-chip, a multi-socket opteron system winds up 
being NUMA-capable (multiple independent memory controllers and banks of 
dimms directly accessible by a specific socket), whereas a 
single-socket/multi-core opteron/athlon64 isn't NUMA-capable, all cores share 
direct access to the same memory.

-- 
Jarod Wilson
[EMAIL PROTECTED]
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Jarod Wilson]

On Wednesday 25 April 2007 14:15:55 Paul Lussier wrote:
> <[EMAIL PROTECTED]> writes:
> > There are also a few dual cores with hyperthreading, Xeon was one
> > product line I noticed had such models.
>
> Yeah, I just checked one of our new machines which is a dual
> dual-core.  It reports 4 "GenuineIntel" CPUs in /proc/cpuinfo, all of
> which have the 'ht' flag set.
>
> So, in theory, if I were to enable HT in my kernel, I might well see 8 CPUs

Enable HT in kernel, and/or BIOS. The box I'm typing this email on has a pair 
of 3GHz dual-core woodcrest xeon procs in it, with HT enabled in the bios 
(and kernel), and I get 8 logical cpus.


-- 
Jarod Wilson
[EMAIL PROTECTED]
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Bill McGonigle]

On Apr 25, 2007, at 12:53, Shawn K. O'Shea wrote:


""ht" in 'flags' field of /proc/cpuinfo indicate that the processor
supports the Machine Specific
Registers to report back HT or multi-core capability. Additional
fields (listed down below) in the
CPU records of /proc/cpuinfo will give more precise information about
the CPU topology as seen
by the operating system. "


Right, 'ht' just reports whether the CPU knows it has hyperthreading  
or knows it doesn't have hyperthreading.  This tripped me up for a  
good hour once!


I have a 2.5GHz P4 here without hyperthreading, but it knows that it  
doesn't:


  $cat /proc/cpuinfo
  processor   : 0
  vendor_id   : GenuineIntel
  cpu family  : 15
  model   : 2
  model name  : Intel(R) Pentium(R) 4 CPU 2.50GHz
  stepping: 9
  cpu MHz : 2500.176
  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 mtrr pge  
mca cmov pat

  pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe up cid xtpr
  bogomips: 5002.39
  clflush size: 64

It's not new enough to know that it doesn't have vt, so there's no vt  
flag.


-Bill

-
Bill McGonigle, Owner   Work: 603.448.4440
BFC Computing, LLC  Home: 603.448.1668
[EMAIL PROTECTED]   Cell: 603.252.2606
http://www.bfccomputing.com/Page: 603.442.1833
Blog: http://blog.bfccomputing.com/
VCard: http://bfccomputing.com/vcard/bill.vcf

For fastest support contact, please follow:
http://bfccomputing.com/support_contact.html


___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Bob King]

On 4/25/07, Shawn K. O'Shea <[EMAIL PROTECTED]> wrote:


""ht" in 'flags' field of /proc/cpuinfo indicate that the processor
supports the Machine Specific
Registers to report back HT or multi-core capability. Additional
fields (listed down below) in the
CPU records of /proc/cpuinfo will give more precise information about
the CPU topology as seen
by the operating system. "

http://oss.intel.com/pdf/mclinux.pdf



The doc indicates that the ht flag is kept for backwards compatibility (for
licensing, etc). Now the mystery is solved.

The dual-core Xeon family does have HT available as well, so you can get 2
real cpu cores each with SMT capability. Of course, you need well structured
software to take advantage of that, but that was a given for everything from
HT onward!

Thanks for the help!
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Paul Lussier]

<[EMAIL PROTECTED]> writes:

> There are also a few dual cores with hyperthreading, Xeon was one
> product line I noticed had such models.

Yeah, I just checked one of our new machines which is a dual
dual-core.  It reports 4 "GenuineIntel" CPUs in /proc/cpuinfo, all of
which have the 'ht' flag set. 

So, in theory, if I were to enable HT in my kernel, I might well see 8 CPUs ?

-- 
Seeya,
Paul
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Ben Scott]

On 4/25/07, Thomas Charron <[EMAIL PROTECTED]> wrote:

Yes, this is exactly why I say the differences from a software
perspective are debatable.


 In the sense of what you have to program to make use of it, yes.
Either way, you have to worry about shared memory concurrency,
locking, re-entrance, all that stuff.  But the performance
characteristics are quite different.

 Dual core and quad core is just SMP (symmetric multi-processor)
implemented in a single IC package.  It used to be this was done at
the motherboard level -- two sockets, two IC packages, and special SMP
chipsets.  Dual core just fits everything into a single IC package,
and makes the SMP chipset standard.  A single dual-core processor is
"exactly" the same thing as a conventional dual processor system.

 (I put "exactly" in quotes because there are some differences.  For
example, I understand some cache sharing is possible on some chips.
But it's really really really close to classic x86 SMP.)

 HyperThreading(TM) is Intel's trademark for SMT (simultaneous
multi-threading).  SMT gives you multiple register files, but re-uses
all the other core logic.  Intel was so interested in SMT on the
Pentium 4 because the P4 has a really looong instruction processing
pipeline.  Keeping that pipeline full of operations was critical to
obtaining good performance on that chip.  By keeping multiple
execution states "hot" in the core, that was more likely to happen.
Ultimately, though, it could still only do one thing at a time.

 Assuming you have good parallelism in your workload (either one
multi-threading app, or multiple separate apps), SMP should "always"
yield much better performance vs SMT.  Dual core has two of
everything; SMT has only one of almost everything.  In particular,
since SMT only has one set of caches, anything that hits the CPU
caches hard can really suffer under SMT.

 Under a workload with with little to no multi-threading, SMT vs SMP
vs just-a-processor is kind of moot; only one running thread really
matters.  Popular wisdom is that most single-user workloads still
correspond to that model.

 Now that Intel has hit a serious wall with clock speed and
diminishing returns, they have decreed multi-core will be the way of
the future (in other news, the sky is blue).  Supposedly, programmers
everywhere are supposed to be changing their programming models to be
more parallel, to take advantage of mutli-core designs, so all those
single-user workloads will benefit from having 42 cores under the
hood.  It remains to be seen if that will actually happen.

-- Ben
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Shawn K. O'Shea]

I was wondering if anyone had heard anything about this? If the cpu really
supports hyperthreading then I might want to invest in a new mobo for that
box.


This got me curiously googling around and I found this in an Intel doc
on the Linux kernel:

""ht" in 'flags' field of /proc/cpuinfo indicate that the processor
supports the Machine Specific
Registers to report back HT or multi-core capability. Additional
fields (listed down below) in the
CPU records of /proc/cpuinfo will give more precise information about
the CPU topology as seen
by the operating system. "

http://oss.intel.com/pdf/mclinux.pdf

-Shawn
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Thomas Charron]

On 4/25/07, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:

also, the table on page 18 of a good review of the D 805 on Tom's Hardware shows the 
D 805 has no HT implementation, but comments on the equivalency of dual-core.  Find 
it at: 


 Yes, this is exactly why I say the differences from a software
perspective are debatable.  They give basically the same thing, ones
just REALLY doing it on two cores, the other, a single core making
pretend it can really doing it, but 'emulating' it.  Some things it
can actually do 'almost in parallel', others, it doesn't do in
parallel, but the software doesn't know, it just knows two tasks took
twice as long as they could have had they a real dedicated cores.

--
-- Thomas
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Bob King]

On 4/25/07, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:


That article also says the D 805 looks like a great overclocker, the
biggest reason I prefer my Core2Duo is power consumption, but with the 805
in hand I might look for a good overclocking mobo to take full advantage of
it, if that's your interest.



That is why I picked up that CPU. The mobo it is on will go from a 533 FSB
up to 1066 , and is fairly flexible. Given that the cpu multiplier is 20
(providing 2.66GHz for that 533MHZ FSB) it could be very interesting. I
don't want to go up to water cooling, but should be able to ratchet the
speed up a fair amount just using fans.

Hope this is helpful!


Yes, it was very helpful. That was more or less what I had thought, but
figured it was worth asking.
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Bob King]

On 4/25/07, Thomas Charron <[EMAIL PROTECTED]> wrote:


>   The differences between Dualcore and Hyperthreading are debatable.
> Is it possible that whatever reported the HT bit simply did so because
> the capabilities from a software perspective are so simular?

  After rereading, I anted to clarify.  Dualcore is better then
Hyperthreading.  A HT processor actually shows as two logical
processors, when there are actually only one that's capable of quickly
switching between two things, looking from the outside like it's
actually doing two at once.



I understood what you meant, and yes Dualcore is better than HT, but I
wanted both -- Just being greedy I guess.

The thought that the HT flag was set for a dualcore was what I initially
thought, but there are differences between a virtual cpu and a real one, so
I was uncertain.
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[bmcculley]

There are also a few dual cores with hyperthreading, Xeon was one product line 
I noticed had such models.

Logically, the HT tech would add some performance at the expense of some added 
hardware logic, dual cores adds more performance but takes more silicon, dual 
cores with HT take the most silicon and give the most performance.  Empirically 
the relative increase in performance and resource costs seems to favor dual 
core, just from observing the current product lineups.  It's not clear if that 
might change as new technologies evolve.

-Brucem

 Original message 
>Date: Wed, 25 Apr 2007 11:08:04 -0400
>From: "Thomas Charron" <[EMAIL PROTECTED]>  
>Subject: Re: Pentium 805D has an interesting surprise  
>To: "Bob King" <[EMAIL PROTECTED]>
>Cc: Greater NH Linux User Group 
>
>On 4/25/07, Thomas Charron <[EMAIL PROTECTED]> wrote:
>> > I was wondering if anyone had heard anything about this? If the cpu really
>> > supports hyperthreading then I might want to invest in a new mobo for that
>> > box.
>>   The differences between Dualcore and Hyperthreading are debatable.
>> Is it possible that whatever reported the HT bit simply did so because
>> the capabilities from a software perspective are so simular?
>
>  After rereading, I anted to clarify.  Dualcore is better then
>Hyperthreading.  A HT processor actually shows as two logical
>processors, when there are actually only one that's capable of quickly
>switching between two things, looking from the outside like it's
>actually doing two at once.
>
>  A Dual-core processor has two real physical cores.  A hyperthreaded
>processor can only do two things at the same time IN CERTAIN CASES.
>In other cases, even though it looks like 2, it's simply switching
>tasks and one core, doing twice as much work, but then taking twice as
>long to do it, negating the performance gain.  It really all depends
>on what the processor is doing.  Two threads making heavy use of
>floating point math = might as well just call it normal non
>hyperthreaded.  Two applications doing simple stuff, it can do then
>basically at the same time on one core, making it look like it was
>actually two cores doing the work.
>
>-- 
>-- Thomas
>___
>gnhlug-discuss mailing list
>gnhlug-discuss@mail.gnhlug.org
>http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[bmcculley]

Interesting question, got me wondering if my Core 2 Duo cpu would look the 
same.  Haven't had the chance to check it yet, but I did some reading and found 
that it seems neither the Pentium D 805 or my Core 2 Duo 6400 have hyperthread 
technology although they're both dual-cores.

I haven't finished studying the Intel® 64 and IA-32 Architecture manuals yet, 
but suspect the software developers reference manual hints at the explanation:  
"The Intel Pentium D processor also features multi-core technology. This 
processor provides hardware multi-threading support with two processor cores 
but does not offer Hyper-Threading Technology. This means that the Intel 
Pentium D processor provides two logical processors in a physical package, with 
each logical processor owning the complete execution resources of a processor 
core."

At a SWAG the HT bit is set because there is more than one logical processor, 
in the case of the D 805 (and my Core2Duo) they are each a physical core but 
for some reason are logically equivalent to a single core with HTT, at least 
enough to justify setting the flag bit that way.  It could be that the flag is 
set in bios or os code that verifies more than one logical processor without 
distingushing physical cores from HTT logic.

ref p.52 of the Intel manual: 


also, the table on page 18 of a good review of the D 805 on Tom's Hardware 
shows the D 805 has no HT implementation, but comments on the equivalency of 
dual-core.  Find it at: 
  

That article also says the D 805 looks like a great overclocker, the biggest 
reason I prefer my Core2Duo is power consumption, but with the 805 in hand I 
might look for a good overclocking mobo to take full advantage of it, if that's 
your interest.

Hope this is helpful!

-Bruce

 Original message 
>Date: Wed, 25 Apr 2007 10:14:05 -0400
>From: "Bob King" <[EMAIL PROTECTED]>  
>Subject: Pentium 805D has an interesting surprise  
>To: "Greater NH Linux User Group" 
>
>   During the Xen discussion at MerriLUG, it was
>   mentioned that the Xen implementation on FC6
>   required PAE on the cpu. I knew that my two Linux
>   boxen at home did not support VT, so I checked
>   the processors for the PAE flag.
>
>   My Xeon system had PAE and also supported
>   hyperthreading, which was hardly a surprise.
>
>   However, when I checked my Pentium D 805 system, it
>   showed PAE and HT as well. I new the 805 was a
>   dual-core, but was surprised to see the HT flag. I
>   cannot find anything on Intel's websites that talk
>   about HT support, nor does the bios for that systems
>   motherboard support it for me to be able to test it.
>   My initial attempts at googling for info have been
>   inconclusive. Most references to the cpu flags are
>   discussing other flags, not the 'ht' flag which is
>   shown but not explicitly discussed.
>
>   I was wondering if anyone had heard anything about
>   this? If the cpu really supports hyperthreading then
>   I might want to invest in a new mobo for that box.
>>___
>gnhlug-discuss mailing list
>gnhlug-discuss@mail.gnhlug.org
>http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Thomas Charron]

On 4/25/07, Thomas Charron <[EMAIL PROTECTED]> wrote:

> I was wondering if anyone had heard anything about this? If the cpu really
> supports hyperthreading then I might want to invest in a new mobo for that
> box.
  The differences between Dualcore and Hyperthreading are debatable.
Is it possible that whatever reported the HT bit simply did so because
the capabilities from a software perspective are so simular?


 After rereading, I anted to clarify.  Dualcore is better then
Hyperthreading.  A HT processor actually shows as two logical
processors, when there are actually only one that's capable of quickly
switching between two things, looking from the outside like it's
actually doing two at once.

 A Dual-core processor has two real physical cores.  A hyperthreaded
processor can only do two things at the same time IN CERTAIN CASES.
In other cases, even though it looks like 2, it's simply switching
tasks and one core, doing twice as much work, but then taking twice as
long to do it, negating the performance gain.  It really all depends
on what the processor is doing.  Two threads making heavy use of
floating point math = might as well just call it normal non
hyperthreaded.  Two applications doing simple stuff, it can do then
basically at the same time on one core, making it look like it was
actually two cores doing the work.

--
-- Thomas
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Pentium 805D has an interesting surprise

[Thomas Charron]

On 4/25/07, Bob King <[EMAIL PROTECTED]> wrote:

During the Xen discussion at MerriLUG, it was mentioned that the Xen
implementation on FC6 required PAE on the cpu. I knew that my two Linux
boxen at home did not support VT, so I checked the processors for the PAE
flag.

My Xeon system had PAE and also supported hyperthreading, which was hardly a
surprise.

However, when I checked my Pentium D 805 system, it showed PAE and HT as
well. I new the 805 was a dual-core, but was surprised to see the HT flag. I
cannot find anything on Intel's websites that talk about HT support, nor
does the bios for that systems motherboard support it for me to be able to
test it. My initial attempts at googling for info have been inconclusive.
Most references to the cpu flags are discussing other flags, not the 'ht'
flag which is shown but not explicitly discussed.

I was wondering if anyone had heard anything about this? If the cpu really
supports hyperthreading then I might want to invest in a new mobo for that
box.


 The differences between Dualcore and Hyperthreading are debatable.
Is it possible that whatever reported the HT bit simply did so because
the capabilities from a software perspective are so simular?

 Now a dualcore hyperthreading processor is both of them combined,
leading to 4 logical processors.  Is that what you where hoping it
meant?  :-)

--
-- Thomas
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/