From: biggy on
Hi Bob,

Thanks for your quick reply!

To answer your question yes indeed I am having problems and here are a
couple
examples, playing back of some HD file formats WMV media player system
stalls or I get lots of drop frames. Extremely long processing times
for convert mpeg files for HDTV (PVR) recordings to standard DVD.

Window HD WMV format playback requirements are:

Minimum Configuration
(to play 720p video)

... Windows XP
... Windows Media Player 9 Series
... 2.4 GHz processor or equivalent
... 384 MB of RAM
... 64 MB video card
... 1024 x 768 screen resolution
... 16-bit sound card
... Speakers

Optimum Configuration
(to play 1080p video with 5.1 surround sound)

... Windows XP
... Windows Media Player 10
... DirectX 9.0
... 3.0 GHz processor or equivalent
... 512 MB of RAM
... 128 MB video card
... 1920 x 1440 screen resolution
... 24-bit 96 kHz multichannel sound card
... 5.1 surround sound speaker system

As you can see HD windows media files (WMV) playback minimum
requirements are 2.4 Ghz processor. The preferred system is 3ghz or
more. I am sure that duo core processor will help with multi tasking,
but as you note the processing speed even with the E6600 is only 2.4
Ghz. That is why I was thinking of going to the single processor
Pentium® 4 -541, x64, 3.2-GHz @ 800Mhz w/ 1Mb (Socket 775). I could go
on & on but I think I have answer your question and I haven't even
started talking about Blue Ray. That's for super computer I hope to
build when the prices drop maybe next year when the quad processors are
online.

Sure wish I could brag about my poor little 2.1 Ghz xp2800 system.
Windows task manage is running out 100 % most of the time and between
frames dropping out and audio out of sync during play back and system
crashes. I am eager to get some good suggestions for my small upgrade!

Thanks again for your response!




Bob Fry wrote:
> >>>>> "big" == biggy <jpoy(a)sympatico.ca> writes:
>
> big> Greetings! I would like to upgrade my HTPC system lowest
> big> cost and get highest performance and use as many components
> big> from my present system.
> . . .
>
> big> What CPU would you select to get best performance at lowest
> big> overall cost? I don't mind paying more money if I really can
> big> get the true performance out of the 2 Duo core E6400
>
> Hmmm...I can't answer your direct question. But I am curious and have
> a question for you: what Home Theater functionality do you want that
> your current system is not giving you? Something deficient with the
> audio or video?
>
> --
> To me, it's always a good idea to always carry two sacks of something
> when you walk around. That way, if anybody says, "Hey, can you give me
> a hand?," you can say, "Sorry, got these sacks."
> - Jack Handey

From: Paul on
biggy wrote:
> Hi Bob,
>
> Thanks for your quick reply!
>
> To answer your question yes indeed I am having problems and here are a
> couple
> examples, playing back of some HD file formats WMV media player system
> stalls or I get lots of drop frames. Extremely long processing times
> for convert mpeg files for HDTV (PVR) recordings to standard DVD.
>
> Window HD WMV format playback requirements are:
>
> Minimum Configuration
> (to play 720p video)
>
> .. Windows XP
> .. Windows Media Player 9 Series
> .. 2.4 GHz processor or equivalent
> .. 384 MB of RAM
> .. 64 MB video card
> .. 1024 x 768 screen resolution
> .. 16-bit sound card
> .. Speakers
>
> Optimum Configuration
> (to play 1080p video with 5.1 surround sound)
>
> .. Windows XP
> .. Windows Media Player 10
> .. DirectX 9.0
> .. 3.0 GHz processor or equivalent
> .. 512 MB of RAM
> .. 128 MB video card
> .. 1920 x 1440 screen resolution
> .. 24-bit 96 kHz multichannel sound card
> .. 5.1 surround sound speaker system
>
> As you can see HD windows media files (WMV) playback minimum
> requirements are 2.4 Ghz processor. The preferred system is 3ghz or
> more. I am sure that duo core processor will help with multi tasking,
> but as you note the processing speed even with the E6600 is only 2.4
> Ghz. That is why I was thinking of going to the single processor
> Pentium� 4 -541, x64, 3.2-GHz @ 800Mhz w/ 1Mb (Socket 775). I could go
> on & on but I think I have answer your question and I haven't even
> started talking about Blue Ray. That's for super computer I hope to
> build when the prices drop maybe next year when the quad processors are
> online.
>
> Sure wish I could brag about my poor little 2.1 Ghz xp2800 system.
> Windows task manage is running out 100 % most of the time and between
> frames dropping out and audio out of sync during play back and system
> crashes. I am eager to get some good suggestions for my small upgrade!
>
> Thanks again for your response!
>

The Core2 Duo clock rate, should be multiplied by at least 1.5x, to get
a better representation of its effective clock rate. So a 2.4GHz Core2 Duo,
is more like 3.6Ghz. And rather than take my word for it, look at some
benchmarks, compared to some P4 processors.

First of all, with this benchmark, the Pentium D 840 Dual core 3.2GHz, has
the same benchmark results as a Pentium 4 540 (or a 640 for that matter).
The two red bars highlight the selected processors.
So this benchmark seems to only use one core, and allows comparing in
a limited way, the performance of one core of the two cores on a dual
processor. (Many of the benchmarks here, were selected to make dual cores
look good, so it is hard to compare single core performance.)

http://www23.tomshardware.com/cpu.html?modelx=33&model1=461&model2=446&chart=181

Now, highlight a E6400 2.13GHz dual core and the Pentium 4 570 3.8GHz single core.
The ratio of 3.8/2.13 is 1.78, so you get more than 1.5x the effective clock
rate. And this is comparing a single core to a single core.

http://www23.tomshardware.com/cpu.html?modelx=33&model1=458&model2=433&chart=181

In the next pair, first I start by comparing the E6400, to the Pentium D 960.
Both are dual cores, the E6400 is 2.13GHz dual and the D 960 is 3.6GHz dual.
The E6400 is ahead by a little this time, because there is no Pentium D family
dual core at 3.8GHz to compare to.

http://www23.tomshardware.com/cpu.html?modelx=33&model1=436&model2=433&chart=181

Keeping the same E6400 and Pentium D 960 processors, and then switching to a
dual core benchmark (in this case, synthetic comparison of integer and floating
point, where both cores are used). This compares dual core to dual core
performance, and on integer the E6400 (2.13GHz) is ahead of the D 960 (3.6GHz).
On float, they are pretty close to one another.

(Integer)
http://www23.tomshardware.com/cpu.html?modelx=33&model1=436&model2=433&chart=158
(Float)
http://www23.tomshardware.com/cpu.html?modelx=33&model1=436&model2=433&chart=159

The Pentium D 960 comes in 130W and a 95W version. The Core2 Duo family are
all rated at 65W.

For a motherboard, I'd be looking for a motherboard that properly supported
any power management features on the processor. At least, for an HTPC, you
want cooler operation when the processor is not busy. EIST and C1E are listed
here. It would take a lot more research though, to determine what is needed to
get it to actually work.

http://processorfinder.intel.com/details.aspx?sSpec=SL9S9

# Dual Core
# Enhanced Halt State (C1E) <--- C1E
# Enhanced Intel Speedstep� Technology <--- EIST
# Execute Disable Bit 1
# Intel� EM64T 2
# Intel� Thermal Monitor 2
# Intel� Virtualization Technology

This utility allows making some changes, but I don't know if this program alone
is enough to make it work.

http://cpu.rightmark.org/products/rmclock.shtml

So a Core2 Duo at "only 2.4GHz", is a lot of processor.

(Note: I had to set the Followup: field above, because the USENET server I use
demands it.)

Paul
From: biggy on
Hi Paul,

Wow, great response just what I was looking for. I also went to Tom’s
website but I wasn’t able to interpret the tests properly. Thanks to
you I understand it better :-)

As I have not used Intel processors for very long time (486DX2) and
only been using AMD single processors, I find it difficult enough task
comparing manufactures apples and oranges and with so many models of
processors (Dou, Qua and more) and new ones coming out it can be mind
boggling. Example the new Intel® Core™2 Duo E4300, 1.80-GHz @ 800Mhz
w/ 2Mb Cache (Socket 775) which I am considering along with the Asrock
MB. To make things worse getting the right balance between older
components and new can be very challenging. With main objective of
increasing the overall performance at minimal $$.

I check the Asrock MB and it looks like it does support (Enhanced Halt
State (C1E) and Enhanced Intel Speedstep® Technology

>From the information you have provided to me it looks like I am on the
right track. Here is some options I leaning towards.

Motherboard:

ASRock 775Dual-Vsta Socket 775, VIA PT880 Pro/Ultra Chipset w/ PCI
Express x16 & 8x AGP (ATX) $79 CAN

… Intel® Core™2 Duo E4300, 1.80-GHz @ 800Mhz w/ 2Mb Cache (Socket
775) $215
I have to wait to see if the MB supports this CPU, but I am pretty sure
it will. Looks like a good overclocker. Then again OClocking could be
a problem I want could keep the stock cooling fan and keep the overall
noise level low as possible.

… Intel® Core™2 Duo E6300, 1.86-GHz @ 1066Mhz w/ 2Mb Cache (Socket
775) $215

… Intel® Core™2 Duo E6600, 2.40-GHz @ 1066Mhz w/ 4Mb Cache (Socket
775) $380
I am not sure if this worth it and can my older DDR PC3200 make use of
the extra 2megs of cashe.

Anymore comments would be appreciated.

I like to thank everyone for their comments and a special thanks to
Paul for his very informative reply.

Jay






> The Core2 Duo clock rate, should be multiplied by at least 1.5x, to get
> a better representation of its effective clock rate. So a 2.4GHz Core2 Duo,
> is more like 3.6Ghz. And rather than take my word for it, look at some
> benchmarks, compared to some P4 processors.
>
> First of all, with this benchmark, the Pentium D 840 Dual core 3.2GHz, has
> the same benchmark results as a Pentium 4 540 (or a 640 for that matter).
> The two red bars highlight the selected processors.
> So this benchmark seems to only use one core, and allows comparing in
> a limited way, the performance of one core of the two cores on a dual
> processor. (Many of the benchmarks here, were selected to make dual cores
> look good, so it is hard to compare single core performance.)
>
> http://www23.tomshardware.com/cpu.html?modelx=33&model1=461&model2=446&chart=181
>
> Now, highlight a E6400 2.13GHz dual core and the Pentium 4 570 3.8GHz single core.
> The ratio of 3.8/2.13 is 1.78, so you get more than 1.5x the effective clock
> rate. And this is comparing a single core to a single core.
>
> http://www23.tomshardware.com/cpu.html?modelx=33&model1=458&model2=433&chart=181
>
> In the next pair, first I start by comparing the E6400, to the Pentium D 960.
> Both are dual cores, the E6400 is 2.13GHz dual and the D 960 is 3.6GHz dual.
> The E6400 is ahead by a little this time, because there is no Pentium D family
> dual core at 3.8GHz to compare to.
>
> http://www23.tomshardware.com/cpu.html?modelx=33&model1=436&model2=433&chart=181
>
> Keeping the same E6400 and Pentium D 960 processors, and then switching to a
> dual core benchmark (in this case, synthetic comparison of integer and floating
> point, where both cores are used). This compares dual core to dual core
> performance, and on integer the E6400 (2.13GHz) is ahead of the D 960 (3.6GHz).
> On float, they are pretty close to one another.
>
> (Integer)
> http://www23.tomshardware.com/cpu.html?modelx=33&model1=436&model2=433&chart=158
> (Float)
> http://www23.tomshardware.com/cpu.html?modelx=33&model1=436&model2=433&chart=159
>
> The Pentium D 960 comes in 130W and a 95W version. The Core2 Duo family are
> all rated at 65W.
>
> For a motherboard, I'd be looking for a motherboard that properly supported
> any power management features on the processor. At least, for an HTPC, you
> want cooler operation when the processor is not busy. EIST and C1E are listed
> here. It would take a lot more research though, to determine what is needed to
> get it to actually work.
>
> http://processorfinder.intel.com/details.aspx?sSpec=SL9S9
>
> # Dual Core
> # Enhanced Halt State (C1E) <--- C1E
> # Enhanced Intel Speedstep® Technology <--- EIST
> # Execute Disable Bit 1
> # Intel® EM64T 2
> # Intel® Thermal Monitor 2
> # Intel® Virtualization Technology
>
> This utility allows making some changes, but I don't know if this program alone
> is enough to make it work.
>
> http://cpu.rightmark.org/products/rmclock.shtml
>
> So a Core2 Duo at "only 2.4GHz", is a lot of processor.
>
> (Note: I had to set the Followup: field above, because the USENET server I use
> demands it.)
>
> Paul

From: CJT on
Paul wrote:

<snip>
> The Core2 Duo clock rate, should be multiplied by at least 1.5x, to get
> a better representation of its effective clock rate. So a 2.4GHz Core2 Duo,
> is more like 3.6Ghz. And rather than take my word for it, look at some
> benchmarks, compared to some P4 processors.
>
> First of all, with this benchmark, the Pentium D 840 Dual core 3.2GHz, has
> the same benchmark results as a Pentium 4 540 (or a 640 for that matter).
> The two red bars highlight the selected processors.
> So this benchmark seems to only use one core, and allows comparing in
> a limited way, the performance of one core of the two cores on a dual
> processor. (Many of the benchmarks here, were selected to make dual cores
> look good, so it is hard to compare single core performance.)
>
> http://www23.tomshardware.com/cpu.html?modelx=33&model1=461&model2=446&chart=181
>
<snip>

What I find interesting about that chart is that ALL of the processors
on it are within a factor of two in speed (i.e. hardly different at all
in my book, since most tasks that use huge amounts of computational
power can be divided up to use multiple processors). But the cheapest
is MUCH cheaper than the most expensive. It seems about time to extend
the RAID concept to processors -- perhaps RAIP? ;-)

--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form che...(a)prodigy.net.
From: Paul on
CJT wrote:
> Paul wrote:
>
> <snip>
>> The Core2 Duo clock rate, should be multiplied by at least 1.5x, to get
>> a better representation of its effective clock rate. So a 2.4GHz Core2
>> Duo,
>> is more like 3.6Ghz. And rather than take my word for it, look at some
>> benchmarks, compared to some P4 processors.
>>
>> First of all, with this benchmark, the Pentium D 840 Dual core 3.2GHz,
>> has
>> the same benchmark results as a Pentium 4 540 (or a 640 for that matter).
>> The two red bars highlight the selected processors.
>> So this benchmark seems to only use one core, and allows comparing in
>> a limited way, the performance of one core of the two cores on a dual
>> processor. (Many of the benchmarks here, were selected to make dual cores
>> look good, so it is hard to compare single core performance.)
>>
>> http://www23.tomshardware.com/cpu.html?modelx=33&model1=461&model2=446&chart=181
>>
> <snip>
>
> What I find interesting about that chart is that ALL of the processors
> on it are within a factor of two in speed (i.e. hardly different at all
> in my book, since most tasks that use huge amounts of computational
> power can be divided up to use multiple processors). But the cheapest
> is MUCH cheaper than the most expensive. It seems about time to extend
> the RAID concept to processors -- perhaps RAIP? ;-)
>

But there are processors that support that. Opterons and Xeons.
Stuff with a server/workstation rating here, is likely to work in
dual (or more) socket boards.

http://www.intel.com/intel/finance/pricelist/processor_price_list.pdf

http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_609,00.html?redir=CPT301

*******
The Opteron interconnect is via Hypertransport. Rather than a single
bus, there are links between processors. An Opteron has three links,
and the max buildout is eight processors. The following picture shows
the wiring pattern for 4-way NUMA. When a processor tries to reach
"distant" memory, the access time is longer. Thus, memory access is
"non-uniform". When you buy an Opteron of the right type, you are
paying extra for the "coherent" links provided on the chip. Cache
coherency, means that all the processors have consistent info stored
in their local caches.

http://www.ixbt.com/cpu/rmma/numa/numa4.png

http://en.wikipedia.org/wiki/Hypertransport

"Multiprocessor interconnect

Another use for HyperTransport is as an interconnect for NUMA
multiprocessor computers. AMD uses HyperTransport with a
proprietary cache coherency extension as part of their
Direct Connect Architecture in their Opteron and Athlon64 line
of processors."

*******
The Xeon uses a common bus, and processing is symmetric, as each
processor has equal access speed to memory. But the common bus
can also be a choke point, and a potential reason to bump up
the cache on the Xeon. (I suspect this figure is now too simplistic,
and doesn't represent what is happening in the latest chips. This
figure is probably most accurate for older single core Xeons.)

http://www.ixbt.com/cpu/rmma/numa/smp.png

http://en.wikipedia.org/wiki/Xeon

"...except for the addition of SMP support, which lets Dempsey
operate in dual-processor systems"

*******
http://en.wikipedia.org/wiki/Symmetric_multiprocessing (Xeon)
http://en.wikipedia.org/wiki/Non-Uniform_Memory_Access (Opteron)

Both of which might be considered "tightly coupled cluster computing".
Only certain classes of problems can be sliced up into a large number
of equal pieces. The last benchmark article I saw for the latest Xeons,
the guys doing the benchmarks had trouble finding enough applications
that could actually scale to use all the cores on their fancy system.
Which is why, at the present time, buying more than a dual core may
be bad economics, for desktop users. It can take a while, for the
software to catch up with the hardware.

http://en.wikipedia.org/wiki/Cluster_computing

Paul