AMD or INTEL ?

  • Thread starter Thread starter Codemutant
  • Start date Start date
Tony said:
?!?! do you think that paying less means higher reliability?!

Quite often it does. One must do some research.
The simple fact of the matter is that the most business users do NOT
buy PCs with raw performance as their #1 concern. They'll easily give
up 10% in performance for a more reliable machine any day, AND they'll
spend a little bit extra for it.

What if they pay much more, and get less reliability?
Even in business software, it just depends on what benchmark you look
at. From the very same review that you keep quoting, check out the
SYSMark Office Productivity benchmark:

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2065&p=5

Sysmark is Bapco benchmark. They have Dragon Naturally Speaking,
and Winzip added to the mix. I don't know anyone who uses those.
Past versions of Sysmark have been very controversial to say the least.
Therefore imo Bapco benchmarks should be ignored.

http://www.vanshardware.com/articles/2001/august/010814_Intel_SysMark/010814_Intel_SysMark.htm

http://www.theinquirer.net/?article=5274
http://www.lostcircuits.com/cpu/northwood/6.shtml

http://www.lostcircuits.com/cpu/prescott/8.shtml
Here the P4 3.2GHz is a full 30% faster than the AthlonXP 3000+. In
your Business Winstone 2004 test the AthlonXP 3000+ is 1% faster than
the P4 3.2GHz. So which benchmark is more better and more applicable
to the applications that Joe business-user runs?! And why?

See above.
 
Sysmark is Bapco benchmark. They have Dragon Naturally Speaking,
and Winzip added to the mix. I don't know anyone who uses those.
Past versions of Sysmark have been very controversial to say the least.
Therefore imo Bapco benchmarks should be ignored.

Of course you want to ignore it because it doesn't show AMD processors
as the end-all, be-all processor. We've all read enough of your posts
to know that you're totally incapable of making an informed opinion
based on facts.

However, for anyone else reading this thread, SYSMark is every bit as
valid as Business Winstone.

I said that finding motherboards for an Athlon64 was tough and you
pointed out that there are cheap boards for the AthlonXP, which really
doesn't help at all, given that the two processors require totally
different motherboards.
 
Tony said:
Of course you want to ignore it because it doesn't show AMD processors
as the end-all, be-all processor.

Did you read these articles? I guess not.

http://www.vanshardware.com/articles/2001/august/010814_Intel_SysMark/010814_Intel_SysMark.htm

http://www.theinquirer.net/?article=5274
http://www.lostcircuits.com/cpu/northwood/6.shtml
http://www.lostcircuits.com/cpu/prescott/8.shtml

We've all read enough of your posts
to know that you're totally incapable of making an informed opinion
based on facts.




However, for anyone else reading this thread, SYSMark is every bit as
valid as Business Winstone.

Valid? What percentage of business users make use of Dragon Naturally Speaking and
Winzip?


I said that finding motherboards for an Athlon64 was tough

Not tough. There are a number of Athlon 64 socket 754 motherboards
at around $80 or so.
and you
pointed out that there are cheap boards for the AthlonXP, which really
doesn't help at all, given that the two processors require totally
different motherboards.

The vast majority of business users don't need an Athlon 64. An Athlon XP3000+
would be more than enough for them.
 
Bitstring <[email protected]>, from the wonderful person JK
The vast majority of business users don't need an Athlon 64. An Athlon XP3000+
would be more than enough for them.

=Nobody= 'needs' an Athlon 64 (like they need air, water, etc.). I'd
have to allow it's =nice= for some people though.
 
In comp.sys.ibm.pc.hardware.chips JK said:
Valid? What percentage of business users make use of
Dragon Naturally Speaking and Winzip?

Please! Bosses are Dragons Naturally Speaking -- fire :)

OTOH, Winzip is _heavily_ used. At work, all email attachments
are translucently [de]compressed. Even PDFs, but intelligence
should not be expected from people who eagerly submit to and
inflict MS-[in]Active Directory. They have other motives. S&M?

-- Robert
 
JK said:
[...]

Valid? What percentage of business users make use of Dragon Naturally Speaking and
Winzip?

Why should 'average business usage' be of any concern for my choice? My
computing is mostly numerical processing, sometimes using the Intel Math kernel
Library, optimised for Intel.
 

Hmm, dated from 3 years ago, before AMD joined BAPco.

2 years old, also from before AMD joined BAPco.

2 and a half years old on this one.

Uhh.. ok.. that lists what is included in the benchmark, though if
anyone wanted to know that they could just go to:

http://www.bapco.com/products/sysmark2004/applications.html
Valid? What percentage of business users make use of Dragon Naturally Speaking and
Winzip?

Damn near every business user I know of makes use of Winzip. I
suppose you could make an argument that Dragon Naturally Speaking
isn't widely used, and in fact that's the one of the few differences
between BAPco SYSMark 2004 and Business Winstone 2004. Here are the
lists of applications in each test:

BAPco SYSMark 2004 Office Productivity

Adobe® Acrobat® 5.0.5
Microsoft® Access 2002
Microsoft® Excel 2002
Microsoft® Internet Explorer 6
Microsoft® Outlook® 2002
Microsoft® PowerPoint® 2002
Microsoft® Word 2002
Network Associates® McAfee® VirusScan® 7.0
ScanSoft® Dragon Naturally Speaking® 6 Preferred
WinZip Computing WinZip® 8.1


Veritest Business Winstone 2004

Microsoft Access 2002
Microsoft Excel 2002
Microsoft FrontPage 2002
Microsoft Outlook 2002
Microsoft PowerPoint 2002
Microsoft Project 2002
Microsoft Word 2002
Norton AntiVirus Professional Edition 2003
WinZip 8.1


One thing I find odd is that Business Winstone does NOT include
Internet Explorer, possibly the most widely used application among
many "business" users these days (web-enabled applications seems to be
the technology that gets PHBs all excited these days). However I
think you'll notice that the difference between these apps is rather
small.
 
Never said:
FALSE prophecies from the archives, Post Replies Here Please <[email protected]> on Tue, 17 Aug 2004 07:51:17 -0500
spoke:




He has observed EXACTLY the opposite of what I've seen.

I have an XP2500 system, an XP2800,and an Intel P4 2.6C system.

The AMDs are MUCH more responsive in most things,
but the P4 is MUCH faster in Seti and a few other
programs, where little human intervention or
interference is necessary.

If you look at the computation orientated benchmarks you will notice
that the P4 core is great at signal processing style workloads. Audio/
video encoding and SETI are signal processing workloads by definition.
By contrast the K7 core is better suited for workloads that tend to do
lots of branches and the like (compiling, database, GUI code).

The approach should be to characterise your workload and then make
your choice rather than making your choice then hacking your workload
to fit. In the case of workloads most PCs run, I suspect that the K7/K8
cores are a far better fit. OTOH if I was planning on encoding video
or audio 24x7 I would probably go for a P4 core.

Cheers,
Rupert
 
JK said:
JK said:
[...]

Valid? What percentage of business users make use of Dragon Naturally Speaking and
Winzip?

Why should 'average business usage' be of any concern for my choice? My
computing is mostly numerical processing,

You bought a Pentium 4 to do numerical processing? An Athlon XP2200+
beats a Pentium 4 2.8 ghz by a large margin in CPU Math Mark 3.0.

http://www.activewin.com/reviews/hardware/processors/intel/p428ghz/benchs.shtml

But not if you look at the MFLOP rates. I suspects that your Math Marks are
biased towards local operations that are locally dependent on each other, such
as e.g. iteration of pi, and not for vector and matrix operations.

And why is the AMD system fitted with 512MB when the P4 system 'only' has
256 MB? This memory may seem sufficient, but it all depends on OS system
& software overheads & autoload, so it's not very clever to use different
memory sizes.
 
Johannes said:
JK said:
JK wrote:

[...]

Valid? What percentage of business users make use of Dragon Naturally Speaking and
Winzip?

Why should 'average business usage' be of any concern for my choice? My
computing is mostly numerical processing,

You bought a Pentium 4 to do numerical processing? An Athlon XP2200+
beats a Pentium 4 2.8 ghz by a large margin in CPU Math Mark 3.0.

http://www.activewin.com/reviews/hardware/processors/intel/p428ghz/benchs.shtml

But not if you look at the MFLOP rates. I suspects that your Math Marks are
biased towards local operations that are locally dependent on each other, such
as e.g. iteration of pi, and not for vector and matrix operations.

And why is the AMD system fitted with 512MB when the P4 system 'only' has
256 MB?

That is a valid criticism. That is an old article(notice the use of RDRAM for the p4)
so the use of twice as much ram for the Athlon XP was to probably to compensate for the very
high RD ram cost. The test should have been repeated with equal ram sizes when DDR ram
motherboards for the P4 became available.
 
Rupert said:
If you look at the computation orientated benchmarks you will notice
that the P4 core is great at signal processing style workloads. Audio/
video encoding and SETI are signal processing workloads by definition.
By contrast the K7 core is better suited for workloads that tend to do
lots of branches and the like (compiling, database, GUI code).

The approach should be to characterise your workload and then make
your choice rather than making your choice then hacking your workload
to fit. In the case of workloads most PCs run, I suspect that the K7/K8
cores are a far better fit. OTOH if I was planning on encoding video
or audio 24x7 I would probably go for a P4 core.

Because of the attention the SPEC benchmarks get and because Intel has
its own compiler, I'm guessing we know what P4 is capable of against
fixed source.

I'm less certain that we really know how well the strategy would work if
code were always tuned against a P4 pipeline at either the source or the
assembly language level.

A different way of characterizing workloads is how much work and how
much specificity you are going to put into the code, and K7 is generally
more friendly to nai-ve code than NetBurst. If you're willing to
settle on Intel and to use its compilers and tools like Vtune and the
Math Kernel Library, then you needn't always be writing code that is
hobbled for NetBurst, but you may be writing code that is less than
optimal for AMD.

That's a prospect that doesn't please advocates of AMD. Writing and
tuning for your own closed universe? AMD if you are writing nai-ve code,
Intel maybe if you are writing code that can make good use of Intel
productivity tools and are willing to fiddle. Writing for the world at
large? I don't know that I've ever seen a discussion of how commercial
developers have really dealt with not knowing whether they are writing
for K7 or P4, but I would assume it's safer to assume you are writing
for P4.

RM
 
Johannes H Andersen wrote:

And PC1066 RDRAM doesn't seem very fast. Luckily I avoided this era from Intel;
the 820 chipset cos Intel ~$250M or so? Just shows that chip making isn't always
plain sailing.

You are confusing 820 and 850? 820 was, well, amazing, but not in a
positive way, unless you really want to be impressed at just how much
Intel marketing really can cope with. 850, in its time, was the chipset
of choice for performance, and it wasn't abandoned because of a lack of
performance, AFAIK, but for reasons having to do with...marketing.

RM
 
JK said:
JK said:
Johannes H Andersen wrote:
[...]

But not if you look at the MFLOP rates. I suspects that your Math Marks are
biased towards local operations that are locally dependent on each other, such
as e.g. iteration of pi, and not for vector and matrix operations.

And why is the AMD system fitted with 512MB when the P4 system 'only' has
256 MB?

That is a valid criticism. That is an old article(notice the use of RDRAM for the p4)
so the use of twice as much ram for the Athlon XP was to probably to compensate for the very
high RD ram cost. The test should have been repeated with equal ram sizes when DDR ram
motherboards for the P4 became available.

And PC1066 RDRAM doesn't seem very fast. Luckily I avoided this era from Intel;
the 820 chipset cos Intel ~$250M or so? Just shows that chip making isn't always
plain sailing.
 
Johannes said:
I was mentioning the 820 because of the RDRAM. The faulty memory translation
chip cost Intel a fortune in replacing motherboards.

Intel fumbled the RDRAM era in a number of ways (probably nodded off
during Steve Ballmer's gratis seminars on how to be a 21st Century
monopolist), but I don't think the 820 fiasco sheds much light on RDRAM.

RM
 
Robert said:
Johannes H Andersen wrote:



You are confusing 820 and 850? 820 was, well, amazing, but not in a
positive way, unless you really want to be impressed at just how much
Intel marketing really can cope with. 850, in its time, was the chipset
of choice for performance, and it wasn't abandoned because of a lack of
performance, AFAIK, but for reasons having to do with...marketing.

RM

I was mentioning the 820 because of the RDRAM. The faulty memory translation
chip cost Intel a fortune in replacing motherboards.
 
Robert Myers wrote:

[SNIP]
A different way of characterizing workloads is how much work and how
much specificity you are going to put into the code, and K7 is generally
more friendly to nai-ve code than NetBurst. If you're willing to
settle on Intel and to use its compilers and tools like Vtune and the
Math Kernel Library, then you needn't always be writing code that is
hobbled for NetBurst, but you may be writing code that is less than
optimal for AMD.

That's a prospect that doesn't please advocates of AMD. Writing and

It should because if you are tuning to a high degree it can increase
the likelihood of taking a hit everytime Intel changes the micro-
architecture (over-specialisation). Intel has announced that the end
is nigh for the NetBurst architecture too... :)

If you look at the grand scheme of things I don't think much code is
tuned to that kind of level. In essence what you're saying is that
K7/K8 is cheaper to tune, and with most software development driven
by cost to the exclusion of quality, something has to give.

NetBurst ain't all bad, Intel pitched it as excelling at multimedia
and it does just that providing the inner loops are tuned to buggery
(which they mostly are).


Cheers,
Rupert
 
That's a prospect that doesn't please advocates of AMD. Writing and
tuning for your own closed universe? AMD if you are writing nai-ve code,
Intel maybe if you are writing code that can make good use of Intel
productivity tools and are willing to fiddle. Writing for the world at
large? I don't know that I've ever seen a discussion of how commercial
developers have really dealt with not knowing whether they are writing
for K7 or P4, but I would assume it's safer to assume you are writing
for P4.

From what I've seen of most commercial code it's mainly a matter of
setting -02 in the compiler and praying that it doesn't break
something! :>

Seriously though, for most commercial code that is going into
widespread use the focus is on it working properly on as many
platforms as possible. This tends to avoid any architecture-specific
optimizations as they can often break your code on different setups,
not just AMD vs. Intel but even current Intel vs. next-generation
Intel. An awful lot of the world are still using PCs with PII and
PIII era processors, and optimizing for the PIII tends to be VERY
different from the P4.

There are, of course, going to be some exceptions to this rule,
particularly for really CPU-bound tasks. Some applications simply
detect what sort of processor you have and use different code paths
depending on what chip you've got. But for the most part you probably
won't see all that much in the way of processor-specific
optimizations.

This is, of course, one of the major complaints against SPEC CPU.
It's so sensitive to the compiler and optimizations used that it often
does not reflect on what you'll encounter in the real world. We've
most recently actually seen AMD and Sun use this to their advantage to
take back the lead as the fastest x86 processor for CFP2000 using the
new Pathscale and PGI compilers along with the newest version of GCC
and AMD's Math Libraries. 3 different compilers and PLENTY of
optimizations flags, but it gave them a rather impressive CFP2000
score (1637/1787 for base/peak).
 
AMD does come out with performance benchmarks higher than intel. But i
find many AMD systems not performing as expected against the intel
counterpart. and almost always its the intel that wins in every
aspect.
Why is the bench mark different from the true story?? and if its the
case of ill-configured systems.. then why is most of the amd
ill-configured??

Intel versus AMD (August 24th,2004)
http://reviews.zdnet.co.uk/hardware/0,39023760,39164010,00.htm
 
AMD, I think it does better on games. Got a 2ghtz Xp 266mhz front
side bus. 512ram, 90 gig hard drive. Abit KX7-333 with raid. Abit
make some soild board. Suttle make some decent boards. Asus, I like
a P3 board i had from them. I build again istead of buy. Then i know
what componets im getting. Run Win ME. Not a big fan of xp yet

==============
Posted through www.HowToFixComputers.com/bb - free access to hardware troubleshooting newsgroups.
 
Back
Top