Intel chipsets are the most stable?

  • Thread starter Thread starter Franklin
  • Start date Start date
No flame intended: what's wrong with Tom's Hardware?

It's been discussed at length here in the past - they specialize in muck
raking and throwing, usually as part of their quest to "follow the money".
Relatively minor, insignificant even, performance result differences get
blown out of proportion as a means to sound authoritative and
controversial... what I call a Hyperbolic Tomvoid.

It's actually kinda amusing how so many of those sites are written by
people who think of themselves as electronic "journalists" now... as well
as the amount of journalistic padding present in their content.. not to
mention the groveling, bowing and general boot-licking to large
multi-nationals. They seem to have forgotten that what got them started...
the whole reason for their success, was that they were enthusiastic
mongrels, who were willing to tell their version of the truth.

The bottom line is that Tom's hardware is now no more credible or
technically competent than PCWorld or PCMag - same old stuff.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
 
No flame intended: what's wrong with Tom's Hardware?

Ohh where to begin? To start with they seem VERY fickle about who
they like and very often draw ridiculous conclusions from their
results just so that they support whatever company they want to look
best (which varies strongly from one week to the next.. my
understanding is that this isn't so much a question of 'bias' as that
Tom has a HUGE ego and he will only promote companies that give him
the full red-carpet treatment). Their benchmarking methodology often
leaves something to be desired, they will usually pick and chose which
benchmark to run depending on which product they want to come out
looking the best (to be fair, pretty much all hardware review sites do
this).

On the more technical side they really just don't have a technical
side. They take a VERY superficial view of running their review but
pretend that they are all super-technical, often drawing conclusions
from extraordinarily limited information (ie something along the lines
of: "If we look at the results of this Quake benchmark we can see that
the P4's memory bandwidth gives it a huge advantage in all OpenGL
gaming", when really all there are hundreds of possible factors
involved). They also tend to make MANY technical errors. This page
from a recent article is classic:

http://www6.tomshardware.com/cpu/20040927/opteron_vs_xeon-02.html

They manage to get virtually every fact on this page completely wrong.
All Opterons have 3 hypertransport connections, all running at 16-bits
in either direction (usually called a 32-bit hypertransport link,
though it's really 2 x 16-bits), running at 800MHz DDR for an
effective speed of 1600MHz. The 1xx series of Opterons have no
cache-coherent Hypertransport connections (ie they can only use the
three HT connections for I/O, not for processor-to-processor
connections), the 2xx series Opterons have 1 cache-coherent HT links
(each chip can connect to up to 1 other processor) and the 8xx series
Opterons have 3 cc-HT links. The one thing they managed to get right
on the entire page were the diagrams at the bottom that were done by
AMD and have been shown many times in AMD publicity info for the
Opteron. Careful observers will even note that those diagrams
contradict what the author of the article said about this chip.


Anyway, I suppose I shouldn't be TOO hard on Tom's Hardware, after
all, the vast majority of other technical websites are total crap as
well, not to mention the mainstream print mags. Tom's has also
improved a fair bit in the past two or three years since Tom himself
pretty much just sites back and waits for the checks to role in, other
employees do all the work. Really this site just gets under my skin
more because people quote this site so regularly when the info listed
on the page is often wrong or at least misleading.
 
The following is an article on the Athlon 64 2800+. But more interesting
is,
the benchmarks included in the article are a GREAT comparison of the 3.2GHz
P4
processors with the Athlon64 3200+. In this article, these two processors
are
pretty evenly matched, with Intel being faster on some benchmarks, and AMD
being faster on others.

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2038&p=1

So here they tested 6 games, of which the Athlon64 3400+ was faster
than the P4 3.4E on 5 of them and they were tied in the (video-card
limited) 6th game.
Now lets look at what Sharky Extreme has to report in their article about
the
3.4GHz Prescott processor. This one has benchmarks that are a great
comparison
of the 3.4GHz Intel chips with the Athlon64 3400+. Here, you have to be
careful,
as Sharky doesn't organize their charts in order of fastest to slowest. And
on
some charts, LOWER scores are better. But if you read all the benchmarks,
you
will again notice that the two chips are pretty evenly matched, with AMD
faster
on some and Intel faster on others.

http://www.sharkyextreme.com/hardware/cpu/article.php/3261_3329681__1


Did you even bother reading this articles before making your post?!
In this article they test 5 games, of which 1 is video card limited
(CPUs all score basically the same) and in the other 4 the Athlon64
was faster! When comparing the P4 3.4E vs. Athlon64 3400+, the latter
is faster by 6%, 3%, 22% and 13%. Even if the video-limited test the
Athlon64 came out 1% faster, though I'm sure that's within the margin
of error for the test.

In this same test the Athlon64 was also faster in their Office
application test and Content Creation test. The only tests in which
the P4 was faster were their synthetic benchmarks (which simply showed
that the Athlon64 they tested did indeed just have a single memory
channel while the P4 had a dual channel memory) and the 2 media
encoding tests.
 
OH I get it now. You are evaluating processors on the basis of who the ****
cares about cost? If I wanted to pay more for a processor than I did for
the last car I purchased, then YES, the benchmarks might support your point
of view. But I have claimed all along that Intel processors are a better
deal at the price point most people are building at, in terms of "bang for
buck". That is still true. -Dave

Uhhh.. Dave?! Are you hitting the crack-pipe here?! The Athlon64
3200+ mentioned above costs $204 at Newegg.com. The slightly slower
P4 3.4GHz Extreme Edition costs $999.

How in the hell can you claim that the slower $999 processor gives you
better bang for your buck than the faster $204 process?!?! Ok, to be
fair there ARE tests in which the P4EE is a faster chip, but it
basically NEVER offers a very good "bang for buck"!
 
On the more technical side they really just don't have a technical
side. They take a VERY superficial view of running their review but
pretend that they are all super-technical, often drawing conclusions
from extraordinarily limited information (ie something along the lines
of: "If we look at the results of this Quake benchmark we can see that
the P4's memory bandwidth gives it a huge advantage in all OpenGL
gaming", when really all there are hundreds of possible factors
involved).

....or what about the infamous (but floor-rolling/gut-bustingly funny)
count-the-capacitors escapade?

Anyway, I suppose I shouldn't be TOO hard on Tom's Hardware, after
all, the vast majority of other technical websites are total crap as
well, not to mention the mainstream print mags. Tom's has also
improved a fair bit in the past two or three years since Tom himself
pretty much just sites back and waits for the checks to role in, other
employees do all the work. Really this site just gets under my skin
more because people quote this site so regularly when the info listed
on the page is often wrong or at least misleading.

Why not, hey *is* the worst of 'em. Remember his rants because he was
stiffed by Via (I think it was, anyway). Dean Kent had some
interesting insights into Tom here a couple of years ago (a few times,
IIRC). Tom should have been a dentist. Oh, wait...
 
cares about cost? If I wanted to pay more for a processor than I did for
the last car I purchased, then YES, the benchmarks might support your
point

You must have a pretty shitty car.
 
Yet, even then there were some boards that,
when set up correctly and with at least marginally good PSU, video,
and NIC, were quite decent and stable performers. Case in point - my
1998-ish FIC VA503+ board that is still alive in my second system with
k6-2+ overclocked to 600MHZ - I use it to browse some iffy sites when
I would not want to risk infecting my main system with some crap.

I've got a K6/2 doing that's been doing server (web server, SSH, VNC, print
server, a few other things) duties for the past 1 1/2 years. It's about the
most "generic" machine I have around at the moment:
1) Motherboard is a a Jetway 530BF (512KB L2 cache)
2) CPU is a K6/2-400, running at stock (not worth overclocking :) ).
3) One 256MB PC133 stick (picked up as 128MB), two 64MB PC100 sticks, all
three generic. Yeah, I know, it can't cache all this, but it's better than
hitting the disk :)
4) El-cheapo AT case with 250W PSU
5) Some 10 gig HDD, cdwriter
6) Windows 2000 Profesional
7) A NIC, an ADSL card, a USB 2.0 card, various other bits

Now, you'd think that with a config like this, it'd be falling over every
few days (though the SiS 530 wasn't too bad of a chipset stability wise).
Not so ... it's been taken out of service perhaps 6 or 7 times in the past
year and a half, all due to power problems or house rewiring (and in one
case to install the cd writer). Up until the power cut two days ago, it had
been up for slightly over 4 months. So even with dirt cheap, generic,
bottom-of-the line components you can have great stability. Sure, the
performance won't blow your socks off, but price (free, as I already had all
the bits) vs performance it comes out quite well :)

[...]
 
"fussiness" ?

Just choose the proper memory.


"problems with new motherboards."

Choose a decent brand of motherboard. If you choose a brand
that makes low quality motherboards, it is your fault, not the fault
of AMD.
 
No flame intended: what's wrong with Tom's Hardware?
[...]
The bottom line is that Tom's hardware is now no more credible or
technically competent than PCWorld or PCMag - same old stuff.

I gave up on Tom's Hardware some time ago because of the way they organize
their content (I like to read "Review of GA-K8NS Pro motherboard", not "14
new motherboards roundup and comparison with a digital cuckoo clock"),
however I still thought them as competent. What other sites would you
recommend?
 
I came across this. Is the guy right?

<QUOTE>
Volumes have been written on this subject, but suffice to say that
Intel chipsets are the most stable. I do not know if this is because
Intel does a better job at manufacturing their chipsets than other
companies, or that software manufacturers test their software more
thoroughly on Intel-based systems, since they are more popular ..
more than they do on systems based upon non-Intel chipsets. Or a
combination of these factors.

Either way, a system based on an Intel chipset will provide you with
the most stable computing experience. This is common knowledge in the
community. Everyone knows it.
<END QUOTE>

http://radified.com/Articles/stability.htm

Fact: This has always been true, and most likely always will be. In
addition, Video encoding, which I use my computer for most often, is much
faster on P4 platforms.
 
JK said:
"fussiness" ?

Just choose the proper memory.

"problems with new motherboards."

Choose a decent brand of motherboard. If you choose a brand
that makes low quality motherboards, it is your fault, not the fault
of AMD.

I don't choose any motherboard at the moment since I already have a well
performing P4 machine. But I've noticed some teething problems reported
for boards for the AMD64, see e.g.

http://groups.google.com/groups?hl=...25&hl=en&lr=&group=uk.comp.homebuilt&start=50

"AMD is beating Intel in desktop performance by such a large margin."

As for benchmarks, even the site you quoted is not a clear win at all for AMD64.

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2065&p=6

Forget about P4EE, nobody buys them for their own money. Compare e.g. Intel
Northwood 3.2 and a AMD64 3200+ . AMD64 win some, but the Intel win other
important benchmarks hands down. Just click on the subsequent pages on the site...
 
Johannes said:
I don't choose any motherboard at the moment since I already have a well
performing P4 machine. But I've noticed some teething problems reported
for boards for the AMD64, see e.g.

http://groups.google.com/groups?hl=...25&hl=en&lr=&group=uk.comp.homebuilt&start=50

That is a Microsoft issue, not a hardware issue. If they don't already have a patch for this, they will probably have one soon.
"AMD is beating Intel in desktop performance by such a large margin."

As for benchmarks, even the site you quoted is not a clear win at all for AMD64.

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2065&p=6

Where on that page do you not see a clear win for AMD when comparably priced
processors are compared?
Forget about P4EE, nobody buys them for their own money. Compare e.g. Intel
Northwood 3.2 and a AMD64 3200+ . AMD64 win some, but the Intel win other
important benchmarks hands down.

Like what? 32 bit video editing benchmarks? Very soon 32 bit video editing
benchmarks on the Athlon 64 won't mean much, as there will be great performing
64 bit software for that.
Just click on the subsequent pages on the site...

The applications where the Athlon 64 doesn't outperform a comparably
priced P4 running 32 bit software will benefit tremendously from a move
to 64 bit software.
 
I don't choose any motherboard at the moment since I already have a well
performing P4 machine. But I've noticed some teething problems reported
for boards for the AMD64, see e.g.

http://groups.google.com/groups?hl=...25&hl=en&lr=&group=uk.comp.homebuilt&start=50

Huh?! Your link points to a message discussing two very specific
problem. The first is strictly a software issue and doesn't even
mention a hardware platform, the second is a very specific issue with
Intel's C0 stepping of Prescott P4's and Celeron-D chips!
"AMD is beating Intel in desktop performance by such a large margin."

As for benchmarks, even the site you quoted is not a clear win at all for AMD64.

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2065&p=6

Forget about P4EE, nobody buys them for their own money. Compare e.g. Intel
Northwood 3.2 and a AMD64 3200+ . AMD64 win some, but the Intel win other
important benchmarks hands down. Just click on the subsequent pages on the site...

Fine, let's do a direct comparison of which chip is faster and by what
percentage (anything less than 1% I'll consider a tie as that's well
within the margin of error)

AMD Athlon64 3200+ wins:
Business Winstone 2004 (11.5%)
Content Creation Winstone 2004 (7.7%)
Unreal Tournament Flyby (8.0%)
Unreal Tournament Botmatch (18.0%)
Warcraft 3 (2.4%)
Quake 3 (4.2%)
Wolfenstein (1.8%)
Jedi Knight (2.8%)
Quake 3 Source Compile (14.5%)


Intel "Northwood" P4 3.2GHz wins:
DivX encoding (21.2%)
Aquamark CPU (9.1%)
3DStudio (23.2%)
Lightwave (17.6%)


Tied:
Aquamark FPS
Gunmetal


So the Athlon64 wins more tests, while when the P4 wins it tends to do
so by a larger margin. If we average all the tests out we get that
the P4 is faster by 0.013% (ie they're tied).

Now if we throw price into the equation, we get that the Athlon64
3200+ costing $204 while the "Northwood" P4 3.2GHz will set you back
$244 (prices care of www.newegg.com) The motherboard cost for boards
used in this test should slightly favor AMD (VIA K8T800 based socket
754 board vs. Intel i875P based Socket 478 board) while all other
components in this test were identical.


Soooo... long story short, if you want to do lots of media encoding or
3D rendering, according to this test at least, you should stick with
the P4. If you want to play games, do general office tasks or compile
code, you should go for an Athlon64. You should also stick with the
Athlon64 if 64-bit code is important to you, or if you want the extra
security offered by non-executable data pages. And finally, sticking
with an Athlon64 will also shave a small amount off the price of your
system.


So where is the better value in the Intel system?
 
Tony said:
Huh?! Your link points to a message discussing two very specific
problem. The first is strictly a software issue and doesn't even
mention a hardware platform, the second is a very specific issue with
Intel's C0 stepping of Prescott P4's and Celeron-D chips!


Fine, let's do a direct comparison of which chip is faster and by what
percentage (anything less than 1% I'll consider a tie as that's well
within the margin of error)

AMD Athlon64 3200+ wins:
Business Winstone 2004 (11.5%)
Content Creation Winstone 2004 (7.7%)
Unreal Tournament Flyby (8.0%)
Unreal Tournament Botmatch (18.0%)
Warcraft 3 (2.4%)
Quake 3 (4.2%)
Wolfenstein (1.8%)
Jedi Knight (2.8%)
Quake 3 Source Compile (14.5%)

Intel "Northwood" P4 3.2GHz wins:
DivX encoding (21.2%)
Aquamark CPU (9.1%)
3DStudio (23.2%)
Lightwave (17.6%)

Tied:
Aquamark FPS
Gunmetal

So the Athlon64 wins more tests, while when the P4 wins it tends to do
so by a larger margin. If we average all the tests out

That is not reasonable to do, as a business for example that only runs
business software doesn't care how fast Divx encoding, 3D Studio,
or Lightwave runs. I don't care how fast those applications would run
on my pc, since I don't use them, and don't plan to use then.
we get that
the P4 is faster by 0.013% (ie they're tied).

Now if we throw price into the equation, we get that the Athlon64
3200+ costing $204 while the "Northwood" P4 3.2GHz will set you back
$244 (prices care of www.newegg.com) The motherboard cost for boards
used in this test should slightly favor AMD (VIA K8T800 based socket
754 board vs. Intel i875P based Socket 478 board) while all other
components in this test were identical.

Soooo... long story short, if you want to do lots of media encoding or
3D rendering, according to this test at least, you should stick with
the P4.

Only if you must use 32 bit software. When using 64 bit software, I
expect the Athlon 64 to be a great performer for those applications.
 
Now if we throw price into the equation, we get that the Athlon64
3200+ costing $204 while the "Northwood" P4 3.2GHz will set you back
$244 (prices care of www.newegg.com) The motherboard cost for boards
used in this test should slightly favor AMD (VIA K8T800 based socket
754 board vs. Intel i875P based Socket 478 board) while all other
components in this test were identical.

So you're going to choose a el-cheapo Athlon 64 board to go against a very
high-end P4 board and conclude that it's cheaper to build AMD? Ummmmm, no,
but we can conclude you are biased against Intel, and that is the ONLY
conclusion that can be reached. -Dave
 
JK said:
That is not reasonable to do, as a business for example that only runs
business software doesn't care how fast Divx encoding, 3D Studio,
or Lightwave runs. I don't care how fast those applications would run
on my pc, since I don't use them, and don't plan to use then.

OK, pick your special application, but then your statement:

"Now AMD is beating Intel in desktop performance by such a large margin."

is unduly generalizing from the specifics. The whole point in increasing
processor performance is to venture into new application arenas such as
photo processing and video. Office computing such as word processing etc.
is a problem that is already satisfactory solved with yesterdays processors.
 
Dave said:
So you're going to choose a el-cheapo Athlon 64 board to go against a very
high-end P4 board and conclude that it's cheaper to build AMD? Ummmmm, no,
but we can conclude you are biased against Intel, and that is the ONLY
conclusion that can be reached. -Dave
Even if you pick a fairly good nForce3 250-board you STILL get it
cheaper than a P4 system and if you are a gamer nothing beats an Athlon64.
 
Even if you pick a fairly good nForce3 250-board you STILL get it
cheaper than a P4 system and if you are a gamer nothing beats an Athlon64.

It's in the Marketing, once you start advertising on TV/radio as much as
Intel has done it's no longer just commercials, it's brainwashing, so
just feel sorry for those people where it fits! ;p

http://yahoo.businessweek.com/magazine/content/04_38/b3900048_mz011.htm
Intel is almost seven times as large, with expected revenues of $34
billion this year. Its projected 2004 profits of $7.35 billion mean that
Intel earns in 11 days what AMD will make all year. And Intel is sitting
on $14 billion in cash, compared with $1.1 billion for AMD, giving Intel
a vast edge in funding research and development and in constructing
cutting-edge manufacturing facilities.

....giving Intel a vast edge in funding research and development and in
constructing cutting-edge manufacturing facilities.

With a statement like that, I'd expect Intel to be light years ahead of
AMD in architecture and performance, what' up with that!?!?


Ed
 
Back
Top