Is the Sempron a budget line that costs more?

[Michael Brown]

Keep in mind, though, that the Sempron 2600+ is on a 333MHz FSB,
whereas the Athlon 2200+ is on a 266MHz FSB. That doesn't give it
quite enough of a boost to justify its higher price, but the
real-world PR rating difference is probably closer to 300 than 400.

FWIW, the slowest 256KB L2, 166MHz FSB Athlon XP was the 2600 at 2083MHz.
The highest Sempron (absolutely identical to the 256KB L2, 166MHz Athlon
XPs), the 2800, has a speed of 2000MHz, which pretty much exactly equates to
a 2500 (if it esisted with the 256/166 spec) under the old scheme. So in
this particular case the rating on the Sempron is very close to 300 points
too high.

At the lower end, we have the Sempron 2200 at 1500MHz. 1500MHz would put you
at an 1800 rating with a 256/100 config, ~1750 (yeah, it's backwards) with a
256/133 config, ~1950 with a 512/100 config, and ~1950 with a 512/133
config. So it looks as though at this speed, AMD thinks that the bus speed
has very little impact on the performance of the CPU, so a 256/166 part at
1500MHz would be somewhere around 1850 rating or so at the most. This means
the Sempron is overrated by about 350 points. Now, there is some variation
at each speed, due to the requirements that the speed must be a half-integer
multiple of the bus speed (hence why the 133MHz FSB part is "slower" than
the 100MHz FSB part at 1500MHz). However, I would say that overall, the
Sempron is over-rated by something in the range of 300 to 350 points.

As for the prices in New Zealand (in order from cheapest to most expensive,
skipping those where the lower rated part is more expensive, and including
the two closest-in-price Athlon XP chips):

Sempron 2400+: $106
XP 2000+: $100
XP 2200+: $116
Should be XP 2100+ at ~$108 (-2%)

Sempron 2500+: $124
XP 2200+: $116
XP 2400+: $133
Should be XP 2200+ at $116 (+7%)

Sempron 2600+: $145
XP 2400+: $133
XP 2500+: $148
Should be XP 2300+ at ~$125 (+16%)

Sempron 2800+: $187
XP 2600+: $155
XP 2700+: $190
Should be XP 2500+ at $148 (+26%)

So, the Sempron 2400+ is priced "about right". However, the 2500, 2600, and
2800 Semprons are clearly overpriced, coming close to the price of an Athlon
XP with a 100 point lower rating.

 

[JK]

Since the Athlon XP chips may start disappearing pretty soon, this might
not be an issue for very long.

 

[Matt]

Are AMD in trouble and need to hike the price up on their products?

I'm sure they set their prices so as to maximize their profit, whether
they "need" it or not.

And I doubt they are in trouble. They make great CPUs!

 

[Matt]

On Fri, 17 Sep 2004 19:40:22 +0100, Aaron S




New products entering a market are often priced a bit too
high, then price normalizes after demand is less than
anticipated, or lower than production. It is still a bit
odd though, even a lowly Duron 1.6 is going for $48 on
pricewatch, seems like it should've dropped to $38 already.

Yes, and the XP 2500+ costs more today than it did a year ago.

 

[Matt]

Dear Lord. What a mess.

But at least now we understand the mess.

 

[David Maynard]

Wes Newell wrote:

On Fri, 17 Sep 2004 00:35:38 +0100, Franklin wrote:



I don't kmow how reliable AMD's "+" figure is in idnicating throughput
power, so I don't know if I can take at face value that these two cpus

are

equivalent:

2500+ Sempron (1.75 GHz, FSB 333, T'bred-B core)
2500+ Athlon (1.92 GHz, FSB 333, Barton core)

[Data taken from page linked above.]

Surely these two are not equivalent in terms of power?


They aren't. The Sempron, if rated with the same suite of benchmarks
Athlon XP's are rated with, would rate as a 2100+. All Semprons are

rated

with a set of benchmarks to compare them to Celerons clock speeds and

the

resulting PR number reflects that. not compared P4's or even XP's. IF

you

subtract 400 from every Semprons number, you will get the approximate
Athlon rating. So a Sempron 2200+ would [erforme the same as an Athlon
1800+, etc. And you can take that to the bank.

Good info and that explains a LOT.

So that makes them even less value for money then!

True, but it explains why I was having such a hard time making sense of the
new numbering scheme.

With computer hardware something new is usually better and often cheaper,
doesn't seem to be the case here.

From their vantage point it is because the Sempron replaces the Duron.

Are AMD in trouble and need to hike the price up on their products?

AMD has always been in price trouble.

 

[Matt]

kony wrote:

If you only sell only one product, you simply set your price so that you
have a few people standing in line _waiting_ to buy. Not too many, but
some. If they don't have to wait, the price is too high. If they have
to wait too long, the price is too low.

It gets more complicated when you sell some products that compete with
each other.

Yes, and the XP 2500+ costs more today than it did a year ago.

Maybe they can't produce the low-end chips fast enough because they've
given over much of their production capacity to the 64-bit chips. The
tide raises all boats, so to speak.

But I am not an economist.

 

[~misfit~]

It does have a faster bus thou 166 not 133 and costs £31
compared to around £40 for the XP

Are you saying that matching "performance for performance" the
Semprons are ever so slightly cheaper?

I have done the data gathering or data comparison but several posters
here seem to say that the Sempron is more expensive than the
equivalent Athlon.

Maybe one new variable is the performance of the Sempron and the
chip's clock speed. The chart on this page matches up the Sempron
and Athon according to their clock speed.

http://www.tomshardware.com/cpu/20040728/sempron-01.html

I don't kmow how reliable AMD's "+" figure is in idnicating throughput
power, so I don't know if I can take at face value that these two
cpus are equivalent:

2500+ Sempron (1.75 GHz, FSB 333, T'bred-B core)
2500+ Athlon (1.92 GHz, FSB 333, Barton core)

[Data taken from page linked above.]

Well, that's a crock of shit for a start, an Athlon XP2500+ Barton is 11 x
166.6Mhz for 1833Mhz

 

[Johannes H Andersen]

kony wrote: [...]
Yes, and the XP 2500+ costs more today than it did a year ago.

Maybe they can't produce the low-end chips fast enough because they've
given over much of their production capacity to the 64-bit chips. The
tide raises all boats, so to speak.

But I am not an economist.

But this has never happened before that a PC CPU is more expensive than
it was a year ago AFAIK.

 

[Ed]

Since the Athlon XP chips may start disappearing pretty soon, this might
not be an issue for very long.

and people who buy retail may like the idea of the NX bit security on
the Sempron and won't mind paying a little more for it.?
Ed

 

[BigBadger]

and people who buy retail may like the idea of the NX bit security on
the Sempron and won't mind paying a little more for it.?
Ed

The Socket A Semprons are Athlon T'bred CPU's down to the last transistor...
They don't have any new features.

 

[Ed]

The Socket A Semprons are Athlon T'bred CPU's down to the last transistor...
They don't have any new features.

Geez, AMD should of just called those Durons then!
Ed

 

[Wes Newell]

But this has never happened before that a PC CPU is more expensive than
it was a year ago AFAIK.

Sure it has. The price of the high end 100MHz (200FSB) Tbirds went up when
the supply got short. Too bad most perople don't realize they really don't
have to have these to upgrade. Here's current pricing on pricewatch.

$94 - Athlon 1.4GHz 266
$149 - Athlon 1.4GHz 200
$42 - Athlon 1.33GHz 266
$56 - Athlon 1.3GHz 200
$43 - Athlon 1.2GHz 266
$57 - Athlon 1.2GHz 200
$50 - Athlon 1.13GHz 266
$61 - Athlon 1.1GHz 200
$54 - Athlon 1GHz 200

Now consider one could buy a $56 XP 2400+ and run it at 1500MHz (without
overclocking the FSB) in those old 100MHz boards like the one below and it
almost becomes comical.

 

[kony]

Sure it has. The price of the high end 100MHz (200FSB) Tbirds went up when
the supply got short. Too bad most perople don't realize they really don't
have to have these to upgrade. Here's current pricing on pricewatch.

$94 - Athlon 1.4GHz 266
$149 - Athlon 1.4GHz 200
$42 - Athlon 1.33GHz 266
$56 - Athlon 1.3GHz 200
$43 - Athlon 1.2GHz 266
$57 - Athlon 1.2GHz 200
$50 - Athlon 1.13GHz 266
$61 - Athlon 1.1GHz 200
$54 - Athlon 1GHz 200

Now consider one could buy a $56 XP 2400+ and run it at 1500MHz (without
overclocking the FSB) in those old 100MHz boards like the one below and it
almost becomes comical.

I agree, except that you keep ignoring that many boards
won't run them. I've tried 100MHz FSB boards (like ECS
K7VZM) that won't run anything Palomino or newer, and even
133MHz FSB boards (like MSI K7T266 Pro2) that won't run
T'Bred or newer... "maybe" in later revisions those boards
would, but actual samples which worked 100% fine, simply
won't run these faster chips even at 1500MHz (or whatever
applied per chip tried), with bios released long after the
CPUs were.

 

[Wes Newell]

I agree, except that you keep ignoring that many boards
won't run them.

I haven't ignored anything. And what makes you think there are many boards
that won't run them. Every time someone has told me the XX board wouldn't
run them, I've proved them wrong. And this list includes boards from about
all manufacturers.

I've tried 100MHz FSB boards (like ECS K7VZM) that won't run anything
Palomino or newer, and even 133MHz FSB boards (like MSI K7T266 Pro2)
that won't run T'Bred or newer... "maybe" in later revisions those
boards would, but actual samples which worked 100% fine, simply won't
run these faster chips even at 1500MHz (or whatever applied per chip
tried), with bios released long after the CPUs were.

What you mean to say is that you couldn't get them to work. There's a big
difference. Most people couldn't get the 2200+ to work in any board that
had multiplier control when it came out. That doesn't mean it didn't work.
it just meant they didn't know what they were doing.

 

[JK]

Geez, AMD should of just called those Durons then!

Not really, since the Sempron 3100+ is K8 based. My guess is that the entire
Sempron line will probably move to K8 based chips by the end of 2005, or that
AMD will eventually outsorce the K7 Semprons. I believe AMD has no plans to
move K7 production to 90nm.

 

[kony]

On Sun, 19 Sep 2004 00:16:50 GMT, Wes Newell

I haven't ignored anything. And what makes you think there are many boards
that won't run them. Every time someone has told me the XX board wouldn't
run them, I've proved them wrong. And this list includes boards from about
all manufacturers.



What you mean to say is that you couldn't get them to work. There's a big
difference. Most people couldn't get the 2200+ to work in any board that
had multiplier control when it came out. That doesn't mean it didn't work.
it just meant they didn't know what they were doing.

OK, then tell me why a K7VZM wouldn't run an XP1800 at
100MHz FSB?

 

[kony]

I haven't ignored anything. And what makes you think there are many boards
that won't run them. Every time someone has told me the XX board wouldn't
run them, I've proved them wrong. And this list includes boards from about
all manufacturers.



What you mean to say is that you couldn't get them to work. There's a big
difference. Most people couldn't get the 2200+ to work in any board that
had multiplier control when it came out. That doesn't mean it didn't work.
it just meant they didn't know what they were doing.

.... and in the case of the K7T266 Pro2, the board had been
running an XP1600 Palomino o'c to 1.7GHz, then system was
returned to default speed, confirmed still working properly.
Next an XP1900 T'Bred A was tried. System DID post, was
running at correct voltage, multiplier, FSB, Mem, but was
severely instable, even bios screens were quickly locking
up. Do tell what should've been changed? There was no
power supply, heatsink/heat/etc, type of problem, other
factors remained constant.

There was a hack at the time, removing a (resistor?), but
attempt was made prior to and post-removal of the resistor
without success. BTW, the board is still running fine,
person it was sold to uses it every day, nothing else wrong
with board.

 

[Wes Newell]

On Sun, 19 Sep 2004 00:16:50 GMT, Wes Newell



OK, then tell me why a K7VZM wouldn't run an XP1800 at
100MHz FSB?

How should I know why the one you tried didn't work. There could be many
reasons. This much I can tell you. It's not because the chipset won't
support it. Specs for the XP cpu are for FSB's as low as 50MHz. It could
be a problem with power, timing or many other things. Now since you don't
have this board any more to try some things, what's the point of me
wasteing my time guessing?

 

[Wes Newell]

... and in the case of the K7T266 Pro2, the board had been
running an XP1600 Palomino o'c to 1.7GHz, then system was
returned to default speed, confirmed still working properly.
Next an XP1900 T'Bred A was tried. System DID post, was
running at correct voltage, multiplier, FSB, Mem, but was
severely instable, even bios screens were quickly locking
up. Do tell what should've been changed? There was no
power supply, heatsink/heat/etc, type of problem, other
factors remained constant.

There was a hack at the time, removing a (resistor?), but
attempt was made prior to and post-removal of the resistor
without success. BTW, the board is still running fine,
person it was sold to uses it every day, nothing else wrong
with board.

Copied from google groups:

From: Homer J. Simpson ([email protected])
Subject: Re: AMD Palomino 2100Xp CPU

View this article only
Newsgroups: alt.comp.periphs.mainboard.msi-microstar
Date: 2004-03-18 14:42:50 PST

If you have the K7T266pro2(-RU) MS-6380 V2, and NOT any of the other boards
like the K7T266pro or the K7T266pro2-A series, you can remove a tiny (1mm x
0.5 mm) surface mount capacitor, C37, which will then allow you to run any
266MHz FSB TBred CPU. I made this mod to my K7T266 Pro2-RU Rev. 2 and I'm
currently running an XP2400+ TBred with complete success.

You should also download BIOS Version 3.7 from MSI's web site and reflash
your BIOS. This BIOS version provides proper identification of the TBred's
during POST. This BIOS version also has 48-bit LBA support for large hard
disks. Don't bother with BIOS Version 3.7 beta 4 as it does not have 48-bit
LBA support.

Check out this link at MSI's Forum for much info on this popular mod:

http://forum.msi.com.tw/thread.php?threadid=5723&boardid=13&styleid=1