'AMD Athlon XP 2600+ (1.92 ghz)'  or 'AMD Sempron 2400+ (1.67ghz)'......which one is better?

Posted on 2005-04-02
Medium Priority
Last Modified: 2008-01-09
Hello Experts,

I am having a confusion in deciding which processor to go for. I have an option of  'AMD Athlon XP 2600+ (1.92 ghz)' and  'AMD Sempron 2400+ (1.67ghz)'.  I am into multimedia and have to work on pretty heavy files. Can anyone help me clear this confusion by suggesting which one of these two i should go for and why? Can you also mention which existing processor is 'AMD Sempron 2400+ (1.67ghz)' equal to?

Thanks in advance
Question by:swiftguy
Welcome to Experts Exchange

Add your voice to the tech community where 5M+ people just like you are talking about what matters.

  • Help others & share knowledge
  • Earn cash & points
  • Learn & ask questions
  • 5
  • 2
  • 2
  • +8
LVL 16

Accepted Solution

samccarthy earned 2000 total points
ID: 13690944
I would definately go with the XP 2600+ processor over the Sempron.  The Sempron is basically in the same class performance wise as the Celeron.  It is a budget processor.  

The name is derived from the Latin word, Semper (meaning "always"), and the tech suffix -ron (which apparently means "budget processor"). Sempron should fall at a lower performance point than equivalent rated parts from other AMD processor lines. As this processor spans two platforms, it makes sense to compare the new Sempron to both the Athlon XP line of processors and current Athlon 64 processors. And as Sempron is a budget processor, it makes sense to compare it to Intel's budget line of processors.

The following can be found here http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2139

The slowest Athlon XP 166MHz FSB Thoroughbred-B was the 2600+ with a 12.5 multiplier (2.083GHz core clock). The highest K7 Sempron clock speed is 2GHz with a 12 multiplier.

What this does mean is that higher model numbered Semprons will be slower than "equivalent" Athlon XP processors. This does make a little bit of sense in that Sempron denotes a budget processor, which will be slower than the mainstream line. This is the same type of thing that we see when looking at a 2.8 GHz Pentium 4 versus a 2.8GHz Celeron. They have the same "speed", but different performance.

Expert Comment

ID: 13691040
I agree with Sam

In the choice of processor.

But it seems your version of the 2600+ is the BARTON version 11,5 X 166,67 Mhz ~ 1,92Ghz

It has the same performance as the thoroughbred because of 512 Kb L2 cache compared to 256 L2 cache

Sempron is a lot of things, but I think its mainly introduced to distinguish between real ATHLONS being 64 bit nowadays.

32 bit parts are all called sempron or duron these days to not "sully " the ATHLON name.

Semprons are thoroughbred/Barton(3000+) in socket A and even newcastles (3100+) in socket 754

check this link:

from this article:


LVL 23

Expert Comment

ID: 13692043
The sempron 2400 is about equal to an athlon XP 1900 or 2000 -- look at the bus speed, that will tell you approximately the equivalence -- sempron is a fine processor, but the faster XP would be better for your needs.  Price is also a fair way to compare.  There is a reasonable correlation between price and bus speed, hence performance, as long as you look at prices per merchant.  You just need to realize that the number following the sempron does not equate to the same number on an athlon XP.
Efficient way to get backups off site to Azure

This user guide provides instructions on how to deploy and configure both a StoneFly Scale Out NAS Enterprise Cloud Drive virtual machine and Veeam Cloud Connect in the Microsoft Azure Cloud.


Expert Comment

ID: 13693259
Something else you should realize is that no one is producing or plans to produce PCIe motherboards for Athlon XP, which severly limits your upgrade options for the future. If you can afford it, I would recommend going out and getting a 939 PCIe motherboard and a cheap Athlon64 3000+, this will give u plenty of performance at a decent price and leave u the options to upgrade in the future(AMD has no plans to change its current 939 chipset for future dual core and quad core processors, plus PCIe is becomming more and more common, especially in regards to grfx cards). My fav is the ASUS A8V Deluxe, goes for about $130-$140, plus the 3000+ for about $160.

Expert Comment

ID: 13693452
Are you planning on changing your motherboard too?
Personally... if you are on a budget... I would still get a S754 or S939 chip... athlon 64 2800(754) or 3000(939)
You get a whopping more performance... plus the chip will last you longer... otherwise you're buying old technology.
Motherboards for 754 are cheap cheap... 939 are a little more.

http://www.newegg.com/app/ViewProductDesc.asp?description=13-123-241&depa=1   (socket 754 board - $67.99)
http://www.newegg.com/app/viewproductdesc.asp?description=13-131-510&DEPA=1   (socket 939 board - $125 best choice)

and then pick yer chip :-)
I have the Asus A8V deluxe board... the 939 socket processor does the 2000 FSB, the 754 does the 1600 FSB.  The Socket 939 chip has better memory management feautures... making it faster... generally supports dual-ddr and stuff.  939 is the way to go.. in fact... I would call socket 754 old technology..lol.


Author Comment

ID: 13694485
Hello Experts,

Thankyou for the comments. I was browsing the net and bumped accross a comparision chart for AMD Athlon XP 2600+ and  AMD Sempron 2400+. It seems that there are only two differences between these two processors. Athlon XP 2600+ has got 512 L2Cache and its processor speed is 1.917 Ghz whereas Sempron 2400+ has 256 L2Cache and the processor speed is 1.667Ghz They both have the same bus speed i.e. 333 mhz.
Speed wise lot of people have said that processor speed doesn't matter its the performance that matters, so though the processor speed of Sempron is lesser than athlon XP, it should'nt matter that much. Bus speed is the same so the only major difference between these two processors is the L2Cache. Does this make a major difference in the performance of the computer? What exactly is this L2Cache? Is it very important to have higher L2cache?  Pricewise, ofcourse, the sempron processor is lesser than athlon xp. I have a limited budget but if it is not worth buying a sempron then i would like to go for XP. What do you think guys?


Expert Comment

ID: 13696058
The Athlon XP 2600+ Barton is easily the better processor, as others have noted. At its rated speed, it will deal well with multimedia work. Also, it will overclock easily to give performance close to a low end Athlon 64, provided you have a good heatsink and PC3200 RAM.

As for socket 939 and a low-end Athlon 64 vs socket A and Athlon XP, I have recently faced the same decision and decided in favour of the Athlon XP - it works a treat! It is true that a low-end Athlon 64 + socket 939 will be more future proof, and give a good upgrade path, but you will pay more:

 Athlon XP 2600+ Barton     $105 http://www.newegg.com/app/viewproductdesc.asp?description=19-103-507&DEPA=0
 Asus A7N8X-E deluxe mobo $92 http://www.newegg.com/app/ViewProduct.asp?submit=property&DEPA=1
 (A7N8X-X costs $63, and is fine if you don't want all the bells and whistles)


 Athlon 64 2800+ $120
 ASUS A8V deluxe $125

You would find little, if any, noticeable difference in performance.

The other plus for Athlon XP and socket A motherboards is that being mature products, all the bugs are out (or well-known and fix-able), and the prices are now dropping rapidly.

Expert Comment

ID: 13696555
An L2 chache is connected to speed.
As you know, the Bandwith of the bus determines how fast data can travel between the various hardware components (motherboard to harddisk to processor ect). Normally, the speed of the PC isnt limited by the number of gigahertz, but by the speed the individual parts can communicate with eachother(the bus). What am i trying to say, and what has this o do with the L2 chache? Basicly, the Speed of your PC relies on the bus width. The faster that is, the faster you can do things with it. But the bussesare limited, so a faster processor wont help. This is where the L2 Cache comes in. This cache stores, just like RAM data. In this case the data is the things that the PC should do (a waiting line (queue)). The bigger this is, the more commands can stand there on hold. As said before, the bus cant deliver commands as fast as a processor can process them, so the bigger the waiting line, the faster it goes.

(hope this made any sense to you, if not, tell me, and i use somewhat easier words. I got carries away with technoterms a bit ^^)
LVL 16

Expert Comment

ID: 13696559
The XP is a higher end processor than the Sempron, just as a full Pentium chip is a higher end processor than the Celeron.  As I said in my first post, the Sempron is basically AMD's equivelant to a Celeron.  It is true that processor speed is not the true measure of performance, but also the efficiency of the processor.  So comparing a faster 3ghz Pentium to an AMD chip running slower, but with the more efficient architechture rated at 3000+, basically gives you equivelant performance.

When comparing AMD chips, 32 bit flavor, a 2600+ XP will always be faster than a 2400+.  

L2 Cache is short for Level 2 cache, cache memory that is external to the microprocessor. In general, L2 cache memory, also called the secondary cache, resides on a separate chip from the microprocessor chip. Although, more and more microprocessors are including L2 caches into their architectures.  This makes them faster and more efficient.  Generally the larger the L2 Cache, the better your performance as the processor is taking data from the cache instead of having to go out to the slower main memory.

If you are going to do general applications: Internet, Word Processing, Home Banking, most general games, etc., the Sempron is a great processor.  The XP is just a little better.  If say you had the choice between the XP with 256mb ram and the Sempron with 512mb of ram for the same money, get the Sempron with the greater amount of memory.  In the grand scheme of life, these two processors are not that far off really in terms of performance.

As John stated, an ASUS A7N8X is cheap.  I have used yads of the A7N8X-E versions and below say 2800+, they are rock solid and stable.  At 3200+ they choke on stability.  I have 13 of these at the higher speed and have had to lower all of them to the equivelant of 2500+ to make them stable.  ASUS still does not have an answer for that behavior.  Also, I do not recommend the ASUS A8V.  I have 3 here in boxes that I have since replaced with Abit AV8 Motherboards.  The A8V's were all part of new systems, but also were not stable with FX-53 processors.  After 6 weeks of troubleshooting I narrowed the problem and could recreate it at will on all 3 systems.  It was 2 months ago that I reported it to ASUS.  Their status is that they have still not assigned it to a technician.
LVL 69

Expert Comment

ID: 13697051
The Sempron is actually an AthlonXP, since as mentioned before, AMD wants to associate the Athlon name with its Athlon64 line and is phasing out the Durn and AthlonXP: http://www.tomshardware.com/cpu/20040728/index.html

The Celeron-D is not to be confused with its predecessors; it actually is a competitive cpu, with a 533 FSB.

Expert Comment

ID: 13697205
Yeah, thats what I mentioned in this similar question: http://www.experts-exchange.com/Hardware/Q_21373828.html#13696825
Also as for the A7N8X-X being stable and mature, I've heard lots of complaints relating to this board, also at EE: http://www.experts-exchange.com/Hardware/Q_21372211.html#13685750
I was interested to hear that samccarthy had lots of trouble with the Asus A8V, I've heard good things about this board, and other people I've spoken to say its a challenge to set up, but well worth the effort. I am curious as to what the general problems with A8V were, and what u think of the Abit AV8, but thats for another question. Getting back to this one, I still think if you're planning to upgrade now, u may as well spend a little extra and buy a platform that will last a while. There is no point investing in harware thats on its way out.

Expert Comment

ID: 13698094
Sorry, I made a mistake in my last comment.

The Athlon 64 2800+ I quoted was socket 754, not 939. If you want to consider a well future-proofed alternative to a Sempron or Athlon XP, that would mean a socket 939 Athlon 64 3000+.

So the cost for a low-end Athlon 64 socket 939 would be:

Athlon 64 3000+ socket 939 $146 (retail pack)
ASUS A8V deluxe mobo       $125


Expert Comment

ID: 13698735
The Athlon XP 2600+ is enough for multimedia using, but for DV edit or other like DVD to DivX compare with Pentium 4 at the same clock speed, will be a bit slowly than P4 platform. But of course, have the better C/P value than the P4 platform and cooler CPU.

If you want to try the 64bit platform and want to upgrade your CPU in future, your choose just have the Athlon64 socket 939 base series.

But for a comment, choose Socket A platform still have it value, like the total of cost lower, and in the near future, all 939 AMD64 CPU will be lower their price, later but a new one also is a good choose.

Just a comment
LVL 12

Expert Comment

ID: 13703661
I think it's kind of funny that all AMD's would be "budget" processors by sammcarthy reasoning.  It's more likely that it's Standard Edition MicroProcessor substrate R with the "on" for whatever else it refers to, but all AMD's seem to be "on" endings.

Maybe it's a decathalon processors for running in the Olympics.

Some engineers just make me laugh.  The promoted 800mHz bus turns out to be the same old top speed 200mHz bus, with the HyperTransport being 400mHz or 800mHz, whichever way you decide to fudge the Data Specification Sheet.  Just like calling the crystal oscillating frequency the "speed of the microprocessor."

L1 Associative Memory cache is ok, but the size of a fly speck; besides being further split up into executeable code and data code.  The L2 cache is small at 512K.

The stated crystal resonant oscillating frequencies stated above, 1.92gHz and 1.67gHz, are absolutely meaningless to the context of the real microprocessor speed.  It's like putting the name Ford Lotus on a Hugo.

You missed the real points, the Data Specification Sheet and the Electrical Specification Sheet.  Specifically, the Timing Diagrams.

These guys can really fudge stuff!

The size of the bus, throughout, it's true speed, the sizes of both the L1 and L2 cache, the real frequency of the microprocessor, and motherboard compatibility are far more important than the marketing numbers of crystal frequency [the "with wax" or insincere] feigning to be the microprocessor speed, and false bus speeds.  Especially when it comes to something called "multimedia."

I also think it's hilarious that Intel has included a Logic Analyzer Interface [LAI], for debugging these chips.  Funny, we use to call it the maintenance bus, long ago, on mainframes, for Tektronics and Biomation.

Both Intel and AMD seem to have eliminated all reference to the Electrical Schematic detailing the actual CPU Clock Cycle Timing Diagram!  Although, after a long search, I did find some really interesting stuff in the Errata Sheets, specifically, all the microprocessor architecture that does not do what Intel says it does.  I can't stop laughing at some of this stuff!

In the IA-32 Specification Updates Errata Sheets, 27283802.pdf, 9600001, this just floored me:  "But in 1-MByte mode only, the vector address for the illegal opcode interrupt is not generated correctly and a random vector address is generated."

Which first, implies that whatever the chip is, it has a 1-MByte mode, which may be read more correctly as a 1MHz mode, a truer indication of chip speed, but secondly, and more importantly, that the architecture is falling all over itself.

While it's just the 8XC196NP or NU, it has the same fancy stuff that's supposed to be in the current 64-bit architecture, things like Indirect Addressing With the Stack Pointer, etc..  They'll show a timing diagram with not real times, just mnemonics for various clocking features.  And if it was true for a 16-bit controller, it makes you wonder what is wrong with tthe 64-bit architecture that they're not telling anyone about.

If you look at the Benchmarks, you'll see that both AMD and Intel are usually within plus or minus ten percent.  That's the spec anyway, +/-10% so it's saying nothing.

One Yalie thinks that modern CPU's can do six adds simultaneously; but he also thinks a 2GHz clock cycle is 0.5 ns, it is not, it is 0.25 ns.  The reason is that the clock has to go both up and down, not just in one direction.  He's thinking look-ahead logic and time-splicing, but that is not quite how they work.  The basic clock is supposedly 100MHz, but even that is doubtful, thereafter, it is supposedly time-spliced.  To do six addes simultaneously you would need at least 18 registers, one for each operand, and one for carry which feeds back on the adder output through yet another adder to arrive at a one clock sum.  Notice that you'd also need another adder.  The microprocessors simply do not have 6 complete microprocessors inside.  And the look-ahead logic for fetches doesn't quite work the way imagined either.  The processor attempts to do a fetch, and if it doesn't have the data, it can execute maybe one or two very different operators which would have no effect on the context of the fetch, but that set is limited.  And if the processor requires a result, it does a restart of the instruction, a hardware designed and builtin microjump back to where it was supposed to do the first instruction, the fetch.  This takes many cpu clocks.

Some of these chips take a thousand cpu cycles to do one memory write.  But they mask the time by burst fetching ahead of time, again, look-ahead logic, which is not as infallible as they make it seem.  And just because you multiple a 100MHz clock by 24, it has nothing to do with how many instructions can be excecuted per second, rather, it has to do with distributing the phase of the clock throughout the microprocessor.  Because it takes time for a junction to react to electricity, different sets of gates at different distances from the Master Clock and its subset of timing clocks, the only way to get every gate the same proximate phase is to splice the Master Clock into discrete phasal distributions.  And it means that everybody gets the clock at the same time, to prevent the "race" problem and it does not mean that it multiplies the number of instructions by 24.

The bus speed of up to 800MHz is achieved not by the clock frequency, but by having four 200MHz buses.  These buses are usually DMA controlled, and that is just another form of memory busrting.  But that bursting puts the buses in the Hold State for a period in the range of those 1000 cpu clocks, wherein, the cpu can no longer request the bus and is on hold.

Some of the errata show that the Hold state doesn't always work right.

Nor do some fetches.

Therefore, the speed starts to get eaten up by error fixing both by hardware and by software.

Some of the Xeon's and Opteron's have large L2 and some L3 caches, 8MB and so on, but the tiny size of the L1 cache and it's being split into two slows the whole thing down anyway.  If you have to do a fetch outside of the associative memory, a lowly 64kX64k L1 for code and data, everybody has to wait, including the data bus, execution pipeline, and so on.  The L1 cache should be at least 1 meg, and 2 meg if it's split into two, which is completely unnecessary if the engineer knows the usage of an Associative Memory in the first place.

Last I checked the Opteron's were about $700.00 each, and you'd need four for a good board.  The boards' also are around $800.00 or so.  I don't know what the Xeon's are, but probably close if they are supposed to be comparable.  So, it's really about $4,000.00 for such a setup.  That is, a true 64-bit setup.  I know you can do one uP and one normal motherboard, but it isn't going to have anywhere near the stated performance of a gigaHertz 64-bit true system.

The MP beats the XP, that's their marketing.  If you have a limited budget, as you say, the Athlon has a very good reputation among server builders.

Try either freeman118 or J-A-L suggestion.  And browse some sites like Intel, AMD, Asus, and Tyan, get the docs and see what these microprocessors and motherboards actually do.  Stop thinking that that ole PR "clock speed 2.4GHz!" means anything other than "McBurger with cheese!"
LVL 16

Expert Comment

ID: 13703892

      I never said that all AMD processors were budget.  I said that the Sempron is inferior to the XP processor as a Celeron is to a full fledged Pentium 4.  What I also said, in response to someone else, is that if you compare the 2 AMD processors, one at 2400+ and one at 2600+, the 2600+ is the faster chip.  I am a big fan of AMD, having 5 machines at home running them and at least 60 at work.

      If you look on the AMD website and at the Sempron processors, AMD compares them to their equivelant Pentium processor, ie, the Celeron.  A simple answer for a simple question.

LVL 12

Expert Comment

ID: 13704782
Sorry, I thought you meant that "on" stood for "budget."  :)

Semper Fi.
LVL 16

Expert Comment

ID: 13705540
All cool...  Even for a Marine

Author Comment

ID: 13860154
Thankyou all for your comments and helping me out in buying the right processor. Ultimately i went for an AMD Athlon XP 2400 processor and its working great.

LVL 16

Expert Comment

ID: 13862702
Great and Thanks!

Featured Post

The Eight Noble Truths of Backup and Recovery

How can IT departments tackle the challenges of a Big Data world? This white paper provides a roadmap to success and helps companies ensure that all their data is safe and secure, no matter if it resides on-premise with physical or virtual machines or in the cloud.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

This is about my first experience with programming Arduino.
New style of hardware planning for Microsoft Exchange server.
Do you want to know how to make a graph with Microsoft Access? First, create a query with the data for the chart. Then make a blank form and add a chart control. This video also shows how to change what data is displayed on the graph as well as form…
In this video, Percona Solution Engineer Dimitri Vanoverbeke discusses why you want to use at least three nodes in a database cluster. To discuss how Percona Consulting can help with your design and architecture needs for your database and infras…

765 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question