Re:Intel Builds In More Built In Obsolescence

[blakerwry]

Re:Intel Builds In More Built In Obsolescence

I think this areticle is really bias against intel and goes against what it itself is saying... the author is critisizing in this statement "Hopefully, next time, the technical community as a whole, both paper and web, will recognise that fact — and not hammer one platform for its lack of an upgrade path, while blithely failing to recognize an identical failure in that platform's successor, simply because the numbers behind the word "Socket" haven't changed"

He has said that each intel p4 chipset allows ~25% raise in clock speed.. then he talks about Athlon chipsets and how the kt133a will run the latest and greatest Athlon XP CPU's an hence is better in terms of longetvity.... what makes so little sence is that the way he compares the intel vs Athlon chipsets differently.. If he compared them the same way he would have to say that there have been 7 VIA chipsets and atleast 2 each from SiS nVidia, and AMD for the Athlon line of CPU's... vs the 4 or so p4 PC chipsets...

If the kt133a could really run the newest T-bred CPU's then all the other chipsets have really been in vain. But of course.. how many kt133a mobo's can really even run an athlon XP? not many... even some kt266a's cant run thoroughbreds.


It is true that intel seems to change socket architecture quickly and has a list of different chipsets with little difference... but I don't neccessarily think it's any worse than going the Athlon route considering that even though the sockets remain more consistant with AMD, there are twice as many chipsets.

 

[cquinn]

What article? :-?

 

[Mercutio]

There are at least two KT133A boards out there that'll accept a 1600+. Of course, a 1600+ with SDRAM is a slightly depressing prospect. Architecturally, the KT133 probably *can* handle a tbred. Why someone would want that is another matter.

... and the fact that an XP CPU of any sort is depressing with plain old SDRAM is the reason for later chipsets such as the KT266.

I think the author made a valid point. Intel seems to have zero concern for backward compatibility. That's not wholly without justification - the overwhelming majority of CPUs and motherboards are never seperated, once assembled, but it's still something of a raw deal.

 

[Buck]

cquinn, the article on our Home page.

Blake, the variety of chipsets isn't the issue presented in the article, instead it is the requirement to upgrade to a new chipset in order to use a newer CPU. This is less of problem with Socket-A.

 

[Cliptin]

cquinn, the article on our Home page.

Blake, the variety of chipsets isn't the issue presented in the article, instead it is the requirement to upgrade to a new chipset in order to use a newer CPU. This is less of problem with Socket-A.

As I see it, the difference is in the way the companies making motherboards have overlapped the technologies. This eases the pain of an upgrade. With the Intel solutions, quite often it is a significant investment.

It is not a problem that the older chipsets don't support the newest chips. The advantage is that the newest chipsets support the older processors.

Action point: If at all possible, lead upgrades with the motherboard.

 

[time]

Blake, as far as I can see the author was simply stating an oft repeated dealer complaint. I'm afraid it's all true.

Yes, some KT133A boards will run XP2600+ (eg Soltek), and many will run at least XP2100+. Which KT266A boards won't run Thoroughbred? I only recently upgraded a couple of KT133 (not A) running Duron 650 to Duron 1300 - which is a different chip design.

There are just as many third party chipsets for P4 as for Athlon, so I don't follow your arithmetic. But Intel is particularly infamous for building obsolesence into their own chipsets.

The article's point was that there is effectively no upgrade path with Intel's platforms. It's not great with AMD either, but he correctly observed that it's not as bad.

 

[blakerwry]

but the Author of course, doesnt state that you can start off with an 845PE and a p4 1.6gHz and get to atleast 3.5gHz(which is what im doing)... that is a double in speed.


We should all wish for the 486 days, eh? go from 33 to 120mHz... over drive processors.... ISA memory add ons... No, not really, but it would be nice to have this kind of upgradability in a new PC...


I think if you shop wisely and stay 1 or 2 steps behind the fastest technology you'll be alot better off (hindsight is 20/20) and you'll be able make your machine go atleast twice as fast as when you first put it together.

Case in point: I bought a XP 1700+ w/ Soyo Dragon plus for ~$360 when they were both pretty new... unfortunately the Dragon + doesnt support t-breds... so i might not be able to upgrade it past 2100+... at the time the t-breds hadn't been released so there was no way for me to know.

I also bought a shuttle AK32L and Duron 1.3... however, the board had been out for a year and I was able to learn from my past bad experience... So I made sure to check what it supports and what it doesn't... the board supports any 133mHz FSB Athlon XP (currently upto 2600+)

The soyo cost me ~$130... the Shuttle cost me $60....

if you're buying the newest things you really have little idea of what it will or wont support in the future... but if you wait and get the next generation.. say 6 months later you will have a much better idea of what will work and what wont.

 

[Buck]

but the Author of course, doesnt state that you can start off with an 845PE and a p4 1.6gHz and get to atleast 3.5gHz(which is what im doing)... that is a double in speed.

Which is, of course, exactly what the author was pointing out. You need a new Intel chipset, the 845PE, in order to get that range of processor support.

 

[Mercutio]

...But if you had purchased an 845G four months ago, you'd be really upset right now.

 

[Tea]

With all due respect, Your Holiness, you've lost the plot. Intel architecture vs AMD archiitecture gives you, in general, an approximate 3X better upgrade prospect with the AMD.

 

[blakerwry]

I haven't lost the plot, i just dont see it.

No kx133 can run a socket AMD... the lifetime was only a few months....

no kt133 can run an Athlon XP... this unstable chipset also only lasted a few months....

very few kt133a boards can run Athlon XP's....

all kt266/266a boards and above can run palomino Athlon XP's, but some can't run thoroughbreds...

kt333 boards can run all 133 FSB XP's.. but may not be able to run 166FSB athlon XP's.... what was the point of this chipset?

kt400 can run them all.

 

[blakerwry]

...But if you had purchased an 845G four months ago, you'd be really upset right now.

yes, but i didn't... what's wrong with the 845g? pc2100 support, hyperthreading support... 533/400FSB support...

not supporting DDR333 is barely worth mentioning considering the performance over DDR266 boost is probably barely, if at all, noticable (although it is benchmarkable)

 

[blakerwry]

not to mention the new southbridge that has audio, LAN, USB 2.0, etc

 

[honold]

ddr333 works fine on 845g boards, almost all of them have the multiplier for it

who gives a good god damn about running an athlon xp on pc133 sdram?

 

[Mercutio]

It's kinda nice when you have piles of PC100 chips laying around. Remember when 256MB SDRAM DIMMs were $15 each?

Someone who bought an MSI K7T two years ago can stick a pretty up-to-date XP1800 on it (and probably faster than that. time's Soltek is an even better example). The guy who bought that 'board didn't have to pay an extra $100 for a new board to move up from his Tbird 800. From the standpoint of upgradability, that seems like a pretty good deal... he has an extra $100 to throw at what's really important, like more hard drives.

What could you do with a 2-year-old P4 board?

The listings on pricewatch for new i845 boards start at $29 including shipping, btw. The cheapest super7, i815 and i810 boards are all more expensive. Not that I'm drawing any conclusions from that. I just find it amusing.

Being able to upgrade Socket 463-based systems has been one of the main reasons I've chosen to use AMD chips for these last two years.

 

[Mercutio]

Upon reflection, everybody remember the stunningly long life 440BX had, right?

I think Intel (largest producer of motherboards) learned a lesson from having a single product outlive its planned product lifetime by 18 months (12 months at the very least).

What followed BX? Alphabet soup. i810 and a couple of variants. Then i820 (rambus!), i815 and some more undifferentiated variants. Then i840 and the million and one different revisions of i845.

Any wonder I don't try to keep Intel chipsets straight?

 

[time]

No kx133 can run a socket AMD... the lifetime was only a few months....

KX133 ruled for only six months or so because Via was incredibly slow releasing it. Slot A was widely available for at least 12 months, thanks to AMD 750.

no kt133 can run an Athlon XP... this unstable chipset also only lasted a few months....

Unstable!? WTF? It ran just fine, and still does. It just didn't overclock very well when compared to the 'A' version, although 15% was easily achievable. The fastest Athlon was 1GHz when KT133 was released; it supports up to 1.4GHz, which is a 40% jump. But most KT133 were sold with CPUs from 600 to 950MHz, so the range was more like 133%.

very few kt133a boards can run Athlon XP's....

Oh yeah, name one that can't!

all kt266/266a boards and above can run palomino Athlon XP's, but some can't run thoroughbreds...

Some manufacturers had glitches, but are you telling me they weren't resolved?

kt333 boards can run all 133 FSB XP's.. but may not be able to run 166FSB athlon XP's.... what was the point of this chipset?

Official support for DDR333. KT333 is a good chipset - from my point of view, the main advantage of KT400 is inbuilt USB 2.0 :-?

 

[LiamC]

Quote:
very few kt133a boards can run Athlon XP's....
Oh yeah, name one that can't!

Abit KT7A Rev 1.0 ~ 1.2

Sometimes boot, sometimes not. Whilst I play Powerball, I dislike lotteries in my computers

Quote:
all kt266/266a boards and above can run palomino Athlon XP's, but some can't run thoroughbreds...
Some manufacturers had glitches, but are you telling me they weren't resolved?

I have an Iwill KA266 that (with latest BIOS) will run a Palomino but not a Thoroughbred. I don't expect Iwill to do anything about it, but it would be nice if they did...

 

[blakerwry]

no kt133 can run an Athlon XP... this unstable chipset also only lasted a few months....

Unstable!? WTF? It ran just fine, and still does. It just didn't overclock very well when compared to the 'A' version, although 15% was easily achievable. The fastest Athlon was 1GHz when KT133 was released; it supports up to 1.4GHz, which is a 40% jump. But most KT133 were sold with CPUs from 600 to 950MHz, so the range was more like 133%.

At the time, this chipset was "known" for instability and is one of the reasons VIA got it's reputation for not being as stable as AMD as far as chipsets are concerned.

very few kt133a boards can run Athlon XP's....

Oh yeah, name one that can't!

Abit KT7A (kt133a)
Asus A7V133 (kt133a)
ASus A7V266 (kt266)
MSI has one that should support XP's Soyo's can support palominos

all kt266/266a boards and above can run palomino Athlon XP's, but some can't run thoroughbreds...

Some manufacturers had glitches, but are you telling me they weren't resolved?

Yes, they weren't resolved. Soyo's flasgship motherboard that was meant to support XP's doesn't support T-breds (Soyo Dragon plus, kt266a) Also, shuttle's AK31(kt266) cant run T-breds either. ... and as earlier mention, the Asus A7V266 (kt266) cant run any XP CPU...


(all my info has been taken off the manufacturer's website, if they provide inacurate information about their specs there is nothing I can do)

 

[blakerwry]

Basically, I started this thread to point out that whether you go Intel or AMD with a particular motherboard, there are no guantees as to the future upgradability of that motherboard.

Given the recent track record of both Intel and AMD motherboards, I don't think you can say that one has necessarily been better than the other in terms of longevity.

Each new motherboard/chipset offers new features that make it worth upgrading to for some people... but that may or may not include the use of a new CPU.

 

[honold]

the kt133/a was really where via's terrible stability/compatibility reputation began

 

[time]

At the time, this chipset was "known" for instability and is one of the reasons VIA got it's reputation for not being as stable as AMD as far as chipsets are concerned.

Any stability problems were not down to Via. I say again, KT133 worked just fine. Not as fast as KT133A, not as overclockable, but still perfectly usable. KT133A implementations were more mature and had less compatibility issues.

Abit KT7A (kt133a)
Asus A7V133 (kt133a)

From Paul's Unofficial ABIT KT7A-RAID FAQ:

KT7A/KT7A-RAID versions 1.0 to 1.2: All Athlon (Thunderbird) processors with both 100MHz and 133MHz FSB. AthlonXP processors are not officially supported due to a minor timing problem that can require you to hit reset during a cold boot. Unofficially, however, few people seem to experience this problem and these processors work fine. One user had to set the CPU Driver value to 1 (instead of 2) to achieve stability with an XP1900+ on a v1.1 motherboard.
KT7A/KT7A-RAID versions 1.3: All Duron, Duron (Morgan core), Athlon (Thunderbird) and AthlonXP processors to 2100+ are supported. AthlonXP processors at 2200+ are not supported.

Yes, they weren't resolved. Soyo's flasgship motherboard that was meant to support XP's doesn't support T-breds (Soyo Dragon plus, kt266a) Also, shuttle's AK31(kt266) cant run T-breds either. ... and as earlier mention, the Asus A7V266 (kt266) cant run any XP CPU...

I confess I was thinking of KT266A here. KT266 implementations such as MSI's just sucked. But in any case, you're right - early KT266A boards turned out to be incompatible with T-bred (due to an error in AMD's preliminary specs, I've read). Some manufacturers renamed their boards to indicate the change (and some like Epox didn't).

However, none of these examples can be attributed to a design decision by either AMD or Via.

 

[Tea]

OK, your Holiness, let's get down to tin tacks. Like Mercutio I don't even try to keep track of Intel's lunatic-pace of change for the sake of change anymore, but see what happened after Socket 7 and try to list the main ones. AMD first:

Super 7
Slot A
Socket A (100 FSB)
Socket A (133 FSB)
Socket A (166 FSB)

That's it. Entire list. Now let's do it for Intel, over the exact same time period

Socket 8
Slot 1 (66 FSB)
Slot 1 (100 FSB)
Socket 370
Socket 370 (Coppermine)
Socket 370 (Tualin)
Socket 423
Socket 478
Socket 478 (533 FSB)
Socket 478 (some new thing that I can't be bothered keeping track of)

Twice as many, and I'm not even counting all the idiocy with RAMBUS and the games with 133MHz Slot 1, and the lunacy of the P4 chipset market, or anything else I forgot.

 

[Mercutio]

Maybe I have a bit longer memory, but I remember many, many knocks at Via prior to the Athlon CPU. I can recall hobbyists bitching and moaning about Apollo Pro (which was a great chipset, As far as I cared), and I'm sure if I could remember clear back to the 486 days, when about 20 different companies made motherboard chipsets, there were people complaining way back then, too.

People who bought Abit KT7x boards and early Soyo Dragon boards were buying dogs. I've got a couple KT7 boards sitting around with low-end Durons on them because they won't take anything that's faster and worthwhile to install. Guess what? I learned my lesson and moved on.

Soyo isn't a top notch company, by any stretch of the imagination. Lately they've been trying to move into the hobbyist market, but prior to that, most of their stuff was utterly middle of the road kit, with features I'd expect from a board targeted at resellers like Buck or Tannin (much like, say, FIC or DFI).

 

[Dr Bombcrater]

the kt133/a was really where via's terrible stability/compatibility reputation began

VIA already had a bad reputation back in the MVP3 days, which it roundly deserved because the MVP3 had some serious problems.

The KX133/KT133(A) chipsets were fine, provided you bought boards from quality manufacturers. Epox, Soltek and some others all built wonderfully stable boards around these chipsets, so there clearly isn't anything intrinsicaly wrong with them. I personally built hundreds of systems around motherboards from these companies, mostly 7KXA, 8KTA2, 8KTA3 and SL-75KAV models, and they were every bit as stable as any Intel-based boards and have proved to have significantly longer lifespans. Because of the cheapness of PC133 I'm still using the 75KAV for builds, and some of the older ones have been upgraded to XP2400 and XP2600 chips.

I find it strange that so many people slam VIA, yet totally ignore the fact Intel, AMD, SiS and ALi all produced their share of buggy chipsets. Perhaps it's because VIA chipsets do tend to out-live everything else.

## Doc

 

[Clocker]

I agree. I don't build hundreds of systems but the only VIA board I had major Major MAJOR problems with was an MVP3 based board (early on). Later I had an Epox MVP3 that was great, however.

KX/KT133/KT133a have been good to me.

My first SiS product (SiS735, I have tow ECS K7S5A's with SDRAM) has been great.

I must say, that my old BX chipset board does hold a special place in my heart.

C

 

[blakerwry]

OK, your Holiness, let's get down to tin tacks. Like Mercutio I don't even try to keep track of Intel's lunatic-pace of change for the sake of change anymore, but see what happened after Socket 7 and try to list the main ones. AMD first:

Super 7
Slot A
Socket A (100 FSB)
Socket A (133 FSB)
Socket A (166 FSB)

That's it. Entire list. Now let's do it for Intel, over the exact same time period

Socket 8
Slot 1 (66 FSB)
Slot 1 (100 FSB)
Socket 370
Socket 370 (Coppermine)
Socket 370 (Tualin)
Socket 423
Socket 478
Socket 478 (533 FSB)
Socket 478 (some new thing that I can't be bothered keeping track of)

Twice as many, and I'm not even counting all the idiocy with RAMBUS and the games with 133MHz Slot 1, and the lunacy of the P4 chipset market, or anything else I forgot.

If i were to nitpick with Tannin, which would surely be against my better judgement, I would have to disagree with this a bit.

On the AMD side, as we've already said there seem to be distinctions between the different XP cores and the ability to run them consistantly on several boards... so I think you should add a segment for Palomino and for T-bred individually...

Next, on the Intel side, I think you should remove several... starting with socket 8. This was around for the pentium pro, which competed with the pentiumMMX... and I think actually came out before the MMX. But surely came out before super 7 (defined as 100mHz or higher FSB socket7 with AGP and SDRAM) It was not really meant for PC use anyway, becuase of it's slower clock and performance compared to the pMMX it was only valuable because of it's large L2 cache and dual/quad configurability.

I also think you should remove the first mention of socket370... all socket370 boards can support coppermine core.. infact, this is the core they were made for. And as most of us know, socket370 is very similar to slot 1 so adapters are available making upgrading possible in almost all situations.

Lastely, I am not convinced that the article is correct in saying that 845E/G owners need to upgrade... according to Intel (and 3 mobo manufacturers), the 845G supports the newest intel chips out. I have also read that the 845E can be upgraded to support HT with a BIOS update. So I think that you should remove the last entry of the socket 478.


That makes my view 6 vs 7

 

[blakerwry]

I'm not sure where to place the RAMBUS people... they knew they were buying PC's that were doomed to obsolesence... but where the fastest thing available at the time.

In my mind this was their choice and they got what they paid for. Besides, anybody who bought a RAMBUS motherboard(knowing what they were buying) probably wants/needs top notch performance and upgrades their motherboard at ~1 year intervals to get that performance.

 

[blakerwry]

im glad I started a hot topic here... i was getting bored

 

[Buck]

Lastely, I am not convinced that the article is correct in saying that 845E/G owners need to upgrade... according to Intel (and 3 mobo manufacturers), the 845G supports the newest intel chips out.

Close, the 845G requires B-1 stepping to support HT. The 845GL does not support HT, whereas the 845GE and 845GV provide full HT support. Unlike the other three G chipsets, the 845GL does not support 533Mhz FSB. The 845E does support 533Mhz FSB, but the 845 (no suffix) does not. The 845 does not support HT either.

 

[Buck]

If anything, this discussion has made me become more familiar with newer Intel chipsets, even though I don't sell them.

 

[Handruin]

It's kinda nice when you have piles of PC100 chips laying around. Remember when 256MB SDRAM DIMMs were $15 each?

Someone who bought an MSI K7T two years ago can stick a pretty up-to-date XP1800 on it (and probably faster than that. time's Soltek is an even better example). The guy who bought that 'board didn't have to pay an extra $100 for a new board to move up from his Tbird 800. From the standpoint of upgradability, that seems like a pretty good deal... he has an extra $100 to throw at what's really important, like more hard drives.

What could you do with a 2-year-old P4 board?

The listings on pricewatch for new i845 boards start at $29 including shipping, btw. The cheapest super7, i815 and i810 boards are all more expensive. Not that I'm drawing any conclusions from that. I just find it amusing.

Being able to upgrade Socket 463-based systems has been one of the main reasons I've chosen to use AMD chips for these last two years.

Woah! So I can put an XP 1800+ on my MSI K7T Pro2A?? Funny enough, I did buy this 2 years ago with the athlon 1.2 GHz!

 

[Mercutio]

blakerwry, Intel had been making statements that RAMBUS was to be the future of PC memory technology for several years before a RAMBUS-based Intel PC became available. As far as anyone knew, Intel (major investor in RAMBUS, after all) was going to ram RDRAM down everyone's throat. There was quite a bit of analyst speculation to support that notion.

Doubtless DDR would never have been adopted had AMD not moved away from 100% intel compatibility.

Socket 8 was a perfectly valid platform at the time it was available. I worked on a lot of Compaq and IBM Pentium Pro desktop machines in 1997 and 1998. On the other hand, I've worked on very few slot 2 or socket 603 desktop machines at any point in time. You'll note that Tannin didn't even get into that hornet's nest, which IMO could rightly make the count even more lopsided.

 

[Buck]

Just to make things more confusing, this is what Intel lists as their present line of chipsets:

Server/Workstation
E7205
860
E7505
E7500
E7501
460GX
E8870

Performance/Mainstream Desktop
850
850E
845
845E
845G
845GE
845GV
845PE

Value Desktop
845 (redundant)
845GL
845GV (redundant)
815
815E
815EP

(Yes, these are all different chipsets, not motherboards. For every chipset listed, there is at least one variety of motherboard.)

 

[blakerwry]

I should note that the 'G' series is just the addition of onboard graphics to the otherwise same chipset.

Intel has to compete with SiS, and on the other side AMD, VIA, and Nvidia... in Server, workstation, and Desktop computers... so they have to have a wide variety of chipsets for every segment.

845 is the only "multi headed hydra" that you have to worry about for new p4 desktops.

 

[honold]

pc1066 still beats dual-channel ddr: http://www.aceshardware.com/read.jsp?id=50000346

the mvp3 had a (deservedly) bad rep too, but via hit critical mass when the athlons got popular and collapsed under the load

i've gotten screwed on countless via boards from countless manufacturers. abit kt7a/sblive/ide/etc (all results of the lack of pci bus parking), iwill kk266 wacko xp install incompatibility (must remove geforce2 agp video card during install, usb must be disabled during install), ethernet issues with the via epia (via manufactured, all via parts), etc

anybody see the anandtech walkthrough of their labs? their qa department looked like the lunch room at a gas station.

 

[honold]

i don't see why everyone's jamming intel for sub-revisions of all things. via is the KING of adding 'a' to a number and calling it new. at least there are solid differentiations in the intel offerings.

you can't compare SOCKETS to CHIPSETS.

'amd cpus have socket a, intel has 845g, 845pe, 845gl, 850e!'

 

[Mercutio]

AMD has socket A. Intel has Socket 370, 423, 478.

Happy?

 

[honold]

they really only have socket 478 for desktop offerings

p3s are for servers only, celerons have been moved to 478, and 423 is unused

 

[mubs]

Cripes! Life has gotten so complex, has it not? I used to know most chipsets quite well, then about 2 years ago, I completely, completely lost it, especially with Intel & Via chipsets. Don't have a clue now. :evil:

My trusty ole BX based dual is chugging along nicely, thank you! When the Hammers and their chipsets have been released and have stabilized, I'll think about moving up.

As somebody pointed out earlier, I think the BX is the best chipset ever released.