PDA

View Full Version : Intel to put the "hammer" on the Hammer & 970!


peter2002
Nov 18, 2002, 07:35 AM
In a surprise announcement, on Monday, Intel announces that a new 3.2GHZ P4 with HyperThreading and a new 800MHZ front side bus will be released in the second quarter of 2003.

Intel's new 800MHZ bus will increase processor bandwidth to 6.4 gigabytes per second, which exceeds the 5.1 gigabyte/p/s bandwidth of the AMD Opteron to be released in Q2 2003, and about matches the bandwidth of the IBM 970 to be released sometime in the third quarter of 2003.

Peter

http://www.theinquirer.net/?article=6286

http://www.theinquirer.net/?article=6285

ddtlm
Nov 18, 2002, 11:22 AM
peter2002:

which exceeds the 5.1 gigabyte/p/s bandwidth of the AMD Opteron to be released in Q2 2003
You are confusing FSB bandwidth with memory bandwidth. The Opteron essentially has no FSB... the question people should ask is what sort of memory is Intel going to be pairing with that FSB?

and about matches the bandwidth of the IBM 970 to be released sometime in the third quarter of 2003
Intel's solution is one 64-bit channel whereas IBM's is two 32-bit channels, so IBM cannot actually reach that number in one-way bandwidth. It would be hard to say which FSB is better at this point.

ffakr
Nov 18, 2002, 12:14 PM
AH... i bet it comes with a new pin-out too...

The other edge of Intels agressive upgrades are the regular and routine obsolescence of hardware. No company has ever pushed 'new' technology onto the scrap heap like Intel.
While AMD has done a good job of keeping compatability while keeping power respectable, Intels processor interfaces have moved from flipchip P3, to P4, to new P4, to various Xeon pinouts... and let's not even talk about what a great idea Slot1 and SlotA were.

Sure, new and faster is better, but it would be great if they could support their older tech a bit longer... perhaps for as long as it is still useful.
I don't want anyone to stop innovating, but don't make changes that seem to push people into buying new hardware just for compatability's sake.

peter2002
Nov 18, 2002, 12:15 PM
The way AMD is losing money, they won't be around in a year. The Opteron release has been pushed back to Q3 2003. It was supposed to be released now. It doesn't matter how good your chip is if you are losing billions after billions after billions. Pretty soon, it adds up to some real money.

Peter :)

http://www.theinquirer.net/?article=6292

ddtlm
Nov 18, 2002, 12:41 PM
ffakr:

and let's not even talk about what a great idea Slot1 and SlotA were
I guess you were being sarcastic... Slot-1 was a very long lived connector, and a great idea. Using adaptors I've heard of chips well over a ghz working in boards 3 years old, quite amazing on the PC side where often mobos are simply replaced with the processor (which can be a good thing).

The other edge of Intels agressive upgrades are the regular and routine obsolescence of hardware.
Yeah, I'm sure you'd rather have PC's that never change, much like Macs never do, so that Macs wouldn't be so far behind? I've heard a lot of limp put-downs for PCs, but this has got to be one of the best.

New switcher add: "With my PC, I was, like, not the fastest on the block in a week! So I just, like, got a Mac and now I don't have to worry about being out of date for, like, 6 months!!"

springscansing
Nov 18, 2002, 12:53 PM
Originally posted by ddtlm
ffakr:


I guess you were being sarcastic... Slot-1 was a very long lived connector, and a great idea. Using adaptors I've heard of chips well over a ghz working in boards 3 years old, quite amazing on the PC side where often mobos are simply replaced with the processor (which can be a good thing).


Yeah, I'm sure you'd rather have PC's that never change, much like Macs never do, so that Macs wouldn't be so far behind? I've heard a lot of limp put-downs for PCs, but this has got to be one of the best.

New switcher add: "With my PC, I was, like, not the fastest on the block in a week! So I just, like, got a Mac and now I don't have to worry about being out of date for, like, 6 months!!"

hahaha, good call.

It's true though. Although with Apple, the updates are like.... KABOOM!

Well.. they used to be. :-(

ffakr
Nov 18, 2002, 02:28 PM
Originally posted by ddtlm
ffakr:


I guess you were being sarcastic... Slot-1 was a very long lived connector, and a great idea. Using adaptors I've heard of chips well over a ghz working in boards 3 years old, quite amazing on the PC side where often mobos are simply replaced with the processor (which can be a good thing).
I'm not entirely sure what you are saying here... Slot 1 was around for a while, but the coppermine effectively killed it. It existed, but it was essentially dead technology halfway through it's life. It ran up to 1GHz processors, but Intel imposed strict limits in their chipsets to force end users to upgrade. LX motherboard? No cards that would run over what? 333? Couldn't make the processors run higher mulitpliers or upgrade the bios to recognize faster processors? No... Intel just wanted to make you buy a new system.
Not to mention that Intel changed the cartridge form making it very difficult for non-techies to figure what heat sink went with their processor.
yea it was great tech... most Slot 1 modules needed two fans and an extra case fan behind the processors.

Yeah, I'm sure you'd rather have PC's that never change, much like Macs never do, so that Macs wouldn't be so far behind? I've heard a lot of limp put-downs for PCs, but this has got to be one of the best.

No, but don't PLAN for obsolescence. Don't design your product line with the intention of FORCING your end users to toss their box every year. How fast can you upgrade an LX machine to? How fast can you upgrade a BX or even a BX2 machine to? Is there some magic limit that prevents you from running a P3 with a 10x multiplier in an LX board? Couldn't you design the chipset from the START to handle that instead just the speeds that are offered at launch?
How is it that you can slap a dual G4 processor card in an Old G3 but you can't put a PIII in a PII motherboard?

that's what I'm talking about... not killing innovation in the computer industy... just making sure that you don't push changes with the intention of screwing over legacy users (and I use the term legacy very loosly... I mean relatively new, useful hardware).

ddtlm
Nov 18, 2002, 04:22 PM
ffakr:

It ran up to 1GHz processors
"Slocket" apaptors allowed socket-370 chips apparenltly well above 1ghz to work on the old mobos.

Intel just wanted to make you buy a new system
You mean, in 3 years when the current ones wear out? Or, like people with iMacs have to do? A charity where I do work just got someone turning in a P-133 (which will be put to use as soon as I get a disk for it). I suspect a lot of people run them into the ground and then get a whole new computer.

most Slot 1 modules needed two fans and an extra case fan behind the processors.
Misimformation. My Linux box at work has a slot-1 1ghz P3, one fan, only other fan in the case is the PSU fan. Upgraded it from a P2-450, not bad.

How fast can you upgrade an LX machine to?
My first slot-1 computer was my first computer that I owned (parents had Macs), and it was a Gateway P2-233 with a 440LX chipset. It has been upgraded to a Athlon 1533, with a GF4Ti. The 440LX chipset did not get in my way cause I just got a series of new mobos each time I wanted to go faster, over the nearly 5 years that I have owned it (new Xmas 1997-1998 I think).

Don't design your product line with the intention of FORCING your end users to toss their box every year.
I have never been compelled to throw out a computer simply because it is no longer the fastest model.

How is it that you can slap a dual G4 processor card in an Old G3 but you can't put a PIII in a PII motherboard?
I have put P3's in mboos built before the P3 existed. Not only that, but my Athlon upgrade to my P2-233 sports much better RAM than the P2 did, and AGP 4x. Compare a Mac to that at your own risk... anyway I'll probably upgrade my G4-800-DP too, once I can get 2 chips more than 25% faster and for less than $1200 or so.

ddtlm
Nov 18, 2002, 04:29 PM
ffakr:

How is it that you can slap a dual G4 processor card in an Old G3 but you can't put a PIII in a PII motherboard?
I wanted to add that you'll be eating your words if Apple ever releases a PPC-970 system. Not even remotely compatible with old systems as far as upgrading goes. G4's will never be upgradeable to a PPC-970.

ffakr
Nov 18, 2002, 05:57 PM
Originally posted by ddtlm
ffakr:


I wanted to add that you'll be eating your words if Apple ever releases a PPC-970 system. Not even remotely compatible with old systems as far as upgrading goes. G4's will never be upgradeable to a PPC-970.

you are still completely missing my point...
You didn't upgrade an LX system...
You put a new computer into the expensive metal box you own. Luckly you got a box that was upgradeable... what about all those HPs and compaqs and others that used proprietary cases and motherboard designs... they were trash long before they needed to be...

I'm not going to argue this point by point... I could go on and on about how I've never seen a PPGA converter card rated much over 500mhz (It isn't reliable to run your leads too long when the chip is too fast), or how I'm talking about 400mhz machines with no upgrade path... not Pentium 133s....

The point is, Intel builds systems that are designed to be obsolete, not in 3 years... but in 12 months.. in 6 months.. This is the biggest problem I have with Intel.

And, who said a 64bit PPC should fit in an old mac? I sure didn't. Did I complain that an Itanium wouldn't fit in a P4 motherboard? NO.
These are different architectures.. even different instruction sets (even if PPC64 is a superset of PPC). I don't expect that to work.
I do expect that Intel should let you buy a faster chip for your LX board than a 333MHz... it isn't like they can't make a processor with a 10x mulitplier. Why scrap the MB, the Memory... when you don't have to.

THAT IS THE POINT.

ddtlm
Nov 18, 2002, 06:10 PM
ffakr:

what about all those HPs and compaqs and others that used proprietary cases and motherboard designs
iMacs? eMacs?

I could go on and on about how I've never seen a PPGA converter card rated much over 500mhz
3rd link on Google: http://www.heatsinkstore.com/supsloccpuca.html They claim it'll do 1.2ghz. I am not going to bother finding more of them. But be my guest, go on and on about it.

And, who said a 64bit PPC should fit in an old mac?
You were very proud of the G3->G4 upgrade, thought I should put things in perspective. The G3 and G4 are damn near the same chip, or at least the original G3 and G4 where, they have diverged somewhat. It is not a big deal that a G3 system can go to a G4, any more than it is that a lot of P2 systems could go to P3's.

Why scrap the MB, the Memory... when you don't have to.
Because they are obsolete and slow, perhaps? Because replacements are cheap?

Why even try to fight this fight when only one model of Mac can be upgraded in either video or processors, and even then Apple won't tell the upgrade makers anything about how to do it. I am amazed that I managed to find someone that you even take your position on the subject.

ddtlm
Nov 18, 2002, 06:31 PM
ffakr:

Oops, I did go find another link. These guys apparently modded their machine to run a 1.4ghz P3 in a BX board... that chipset is from 1998.

http://www.overclockers.com/tips965/index02.asp

Pretty funny bit of work they did there.

Here's a review of an actual product that they tested with a 1.2ghz Celeron (256k), an updrade to a P2-400 Dell looks like:

http://www.duhvoodooman.com/powrleap/PLintro.htm

These guys apparently tested a 1.4ghz 512k P3 in such an adaptor.

http://www.meetthegeeks.org/ourreview/2002/powerleapadaptors/

ffakr
Nov 19, 2002, 09:31 AM
Originally posted by ddtlm
ffakr:

Oops, I did go find another link. These guys apparently modded their machine to run a 1.4ghz P3 in a BX board... that chipset is from 1998.

It's nice to see this, my friend may actually squeeze some more life out of his Dell.

It is interesting though, that Intel basically gave two reasons for scrapping the slot.. two main reasons at the time at least: the inability to run at higher frequencies because of the length of leads to the MB, and extra cost associated with the card.

Its a shame that they foisted an architecture on the community only to force a massive change not too far down the road (which has been my point since the start).

You still aren't showing me any evidence that I'm incorrect in asserting that Intel doesn't go out of it's way to force the premature obscelesence of equipment. It is only through clever hacks that Intel hardware runs as long as it does (in a useful manner).

It's too bad that slot one wasn't able to pull the same tricks that Apple's cpu slot could... namely an upgrade path from a single CPU card to a quad cpu card. That would have really been useful, though I suppose that would have further reduced the need to throw out an entire PC.
You were very proud of the G3->G4 upgrade, thought I should put things in perspective. The G3 and G4 are damn near the same chip, or at least the original G3 and G4 where, they have diverged somewhat. It is not a big deal that a G3 system can go to a G4, any more than it is that a lot of P2 systems could go to P3's.
Yes... but a PII and a P3Cu are pretty much the same chip too, aren't they? People put 100MHz bus celerons in BX1 boards (as you point out), but pop one in an LX and see how fast it runs. I've owned dualcpu LX boards with built in SCSI. Nice boxes... Why can't I put 466MHz processors in that board? Do you really believe the LX chipset couldn't possibly handle a cpu with a mulitplier that high? It couldn't have been engineered to? Or do you think that Intel designed the LX as a transition to the P2 with the intention of making it obsolete within a year?

yes, I conceed that you can, on ocassion thwart Intel's plans and upgrade a BX or BX2 motherboard farther than INTEL had ever intended... You've provided the links.
BUT!
Maybe you should read a bit more into them....
Overclockers.com:
"There are numerous other combinations that may or may not work."
"I have not heard of anyone being successful with generic slockets. Having a brand-name slocket is no guarantee, either. I was unsuccessful with the mod using an Iwill Slocket II ver 1.0, but I believe someone in one of the threads was successful using ver 1.2 of the Iwill Slocket II."
"I actually tried 2 different ASUS S370-133 units but was unsucessful with both of them. I also unsuccessfully tried a Gigabyte GA-6R7+."

Hmn... I wonder how many of those dual G4 upgrades have reported..."well, I was unsuccessful with Serial# xxyyssxx of the Powermac 7300 but the machines in the batch with serials starting with xxyyxt were just peachy"
overclockers.com
"To see if anyone else has been successful with your board, your best bet would be to check the above threads rather than asking me since I will probably have no idea. "

Well, that should certainly make the average joe confident about the whole proposition. Search the web to see if anyone has been able to pull it off... or just keep trying different upgrade hardware till it works, or you are confident that there is no way to get your machine upgraded.

So... you've give me two main bodies of evidence that Intel has not forced obsolescence.
The first is your ability to remove all the guts from an ATX case and replace them with new components.
The second is the ability of a few hard core geeks to upgrade a small number of BX motherboards... if they can find the right combination of motherboard model and revision that will work with the right slocket card.

Well, I guess I have to conceed to you.

BTW... I'm running a Duron in a MacIIci case. I suppose that, by your definition, Apple intended the old IIci architecture to be flexible enough to run seventh generation AMD processors... right?

ddtlm
Nov 19, 2002, 10:46 PM
ffakr:

The second is the ability of a few hard core geeks to upgrade a small number of BX motherboards... if they can find the right combination of motherboard model and revision that will work with the right slocket card
Apple has done no better, since they too have very few chipsets and very few motherboards that are upgradable. As I pointed out, those that are upgradable are only upgradable because people reverse engineer what Apple has done, not because Apple wants upgrades. Also, the upgradability of Apple computers has been helped by the insignificant technology changes from generation to generation, whereas Intel and AMD regularly add faster RAM and FSB's that make backwards compatibility difficult.

To restate: the vast majority of modern Macs have soldered-on processors, and those where this is not the case, they are upgradable only because people reverse engineer what Apple has done. These upgrades sell only because Apple does not progress quickly enough to make them uninteresting to Apple owners. If Intel and AMD where still selling products that are nothing more than warmed-over versions of the same stuff they were selling when Apple rolled out the G4, then I would expect a large upgrade market there as well, since there would be no reason to upgrade anything except the processor, since it is essentially all that changes.

Further, if/when Apple progresses to the PPC-970, the upgrade market will probably be essentially dead, since the FSB on the PPC-970 is much more complicated than that on the G4, and because there is little or no need for there to be a daughter card for the PPC-970, since it has no off-die cache as the slot-1, slot-A, G4 and G3 processors on daughter cards usually did. This means that unless people can buy naked PPC-970's by themselves, and unless the chipset happens to support the FSB at half the clock speed of the new PPC-970, then these machines will not be upgradable. We will then find that no Macs are upgradeable. And thats what Apple wants.

ffakr
Nov 20, 2002, 10:02 AM
Originally posted by ddtlm
ffakr:
Apple has done no better, since they too have very few chipsets and very few motherboards that are upgradable. As I pointed out, those that are upgradable are only upgradable because people reverse engineer what Apple has done, not because Apple wants upgrades. Also, the upgradability of Apple computers has been helped by the insignificant technology changes from generation to generation, whereas Intel and AMD regularly add faster RAM and FSB's that make backwards compatibility difficult.

There is a subtle difference here that you do not realize. I shudder to mention this because this point is very subjective.
Intel and AMD do not make computers. They make computer components. In fact, they make computer components that are designed to be modular and upgradeable. This is their business model, this is what they do. ...so much so, that you can go to any major electronics/computer store and purchase a retail version of their chips that are marketed to lay-users.
Apple makes computers... from the ground up, plain and simple. They are not in the business of selling upgrades, they are not in the business of selling parts. If you want a Mac, you buy a mac.
This doesn't mean that Apple has a right to screw users, this is just the way they do business... black and white.
Apple, however, doesn't seem to consistently go through systematic and regular changes that lock out old hardware... Notice I say regular changes. There are exceptions to this rule which I will note:
Apple's Bad Design Decisions

Putting the firmware on the processor card for older iMac models
possibly locking out upgrade vendors from receiving processors faster than 500MHz
Locking out Pre-G3 Macs from OS X installations

Of these... I can't verify the second item, but i suggest Apple and Motorola conspired against upgrade vendors.
I also understand and support the last point. This was a decision based on support, allocation of limited resources, and of course the desire to push upgrades. OS 9 was (and still is) supported long after the OS X release however so Apple did not simply abandon older machines.


To restate: the vast majority of modern Macs have soldered-on processors,
this is NOT the case. All G4s, Blue and white G3s, Beige G3s, CRT iMacs, All PCI pre-g3s except the 7200 and some performas (7300, 7500, 7600, 8500, 8600, 9500, 9600) had either socketed or slotted processors. This is a fact. I haven't opened an eMac up and I haven't opened an LCD iMac in a long time so I can't comment on those. I think you are incorrect in stating that "the vast majority" of modern Macs have soldered on processors.

and those where this is not the case, they are upgradable only because people reverse engineer what Apple has done.

yes, because Apple isn't in the business of selling parts or upgrades... doh. When Apple starts selling upgrade processor cards that only work in very recent Macs, then you can make a valid comparison to Intel.

These upgrades sell only because Apple does not progress quickly enough to make them uninteresting to Apple owners.
And of course because people don't want to toss their hardware investment just to have a reasonable word processor around

So, you can't argue that Intel doesn't try to intentionally make their hardware obsolete at regular intervals so you've moved the argument to ... 'well Apple is just as bad'.

My assertion still stands... Intel designs hardware with the full knowledge and intention to make a non-compatable upgrade in the near future.

Apple doesn't plan to make their hardware upgradeable, but the EASE with which you can upgrade an older Macintosh makes a VERY strong argument that they do not intentionally design hardware so that it can't be upgraded later on. Ah... do you see the implications here?

Apple's didn't nescessarily design their machines for upgrades, but they haven't consistently designed their machiens so that they CAN'T be upgraded... quite the contrary, older Macs have proven to be very easy to upgrade. Architectures that support multiple processors on one card or zif module... Architectures that allow users to drop in upgrades without intentional bios incompatabilities...

Irregardless, the entire Apple issue is moot... the 970 issue is completely outside the realm of my contentions since that is an entirely ISA, it is only a cousin of the existing PPCs. The point has always been whether or not Intel has been screwing people over... I still haven't seen an argument to refute this contention.

My Asus Athlon motherboard (athlon) is great, but the last bios revision won't support any clock speed over 1300MHz. Other boards with my chipset support higher multipliers. Did they design it with the expectation that the K7 would never scale past 1.3GHz? Is there a limitation in that rev of the chipset? Is this an intentional decision in BIOS design to force me to upgrade? I can't say for certain, but it is suspicious, and I did almost upgrade it the last time I added new components to my box.

ddtlm
Nov 20, 2002, 03:36 PM
ffakr:

Intel and AMD do not make computers. They make computer components. In fact, they make computer components that are designed to be modular and upgradeable. This is their business model, this is what they do. . . .
Apple makes computers... from the ground up, plain and simple. They are not in the business of selling upgrades, they are not in the business of selling parts. If you want a Mac, you buy a mac.
I have to admit that this looks like a valid foundation for holding the companies to different standards.

this is NOT the case. All G4s...
Depends on your definition of modern.

My assertion still stands... Intel designs hardware with the full knowledge and intention to make a non-compatable upgrade in the near future.
A good deal of these incompatabilities are because of the march of technology, the three exceptions that I can think of are that Intel didn't support the 66mhz FSB P2's very well, that the first socket-370 standard wasn't good enough for later chips, and that the first P4 socket was replaced by one with the same functionality but different pins. Twice now (at least) intel has even changed production technology and cache arrangement while maintaining full compatability, when they could have broken it. Intel usually makes a handul of new models supporting slower FSB's even after the newer FSB is supported, but I can't blame them for moving on since they already have so darn many different processors for sale, and since new motherboards are so cheap. I've done upgrades before in such a way that I can use my old RAM and keep the costs very reasonable.

My Asus Athlon motherboard (athlon) is great, but the last bios revision won't support any clock speed over 1300MHz. Other boards with my chipset support higher multipliers. Did they design it with the expectation that the K7 would never scale past 1.3GHz?
Is that a KT133 chipset? I think that your problem may be that the FSB is only 200mhz, and the fastest processor to use that was the 1300mhz chip. Who knows, maybe a higher-multiplyer chip will work in there, even if the motherboard doesn't know the correct clock speed for it.

ddtlm
Nov 20, 2002, 03:45 PM
ffakr:

Hmm, I just read that the FX-chipset was used by the PPro as well, so it had a longer lifespan than it would at first appear.