AMD Channel Stuffing In 4Q06

[Phynaz]

Text

 

[Viditor]

How many times must we read this Doug Freedman article?

 

[formulav8]

I'm building MANY more AMD based chips than Core2. AMD not selling well is absolutely wrong. There is not near the so-called customer demand for Core2's as the enthusiast. AMD is selling many Semprons and X2's. And with more AMD price cuts coming during the beginning of next month, I will most likely continue to build many AMD based systems.

Just my real world opinion and observation.


Jason

 

[nyker96]

I have to say in the OEM market formulav8 is right, AMD is going strong. I think core only makes sense when it comes to OCed setup if not those 85/95 dollar X2s are incredible values for end users.

 

[Dadofamunky]

All I can say is, root for AMD! Does anybody really want a return to the days of the Intel monopoly? I remember when a stupid 386 motherboard with the Intel chipset cost $700. The CPU was even more. As a result I stuck with 286s for a really long time.

This ain't fanboyism. It's self-interest. I want two strong competitors in the CPU market. I sure hope AMD manages to get its chestnuts out of the fire.

Also remember that most of Intel's manufacturing capacity is still devoted to the horrible NetBust CPUs. If AMD dies or is bought out by private capital thugs, Intel will likely lose whatever incentive it had to innovate and provide better products, and go back to the days of overinflated crappy silicon (though nothing now could be as bad as the sh*t they pumped out in those days). We all lose if that happens.

 

[Phynaz]

though nothing now could be as bad as the sh*t they pumped out in those days

This sh*t you talk about would include:

8086
8088
80286
80386
80486
Pentuim I, II, III

Funny, I don't recall any of those being "bad".

 

[Midnight Rambler]

Phynaz : That's because many of these fanboyz weren't even born yet !

It's just like their other tired old complaint that Intel has always overcharged people for their CPUs and if it weren't for AMD, Intel would still be doing so. Fact is, economies of scale is what has lead to CPU price drops, not AMD. Sales numbers today approach 25M units a year vs. maybe 0.5M tops back in the day.

 

[Viditor]

Originally posted by: Midnight Rambler
Phynaz : That's because many of these fanboyz weren't even born yet !

It's just like their other tired old complaint that Intel has always overcharged people for their CPUs and if it weren't for AMD, Intel would still be doing so. Fact is, economies of scale is what has lead to CPU price drops, not AMD. Sales numbers today approach 25M units a year vs. maybe 0.5M tops back in the day.

Rambler, if you're gonna lecture, you should get your facts straight...

X86 sales are closer to 200 Million/year+...

 

[DrMrLordX]

Originally posted by: Phynaz

though nothing now could be as bad as the sh*t they pumped out in those days

This sh*t you talk about would include:

8086
8088
80286
80386
80486
Pentuim I, II, III

Funny, I don't recall any of those being "bad".

Actually, there were some bad things about a few of those chips . . .

80386: SX is fur sux.
80486: SX is for sux.
Pentium: floating point error, yuck. At least Intel replaced bugged chips.
Pentium II: When Intel introduced this CPU to the desktop market, this represented the first Intel CPU in a long time (possibly first ever) that had a platform that was completely incompatible with "clone" chips. AMD was stuck using older Socket 7 boards until Super 7 shlepped onto the scene. That move was the first thing Intel ever did to genuinely upset me, and I wasn't the only one. Also, the original Klamaths ran too damn hot.
Pentium III: In a move even more obnoxious than their switch to Slot 1, Intel tried again to leverage the ubiquity of their platform to wipe out "clone" competitors (though by this time, AMD was no clone manufacturer) by trying to get the market to adopt RDRAM which AMD platforms did not (and due to licensing agreements, could not) support. Say what you will about the RAMBUS/JEDEC shenanigans, but it was obvious that Intel pushed RDRAM back then as a way to make life hell on AMD rather than promote a memory technology that actually made sense for Pentium IIIs. Had Intel stuck with the old 440BX and added official support for PC133, they could have supported their entire line of Pentium IIIs with a cheap PC100/PC133 platform. Instead we got the i820 which was the devil's spawn.

 

[myocardia]

Originally posted by: DrMrLordX
Actually, there were some bad things about a few of those chips . . .

Pentium: floating point error, yuck. At least Intel replaced bugged chips.
Pentium II: When Intel introduced this CPU to the desktop market, this represented the first Intel CPU in a long time (possibly first ever) that had a platform that was completely incompatible with "clone" chips. AMD was stuck using older Socket 7 boards until Super 7 shlepped onto the scene. That move was the first thing Intel ever did to genuinely upset me, and I wasn't the only one. Also, the original Klamaths ran too damn hot.
Pentium III: In a move even more obnoxious than their switch to Slot 1, Intel tried again to leverage the ubiquity of their platform to wipe out "clone" competitors (though by this time, AMD was no clone manufacturer) by trying to get the market to adopt RDRAM which AMD platforms did not (and due to licensing agreements, could not) support. Say what you will about the RAMBUS/JEDEC shenanigans, but it was obvious that Intel pushed RDRAM back then as a way to make life hell on AMD rather than promote a memory technology that actually made sense for Pentium IIIs. Had Intel stuck with the old 440BX and added official support for PC133, they could have supported their entire line of Pentium IIIs with a cheap PC100/PC133 platform. Instead we got the i820 which was the devil's spawn.

It seems as if you've forgotten all of the good things that came along with these chips. For instance, the Pentium brought on-die L1 cache. The Pentium II brought on-die L2 cache to non-business users. Although the Pentium Pro was the first chip to have on-die L2, it was riddled with problems, and was extremely expensive, so add it to your list.

And besides having two different Socket 370's, I don't remember anything being wrong with the P3, besides the fact that it wasn't as fast as the Athlons; all of them used either PC100 or PC133. It was the Williamette P4 that was so horrible, and used RAMBUS, not the P3.

 

[3chordcharlie]

Originally posted by: myocardia

Originally posted by: DrMrLordX
Actually, there were some bad things about a few of those chips . . .

Pentium: floating point error, yuck. At least Intel replaced bugged chips.
Pentium II: When Intel introduced this CPU to the desktop market, this represented the first Intel CPU in a long time (possibly first ever) that had a platform that was completely incompatible with "clone" chips. AMD was stuck using older Socket 7 boards until Super 7 shlepped onto the scene. That move was the first thing Intel ever did to genuinely upset me, and I wasn't the only one. Also, the original Klamaths ran too damn hot.
Pentium III: In a move even more obnoxious than their switch to Slot 1, Intel tried again to leverage the ubiquity of their platform to wipe out "clone" competitors (though by this time, AMD was no clone manufacturer) by trying to get the market to adopt RDRAM which AMD platforms did not (and due to licensing agreements, could not) support. Say what you will about the RAMBUS/JEDEC shenanigans, but it was obvious that Intel pushed RDRAM back then as a way to make life hell on AMD rather than promote a memory technology that actually made sense for Pentium IIIs. Had Intel stuck with the old 440BX and added official support for PC133, they could have supported their entire line of Pentium IIIs with a cheap PC100/PC133 platform. Instead we got the i820 which was the devil's spawn.

It seems as if you've forgotten all of the good things that came along with these chips. For instance, the Pentium brought on-die L1 cache. The Pentium II brought on-die L2 cache to non-business users. Although the Pentium Pro was the first chip to have on-die L2, it was riddled with problems, and was extremely expensive, so add it to your list.

And besides having two different Socket 370's, I don't remember anything being wrong with the P3, besides the fact that it wasn't as fast as the Athlons; all of them used either PC100 or PC133. It was the Williamette P4 that was so horrible, and used RAMBUS, not the P3.

P4 was the second run at rambus, and at least had some validity, as the P4 was bandwidth starved, while the P3 never ran at a faster bus speed than could be handled nicely by sdram. DDR ram came along and fixed this bandwidth problem long before intel gave up on rdram though.

 

[DrMrLordX]

Originally posted by: myocardia


It seems as if you've forgotten all of the good things that came along with these chips.

Nah, I was simply listing problems that they had (or problems they created). There were plenty of good things about them.

Although the Pentium Pro was the first chip to have on-die L2, it was riddled with problems, and was extremely expensive, so add it to your list.

I rather liked the Pentium Pro. I also remember people holding on to quad PPro 233mhz systems for small x86 server use for a long time, because early Xeons weren't that much better and were restricted to 2p configurations in most cases.

In fact, everyone I know who could afford Pentium Pros and had a use for them, really enjoyed owning them.

And besides having two different Socket 370's, I don't remember anything being wrong with the P3, besides the fact that it wasn't as fast as the Athlons; all of them used either PC100 or PC133. It was the Williamette P4 that was so horrible, and used RAMBUS, not the P3.

Huh? RDRAM was originally introduced for use with Pentium IIIs using Intel's i820 and i840 chipsets. To make matters worse, the i820 had severe bugs:

http://www.tomshardware.com/1999/10/05/intel_i820_chipset_review/index.html
http://www.theregister.co.uk/2000/05/15/intel_dithers_over_i820_recall/

Getting an SDRAM platform for 133 mhz FSB Pentium IIIs from Intel was a nightmare until they finally released the i815, and SDRAM performance on the i815 was arguably worse than SDRAM performance on the venerable 440BX:

http://www.anandtech.com/showdoc.html?i=1265
http://www.anandtech.com/showdoc.aspx?i=1265&p=13

Otherwise you had to use Via chipsets or the i810e and . . . well let's not go there.

 

[zsdersw]

Ahh, yes.. the Pentium Pro. We used them at work until sometime in 2002-3. They ran Win2000 Server pretty well. And we didn't even have the 1MB L2 cache version.. just the 200MHz 256k L2 version. Pretty neat looking too, once you get that black ceramic (or whatever it was) lid off of the underside of the chip.

 

[SuperSix]

Originally posted by: formulav8
I'm building MANY more AMD based chips than Core2. AMD not selling well is absolutely wrong. There is not near the so-called customer demand for Core2's as the enthusiast. AMD is selling many Semprons and X2's. And with more AMD price cuts coming during the beginning of next month, I will most likely continue to build many AMD based systems.
Jason

I will have to disagree.

My experience is from working for a tier1 AMD distributor, and I can say with confidence, MANY more small systems houses are building with Intel now. The primary reason was the severe drought of AMD product in Q4 last year, duy primarily to AMD shipping nearly all of their processors to Tier1 OEMS (Dell primarily). The severe shortage (I'm tallking from REAL distributors, not Newegg, and other ecom's that buy gray) caused a lot of sysem builders to go to Intel (No real choice, they had product available at the time), and shun AMD for turning their back on the channel that helped get them towhere they are today.

Ruiz finally admitted they made a grave error favoring Dell (And other OEM's) and shorting the channel, and part of the reason for the massive price drops is to attempt to win back the resellers.

I'm not sure if it will work, with Intel's massive cheap Netburst inventory (Not everyone needs a C2D, and Intel's brisk C2D sales, combined with their 04/07 price drop - AMD has a very hard road ahead of them. The goods news is that there's a ton of AM2 motherboards in the channel, (With a lot of manufacturer rebates due to excess inventory) and DDR2 is slated to continue to drop.

My credits? While I don't like to say where I am employed, or exactly what I do, let's just say that if you bought an AMD or Intel proc from all major etailers, there's a VERY good chance it came from my employer.

 

[Special K]

So has AMD's problem always been inability to meet demand, rather than having a low demand for their products?

Is it because they only have 1 fab?

 

[SuperSix]

Originally posted by: Special K
So has AMD's problem always been inability to meet demand, rather than having a low demand for their products?

Is it because they only have 1 fab?

They have always had SOME capacity issues - unavoidable with the small amout of fabs they have. BUt it is amazing how far they have made it and the ass whuppin they gave Intel before the Core2Duo came out.

 

[formulav8]

Originally posted by: SuperSix

Originally posted by: formulav8
I'm building MANY more AMD based chips than Core2. AMD not selling well is absolutely wrong. There is not near the so-called customer demand for Core2's as the enthusiast. AMD is selling many Semprons and X2's. And with more AMD price cuts coming during the beginning of next month, I will most likely continue to build many AMD based systems.
Jason

I will have to disagree.

My experience is from working for a tier1 AMD distributor, and I can say with confidence, MANY more small systems houses are building with Intel now. The primary reason was the severe drought of AMD product in Q4 last year, duy primarily to AMD shipping nearly all of their processors to Tier1 OEMS (Dell primarily). The severe shortage (I'm tallking from REAL distributors, not Newegg, and other ecom's that buy gray) caused a lot of sysem builders to go to Intel (No real choice, they had product available at the time), and shun AMD for turning their back on the channel that helped get them towhere they are today.

I don't know about the small home customers. I'm talking about where the money and builds are. Thats with businesses and resellers and such. Companys that buy real computers in higher quantitys. Sure I do quite a few single system builds and 2x system builds. But No home user is going to do 100x that we did in January and 60x builds that will be done this Friday. This is where the real volume and money is at.

Just my experience in the oem style market since I don't do house builds except on the side and don't know what the market is like. But what I do know, is that I build more X2's than I do Core2's.


Jason

 

[myocardia]

Originally posted by: DrMrLordX

Originally posted by: myocardia

Although the Pentium Pro was the first chip to have on-die L2, it was riddled with problems, and was extremely expensive, so add it to your list.

I rather liked the Pentium Pro. I also remember people holding on to quad PPro 233mhz systems for small x86 server use for a long time, because early Xeons weren't that much better and were restricted to 2p configurations in most cases.

In fact, everyone I know who could afford Pentium Pros and had a use for them, really enjoyed owning them.

All I really remember about the Pentium Pro's were the complaints. I can't remember exactly what the people who bought them were complaining about (maybe you do), but I do remember a considerable amount of complaining.

And besides having two different Socket 370's, I don't remember anything being wrong with the P3, besides the fact that it wasn't as fast as the Athlons; all of them used either PC100 or PC133. It was the Williamette P4 that was so horrible, and used RAMBUS, not the P3.

Huh? RDRAM was originally introduced for use with Pentium IIIs using Intel's i820 and i840 chipsets. To make matters worse, the i820 had severe bugs:

http://www.tomshardware.com/1999/10/05/intel_i820_chipset_review/index.html
http://www.theregister.co.uk/2000/05/15/intel_dithers_over_i820_recall/

Getting an SDRAM platform for 133 mhz FSB Pentium IIIs from Intel was a nightmare until they finally released the i815, and SDRAM performance on the i815 was arguably worse than SDRAM performance on the venerable 440BX:

http://www.anandtech.com/showdoc.html?i=1265
http://www.anandtech.com/showdoc.aspx?i=1265&p=13

Otherwise you had to use Via chipsets or the i810e and . . . well let's not go there.

Haha, the i820 and i840. Now that you mention them, I remember them. The i820 was the chipset that was making people consider a Via chipset, especially since they didn't have to get a second mortgage on their house, so they could buy RDRAM.

 

[DrMrLordX]

I'm sure Intel is quite happy to have gotten well past the bad old days of the i820.

 

[Acanthus]

i850 + P4C with overclocked rambus @ 1200mhz was great though.

 

[crimson117]

Originally posted by: formulav8
I'm building MANY more AMD based chips than Core2. AMD not selling well is absolutely wrong. There is not near the so-called customer demand for Core2's as the enthusiast. AMD is selling many Semprons and X2's. And with more AMD price cuts coming during the beginning of next month, I will most likely continue to build many AMD based systems.

For $95 you can get a Athlon x2 3800+, or a Intel Pentium D 915. Both dual-core, both take DDR2. Why do you recommend the athlon over the intel? Is 3800+ performance better than the 915? Does motherboar dcost become a factor?

 

[MercenaryForHire]

Originally posted by: crimson117
Is 3800+ performance better than the 915?

Does a highly-renowned religious figure defecate in the woods?

- M4H

 

[crimson117]

Originally posted by: crimson117
Is 3800+ performance better than the 915?

Found an answer: "If we compare the performance of [AMD Athlon 64 X2 3600+] against Intel Pentium D 915, which will most likely be its primary competitor from the pricing standpoint, the AMD solution turns out a definite winner. Just like before, the new AMD processor on K8 microarchitecture outperforms the competitor built on NetBurst. Even the recent price wars didn?t change anything, and the recently announced Core 2 Duo processors are still in a different price category."

 

[formulav8]

There is absolutely NO reason to buy any of that PD crap. The only thing I can think of is if you absolutely plan on doing a Core2 in the very near future and don't have the few bucks it takes to get Core2. Sadly alot of my customers are buying into the PD's, but nothing I can do about it. Core2 and especially X2's are low enough that PD's shouldn't even be a contender. They need to die already. Just my opinion of course



Jason