AMD vs. Intel: Refuting Historical Inaccuracies

[ElMoIsEviL]

First off it's great to be posting here again. I haven't been on TomsHardware for over 2 years now for reasons I won't disclose as they are unimportant. Recently I had a run-in with some folks over at AMDZone who seemed to live in a twilight zone of sorts. They had these odd (and false) beliefs regarding historical events involving AMD & Intel and were spreading this "FUD" around unhinged

That having been said I tackled their belief structures and refuted their arguments with clear evidence and I was met with... a ban. I remained respectful throughout the exchanges (something I rarely do) but even that was not enough. After I had defeated the best "minds" they had to offer the admin "The_Ghost" decided that I was a threat to their "alternate point of view" and banned me.

Slightly frustrated from the encounter I decided that the lies end here. So open up your textbooks friends we're going to have a look at some inaccurate claims made by AMD fanbois.

1.

Hypertransport was an AMD Innovation and QPi is a copy cat

This is quite a misleading statement but I see it parroted all around the web as though it were fact. The truth is a tad more complex and requires some clear and concise historical knowledge regarding a former large player in the CPU business known as DEC.

Digital Equipment Corporation was a pioneering American company in the computer industry. It is often referred to within the computing industry as DEC (this acronym was frequently officially used by Digital itself, but the trademark was always DIGITAL).

In May 1997, DEC sued Intel for allegedly infringing on its Alpha patents in designing the Pentium chips. As part of a settlement, DEC's chip business was sold to Intel. This included DEC's StrongARM implementation of the ARM computer architecture, which Intel sold as the XScale processors commonly used in Pocket PCs. (read more about it here: http://www.wired.com/techbiz/media/news/1997/10/8024).

Shortly after (in 1998) DEC sold it's Alpha CPU division to Compaq Computers: http://news.cnet.com/2100-1001-207442.html

This is of particular importance as now Compaq began working closely with AMD and had now licensed two bus technologies to AMD. The Alpha EV6 and EV7 buses. The Alpha EV6 bus was the DDR bus employed on the AMD K7 (Athlon/XP) while the EV7 was the building block for a scaled down point to point interconnect to be used on a future AMD project known as K8 (yes talking about HyperTransport here).

Then in 2001 it all changed. Intel bought the Alpha CPU division from Compaq: http://news.zdnet.co.uk/hardware/0,1000000091,2089920,00.htm

Of course Monopoly laws came into play as Intel would now control far too much Intellectual Property therefore a deal was brokered: Samsung, IBM and AMD would all receive rights to the Alpha Intellectual Property and thus HyperTransport was born.

What is important to note here is that INTEL not AMD owns the true Intellectual Property for the EV7 bus with AMD being a licensee. Therefore QPi did not copy HyperTransport but rather both are based to a large degree on the same IP (EV7 bus).



2.

AMD were the first to Integrate the Memory Controller onto the CPU die for an x86 processor

This one is just false. Intel were actually the first to integrate a memory controller onto an x86 architecture based processor. In fact Intel did so on the mobile variant of their 386 and 486 processors as seen here: http://smithsonianchips.si.edu/intel/i386sx.htm



3.

AMD had the first Native Quad Core and Six Core CPUs

Partially False. To start off, having a "native" anything is not a large accomplishment as it means nothing for performance. That having been said AMD were the first to introduce a "native" quad core part but not the first to introduce a native six core part. That award goes to the Intel Dunnington based processors: http://news.cnet.com/8301-13579_3-9878131-37.html


Anyways.. that is all for now Next time I'll tackle QPi vs. HyperTransport.

Peace.

 

[sminlal]

Interesting post, thanks for sharing.

The Alpha was a tremendously innovative design - unfortunately it was yet another COUNTER example to the old saying: "If you build a better mousetrap, the world will beat a path to your door"...

 

[apache_lives]

Nice work thumbs up from me

 

[EXT64]

Nice post. As usual, the truth is more interesting than the fiction most make up. And I, even being an AMD fan/supporter, don't see what problem those AMDzoners had with your post.

 

[dokk2]

Weeell ,,you all know the old saw about taking a horse to water,you can't force him to drink,unless,of course you gave him lots of salt first,it truly is amazing to see the number of people who wrap themselves up in their fantasies and will not come out "for hell or high water",, I say "Who's that mowing the lawn, over there?? All I can see is their hat???

 

[randomizer]

Are you kidding me? "Native" means it's smoother in the real world!

 

[chowmanga]

First off, great post. I'm glad there are people out there willing to find out things for themselves rather than just repeating what they hear being said. Just one thing, dunnington was not the first native six core processor because dunnington isn't a native six core. In that article you posted, it said dunnington was a compilation of 3 penryn cores one one die. That conflicts the very definition of being 'native'. Penryn is a native dual core, combining them together doesn't make dunnington a native six core. This is the same argument used as to why core2 quad isn't a native quad core - because its 2 penryn cores on one die. So the award doesn't go to intel, or amd because as far as I know, neither company has made a native six core processor.

edit: note: Technical error - forgot about conroe. c2q can be combinations of penryn or conroe. Conroe core was used in 65nm variations.

 

[amnotanoobie]

Just one thing, dunnington was not the first native six core processor because dunnington isn't a native six core.

"Native" anything really doesn't mean squat as intel has proven with the C2Q. It still is a processor with x number of cores, which is addressed by software the same way.

At the end of the day, benchmarks and actual usage wins the day rather than observing how your piece of silicon was made.


+1 for ElMoIsEviL

 

[chowmanga]

"Native" anything really doesn't mean squat as intel has proven with the C2Q. It still is a processor with x number of cores, which is addressed by software the same way.

At the end of the day, benchmarks and actual usage wins the day rather than observing how your piece of silicon was made.


+1 for ElMoIsEviL

Okay but, he said that dunnington was the first native six core and its not. We're not talking benchmarks or performance here. OP tried to clear out information on technical terms and I just corrected him.

Edit: I wrote quad instead of six above

 

[ElMoIsEviL]

First off, great post. I'm glad there are people out there willing to find out things for themselves rather than just repeating what they hear being said. Just one thing, dunnington was not the first native six core processor because dunnington isn't a native six core. In that article you posted, it said dunnington was a compilation of 3 penryn cores one one die. That conflicts the very definition of being 'native'. Penryn is a native dual core, combining them together doesn't make dunnington a native six core. This is the same argument used as to why core2 quad isn't a native quad core - because its 2 penryn cores on one die. So the award doesn't go to intel, or amd because as far as I know, neither company has made a native six core processor.

edit: note: Technical error - forgot about conroe. c2q can be combinations of penryn or conroe. Conroe core was used in 65nm variations.

I think you're confused as to what a "CPU die" is.

This is a shot of a Dunnington CPU die:

This is a shot of a Core 2 Quad CPU die:

As you can see a Core 2 Quad is comprised of two Core 2 Duo CPU dies on a single LGA 775 package whereas Dunnington is a single CPU die on a single mPGA604 package.

Dunnington is a native hexa-core processor as Toms hardware also states here: http://www.tomshardware.com/news/intel-core-coming-2008,4848.html

However, before the FSB will disappear for good, there will be one more siginificant chip using this structure: Meet Dunnington, Intel's first native multi-core chip. If AMD's definition of "multi-core" is correct, then "multi" indicates a number greater than two (three-core AMD Phenom 8000 series, four-core Phenom 9000, etc). In that case, Intel is set to offer 50% more cores than AMD currently has.

You can see a few Intel slides on Dunnington here: http://forums.vr-zone.com/news-around-the-web/246929-intel-6-core-dunnington-die-shot.html

I hope this clarifies the misunderstanding

 

[chowmanga]

I think you're confused as to what a "CPU die" is.

This is a shot of a Dunnington CPU die:
http://dorgas.files.wordpress.com/2009/05/dunnington.jpg

This is a shot of a Core 2 Quad CPU die:
http://www.gearfuse.com/wp-content/uploads/nik/07_jan/intel_quad_chip.jpg

As you can see a Core 2 Quad is comprised of two Core 2 Duo CPU dies on a single LGA 775 package whereas Dunnington is a single CPU die on a single mPGA604 package.

Dunnington is a native hexa-core processor.

You can see a few Intel slides on Dunnington here: http://forums.vr-zone.com/news-around-the-web/246929-intel-6-core-dunnington-die-shot.html

I hope this clarifies the misunderstanding

Yes, you bet. Guess it shows how unfamiliar I am with server processors. I wasn't sure if istanbul was native either but I couldn't find much information about the architecture online so I made an educated guess.

 

[ElMoIsEviL]

Yes, you bet. Guess it shows how unfamiliar I am with server processors. I wasn't sure if istanbul was native either but I couldn't find much information about the architecture online so I made an educated guess.

Yessir. Istanbul is also a native Hexa-core processor but it came after Dunnington. What you will often hear is that Istanbul was the first Native Hexa-core (Baron Matrix is notorious for spreading such FUD) and it is just not true.

Thanks for commenting though, it forced me to grab some links and enforces/cements my above comments

Peace.

 

[JAYDEEJOHN]

First of all, good to see you Elmo, loooooong time no see.
I always claim that AMD may not have introwed alot of these things, but they did make them popular, and brought them to the forefront of DT, making their chips the fastest at the time.
I agree, because at the time, AMD did have arguably the fastest things out, some AMD fans prefer to live in that past, and also alter it

 

[JAYDEEJOHN]

When I first saw the OP of this thread, I had to double check for a necro heheh
Like I said, welcome back

 

[Upendra09]

very informative for a noob like me

This thread is definitely the next 100 page thread

oops i jinxed it

 

[kwalker]

No you didnt.
I even stopped in to monitor this thread

 

[uncfan_2563]

umm i thought having a native quad core is better because it reduces latency between the two dual cores "glued" together

 

[uncfan_2563]

can we see the thread that got you banned by any chance?

 

[Upendra09]

^ please?

 

[JAYDEEJOHN]

Alot of those old threads were deleted, and some werent picked up on the "new" system.
Personally, Im glad to see some folks back, and let those "old" days die, as well as the reasons why they happened

 

[Mousemonkey]

Personally, Im glad to see some folks back, and let those "old" days die, as well as the reasons why they happened

Hear hear.

 

[ElMoIsEviL]

First of all, good to see you Elmo, loooooong time no see.
I always claim that AMD may not have introwed alot of these things, but they did make them popular, and brought them to the forefront of DT, making their chips the fastest at the time.
I agree, because at the time, AMD did have arguably the fastest things out, some AMD fans prefer to live in that past, and also alter it

Indeed, I still run 4 AMD based rigs. I still use an AMD AthlonXP 3200+ (typing on it as we speak), AMD Athlon64 X2 3800+, AMD Phenom II X3 720 (unlocked) and AMD Phenom II X4 955BE.

AMD still makes very good machines (especially for the price) but some folks do still live in the past (as you mentioned). In fact I'd even go as far as claim that these folks are delusional to a large extent.

When I first saw the OP of this thread, I had to double check for a necro heheh
Like I said, welcome back

Yeah, it's been a long time. Randomizer coaxed me into coming back. Not sure how long I'll last this time around Nice to see some familiar faces still around

 

[Cryslayer80]

First off it's great to be posting here again. I haven't been on TomsHardware for over 2 years now for reasons I won't disclose as they are unimportant. Recently I had a run-in with some folks over at AMDZone who seemed to live in a twilight zone of sorts. They had these odd (and false) beliefs regarding historical events involving AMD & Intel and were spreading this "FUD" around unhinged

That having been said I tackled their belief structures and refuted their arguments with clear evidence and I was met with... a ban. I remained respectful throughout the exchanges (something I rarely do) but even that was not enough. After I had defeated the best "minds" they had to offer the admin "The_Ghost" decided that I was a threat to their "alternate point of view" and banned me.

Slightly frustrated from the encounter I decided that the lies end here. So open up your textbooks friends we're going to have a look at some inaccurate claims made by AMD fanbois.

1.

Hypertransport was an AMD Innovation and QPi is a copy cat

This is quite a misleading statement but I see it parroted all around the web as though it were fact. The truth is a tad more complex and requires some clear and concise historical knowledge regarding a former large player in the CPU business known as DEC.

Digital Equipment Corporation was a pioneering American company in the computer industry. It is often referred to within the computing industry as DEC (this acronym was frequently officially used by Digital itself, but the trademark was always DIGITAL).

In May 1997, DEC sued Intel for allegedly infringing on its Alpha patents in designing the Pentium chips. As part of a settlement, DEC's chip business was sold to Intel. This included DEC's StrongARM implementation of the ARM computer architecture, which Intel sold as the XScale processors commonly used in Pocket PCs. (read more about it here: http://www.wired.com/techbiz/media/news/1997/10/8024).

Shortly after (in 1998) DEC sold it's Alpha CPU division to Compaq Computers: http://news.cnet.com/2100-1001-207442.html

This is of particular importance as now Compaq began working closely with AMD and had now licensed two bus technologies to AMD. The Alpha EV6 and EV7 buses. The Alpha EV6 bus was the DDR bus employed on the AMD K7 (Athlon/XP) while the EV7 was the building block for a scaled down point to point interconnect to be used on a future AMD project known as K8 (yes talking about HyperTransport here).

Then in 2001 it all changed. Intel bought the Alpha CPU division from Compaq: http://news.zdnet.co.uk/hardware/0,1000000091,2089920,00.htm

Of course Monopoly laws came into play as Intel would now control far too much Intellectual Property therefore a deal was brokered: Samsung, IBM and AMD would all receive rights to the Alpha Intellectual Property and thus HyperTransport was born.

What is important to note here is that INTEL not AMD owns the true Intellectual Property for the EV7 bus with AMD being a licensee. Therefore QPi did not copy HyperTransport but rather both are based to a large degree on the same IP (EV7 bus).



2.

AMD were the first to Integrate the Memory Controller onto the CPU die for an x86 processor

This one is just false. Intel were actually the first to integrate a memory controller onto an x86 architecture based processor. In fact Intel did so on the mobile variant of their 386 and 486 processors as seen here: http://smithsonianchips.si.edu/intel/i386sx.htm



3.

AMD had the first Native Quad Core and Six Core CPUs

Partially False. To start off, having a "native" anything is not a large accomplishment as it means nothing for performance. That having been said AMD were the first to introduce a "native" quad core part but not the first to introduce a native six core part. That award goes to the Intel Dunnington based processors: http://news.cnet.com/8301-13579_3-9878131-37.html


Anyways.. that is all for now Next time I'll tackle QPi vs. HyperTransport.

Peace.

Bravo! Now, if you post the same thing for Intel fanboys, I will support you to the full.

 

[ElMoIsEviL]

Bravo! Now, if you post the same thing for Intel fanboys, I will support you to the full.

Well I'm not quite familiar with Intel fanbois as I don't know many of them right now (I can't tell them apart from the general public because right now they do support the better processor).

But I do remember Intel fanbois back during the days of the Pentium D (I ran AMD exclusively back then). And they would make outlandish claims but never delusional to the point of creating their own website and kicking anyone out who disagrees (referring to AMDZone).

But by all means, if you have certain allegations Intel fans are making which you think are not true, run them by me. I will objectively do the research and then give you the facts (I won't lie to you that's a promise).

 

[SpidersWeb]

Good post.
Good overclock.
Love you long time.

(and yeah if you find more CPU mythbuster stuff, definately share, irrelevant of who the company is (Intel/AMD/Ati/nVidia etc))