"AMD: We Innovate, Intel Copies"

Joined
Jun 10, 2004
Messages
3,954
I have had that in my sig since about an hour after it was first reported in the press.Hector is losing his mind...again ! :p Only recently took it out,as it was just to easy a cheap shot.Obviously the man is mentally unstable ! :D
 

infin@

Gawd
Joined
Oct 18, 2006
Messages
812
My kind of CEO. :D
Must be little upset at his companies last few quarters. When I found this I almost fell off my chair.
 

MrGuvernment

Fully [H]
Joined
Aug 3, 2004
Messages
20,565
lol

Yes, that is why Intels "non-native quad core" is kicking phenom to the moon and back. cause they copied what AMD did, and that IMC, ya, that is helping intel,,,,?


he has lost it, depseration
 

Yiffy

[H]ard|Gawd
Joined
Nov 5, 2001
Messages
1,105
Hector goes on to effectively blame AMD's poor sales on their competitor Intel. Now that makes sense. AMD needs to fire this douche.
 

Glow

Supreme [H]ardness
Joined
Jul 20, 2002
Messages
4,834
You'd figure that if they were so innovative everyone would have been running something other that upgraded 939s (AM2) when many people were upgrading to c2ds. I love AMD but this is just idiotic
 

CptFalcon

[H]ard|Gawd
Joined
Dec 13, 2006
Messages
2,012
I think that AMD and Intel need to do a joint venture. Would be nice if the next socket was universal.
 

quadnad

Supreme [H]ardness
Joined
Oct 24, 2005
Messages
7,656
Guess amd should stop innovating and start copying... no?

ZING!

but seriously folks. In my opinion he's just referring to the IMC and 64-bit instructions. I can't come up with anything else specific to AMD in terms of innovation (although you could certainly argue that the IMC had been done before and that AMD copied the DEC Alpha).
 

pxc

Extremely [H]
Joined
Oct 22, 2000
Messages
33,064
Pretty lame, but I already said that in the previous thread on the subject.

Read one of the better Barcelona reviews, like at RWT or Ars, you'll see how many of the Barcelona "innovations" first appeared in older Intel chips like the Banias, Yonah and Conroe. It's ridiculous for AMD to claim copying when AMD and Intel have cross-licensing agreements and both have added the other's technologies, and the whole point of conferences like ISSCC is to share those ideas.

IMO, Ruiz is simply trying to blunt AMD's next problem: Nehalem. While CSI (QuickPath) is almost nothing like HyperTransport (AMD didn't invent point to point bus protocols), it has much more bandwidth and is more robust. Neither should be surprising since CSI was designed years after HT. Also, by attempting to paint Nehalem as a copy, Ruiz might hope to take away prestige from what is shaping up to be a monster in Intel's traditionally weak segments: 4S and higher and cheap HPC nodes.

Edit: ODMC is only unique in that it was part of the NB moved back to the chip... all x86 CPUs used to connect directly to memory, until the memory controller was moved to the NB. :p
 

john2k

n00b
Joined
Mar 30, 2004
Messages
39
Weren't AMD motherboards, via the Nforce 2 chipset, first to have dual channel RAM?

AMD was first to move away from FSB to Hypertransport technology.

AMD was first to release 64bit (in the consumer processor market).

AMD was first to release dual core processors, although Intel was first to release *virtual* dual core tech via hyper threading.

AMD was first to release native/true quad core technology.


AMD does have a lot of "firsts" in comparison to Intel, so the statement doesn't appear to be all that incorrect. It just so happens, though, that Intel seems to outperform AMD despite all of that.
 

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
52,943
If AMD made cars we could all drive the most technologically advanced super car on the road...it would just only go 20 miles per hour.
 

j0j081

Gawd
Joined
Aug 29, 2007
Messages
668
lol

Yes, that is why Intels "non-native quad core" is kicking phenom to the moon and back. cause they copied what AMD did, and that IMC, ya, that is helping intel,,,,?


he has lost it, depseration

yes but then what happens when Phenom's design does start working properly?
 
Joined
Jun 10, 2004
Messages
3,954
yes but then what happens when Phenom's design does start working properly?


If and when it does start 'working' properly,INTC will have had penryn in full swing for months,and it will likely have gone through a revision or three and be running even cooler (and getting even higher oc's ?),and we'll likely also have early Nehalem benchmarks via Anand.

Intel has one interesting innovation: PROFITS.


That is my fave innovation ! :)
 

DemonDiablo

[H]ard|Gawd
Joined
Jun 7, 2004
Messages
1,767
ZING!

but seriously folks. In my opinion he's just referring to the IMC and 64-bit instructions. I can't come up with anything else specific to AMD in terms of innovation (although you could certainly argue that the IMC had been done before and that AMD copied the DEC Alpha).

I was thinking that along with multi-core processing and the huge crave for IPC over raw power.
 

DemonDiablo

[H]ard|Gawd
Joined
Jun 7, 2004
Messages
1,767
AMD does have a lot of "firsts" in comparison to Intel, so the statement doesn't appear to be all that incorrect. It just so happens, though, that Intel seems to outperform AMD despite all of that.

Now thats a stretch.

Me personally I would count every process/die shrink as an innovation. In which case Intel would have quite a few wins under the belt. HT had its uses pre-multi-core and I'm sure it'll have its uses with Nethalem. I'm sure theres some other stuff they've done that just escape me.

Oh I guess you could make mention of all the different die process technologies they've come up with and matured but then again I guess that goes both ways with AMD/IBM and the new techniques they come up with as well, for silicon and transistors.
 

pxc

Extremely [H]
Joined
Oct 22, 2000
Messages
33,064
Weren't AMD motherboards, via the Nforce 2 chipset, first to have dual channel RAM?

AMD was first to move away from FSB to Hypertransport technology.

AMD was first to release 64bit (in the consumer processor market).

AMD was first to release dual core processors, although Intel was first to release *virtual* dual core tech via hyper threading.

AMD was first to release native/true quad core technology.
The first PC dual channel chipset was the i840 from 1999 (dual channel RDRAM) The i845E (dual channel DDR) came out in 2002. nforce2 also came out in 2002, and nvidia is not AMD. :p

HT is not the first point to point bus, but HT was the first modern move away from FSB on desktop PCs.

Arguably there have been workstation/high end 64-bit PCs before, but AMD had the first 64-bit x86 extension, which funny for your limitation has a very tiny percentage of usage in the consumer market.

Yeah, AMD was first with a "native" quad core, which is getting beaten silly by a FSB based processor without an ODMC. :p

What's funny about your list is that it highlights the few firsts AMD actually had. Who was first to virtually every other PC technology and microarchitecture improvement to x86, not to mention nearly every major improvement to the x86 platform (ATX, USB, PCI, AGP, PCI-E, decoupled CISC instructions, etc). It starts with an "i". :p

AMD is just being silly, and to feed the ostridges what they want to hear. It apparently works.
 

Grentz

Fully [H]
Joined
May 5, 2006
Messages
17,224
Seems more like "AMD inoovates, Intel makes it work and sell" :p

I think one of AMD's problems came cause they were always geared towards the enthusiasts and awesome overclocking crowd, but then C2D blew the socks of AMD in the speed and overclocking departments for the time being. That got a lot of die hard AMD overclocking fans to move over to intel for this round.

And if Intel made cars we could all drive the same high priced low mpg cars with new models coming every 5 years instead of every year. :D

and they would never break down or have any issues :D
 
Joined
Jun 10, 2004
Messages
3,954
AMD is just being silly, and to feed the ostridges what they want to hear. It apparently works.


ROLF !!!


Seems more like "AMD inoovates, Intel makes it work and sell" :p

I think one of AMD's problems came cause they were always geared towards the enthusiasts and awesome overclocking crowd, but then C2D blew the socks of AMD in the speed and overclocking departments for the time being. That got a lot of die hard AMD overclocking fans to move over to intel for this round.



and they would never break down or have any issues :D


:D Low blow but I have to admit it made me laugh.I do agree,that C2D/Q has certainly led many to stray from the Church of Hector. :eek:
 

jvrobert

Limp Gawd
Joined
Jun 7, 2004
Messages
173
The first PC dual channel chipset was the i840 from 1999 (dual channel RDRAM) The i845E (dual channel DDR) came out in 2002. nforce2 also came out in 2002, and nvidia is not AMD. :p

HT is not the first point to point bus, but HT was the first modern move away from FSB on desktop PCs.

Arguably there have been workstation/high end 64-bit PCs before, but AMD had the first 64-bit x86 extension, which funny for your limitation has a very tiny percentage of usage in the consumer market.

Yeah, AMD was first with a "native" quad core, which is getting beaten silly by a FSB based processor without an ODMC. :p

What's funny about your list is that it highlights the few firsts AMD actually had. Who was first to virtually every other PC technology and microarchitecture improvement to x86, not to mention nearly every major improvement to the x86 platform (ATX, USB, PCI, AGP, PCI-E, decoupled CISC instructions, etc). It starts with an "i". :p

AMD is just being silly, and to feed the ostridges what they want to hear. It apparently works.

I'm not sure why people give AMD credit for "innovating" x86-64. Look at how Intel moved from 16->32 bits. Then look at how AMD moved from 32->64 bits. Now, tell me exactly how AMD "innovated" anything? The only thing they "innovated" was the marketing push to make people think they needed 64 bits, which they are only _just now_ starting to actually need.

As for HT/IMC - they certaintly didn't "innovate" IMC, that was around long before. They did innovate HT, and it helped them in the high end MP space, but nowhere else.
 

MrGuvernment

Fully [H]
Joined
Aug 3, 2004
Messages
20,565
Weren't AMD motherboards, via the Nforce 2 chipset, first to have dual channel RAM?

AMD was first to move away from FSB to Hypertransport technology.

AMD was first to release 64bit (in the consumer processor market).

AMD was first to release dual core processors, although Intel was first to release *virtual* dual core tech via hyper threading.

AMD was first to release native/true quad core technology.


AMD does have a lot of "firsts" in comparison to Intel, so the statement doesn't appear to be all that incorrect. It just so happens, though, that Intel seems to outperform AMD despite all of that.


And who is to say Intel didnt already have these ideas on the board, just because AMD was first to the market with it, doesnt mean they thought of it, same for Intel.

how uselful were 64bit processors when they were released?\

How useful has the IMC been? (seem performance wise, intel doesnt need it) now you need a new socket for a new CPU for new ram....

Native / true thing is just marketing BS and grasping at straws there.
 

quadnad

Supreme [H]ardness
Joined
Oct 24, 2005
Messages
7,656
How useful has the IMC been? (seem performance wise, intel doesnt need it)

!

The IMC was one of the major reasons AMD was able to compete with Intel in the A64/P4 gen. Don't underestimate its critical contribution to the A64s performance.
 

pxc

Extremely [H]
Joined
Oct 22, 2000
Messages
33,064
I'm not sure why people give AMD credit for "innovating" x86-64. Look at how Intel moved from 16->32 bits. Then look at how AMD moved from 32->64 bits. Now, tell me exactly how AMD "innovated" anything?
I agree. "AMD64" was a small step to move 32->64. When Intel moved from 16 to 32-bit, there were very large extensions to the ISA that went along with it, the biggest being V86 mode which made real multi-tasking possible.

People don't realize that AMD extended the register set from 32->64 bits, added 8 more registers and put several limitations on the existing instruction set to operate on the new registers. It's not a new instruction set, just added opcodes for existing instructions that operate under a new 64-bit mode. That's it. If Intel did that, they would be called lazy. For AMD, it's a "revolution". :p
 

Silus

Supreme [H]ardness
Joined
Aug 21, 2006
Messages
6,477
Intel has one interesting innovation: PROFITS.

Which is the only thing that matters, when you're a money making company...

Hector seems to think that innovation, just for the sake of innovation, will make AMD come out of its dire financial situation.
 

Grentz

Fully [H]
Joined
May 5, 2006
Messages
17,224
Which is the only thing that matters, when you're a money making company...

Hector seems to think that innovation, just for the sake of innovation, will make AMD come out of its dire financial situation.

Heck, it worked for Apple....and they didnt even really innovate much!

But then again, AMD does not have this consumer cult following like Steve does for some reason...
 

Simpson5774

2[H]4U
Joined
Feb 1, 2005
Messages
2,937
I seem to remeber Intel developing SSE1-3, then AMD taking that from them, people dont care about that?. Then AMD ripps the name off with SSE4A which isn't even SSE4 instructions to start off with, their proprietary instructions for the phenom. Then they take SSE5 name, even tho intel has made a habit of reserving those nunbers for when they make the tech.

Intel had the first dual (double) core out the door before AMD, but that was an attempt to spite them.

I believe intel had dual channel first via its RDRAM chipsets, NF2 may of been the first to support DDR dual channel, but it wasn't a true dual channel implementation, more of a cob job.

Although the integrated memory controller is a better design, intel still can pump out more memory bandwidth with its FSB, the IMC does have better latency, but it just hasn't been a needed tech for intel up to now. The Idea OF hyper transport was out there way before AMD put it into motherboards, its even used today on nforce 5-6-7 intel chipsets to connect the north and southbridges.

I think hector was out of his gourd when he said that.
 

nobody_here

[H]F Junkie
Joined
Feb 1, 2006
Messages
8,878
Imitation, the highest form of flattery

however also consider the companies that are big in their respective parts of the market typically took an idea that someone else came up with and simply copied it with minor tweaks to call it their own, but just did a better job at marketing it, better agreements with partners, etc.....and are making more money and better off than the people that actually came up with the ideas to begin with....take a look at China, they are professionals at taking something an American company has made, and copying it, making them in droves, and selling them back to us from overseas via Worstmart and the like for half the cost....you can find almost anything created here in China for pennies on the dollar, inferior product? sometimes

in the end, do i give a crap if AMD came up with something if Intel just did it better...or the other way around??? nope, only people that would let that affect them are fanboys
 

MrWizard6600

Supreme [H]ardness
Joined
Jan 15, 2006
Messages
5,779
edit: quoted the wrong person

The first customer in line for a new gen proc on its launch day would be an engineer from the rival company.

"reverse engineering" isn't illeagal, and when one company makes an innovation that brings better performance, the 2nd company will follow suit. The net result is that they copy each other equally.
 

APOLLO

[H]ard|DCer of the Month - March 2009
Joined
Sep 17, 2000
Messages
9,089
So we can have zero competition and massive price fixing? No thank you...
Well, during the Socket-7 era, there was a 'universal' socket design so to speak. Both Intel and AMD processors worked on this socket and there was no price fixing but competition as always. Not advocating going back to that, which is impossible at this point, but the two companies moved their separate ways after Socket-7.
 

MrGuvernment

Fully [H]
Joined
Aug 3, 2004
Messages
20,565
!

The IMC was one of the major reasons AMD was able to compete with Intel in the A64/P4 gen. Don't underestimate its critical contribution to the A64s performance.


I am not underestimating how it helped AMD at all, obviously Intel doesnt feel it would be so great to them or as useful, or they would of done it years ago, but havent and it shows that the IMC isnt the end all for performance, at least for intel.
 

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
60,246
I am not underestimating how it helped AMD at all, obviously Intel doesnt feel it would be so great to them or as useful, or they would of done it years ago, but havent and it shows that the IMC isnt the end all for performance, at least for intel.

Actually the Pentium 4's Netburst architecture was in fact very dependent on memory bandwidth. An integrated memory controller would have contributed to a large performance boost. (In theory anyway.)

The reasons Intel doesn't use an Integrated memory controller are numerous. The first of which is the lack of flexibility. The Pentium 4 can be used on various motherboards using anything from SDRAM to DDR3 RAM and everything in between. The Core 2 Duo was designed with high IPC in mind so that memory performance wouldn't be as crucial, but I agree with the poster that you were responding to. The integrated memory controller is responsible for nearly 100% of the improvements found between the Athlon XP and the Athlon 64. The integrated memory controller and Hypertransport bus are what got the Opteron space in the x86 server market place. Without their scalability and raw memory bandwidth the Opteron wouldn't have been taken seriously.

Again that all stems back to the integrated memory controller.
 

MrWizard6600

Supreme [H]ardness
Joined
Jan 15, 2006
Messages
5,779
They are right to blame their poor sales on Intel. How dare Intel bring a superior product to market and sell it for a reasonable price?

hehe, exactly.

Not going to happen as long as Intel CPUs lack an integrated memory controller.

its in the cards man, nehalem! You should know that! It's next (once penryn is ramped into full speed)!

Yeah AMD needs to knock that crap off. I am certain Intel can integrate four cores on one die but they choose not to for reasons known only to them. Reasons which obviously make sense and the truth is that native solutions have proven to be of little value in the past and continue to prove unecessary today. Smithfield vs. Pressler and Phenom vs. Kentsfield prove this point.

Seems like you, like me, are itching to see a cost sheet per core and total manufacturing costs of a phenom vs a kentsfield.

What do you mean by "smithfield vs presler"? both of those are intel chips, the former based on two prescotts on a single wafer :eek: and the latter based on two ceader mill's on a single wafer. Smithfield was nasty... I mean really, REALLY nasty.

Had AMD not created the Athlon 64 Intel wouldn't have needed to develope the Core 2 Duo/Quad processors and that would have been unfortunate.

oh god, netburst in 2007? Thats a nasty thought. Maybe they'd finally be at the 10Ghz they promised us tho... probibly not :p

Intel hasn't gone with an integrated memory controller for a number of reasons. As they've shown with Core 2 Duo, the integrated memory controller isn't the only way to great performance and it lacks flexibility. Pentium 4 processors were compatible with RDRAM, SDRAM, DDR and DDR 2 memory. Technically you can use a modern Pentium 4 on a DDR3 chipset. So you can see how not having a memory controller built into your processor can help. Basically motherboard, memory and processor technology can evolve seprately which has obvious advantages. AMD always lags behind memory technology and the reason why is simple. They have to redesign their processors and most likely change sockets everytime new memory technology becomes available. Intel doesn't have to. All they need is a new north bridge for their motherboard chipsets and a short time later you can drop Intel's best on the latest and greatest motherboard.

Another side effect of this seems to be reported on Phenom equipped systems. Apparently raising your memory voltage too high for overclocking can kill or at least damage your processor. That's something you can't do on your Intel based systems. You can kill RAM or the board, but not your processor in that manner.

If that were true, wouldn't OEMs and end users looking for an upgrade flock to asrock and there massive lineups of multiple ram type supporting mobos (I still dont know how they do it but I bet its an ugly ass bridge)?
 
Top