AMD Stock Broke All-Time Record for the Company, Peaked at $49.10 per Share

[SL2]

AMD - first to dual core, AMD64, IMC, HyperTransport, IF, taking IMC off die for MCM etc.

You can add other achievements, but these are the major ones that come to mind.

First dual core is not true. Intel beat AMD to it and launched Pentium D one week earlier. The fact that Pentium D was hot running and inferior doesn't change that..

 

[R0H1T]

Alright, native dual core ~ does that sound right?

 

[moproblems99]

I sold my stock at 42.71 to help get us in line to buy a house, I've held off buying any more simply because I'm weary how hard the stock market will tank once the next recession hits.

If you are worried about a recession, dumping money into real estate isn't exactly the best idea.

 

[division2Ubisoft]

First dual core is not true. Intel beat AMD to it and launched Pentium D one week earlier. The fact that Pentium D was hot running and inferior doesn't change that..

intel website says that pentium D 820 has 2 cores and 2 threads , but if i remember , D was singlecore with hyperthreading, maybe i am wrong.

 

[R0H1T]

That's what I remember as well, but tbf I didn't even have a PC at that time so no idea about the launch dates.

 

[division2Ubisoft]

That's what I remember as well, but tbf I didn't even have a PC at that time so no idea about the launch dates.

first real dualcores was Core2 , is that right? then D was fake like whole intel.

 

[SL2]

intel website says that pentium D 820 has 2 cores and 2 threads , but if i remember , D was singlecore with hyperthreading, maybe i am wrong.

That's some of the Pentium Northwood a couple of years earlier, maybe those with 800 FSB, not sure.

 

[thesmokingman]

That's what I remember as well, but tbf I didn't even have a PC at that time so no idea about the launch dates.

2005	AMD released their first dual-core processor, the Athlon 64 X2 3800+ (2.0 GHz, 512 KB L2 cache per core), on April 21, 2005.

Computer Processor History

The complete history of the computer processor or CPU (Central Processing Unit) and major events and releases of computer processors.

www.computerhope.com

No to fakes!

 

[SL2]

first real dualcores was Core2 , is that right? then D was fake like whole intel.

Pentium D was utter crap, but I'm not sure that they were fake dual cores because of that. Someone might call Ryzen 3950 fake just because it doesn't have all the cores on one die.
Hypocrisy.

2005	AMD released their first dual-core processor, the Athlon 64 X2 3800+ (2.0 GHz, 512 KB L2 cache per core), on April 21, 2005.

Computer Processor History

The complete history of the computer processor or CPU (Central Processing Unit) and major events and releases of computer processors.

www.computerhope.com

No to fakes!

I do remember *Anandtech reviewing the PD first, and they pointed out that Intel was first. Can't find it now tho.

Other sites show different dates, with the 25th for PD, and the 31st for X2.

X2: https://www.theregister.co.uk/2005/05/31/amd_athlon_64_x2/
X2: https://www.computerworld.com/article/2556702/amd-launches-dual-core-athlon-64-x2.html

PD: https://www.extremetech.com/extreme/75294-intels-pentium-d-price-half-that-of-amds-x2

 

[division2Ubisoft]

That's some of the Pentium Northwood a couple of years earlier, maybe those with 800 FSB, not sure.

ok. no matter what. but core2 was big imprevement in cpu architecture. and ryzen too. for few hundred bucks 8core 16 threads with great IPC.

 

[SL2]

ok. no matter what. but core2 was big imprevement in cpu architecture. and ryzen too. for few hundred bucks 8core 16 threads with great IPC.

Yeah for sure. Intel was on a roll back then, with Core (1), Core 2, and 45 nm Core 2 being launched within 15 months or so, and with the Core 2 Quad following only six months later (and then no more cores for ten years..).
The situation is pretty much the opposite now with AMD's new models.

 

[s3thra]

This strange alternate dimension we seemed to slip into around 2016 has had some positives!

 

[R0H1T]

but I'm not sure that they were fake dual cores because of that

They weren't, just not what dual cores are today. More like dual socket independent chips ~

Intel's Pentium D was literally glued together as there's not connection between the two Pentium 4 dies. They both connect to NB independently and share the FSB as a 2-way/2-socket server system. It was just the RAM controller was in the NB made it UMA instead of NUMA. At that point, Intel didn't have QPI or HyperTransport equivalent connects. So to share data Pentium D use FSB to sync the L2 cache just like how SLi sync VRAM for two graphic card. That significantly reduce the memory performance and Pentium D at that time was way slower than a single core Pentium 4.

One of the reasons why Intel with "native" dual cores is this. There's no way they could compete without making one, their glued disaster though was a learning curve & I'm sure they pulled a fast one back then just like with the launch of TR3 recently.

 

[trparky]

Wasn't the Pentium D simply two Prescott Pentium 4's glued together?

 

[RandallFlagg]

Long term, the formula is the same: AMD will be reaching the end of the Zen product stack in a year, Intel is recovering from their 14nm+++++++ (+?) era, and there are rumors about Keller's new stacked/ultra-dense designs swirling around. Intel will want a replay of 2006 in 2021 (and I feel very old saying that).

View attachment 141184

Moore's Law is not dead insists Intel's Jim Keller

And the 71st anniversary of the announcement of the first transistor was this weekend.

hexus.net

You guys are comparing Apples to Oranges and coming up with a Bannana.

In 2006 AMD was like Intel. It not only did its own chip design, it had its own Fabs. Even after GloFlo split off, AMD was locked into using their process nodes. A big part of the reason AMD faltered was due to falling behind in fab process technology. No more.

AMD focus' almost entirely on chip design now. TSMC, Samsung, and GloFlo have to worry about the process tech not AMD. These companies are not flyweights. Samsung overall has 3x Intel's revenue (~$210B). TSMC, more than half of Intel's revenue (> 35B). TSMC is far more focused on a particular service than Intel , Samsung much less so. Other much larger companies have vested interest in the success of partners like TSMC (see Apple).

You have mobile / ARM SoCs which continue to pour money and resources into these fabs. They are not in the slightest dependent on AMD, Nvidia, or any other single customer.

This is an entirely different landscape and Intel hasn't shown any capability for quite some time, not only failing in execution (process fab) but in strategy. Core was a derivative of the Pentium-M, which was downplayed vs NetBurst until it became obvious that NetBurst was a failure. Then they sort of dug up Pentium-M and reworked it into Core. It wasn't strategy, they just bumbled into something good that saved their tails.

Highly unlikely that'll happen again.

 

[Fluffmeister]

Bingo... "Real men have fabs" until they can't afford them anymore.

 

[mtcn77]

Against the farewell wishes of the OP, I would like to add I hope they don't drink and drive. Because that is how Sanitarium begins...

 

[Schmuckley]

If only I had the money to buy when it was $2 a share. It was not that long ago, and I knew they would come through.

 

[Zubasa]

Wasn't the Pentium D simply two Prescott Pentium 4's glued together?

Yes they were.
Glue together desktop dies, according to Intel own marketing slides should be completely useless
What is even more ironic is the 28-core XCC dies are technically HEDT dies now,
so Intel also glued together 2 of them to make the 56-core Xeon 9200 series.

 

[phanbuey]

You guys are comparing Apples to Oranges and coming up with a Bannana.

In 2006 AMD was like Intel. It not only did its own chip design, it had its own Fabs. Even after GloFlo split off, AMD was locked into using their process nodes. A big part of the reason AMD faltered was due to falling behind in fab process technology. No more.

AMD focus' almost entirely on chip design now. TSMC, Samsung, and GloFlo have to worry about the process tech not AMD. These companies are not flyweights. Samsung overall has 3x Intel's revenue (~$210B). TSMC, more than half of Intel's revenue (> 35B). TSMC is far more focused on a particular service than Intel , Samsung much less so. Other much larger companies have vested interest in the success of partners like TSMC (see Apple).

You have mobile / ARM SoCs which continue to pour money and resources into these fabs. They are not in the slightest dependent on AMD, Nvidia, or any other single customer.

This is an entirely different landscape and Intel hasn't shown any capability for quite some time, not only failing in execution (process fab) but in strategy. Core was a derivative of the Pentium-M, which was downplayed vs NetBurst until it became obvious that NetBurst was a failure. Then they sort of dug up Pentium-M and reworked it into Core. It wasn't strategy, they just bumbled into something good that saved their tails.

Highly unlikely that'll happen again.

Except they did it again with Nehalem, and then again with Sandy Bridge... Also, they have a guy working for them who is known to do these types of things, and was the guy that came up with Zen.

I don't really think it's as unlikely as you think it is... AMD was unlikely with Zen - that was a rabbit out of a hat; but again, the guy that did that has been at intel for the last 3 years.

 

[R0H1T]

Except they did it again with Nehalem, and then again with Sandy Bridge... Also, they have a guy working for them who is known to do these types of things, and was the guy that came up with Zen.

I don't really think it's as unlikely as you think it is... AMD was unlikely with Zen - that was a rabbit out of a hat; but again, the guy that did that has been at intel for the last 3 years.

And again AMD was stuck with 90nm & then 65nm, before moving to 32nm for *dozer which in fact was going backwards for IPC. Yes Jim Keller did a lot of good work at AMD, but you'd be naive to think that he did wonders (all by himself) & that Zen is not a culmination of whatever AMD had cooking up from their experiences with FX, Cat cores.

Simply put don't expect miracles from the Man, while a double digit IPC is likely with one or both of the upcoming *coves but again you've also seen absolute frequency go down on 10nm & will likely go down further on 7nm. Intel & AMD will be fighting a bitter freq vs IPC battle as the nodes get smaller, so any potential rabbit out of the hats will be rarer still.

 

[hat]

Overall performance is what matters. Nobody cared that Core 2 was clocked slower than their predecessors when they saw how well it actually performed.

 

[thesmokingman]

Except they did it again with Nehalem, and then again with Sandy Bridge... Also, they have a guy working for them who is known to do these types of things, and was the guy that came up with Zen.

I don't really think it's as unlikely as you think it is... AMD was unlikely with Zen - that was a rabbit out of a hat; but again, the guy that did that has been at intel for the last 3 years.

You know he wasn't the chief architect right and that there was a small team there specifically for Zen? Fyi, the chief architect is Michael Clark. Jim was the team leader. And Keller has been at Intel since 2018 so not 3 years. Keller does have a team that's more than double the team below. That's the Austin design team btw. I love Keller but he definitely wasn't the the chief behind Zen.

 

[Flanker]

Yes they were.
Glue together desktop dies, according to Intel own marketing slides should be completely useless
What is even more ironic is the 28-core XCC dies are technically HEDT dies now,
so Intel also glued together 2 of them to make the 56-core Xeon 9200 series.

The Q6600 was also glued together wasn't it? I remembered all the hype of it being the first quad core
But then the epic fail of that era was probably AMD Quad Platform which didn't even have the glue

 

[Zubasa]

The Q6600 was also glued together wasn't it? I remembered all the hype of it being the first quad core
But then the epic fail of that era was probably AMD Quad Platform which didn't even have the glue

Yes it was, and it was one of the best Intel product for its time.
It also overclocked like a champ.
The issue with how Intel did the Pentium D and Core 2 Quads were the 2 dies were connected via the external Northbridge.
So as you expect there would be a latency hit for doing that. Ironically the Zen 2 IO die is basically an internal Northbridge.
So we went from external to on die, then to off the die but still in the CPU package.