Very Excited About Upcoming 7nm Ryzen, but...

[Zarathustra[H]]

...I am worried.

First off, let me be excruciatingly clear.

I have been a long time AMD fan. Some of my favorite times in this computer hobby for me came when I started college, which perfectly coincided with the 1999 Athlon launch. Having been a long time PC builder (since I was 11 and had a 286) I finally had part time work and a budget to build a real system, and I had a long string of AMD systems.

Ever since 2006, I've been excited about the prospect that AMD would come back and once again be competitive per core with Intel.

- Phenom was a disappointment with its TLB bug
- Phenom II couldn't quite keep up.
- Bulldozer was an utter disaster, and none of the follow-on designs helped much.
- The first two Ryzen releases almost got us there.
- 7nm may finally seal the deal.

Depending on which rumor you believe, 7nm Ryzen will either catch up, just barely miss, or just barely beat beat Intel's current offerings on a per core performance basis. This is great! it's finally happening! But...


There is an elephant in the room no one is talking about

Why is it AMD needs a 7nm process in order to be competitive with Intel's 14nm chips? By all accounts all else being equal, a 7nm chip should be crushing a 14nm chip. Smaller process size means less heat and power, which means you can crank things up more. The fact that AMD is only catching up to Intel's 14nm chips on per core performance on a 7nm process strongly suggests that all things are not equal.

AMD's current architecture may be better than the highly flawed Bulldozer architecture, but the above means that the architecture itself is still WELL behind Intel's, and that they are competitive solely because they currently have the process advantage.



Lets take a historical perspective

Last time AMD was successful with the Athlon launch it was through a combination of great efforts on AMD's part, some key acquisitions (NexGen), Technology Licenses (DEC Alpha EV6 bus) and strategic hires (layoffs from DEC), but also in HUGE part due to Intel's spectacular failure with Netburst and the Pentium 4.

This gave AMD a limited opportunity to break in, and try to cement themselves in the industry, before Intel came roaring back from the Netburst mistake. Through a combination of mismanagement on AMD's part, different priorities and a ton of unfair (and illegal) business practices on Intel's part including exclusivity bribes to OEM's and the Intel compiler intentionally sabotaging AMD performance (later resulting in a $1B settlement paid by Intel to AMD) they were not successful in this regard. It was a desperate and damned near insolvent AMD that finally accepted a lowball $1B settlement from Intel because they didn't have any other choice. When they received it, acquiring ATi and entering the GPU market took money away from R&D resulting in a string of disappointing CPU releases over the next decade.



History Repeats Itself

Similarly, this time around a combination of an internal effort (Zen development under Jim Keller) and Intel screwup (the spectacular failure of Intel's 10nm process) have given AMD an opportunity to break in and cement themselves once again. Unless something drastic changes - however - they look set to fail this time as well.

The fact that AMD is only competitive with Intel's 14nm chips because they are at 7nm means that once Intel fixes their process (and they will eventually, probably skipping 10nm and moving on to the next smaller one at this point, I'm guessing in 2021) Intel will come roaring back and crush AMD.

AMD really needs to not rest on their laurels with 7nm Ryzen, and not spread themselves too thin with acquisitions and adventures in other tech and really focus on improving the core Zen architecture to the point where it is competitive with Intel AT THE SAME PROCESS NODE. They have ~2 years to get there. If they don't, they are toast.


I'd like to hear your thoughts.

 

[gigaxtreme1]

Global foundries is just not that good. TSMC and Samsung are higher echelon. NVDA use both too. The design seems to work well. 5 Ghz seems to be the upper limits for silicon without extreme cooling.. Wait and see.

 

[Dan_D]

This is a very well thought out post. I don't agree with all of it, but here are my thoughts:

...I am worried.

First off, let me be excruciatingly clear.

I have been a long time AMD fan. Some of my favorite times in this computer hobby for me came when I started college, which perfectly coincided 1999 Athlon launch. Having been a long time PC builder (since I was 11 and had a 286) I finally had part time work and a budget to build my own, and I had a long string of AMD systems.

One of my favorite times and favorite systems over the years was my dual Opteron 254 rig. I remember going to an AMD press event at the time and being able to buy an Athlon 3800+ motherboard and CPU combo, as well as a Tyan board and dual CPU combo. We were also able to get FarCry 64bit and both Windows XP Professional x64 Edition and Windows Server 2003 x64 Edition. Those were great times. I got all that stuff and enjoyed the crap out of it. I wouldn't call myself an AMD fanboy though. I like AMD, but they've made more mistakes and bonehead decisions than anything else. You want to like the company, but sometimes they make it very difficult to do so.

Ever since 2006, I've been excited about the prospect that AMD would come back and once again be competitive per core with Intel.

- Phenom was a disappointment with its TLB bug
- Phenom II couldn't quite keep up.
- Bulldozer was an utter disaster, and none of the follow-on designs helped much.
- The first two Ryzen releases almost got us there.
- 7nm may finally seal the deal.

I have doubted AMD would ever reach any kind of parity with Intel again without very specific circumstances coming into play to make that happen. I'm not sure Intel's 7nm failures will be enough to make that happen.

That said, I disagree with your above points.

- Phenom was a disappointment for many reasons. The TLB bug made things worse, but it wasn't a great chip to begin with. It was slower than Athlon x2 per clock and it ran hot etc. I had an early sample of one of these and the only thing I can say for it was that it was nearly impossible to kill. I had no guidelines from AMD and I pushed a stupid amount of voltage through it. I ended up burning out half the phases on the GIGABYTE test board, but the chip was fine. It wouldn't clock for anything either.
- Phenom II was less bad, but couldn't keep up with anything Intel had to offer.
- The point on Bulldozer, I think is spot on.
- I wouldn't say the first two Ryzen releases quite hit that mark. They are certainly much closer to Intel than we would have imagined, but they still come up short in IPC and clock speeds. The saving graces are what bridge the gap for many people. You have good enough performance in games and superb performance in applications which leverage the extra CPU cores AMD offers at some price points. On the HEDT side, this is certainly true where AMD not only offers more cores period, but does so for less money while allowing the user to fully leverage its platform without additional nonsense in the form of a vROC license key.
- I disagree with the idea that 7nm may seal the deal. It will close the gap to an extent, but what extent that is remains to be seen. I am not sure it will match or surpass Intel. I don't think it will.

Depending on which rumor you believe, 7nm Ryzen will either catch up, just barely miss, or just barely beat beat Intel's current offerings on a per core performance basis. This is great! it's finally happening! But...

I think a 7nm Ryzen will be pretty close in terms of IPC. I still think Intel will hold a slight but not insignificant IPC advantage. Even if it doesn't, I think Intel will hold a reasonably large clock speed advantage just as they do now. Where AMD will shine is offering more cores for the same price points, or lower price points than Intel does.

There is an elephant in the room no one is talking about

Why is it AMD needs a 7nm process in order to be competitive with Intel's 14nm chips? By all accounts all else being equal, a 7nm chip should be crushing a 14nm chip. Smaller process size means less heat and power, which means you can crank things up more. The fact that AMD is only catching up to Intel's 14nm chips on per core performance on a 7nm process strongly suggests that all things are not equal.

AMD's current architecture may be better than the highly flawed Bulldozer architecture, but the above means that the architecture itself is still WELL behind Intel's, and that they are competitive solely because they currently have the process advantage.

Well, we knew this. AMD's architecture lags Intel by a significant margin. With Ryzen's release, AMD only managed to just match or edge out Haswell CPU's which were ancient in comparison. Intel often overstates its advantages, but there were many generations of incremental improvements over the years in between Haswell and its latest architectures. Not only that, but Intel was able to go back to Sandy Bridge era clock speeds. AMD may also face problems at 7nm that keep them from fully leveraging the advantages the smaller process brings. We don't really know and may never fully understand the reason for this.

Lets take a historical perspective

Last time AMD was successful with the Athlon launch it was through a combination of great efforts on AMD's part, some key acquisitions (NexGen), Technology Licenses (DEC Alpha EV6 bus) and strategic hires (layoffs from DEC), but also in HUGE part due to Intel's spectacular failure with Netburst and the Pentium 4.

This gave AMD a limited opportunity to break in, and try to cement themselves in the industry, before Intel came roaring back from the Netburst mistake. Through a combination of mismanagement on AMD's part, different priorities and a ton of unfair (and illegal) business practices on Intel's part including exclusivity bribes to OEM's and the Intel compiler intentionally sabotaging AMD performance (later resulting in a $1B settlement paid by Intel to AMD) they were not successful in this regard. It was a desperate and damned near insolvent AMD that finally accepted a lowball $1B settlement from Intel because they didn't have any other choice. When they received it, acquiring ATi and entering the GPU market took money away from R&D resulting in a string of disappointing CPU releases over the next decade.

Indeed. People often fail to realize that AMD's success with the Athlon required a lot of things to happen. Most of which were acquisitions of technologies through the purchase of other businesses. For that to happen, Compaq had to acquire DEC and shut it down, AMD had to buy NexGen Systems, etc. Most importantly, while all this was going on, Intel had to make its gamble with Netburst and lose. It isn't so much that Athlon was that great, it was simply well timed because Intel faltered with a product direction that ultimately didn't pan out. From a business perspective, AMD has made some missteps and that's putting it mildly. AMD, for whatever reason wants to be a premiere CPU company even though they've rarely shown all that much aptitude at doing so. It was far more successful as a flash memory company and in making a few other things. It has a habit of selling off profitable divisions to save its floundering CPU business with quick cash infusions. No other business would have done this. They'd have focused on what they were good at and shut down or sold off divisions that weren't profitable.

AMD's acquisition of ATI hasn't really born the kind of fruit I think AMD was hoping for. Its been profitable at times but honestly, AMD hasn't been all that competitive in its market either. Products aren't priced to be competitive with NVIDIA, quantities of GPU's are low etc. The only thing that's keep cards going out the door was mining. Again, AMD desperately wants to be a premiere GPU company, but the fact is that they haven't been able to compete with NVIDIA on any real level very often.

History Repeats Itself

Similarly, this time around a combination of an internal effort (Zen development under Jim Keller) and Intel screwup (the spectacular failure of Intel's 10nm process) have given AMD an opportunity to break in and cement themselves once again. Unless something drastic changes - however - they look set to fail this time as well.

The fact that AMD is only competitive with Intel's 14nm chips because they are at 7nm means that once Intel fixes their process (and they will eventually, probably skipping 10nm and moving on to the next smaller one at this point, I'm guessing in 2021) Intel will come roaring back and crush AMD.

AMD really needs to not rest on their laurels with 7nm Ryzen, and not spread themselves too thin with acquisitions and adventures in other tech and really focus on improving the core Zen architecture to the point where it is competitive with Intel AT THE SAME PROCESS NODE. They have ~2 years to get there. If they don't, they are toast.


I'd like to hear your thoughts.

I was wondering what you meant by being worried earlier in your post, but now I see it. I think you are right to be worried in that context. Of course, Intel will get their next process together and have that going for them but I don't think it needs to right away in order to dominate AMD. Intel's problem is that its a big bureaucratic pig that doesn't move very fast. When it comes to desktop parts, which frankly are more or less enthusiast parts at this point, it doesn't know what its doing. Its been reacting to AMD because it doesn't understand enthusiasts and / or gamers at all. Its hired guys like Ryan Shrout and Kyle Bennett, which is a reason to be worried for AMD. I think Intel is trying to understand those markets. If it ever does, it has an industrial and production might that AMD can't hope to match. Intel can sell its products at a lower price and still make more money than AMD. Even if it couldn't, it can afford to play a longer game than AMD can.

Now, if Ryzen 2 is some miracle monster that exceeds Intel's current IPC and gets within 200MHz of the same clock speeds at a lower price point, that would be difficult for Intel to overcome any time soon. However, at that point my biggest concern would be AMD being able to maintain that lead in future generations. Given their track record and history of business decisions, it would seem unlikely.

 

[owcraftsman]

Zarathustra[H] You make good points that can't be ignored but there is a reason Intel is having issues shrinking the die that I'm sure we aren't aware of at this time. Even if they come roaring back I intend to support Zen 2 becasue I love supporting the underdog.
In any case I'm sure the image below will be a nice blast from the past for you.

 

[mvmiller12]

This is a very well thought out post. I don't agree with all of it, but here are my thoughts:



One of my favorite times and favorite systems over the years was my dual Opteron 254 rig. I remember going to an AMD press event at the time and being able to buy an Athlon 3800+ motherboard and CPU combo, as well as a Tyan board and dual CPU combo. We were also able to get FarCry 64bit and both Windows XP Professional x64 Edition and Windows Server 2003 x64 Edition. Those were great times. I got all that stuff and enjoyed the crap out of it. I wouldn't call myself an AMD fanboy though. I like AMD, but they've made more mistakes and bonehead decisions than anything else. You want to like the company, but sometimes they make it very difficult to do so.




I have doubted AMD would ever reach any kind of parity with Intel again without very specific circumstances coming into play to make that happen. I'm not sure Intel's 7nm failures will be enough to make that happen.

That said, I disagree with your above points.

- Phenom was a disappointment for many reasons. The TLB bug made things worse, but it wasn't a great chip to begin with. It was slower than Athlon x2 per clock and it ran hot etc. I had an early sample of one of these and the only thing I can say for it was that it was nearly impossible to kill. I had no guidelines from AMD and I pushed a stupid amount of voltage through it. I ended up burning out half the phases on the GIGABYTE test board, but the chip was fine. It wouldn't clock for anything either.
- Phenom II was less bad, but couldn't keep up with anything Intel had to offer.
- The point on Bulldozer, I think is spot on.
- I wouldn't say the first two Ryzen releases quite hit that mark. They are certainly much closer to Intel than we would have imagined, but they still come up short in IPC and clock speeds. The saving graces are what bridge the gap for many people. You have good enough performance in games and superb performance in applications which leverage the extra CPU cores AMD offers at some price points. On the HEDT side, this is certainly true where AMD not only offers more cores period, but does so for less money while allowing the user to fully leverage its platform without additional nonsense in the form of a vROC license key.
- I disagree with the idea that 7nm may seal the deal. It will close the gap to an extent, but what extent that is remains to be seen. I am not sure it will match or surpass Intel. I don't think it will.



I think a 7nm Ryzen will be pretty close in terms of IPC. I still think Intel will hold a slight but not insignificant IPC advantage. Even if it doesn't, I think Intel will hold a reasonably large clock speed advantage just as they do now. Where AMD will shine is offering more cores for the same price points, or lower price points than Intel does.



Well, we knew this. AMD's architecture lags Intel by a significant margin. With Ryzen's release, AMD only managed to just match or edge out Haswell CPU's which were ancient in comparison. Intel often overstates its advantages, but there were many generations of incremental improvements over the years in between Haswell and its latest architectures. Not only that, but Intel was able to go back to Sandy Bridge era clock speeds. AMD may also face problems at 7nm that keep them from fully leveraging the advantages the smaller process brings. We don't really know and may never fully understand the reason for this.



Indeed. People often fail to realize that AMD's success with the Athlon required a lot of things to happen. Most of which were acquisitions of technologies through the purchase of other businesses. For that to happen, Compaq had to acquire DEC and shut it down, AMD had to buy NexGen Systems, etc. Most importantly, while all this was going on, Intel had to make its gamble with Netburst and lose. It isn't so much that Athlon was that great, it was simply well timed because Intel faltered with a product direction that ultimately didn't pan out. From a business perspective, AMD has made some missteps and that's putting it mildly. AMD, for whatever reason wants to be a premiere CPU company even though they've rarely shown all that much aptitude at doing so. It was far more successful as a flash memory company and in making a few other things. It has a habit of selling off profitable divisions to save its floundering CPU business with quick cash infusions. No other business would have done this. They'd have focused on what they were good at and shut down or sold off divisions that weren't profitable.

AMD's acquisition of ATI hasn't really born the kind of fruit I think AMD was hoping for. Its been profitable at times but honestly, AMD hasn't been all that competitive in its market either. Products aren't priced to be competitive with NVIDIA, quantities of GPU's are low etc. The only thing that's keep cards going out the door was mining. Again, AMD desperately wants to be a premiere GPU company, but the fact is that they haven't been able to compete with NVIDIA on any real level very often.



I was wondering what you meant by being worried earlier in your post, but now I see it. I think you are right to be worried in that context. Of course, Intel will get their next process together and have that going for them but I don't think it needs to right away in order to dominate AMD. Intel's problem is that its a big bureaucratic pig that doesn't move very fast. When it comes to desktop parts, which frankly are more or less enthusiast parts at this point, it doesn't know what its doing. Its been reacting to AMD because it doesn't understand enthusiasts and / or gamers at all. Its hired guys like Ryan Shrout and Kyle Bennett, which is a reason to be worried for AMD. I think Intel is trying to understand those markets. If it ever does, it has an industrial and production might that AMD can't hope to match. Intel can sell its products at a lower price and still make more money than AMD. Even if it couldn't, it can afford to play a longer game than AMD can.

Now, if Ryzen 2 is some miracle monster that exceeds Intel's current IPC and gets within 200MHz of the same clock speeds at a lower price point, that would be difficult for Intel to overcome any time soon. However, at that point my biggest concern would be AMD being able to maintain that lead in future generations. Given their track record and history of business decisions, it would seem unlikely.

I generally agree with your points but one thing I think you have both glazed over a bit is the original Athlon competing against the Pentium III, not the Pentium 4. The Athlon itself at the time was generally considered to be at least equal to, if not flat-out better than the Pentium III. The issues came down to motherboards. It was almost impossible to get quality mainboards for a very long time. Hell, the Asus mainboard (one of the best available for the original Athlon) came without the Asus logo on the board itself in a plain white box, and the build quality was merely competent. Despite the fact the boards available for the Athlon were by and large truly terrible, AMD still managed to make some headway.

The Pentium 4 came a fair bit later in the Athlon's life and was Intel's attempt to cleanly leapfrog the Athlon's performance. It was also Intel's attempt to drive the market towards very expensive Rambus memory by saying that the Pentium 4 "required" it. Of course, they then clearly proved that it did not when they started releasing DDR chipsets for their Northwood (mind you, the 2nd gen) Pentium 4's. Even so, their initial DDR chipsets were deliberately performance crippled to artificially advantage their still-selling Rambus chipsets. When Pentium 4's were beating the Athlon in performance, it was by running at stupid-high (for the time) clock speeds at a much higher power draw and with greater heat output. Ironically, this is pretty much the same approach AMD has themselves taken with both Bulldozer and GCN versus their respective competitors. The performance race did not tilt cleanly into Intel's favor until they brought out the original Core M and Core Solo processors, which are themselves a retreat to and refinement of the Pentium III architecture that was a side project relegated to their Israel office.

tl;dr: The Athlon was initially pitched against Pentium III and it performed VERY well in that position despite Intel's fingers on the scale. The Pentium 4 was Intel's response to Athlon. It just happened to fail...

 

[mvmiller12]

Unrelated followup... What happened to all of that Athlon money from when AMD had what was arguably the best processor money could buy at the time?

They pissed it away trying to act like Intel. They hosted "enthusiast" parties at major hotels in a bunch cities, sponsored sports teams, and other stupidity. I've attended Intel and AMD hosted events here in Hampton Roads in the past, and they were relatively lavish for free events being complete with catered lunches. Intel realized they were uselessly pissing cash away and laid off of a lot of this nonsense and AMD eventually stopped being able to afford this stuff.

We can only imagine how things * might * have been different if they had concentrated that money into their R&D efforts... The Intel Core series came seemingly out of nowhere and that was it.

edit: Typos

 

[drescherjm]

What happened to all of that Athlon money from when AMD had what is arguably the best processor money could but at the time?

They paid way way too much for the ATI acquisition.

 

[mvmiller12]

They paid way way too much for the ATI acquisition.

That, too

 

[Dan_D]

I generally agree with your points but one thing I think you have both glazed over a bit is the original Athlon competing against the Pentium III, not the Pentium 4. The Athlon itself at the time was generally considered to be at least equal to, if not flat-out better than the Pentium III. The issues came down to motherboards. It was almost impossible to get quality mainboards for a very long time. Hell, the Asus mainboard (one of the best available for the original Athlon) came without the Asus logo on the board itself in a plain white box, and the build quality was merely competent. Despite the fact the boards available for the Athlon were by and large truly terrible, AMD still managed to make some headway.

The Pentium 4 came a fair bit later in the Athlon's life and was Intel's attempt to cleanly leapfrog the Athlon's performance. It was also Intel's attempt to drive the market towards very expensive Rambus memory by saying that the Pentium 4 "required" it. Of course, they then clearly proved that it did not when they started releasing DDR chipsets for their Northwood (mind you, the 2nd gen) Pentium 4's. Even so, their initial DDR chipsets were deliberately performance crippled to artificially advantage their still-selling Rambus chipsets. When Pentium 4's were beating the Athlon in performance, it was by running at stupid-high (for the time) clock speeds at a much higher power draw and with greater heat output. Ironically, this is pretty much the same approach AMD has themselves taken with both Bulldozer and GCN versus their respective competitors. The performance race did not tilt cleanly into Intel's favor until they brought out the original Core M and Core Solo processors, which are themselves a retreat to and refinement of the Pentium III architecture that was a side project relegated to their Israel office.

tl;dr: The Athlon was initially pitched against Pentium III and it performed VERY well in that position despite Intel's fingers on the scale. The Pentium 4 was Intel's response to Athlon. It just happened to fail...

The Athlon's biggest claim to fame was beating Intel to the 1GHz mark. As I recall, AMD wasn't clearly victorious across the board. It's somewhat debatable that it was better than Intel's Pentium III at the time. It was certainly the closest race we've seen up to this point where one side didn't clearly have an edge over the other. The Athlon is still a product of AMD's acquisition of NexGen Systems and hiring specific people from the DEC buyout by Compaq etc.

I certainly agree with you in that the motherboards and the overall platform weren't up to par on the AMD side of things. I replaced motherboards with burnt out and blown VRMs on a daily basis there for awhile. People are quick to point out that this isn't necessarily AMD's fault, but the specifications that motherboard makers built to are in fact, AMD's. However, since AMD doesn't have the same kind of control over its motherboard partners that Intel does, its difficult to know precisely where the fault lies.

Unrelated followup... What happened to all of that Athlon money from when AMD had what was arguably the best processor money could buy at the time?

They pissed it away trying to act like Intel. They hosted "enthusiast" parties at major hotels in a bunch cities, sponsored sports teams, and other stupidity. I've attended Intel and AMD hosted events here in Hampton Roads in the past, and they were relatively lavish for free events being complete with catered lunches. Intel realized they were uselessly pissing cash away and laid off of a lot of this nonsense and AMD eventually stopped being able to afford this stuff.

We can only imagine how things * might * have been different if they had concentrated that money into their R&D efforts... The Intel Core series came seemingly out of nowhere and that was it.

edit: Typos

You also have to keep in mind that while AMD held the performance crown in the Athlon 64 days, it never made Intel money. Intel's been acused, tried, convicted and fined for its business practices concerning those days. However, it doesn't change the fact that AMD never had the production capacity to come near Intel's sales even if Intel's anti-competitive practices weren't in play.

 

[Zarathustra[H]]

Unrelated followup... What happened to all of that Athlon money from when AMD had what was arguably the best processor money could buy at the time?

They pissed it away trying to act like Intel. They hosted "enthusiast" parties at major hotels in a bunch cities, sponsored sports teams, and other stupidity. I've attended Intel and AMD hosted events here in Hampton Roads in the past, and they were relatively lavish for free events being complete with catered lunches. Intel realized they were uselessly pissing cash away and laid off of a lot of this nonsense and AMD eventually stopped being able to afford this stuff.

We can only imagine how things * might * have been different if they had concentrated that money into their R&D efforts... The Intel Core series came seemingly out of nowhere and that was it.

edit: Typos

My understanding is that there wasn't much money. Sure, the Athlon's were popular among enthusiasts and gamers like ourselves, but because of Intel's illegal business practices, bribing OEM's to not buy AMD CPU's, AMD was almost entirely robbed of the volume market, and without volume even the expensive Athlon X2's in the first half of the 2000's were not enough to drive a huge profit margin.

Of course, Intel settled with them out of court for $1B, but this was a fraction of the true damage done to AMD by their illicit behavior. AMD was desperate for cash at that point though, so they more or less had to take it.

 

[Zarathustra[H]]

The Athlon's biggest claim to fame was beating Intel to the 1GHz mark. As I recall, AMD wasn't clearly victorious across the board. It's somewhat debatable that it was better than Intel's Pentium III at the time. It was certainly the closest race we've seen up to this point where one side didn't clearly have an edge over the other. The Athlon is still a product of AMD's acquisition of NexGen Systems and hiring specific people from the DEC buyout by Compaq etc.

I certainly agree with you in that the motherboards and the overall platform weren't up to par on the AMD side of things. I replaced motherboards with burnt out and blown VRMs on a daily basis there for awhile. People are quick to point out that this isn't necessarily AMD's fault, but the specifications that motherboard makers built to are in fact, AMD's. However, since AMD doesn't have the same kind of control over its motherboard partners that Intel does, its difficult to know precisely where the fault lies.



You also have to keep in mind that while AMD held the performance crown in the Athlon 64 days, it never made Intel money. Intel's been acused, tried, convicted and fined for its business practices concerning those days. However, it doesn't change the fact that AMD never had the production capacity to come near Intel's sales even if Intel's anti-competitive practices weren't in play.

I remember ABIT being awesome That was later with Socket A though I think. These first shitty motherboards came with the Slot A chips, right?

 

[mvmiller12]

The Athlon's biggest claim to fame was beating Intel to the 1GHz mark. As I recall, AMD wasn't clearly victorious across the board. It's somewhat debatable that it was better than Intel's Pentium III at the time. It was certainly the closest race we've seen up to this point where one side didn't clearly have an edge over the other. The Athlon is still a product of AMD's acquisition of NexGen Systems and hiring specific people from the DEC buyout by Compaq etc.

While it is true to a point, it is also a little disingenuous. AMD bought NexGen systems and released their chip as the AMD 5x86-133-P75 - a 486 class chip. The Nexgen team was integral to creating the K6 line (tanking the embarrassing mess that was AMD's internally designed K5) and it's follow-ups. They were well and truly fully integrated into AMD by the time the Athlon was released. The DEC stuff - THAT was still really new.

 

[mvmiller12]

I remember ABIT being awesome That was later with Socket A though I think. These first shitty motherboards came with the Slot A chips, right?

Abit made awesome boards with shitty capacitors. AMD was also not helped by their own reliance on sub-par chipsets.

 

[Dan_D]

While it is true to a point, it is also a little disingenuous. AMD bought NexGen systems and released their chip as the AMD 5x86-133-P75 - a 486 class chip. The Nexgen team was integral to creating the K6 line (tanking the embarrassing mess that was AMD's internally designed K5) and it's follow-ups. They were well and truly fully integrated into AMD by the time the Athlon was released. The DEC stuff - THAT was still really new.

Negative. AMD released its Am5x86 133 in 1995. AMD didn't purchase NexGen Systems until 1996. NexGen's NX586 was definitely Pentium class. Its biggest flaw was that it initially had no mathCo processor. This was at a time when the floating point calculation started to get used in game engines. The NX586 was relatively close to Intel's levels of performance outside of where an FPU was needed. In contrast, the Am5x86 was nothing more than a 486 DX4 133MHz with some tweaks. It was a 486 CPU through and through. As I understand it the NextGen team was well under way on its NX686 chip and even had some preliminary work done on the NX786 when they were bought by AMD. The NX686 was refined and altered to eventually become the K6.

What's interesting to me is the whole transition into the Pentium era. The Pentium 60/66MHz and later CPUs had lower clock speeds than their 486 counterparts but were roughly twice as fast per clock cycle if not more so. We saw 486's that had double the clock speed of their Pentium counterparts often keep up with the later CPU, but not always, and certainly didn't definitively win. Pentium Overdrive processors were a thing back then, but were often hampered by older 486 motherboards and pre-PCI buses. Even the average computer enthusiast and even people with passing hardware knowledge seemed to realize the Pentium was faster.

Yet, when AMD and Cyrix released true Pentium class CPUs a couple years later, the public seemingly forgot that raw clock speed wasn't everything, which led to the whole nightmarish and confusing performance rating system or "PR" system that AMD and Cyrix both used.

 

[Zarathustra[H]]

Abit made awesome boards with shitty capacitors. AMD was also not helped by their own reliance on sub-par chipsets.

It's been really long. My memory is hazy. I don't remember ever losing a k7 motherboard though. I did upgrade when they moved from 100Mhz to 133Mhz FSB.

 

[Nobu]

I'm not too worried. Intel is in the process of totally overhauling their processor tech to get 2.5/3d stacking working, and at the same time a node shrink. I'm not going to say they won't be able to match zen2/3 performance, but I'm starting to wonder whether they will have a competing product out before AMD is ready with their next (zen2+?). We don't have any high performance 10 or 5nm products from intel demoed yet, only some low power 10nm chips. Meanwhile AMD is ready to release 7nm zen2 soon (already shipping rome chips), and is working on the next step (which should also be 7nm, so less pain there).

 

[mvmiller12]

Negative. AMD released its Am5x86 133 in 1995. AMD didn't purchase NexGen Systems until 1996.

Apologies. I was going off of memory, and apparently in this case, my apparently non-ECC memory flipped a bit. But still, AMD made the K6 line for a very long time before the Athlon was introduced. I feel my wider point still stands.


Edit to clarify: While I acknowledge that the Athlon may have started in it's early design phases as a NexGen chip, the DEC licensing and AMD's input would have been more than enough to shift it from being a "NexGen design" to an "AMD/DEC design."

 

[mvmiller12]

What's interesting to me is the whole transition into the Pentium era. The Pentium 60/66MHz and later CPUs had lower clock speeds than their 486 counterparts but were roughly twice as fast per clock cycle if not more so. We saw 486's that had double the clock speed of their Pentium counterparts often keep up with the later CPU, but not always, and certainly didn't definitively win. Pentium Overdrive processors were a thing back then, but were often hampered by older 486 motherboards and pre-PCI buses. Even the average computer enthusiast and even people with passing hardware knowledge seemed to realize the Pentium was faster.

Yet, when AMD and Cyrix released true Pentium class CPUs a couple years later, the public seemingly forgot that raw clock speed wasn't everything, which led to the whole nightmarish and confusing performance rating system or "PR" system that AMD and Cyrix both used.

Quake and it's hunger for FPU power well and truly pushed K6 (and Cyrix) into the "budget" category, and that's a crying shame. What I personally find interesting is that is how K6 and Super Socket 7 were a preview of the horrible mess chipsets were about to become for AMD in the Athlon era. 66Mhz bus K6 CPUs were rock solid on Intel TX chipset boards. The system instability and weird incompatibilities that K6 became infamous for were solely with those 3rd-party chipsets, in my experience. Heck, the K6-2-233MHz CPU on a Shuttle motherboard using the Intel TX chipset was an amazing performer in my personal system for a good little while, and it ran nearly stone-cold as well.

(Fond memories of Wing Commander: Prophecy on that system, using my Voodoo Rush card...)

 

[MMitch]

ah wow sorry stopped reading half way of OP.

Intel have so many architecture flaws that once patched, when possible, reduce that performance advantage. We've only seen the tip of the iceberg of this spectre and meltdown problems.
(The above are what I thinks and that's it)

Also, Intel did resort to shady business to "almost" kill AMD so yeah, R&S isn't there.

My first PC (Bought with my hard teen earned money) was a Thunderbird 1GHz with if I recall correctly an ABIT KT6A-Raid (need validation lol) !! AMD was ahead once, let's see what happens now and anyway if you remove gaming from the picture (And obviously perf/money), AMD is already very competitive !!
In the end we need competition so the main player don't sits on its *** and release rebadged products.

Well I cross read the above before posting and yeah... already said but who cares I push post reply anyway !

 

[N4CR]

They already stomped Intel in perf watt where it counts in datacentre. Now they're coming for the smaller desktop crown.
They have them beat everywhere but the 9999kTi. So for majority of places they are better and cheaper options..
14nm with twice the power use or more beating 14 or 7nm isn't impressive. It's just desperation and marketing strategy washing you over.

 

[noko]

I tend towards AMD for the biggest reason, much better Performance/buck. Other good reasons is upgrades are a hell a lot cheaper at times over Intel. I see nothing that Intel has that would dramatically or even significantly enhance my PC experience. The usual reasons some use to go Intel I just don't see as good reasons for me. Single thread performance, yep Intel is better, multi thread performance? Take Ryzen at 4ghz and Intel at 4ghz and run one core with SMT or Hyperthreading (one core two threads) and see who comes out on top (core to core performance AMD is better at same frequency) - AMD creamates Intel. Anyways one just needs to go with what is best for them with accurate reviews and research. I am more interested in TR this time around which may be awhile before the Zen2 version to come out but if the X570 does have 40pcie lanes and 16 core chips are out - sign me up. I will sorely miss the reviews,excitement and cadence that HardOCP use to have with intelligent conversations.

 

[VanGoghComplex]

I see the rationale for the concern, but I don't understand the reason. Do we want AMD to stomp Intel and then stay on top forever? Isn't that what lead us to a decade of milquetoast generational "improvements" from Intel in the first place - no competition?

I too am excited for 7nm Ryzen and will probably be buying one. I like rooting for the underdog, and everyone likes a good comeback story. But from where I'm standing, it doesn't matter where the competitive edge lies, as long as no one has it for very long.

Right now Intel's competitive edge is architecture maturity. AMD will hopefully make the next competition about process, and something makes me think that a successful 7nm launch well get Intel working overtime on 10nm. But guess what? 5nm is already on AMD's roadmap.

I want the two of them in the ring, trading blows and just barely winning each alternating generation. As long as they're winning by improving their products and not by sabotaging one another, I don't really care which specific part of their product they each have an expertise in. The maturity of the architecture in a 9900K is not enough of a selling point if the 3700X performs on par or better for half the price.

 

[Wiseguy2001]

It's very noble that AMD has honoured the lifetime of AM4 (compared with Intel's shenanigans of a new socket for every other CPU), and it's great that the core-count has rizen so fast, but the memory channels haven't increased, nor has memory speed kept pace. Is dual channel RAM going to heavily bottleneck a 16 core CPU? Personally, I think AMD should have released the 3000 series on two sockets, with the upper range (on AM5?) with quad channel memory, but that would mean cannibalising Thread-Ripper to an extent. At least going forward they can just upgrade the IO chip, not the whole CPU.

I'll probably buy one anyway, if only to marvel at 32 threads in task manager.

I haven't owned an AMD CPU since Barton/ AthlonXPs (although most of my GPUs have been AMD..). I've wanted to switch from Intel but performance hasn't been there. Intel also has the opposite problem of not increasing cores (quad-core Q6600/ i7's for a f**king decade!?). Now AMD has a chance of competing, even beating Intel!

 

[Gideon]

Unless Intel gives up on Monolithic chip design they wont be a threat to AMD as the failure rate for parts for Intel will be much higher. Intel 10nm failure should have been a warning to them, but corporate giants are extremely slow learners. As for 7nm vs 14nm that not quite true, the 14nm process is quite mature and well refined while the 7nm is closer to 10nm and is a brand new process which is likely to get better refined after a year or so of production. Shrinking silicon is no longer the huge gain it used to be. Also what hurt AMD the most when it came to the Athlon was the illegal business practices of Intel. The best thing that can happen for AMD is for business to start buying lots of it's server cpu's as that is where the money is.

 

[RamonGTP]

Well you can look at it a couple different ways.

"Why does AMD need 7nm to compete with 14nm"

Or, on the bright side

"Why is it taking Intel so long to get to 7nm when AMD is already there"

Or, and below is what I subscribe to.

Who cares about process node, and buy what performs the best.

 

[Nobu]

Well you can look at it a couple different ways.

"Why does AMD need 7nm to compete with 14nm"

Or, on the bright side

"Why is it taking Intel so long to get to 7nm when AMD is already there"

Or, and below is what I subscribe to.

Who cares about process node, and buy what performs the best.

There's always, "TSMC's 7nm is more like Intel's 10nm"...which would make AMD equal or better on the "same" node.

Edit: Oh wait, Intel doesn't have any 10nm chips out yet, so I guess you can't say that. Well, they do, but it's not really a fair fight right now with the crippled 10nm node and AMD not having anything released to consumers on 7nm yet.

 

[Pieter3dnow]

You posted a chronological history of events what you forgot to post is the R&D budget for each process (from each camp).

Your feelings about the matter maybe justified but the reason for things happening do have a background with money available for research and development which is telling.

Your assessment when you apply the 7nm process characteristics to Intel counterpart you will see it is better but I would say that if Intel should have the same shoestring budget AMD has had (last 10 years) the tables might as well have been turned.

If AMD can do well next few years the budget would increase and the gap would get smaller and the importance of getting your design to work on 7+nm or 5nm is more important then having a processor architectural advantage.

It is not an elephant it is how well Lisa Su can do with AMD when there is competition (chiplets more or less shows that AMD can adapt).

 

[chameleoneel]

...I am worried.

First off, let me be excruciatingly clear.

I have been a long time AMD fan. Some of my favorite times in this computer hobby for me came when I started college, which perfectly coincided with the 1999 Athlon launch. Having been a long time PC builder (since I was 11 and had a 286) I finally had part time work and a budget to build a real system, and I had a long string of AMD systems.

Ever since 2006, I've been excited about the prospect that AMD would come back and once again be competitive per core with Intel.

- Phenom was a disappointment with its TLB bug
- Phenom II couldn't quite keep up.
- Bulldozer was an utter disaster, and none of the follow-on designs helped much.
- The first two Ryzen releases almost got us there.
- 7nm may finally seal the deal.

Depending on which rumor you believe, 7nm Ryzen will either catch up, just barely miss, or just barely beat beat Intel's current offerings on a per core performance basis. This is great! it's finally happening! But...


There is an elephant in the room no one is talking about

Why is it AMD needs a 7nm process in order to be competitive with Intel's 14nm chips? By all accounts all else being equal, a 7nm chip should be crushing a 14nm chip. Smaller process size means less heat and power, which means you can crank things up more. The fact that AMD is only catching up to Intel's 14nm chips on per core performance on a 7nm process strongly suggests that all things are not equal.

AMD's current architecture may be better than the highly flawed Bulldozer architecture, but the above means that the architecture itself is still WELL behind Intel's, and that they are competitive solely because they currently have the process advantage.



Lets take a historical perspective

Last time AMD was successful with the Athlon launch it was through a combination of great efforts on AMD's part, some key acquisitions (NexGen), Technology Licenses (DEC Alpha EV6 bus) and strategic hires (layoffs from DEC), but also in HUGE part due to Intel's spectacular failure with Netburst and the Pentium 4.

This gave AMD a limited opportunity to break in, and try to cement themselves in the industry, before Intel came roaring back from the Netburst mistake. Through a combination of mismanagement on AMD's part, different priorities and a ton of unfair (and illegal) business practices on Intel's part including exclusivity bribes to OEM's and the Intel compiler intentionally sabotaging AMD performance (later resulting in a $1B settlement paid by Intel to AMD) they were not successful in this regard. It was a desperate and damned near insolvent AMD that finally accepted a lowball $1B settlement from Intel because they didn't have any other choice. When they received it, acquiring ATi and entering the GPU market took money away from R&D resulting in a string of disappointing CPU releases over the next decade.



History Repeats Itself

Similarly, this time around a combination of an internal effort (Zen development under Jim Keller) and Intel screwup (the spectacular failure of Intel's 10nm process) have given AMD an opportunity to break in and cement themselves once again. Unless something drastic changes - however - they look set to fail this time as well.

The fact that AMD is only competitive with Intel's 14nm chips because they are at 7nm means that once Intel fixes their process (and they will eventually, probably skipping 10nm and moving on to the next smaller one at this point, I'm guessing in 2021) Intel will come roaring back and crush AMD.

AMD really needs to not rest on their laurels with 7nm Ryzen, and not spread themselves too thin with acquisitions and adventures in other tech and really focus on improving the core Zen architecture to the point where it is competitive with Intel AT THE SAME PROCESS NODE. They have ~2 years to get there. If they don't, they are toast.


I'd like to hear your thoughts.

In short, it doesn't matter. Intel doesn't have a 7nm chip right now. What matters is, the products available. And AMD's available product, will compete. By the time Intel has a 7nm chip, AMD will have a refined process. Maybe even a smaller, to 5nm. So, we should still see competitive products.

AMD is actually a little ahead of the curve, right now. Intel is scrambling. We got mainstream 6 and 8 cores from Intel, almost 2 years sooner than their plans. Because AMD forced them to, with Ryzen.

 

[Revenant_Knight]

The nanometer specs are mostly marketing fluff now. You have to look at transistor density. Intels 10nm is actually about the same as the Global Foundries / TSMC AMD 7nm.

https://www.google.com/amp/s/wccftech.com/analysis-about-intels-10nm-process/amp/

https://semiengineering.com/nodes-vs-node-lets/

 

[notarat]

...I am worried.

First off, let me be excruciatingly clear.

I have been a long time AMD fan. Some of my favorite times in this computer hobby for me came when I started college, which perfectly coincided with the 1999 Athlon launch. Having been a long time PC builder (since I was 11 and had a 286) I finally had part time work and a budget to build a real system, and I had a long string of AMD systems.

Ever since 2006, I've been excited about the prospect that AMD would come back and once again be competitive per core with Intel.

- Phenom was a disappointment with its TLB bug
- Phenom II couldn't quite keep up.
- Bulldozer was an utter disaster, and none of the follow-on designs helped much.
- The first two Ryzen releases almost got us there.
- 7nm may finally seal the deal.

Depending on which rumor you believe, 7nm Ryzen will either catch up, just barely miss, or just barely beat beat Intel's current offerings on a per core performance basis. This is great! it's finally happening! But...


There is an elephant in the room no one is talking about

Why is it AMD needs a 7nm process in order to be competitive with Intel's 14nm chips? By all accounts all else being equal, a 7nm chip should be crushing a 14nm chip. Smaller process size means less heat and power, which means you can crank things up more. The fact that AMD is only catching up to Intel's 14nm chips on per core performance on a 7nm process strongly suggests that all things are not equal.

AMD's current architecture may be better than the highly flawed Bulldozer architecture, but the above means that the architecture itself is still WELL behind Intel's, and that they are competitive solely because they currently have the process advantage.



Lets take a historical perspective

Last time AMD was successful with the Athlon launch it was through a combination of great efforts on AMD's part, some key acquisitions (NexGen), Technology Licenses (DEC Alpha EV6 bus) and strategic hires (layoffs from DEC), but also in HUGE part due to Intel's spectacular failure with Netburst and the Pentium 4.

This gave AMD a limited opportunity to break in, and try to cement themselves in the industry, before Intel came roaring back from the Netburst mistake. Through a combination of mismanagement on AMD's part, different priorities and a ton of unfair (and illegal) business practices on Intel's part including exclusivity bribes to OEM's and the Intel compiler intentionally sabotaging AMD performance (later resulting in a $1B settlement paid by Intel to AMD) they were not successful in this regard. It was a desperate and damned near insolvent AMD that finally accepted a lowball $1B settlement from Intel because they didn't have any other choice. When they received it, acquiring ATi and entering the GPU market took money away from R&D resulting in a string of disappointing CPU releases over the next decade.



History Repeats Itself

Similarly, this time around a combination of an internal effort (Zen development under Jim Keller) and Intel screwup (the spectacular failure of Intel's 10nm process) have given AMD an opportunity to break in and cement themselves once again. Unless something drastic changes - however - they look set to fail this time as well.

The fact that AMD is only competitive with Intel's 14nm chips because they are at 7nm means that once Intel fixes their process (and they will eventually, probably skipping 10nm and moving on to the next smaller one at this point, I'm guessing in 2021) Intel will come roaring back and crush AMD.

AMD really needs to not rest on their laurels with 7nm Ryzen, and not spread themselves too thin with acquisitions and adventures in other tech and really focus on improving the core Zen architecture to the point where it is competitive with Intel AT THE SAME PROCESS NODE. They have ~2 years to get there. If they don't, they are toast.


I'd like to hear your thoughts.

TL;DR buy what you want

 

[ManofGod]

Having a smaller manufacturing process does not automatically mean that a cpu, gpu or whatever is going to be better. Even Intel is not bulletproof with their process, after all.

 

[Pieter3dnow]

The nanometer specs are mostly marketing fluff now. You have to look at transistor density. Intels 10nm is actually about the same as the Global Foundries / TSMC AMD 7nm.

https://www.google.com/amp/s/wccftech.com/analysis-about-intels-10nm-process/amp/

https://semiengineering.com/nodes-vs-node-lets/

Fluff trumps vaporware every day of the week

I mean one has a marketing designation but the company using it can get it working and the other is technically superior but can not be used for anything but laptop cpu/igp.

At this point it is trivial what the name is or technical status is as long as you can get yields out of it that are decent (on parts that actually matter to us enthusiasts).

 

[dvsman]

OP brings up a good point. But while Intel does seem to have advantages stemming from their superior architecture design - how much of that advantage will remain after they redesign to mitigate Meltdown and Spectre and all the variants. Apparently the less efficient AMD chip design is not as vulnerable, and stands to lose less in what mitigations are necessary.

 

[Ocellaris]

[QUOTE="Nobu”]AMD not having anything released to consumers on 7nm yet.[/QUOTE]

Radeon 7 is on 7nm tech right now...

 

[Nobu]

Radeon 7 is on 7nm tech right now...

This is in the processor subforum in a thread about ryzen processors...

 

[VanGoghComplex]

Radeon 7 is on 7nm tech right now...

This is in the processor subforum in a thread about ryzen processors...[/QUOTE]
You did say "anything," not "any processor."

 

[Nobu]

You did say "anything," not "any processor."

I'm sorry for not being more specific? Why bring up vega in this thread except to be contrary?

 

[VanGoghComplex]

You did say "anything," not "any processor."

I'm sorry for not being more specific? Why bring up vega in this thread except to be contrary?[/QUOTE]
Why claim "AMD hasn't delivered anything to consumers on 7nm" except to be misleading?

 

[tangoseal]

It's very noble that AMD has honoured the lifetime of AM4 (compared with Intel's shenanigans of a new socket for every other CPU), and it's great that the core-count has rizen so fast, but the memory channels haven't increased, nor has memory speed kept pace. Is dual channel RAM going to heavily bottleneck a 16 core CPU? Personally, I think AMD should have released the 3000 series on two sockets, with the upper range (on AM5?) with quad channel memory, but that would mean cannibalising Thread-Ripper to an extent. At least going forward they can just upgrade the IO chip, not the whole CPU.

I'll probably buy one anyway, if only to marvel at 32 threads in task manager.

I haven't owned an AMD CPU since Barton/ AthlonXPs (although most of my GPUs have been AMD..). I've wanted to switch from Intel but performance hasn't been there. Intel also has the opposite problem of not increasing cores (quad-core Q6600/ i7's for a f**king decade!?). Now AMD has a chance of competing, even beating Intel!

ABSOLUTELY NOT!

I have a 16 core 2950x. I can't tell a difference in performance whether it be quad or dual channel. None whatsoever and I have ran both.

 

[tangoseal]

Radeon 7 is on 7nm tech right now...

This is in the processor subforum in a thread about ryzen processors...[/QUOTE]

Guess you have a broken quote.




Radeon VII is an AMD Processor.

It might be stuck on a video card but it still processes data.