Jim Keller leaves AMD

Rizen

[H]F Junkie
Joined
Jul 16, 2000
Messages
9,487
http://hexus.net/tech/news/cpu/86585-legendary-cpu-architect-jim-keller-leaves-amd/

Ooof. This isn't good in my opinion. I know this is the GPU forum but the AMD CPU forum is dead. I think this doesn't bode well for the future.

Jim Keller, former Chief Architect of Microprocessor Cores, will leave the company today to pursue other opportunities.

Well known for his work during AMD's heyday, Keller (pictured) was involved in the creation of the original Athlon architecture, K7, and then served as a lead architect on K8. After playing an instrumental role in developing the world's first native x86-64 bit architecture, Keller later joined Apple and helped develop the company's A4 and A5 SoCs before rejoining AMD in 2012 to spearhead the firm's upcoming Zen architecture.
 
Well, the damage won't be felt for a few years. :D

The smart ones leave early.
 
Yeah well, guess I better upgrade early on before the inevitable Intel price hikes occur. (I do not live near a Micro Center so that does not count for me.)
 
High-end x86 is dying. It has been dying since 1992. Better architectures are available. I cannot wait do see AMD to go bankrupt. They are just prolonging their certain demise. Intel has had a monopoly on x86 for years but nobody will do shit about it until is is obvious. Mostly because most consumers want x86, Windows-basedhardware and nobody by Intel or AMD can make a x86 CPU. General PC consumers don't care about performance or specs and have no need for a desktop or laptop other than an Ultrabook. Let Intel become a manufacturer for Windows-based, x86 junk. Hope that AMD spins off their GPU department. And hope that PowerPC CPUs make a comeback. IBM and Freescale have the resources to make enthusiast hardware. That is what we should hope for.
 
What? There is nothing wrong with x86 such that IBM or even ARM will come in and make better. Without AMD it will only be Intel in the enthusiast space. Period. There isn't enough cash available to anyone else to become a meaningful entrant.
 
High-end x86 is dying. It has been dying since 1992. Better architectures are available. I cannot wait do see AMD to go bankrupt. They are just prolonging their certain demise. Intel has had a monopoly on x86 for years but nobody will do shit about it until is is obvious. Mostly because most consumers want x86, Windows-basedhardware and nobody by Intel or AMD can make a x86 CPU. General PC consumers don't care about performance or specs and have no need for a desktop or laptop other than an Ultrabook. Let Intel become a manufacturer for Windows-based, x86 junk. Hope that AMD spins off their GPU department. And hope that PowerPC CPUs make a comeback. IBM and Freescale have the resources to make enthusiast hardware. That is what we should hope for.

You cannot wait to see people lose the jobs? PowerPC make a comeback? Also, Windows based, X86 is far from junk and you know it. I assume you don't use any X86 hardware then, right? Right?
 
Maybe this actually means that Zen is finished and he doesn't need to be there now?
 
High-end x86 is dying. It has been dying since 1992. Better architectures are available. I cannot wait do see AMD to go bankrupt. They are just prolonging their certain demise. Intel has had a monopoly on x86 for years but nobody will do shit about it until is is obvious. Mostly because most consumers want x86, Windows-basedhardware and nobody by Intel or AMD can make a x86 CPU. General PC consumers don't care about performance or specs and have no need for a desktop or laptop other than an Ultrabook. Let Intel become a manufacturer for Windows-based, x86 junk. Hope that AMD spins off their GPU department. And hope that PowerPC CPUs make a comeback. IBM and Freescale have the resources to make enthusiast hardware. That is what we should hope for.

Wow.. someone is stuck in the late 80s, early 90s...

High End X86 has dominated the market, including server space for a long while... not so much in 1992... but pretty much since the Xeons came out.

Look at IBMs market share, and same for Sun(now Oracle).... Hp is toast as well...
 
Last edited:
Wow.. someone is stuck in the late 80s, early 90s...

High End X86 has dominated the market, including server space :p Look at IBMs market share, and same for Sun(now Oracle).... Hp is toast as well...

Shit. Have you not seen the man go on and on and on and on and on

and on and on and on and on and on and on and on and on and on

and on and on and on and on and on about CRTs and G4 workstations?

I hope nobody breaks in and finds old mid-90s grunge CDs scattered around empty bottles of Crystal Pepsi.
 
High-end x86 has only dominated because Intel chose to kill any opposition through illegal trust deals, whether it was AMD or other CPU architectures. x86 is terrible 32-bit, CISC architecture that has never been changed for 20+ years. Intel and AMD have figured out that x86 sucks. Their CPUs are only x86 by name and support. The actual internal µarch of modern AMD and Intel CPUs is a 64-bit, RISC-based proprietary design that is similar to PowerPC. That's what AMD64, AVX and other "extensions" are. However, having a CPU that has an incredibly complex microcode and translates its data into an internal architecture while not directly execute code from the OS is terrible for performance. You can do these sorts of designs as long as you don't have any competition, but if a company is legitimately competing against you in the same market, you are going to get destroyed. Look at Intel's best pieces of silicon versus IBM's. The best Power8 dies can go up to 5GHz, have 12 independent hardware cores, 8-way multithreading for a total of 96 threads and much better IPC than Haswell. Intel's best Haswell dies go up to 3GHz, have 18 independent hardware cores, 36-threads and IPC that crippled because of the architecture. If IBM wanted to, they could wipe Intel out of the enthusiast hardware market. Both chips are 22nm and have similar die sizes, but IBM's is faster.
 
Anyways, this looks like a source of concern for AMD since I thought Jim Keller was the "processor man" that was supposed to help AMD turnaround their processor line. It looks like Zen is the fruit of that endeavor and they have a roadmap to speak of, but what about in five years time?
 
I don't think it's time for sky is falling, but there are quite a few ways this can be taken.

If this guy is indeed a superstar wouldn't you at least want his name on the payroll? Just the simple subtraction of his name at a time like this is enough to raise eyebrows.
 
High-end x86 is dying. It has been dying since 1992. Better architectures are available. I cannot wait do see AMD to go bankrupt. They are just prolonging their certain demise. Intel has had a monopoly on x86 for years but nobody will do shit about it until is is obvious. Mostly because most consumers want x86, Windows-basedhardware and nobody by Intel or AMD can make a x86 CPU. General PC consumers don't care about performance or specs and have no need for a desktop or laptop other than an Ultrabook. Let Intel become a manufacturer for Windows-based, x86 junk. Hope that AMD spins off their GPU department. And hope that PowerPC CPUs make a comeback. IBM and Freescale have the resources to make enthusiast hardware. That is what we should hope for.
My god, kid get off the forums. You have absolutely no idea what the hell youre talking about. If AMD goes out of business then you can be SURE Intel CPU now costing $350 will cost $500+, along with motherboards, GPUs and everything else. It gives manufacturers no reason to compete naturally so the prices they make will be there to stay.
 
My god, kid get off the forums. You have absolutely no idea what the hell youre talking about. If AMD goes out of business then you can be SURE Intel CPU now costing $350 will cost $500+, along with motherboards, GPUs and everything else. It gives manufacturers no reason to compete naturally so the prices they make will be there to stay.


That has already happened, when was the last time you have seen $8000 high end workstation CPU's?

On the lower consumer end, still looking at $1500 for top end CPU's, it wouldn't matter of AMD is there or not, at this point, AMD isn't there in most of the CPU market, even on the low end because they don't compete well because of power usage.

Anyways, RISC architecture died out in the PC market because of lack of software and CISC performance did start to catch up and surpass RISC (remember the MIP wars), and this is where servers started switching over to Intel processors. (if I remember correctly it was right around the time of Pentium 2)
 
Last edited:
My god, kid get off the forums. You have absolutely no idea what the hell youre talking about. If AMD goes out of business then you can be SURE Intel CPU now costing $350 will cost $500+, along with motherboards, GPUs and everything else. It gives manufacturers no reason to compete naturally so the prices they make will be there to stay.

AMD cannot stay in business. If they really could make CPUs that were comparable to Intel's, that would be great. But it is clear that they are just dragging out their demise. If they make a comeback to their golden age, that would be great. But in reality, AMD is just staying alive. It's enough to keep Intel in total control of the high-end, but not enough for Intel to jack prices up through the roof and stop innovating. What we have right now is something Intel wants. Intel doesn't want AMD dead. If AMD went out of business, Intel would have to make some very hard choices: sell their junk i5s for $750 and entry-level motherboards for $500 or continue what they are doing now. Either way, they would be at risk of prosecution via anti-monopoly and anti-trust laws. Pumping prices through the roof would give them an immediate surge of income, but it would ease makers of non-x86 CPUs into competition. Pumping prices would also give more attention from regulatory industries.

If AMD goes under and Intel chooses to pump prices, it would suck for a few years, but it would give other CPU manufacturers a huge chance to truly compete. It would give makers other CPUs based on other architectures a chance to end x86 forever.
 
Anyways, RISC architecture died out in the PC market because of lack of software and CISC performance did start to catch up and surpass RISC (remember the MIP wars), and this is where servers started switching over to Intel processors. (if I remember correctly it was right around the time of Pentium 2)

What happened was that Intel and AMD basically turned their CPUs into RISC chips internally while still holding a monopoly on the market due to the fact that they were x86 on the outside. No modern CPU is pure x86. RISC will come back. PowerPC has always been the architecture of the future. Linux will surpass Windows for gaming and enthusiast use. Today's PowerPCs are so powerful that they can fully emulate Windows and x86 without a gaming-significant performance hit. The hit would likely be about enough to reduce a (theoretical) CPU that was as fast as a 12-core Haswell at 4-5 GHz to a 4560K at ~3 GHz. But people would not need to emulate most things. People who game or use other windows-only software would do that in the emulator and it would work comparably to a x86 CPU of the same price. Everything else can run without emulation. And if people start switching to PowerPC in significant numbers, it would probably get game developers and software makers to port their work.
 
Well I kinda of agree with you, yes, 64 bit chips are more RISC based now. But it was still a software issue, with Windows having 80+ percent of the market at the time which put RISC based systems at a disadvantage. now with Windows working on ARM cpu's we can hope that PowerPC's will capitalize but since the landscape is saturated for Windows based software, it will be much harder for them to make in roads. I can see it sometime in the future, but it will be a slow growth.
 
What happened was that Intel and AMD basically turned their CPUs into RISC chips internally while still holding a monopoly on the market due to the fact that they were x86 on the outside. No modern CPU is pure x86. RISC will come back. PowerPC has always been the architecture of the future. Linux will surpass Windows for gaming and enthusiast use. Today's PowerPCs are so powerful that they can fully emulate Windows and x86 without a gaming-significant performance hit. The hit would likely be about enough to reduce a (theoretical) CPU that was as fast as a 12-core Haswell at 4-5 GHz to a 4560K at ~3 GHz. But people would not need to emulate most things. People who game or use other windows-only software would do that in the emulator and it would work comparably to a x86 CPU of the same price. Everything else can run without emulation. And if people start switching to PowerPC in significant numbers, it would probably get game developers and software makers to port their work.


PowerPC will never come back, for the same reason it didn't win the market over in the first place, it's too expensive.
 
Biggest issue for intel is no one wants to upgrade so pricing too high will just result in less customers until the used cpu market depletes and more old cpus die
 
PowerPC will never come back, for the same reason it didn't win the market over in the first place, it's too expensive.

I would rather see the enthusiast desktop market crash and burn quickly than slowly die from a perpetual, Intel-run monopoly that does not innovate. PowerPC is no more expensive than high-end x86 hardware. IBM can't crank out $99 Haswell Pentiums for consumers but they can do amazing, x86-competitive hardware for the $300+ range. Today is the time for PowerPC to come back. As long as Intel around and x86 is common, PowerPC will never be a general-purpose architecture for office PCs, ultrabooks and junk. General consumers don't give a shit about their devices' performance. They care about two things: is it cheap and can it handle Facebook with stable, reliable performance. PowerPC is not meant to compete there. But it could destroy x86 in markets where it can compete.
 
Last edited:
I would rather see the enthusiast desktop market crash and burn quickly than slowly die from a perpetual, Intel-run monopoly that does not innovate. PowerPC is no more expensive than high-end x86 hardware. IBM can't crank out $99 Haswell Pentiums for consumers but they can do amazing, x86-competitive hardware for the $300+ range.

Maybe one day we'll have a competitive ARM desktop CPU that runs on Windows. Once we do, it will be the end of Intel's monopoly. I see more potential in that happening than AMD ever making a comeback.
 
I would rather see the enthusiast desktop market crash and burn quickly than slowly die from a perpetual, Intel-run monopoly that does not innovate. PowerPC is no more expensive than high-end x86 hardware. IBM can't crank out $99 Haswell Pentiums for consumers but they can do amazing, x86-competitive hardware for the $300+ range.

Don't forget about SPARC or Alpha, either. Sun and DEC workstations were blazingly fast (especially in regards to UltraSPARC in terms of system I/O, FPU performance, and scalability) and had colossal amounts of memory for their time.

Hell, let's get Motorola in on things again as a second source supplier for them both :D
 
Last edited:
Hell, let's get Motorola in on things again as a second source supplier for them both :D

Moto is still around. They spun off their CPU department and it is now Freescale. They make good stuff that would work well in laptops and is comparable to mobile i7s, but it is not high-performance stuff like IBM.
 
Biggest issue for intel is no one wants to upgrade so pricing too high will just result in less customers until the used cpu market depletes and more old cpus die

You're right, but the issue is again competition. Intel is simply rehashing old architectures without really putting too much effort into anything new. Ever since i7 chips launched there was little reason to upgrade.
 
Moto is still around. They spun off their CPU department and it is now Freescale. They make good stuff that would work well in laptops and is comparable to mobile i7s, but it is not high-performance stuff like IBM.

I know they're still around, and they spun off their CPU division, but it'd be fun to see another Moto branded CPU after all these years.

*weeps over the 68060's coffin*

You're right, but the issue is again competition. Intel is simply rehashing old architectures without really putting too much effort into anything new. Ever since i7 chips launched there was little reason to upgrade.

Last big thing we got was in Sandy Bridge. One is still has a happy home in my system, and I see no huge rush to upgrade yet. I couldn't say that about my Core 2 Quad Q9550 three years after I got one.
 
I'm relatively certain rabidz7 is yanking y'alls chains. His template is to take 20+ year old hardware and trumpet how superior it is to today's. Search his posts on CRT superiority for lulz.
 
I'm relatively certain rabidz7 is yanking y'alls chains. His template is to take 20+ year old hardware and trumpet how superior it is to today's. Search his posts on CRT superiority for lulz.

He may be whatever you're painting him as, but the CRT is a poor example to show it since it is still a superior display tech in many ways.
 
I'm relatively certain rabidz7 is yanking y'alls chains. His template is to take 20+ year old hardware and trumpet how superior it is to today's. Search his posts on CRT superiority for lulz.

Nope. What do displays have to do with a thread on CPU architecture? PowerPC and CRTs are both great technologies that have been forgotten.
 
I'm relatively certain rabidz7 is yanking y'alls chains. His template is to take 20+ year old hardware and trumpet how superior it is to today's. Search his posts on CRT superiority for lulz.


Actually no, I remember my first system I used in college, it was a next system, dude that thing blew away any top end PC at the time 6 times over, of course it was 6 times the price too, 30k vs 5k. This was when the 486 66mhz was top of the line chip and that's the pc I got a month before going to college.
 
I actually doubt prices would raise too much for Intel CPUs if AMD did indeed go out of business. They are mostly competing with themselves, selling upgrades to users who already have Intel CPUs. The biggest competition right now is in the Mobile space (tablets and phones) where Intel is focusing most of their R&D, so it's not like the desktop is even a huge priority for them outside of server processors.
 
What? There is nothing wrong with x86 such that IBM or even ARM will come in and make better. Without AMD it will only be Intel in the enthusiast space. Period. There isn't enough cash available to anyone else to become a meaningful entrant.

give me some of what are you smoking, clearly is very good! :mad:
 
He may be whatever you're painting him as, but the CRT is a poor example to show it since it is still a superior display tech in many ways.

Tru dat. If they still sold them new, I'd probably be using a 16:10 model myself. The only things I don't miss are the imperfect screen geometry and weight/depth.

Nope. What do displays have to do with a thread on CPU architecture? PowerPC and CRTs are both great technologies that have been forgotten.

Nothing at all. I used to actually want a Macintosh back in the 90's because PowerPC wiped the floor with x86 in graphics intensive stuff. Now they're just outrageously priced PCs with a fruit logo on them (with the exception of the tower Mac Pros - where they were pretty good value for a workstation compared to the Dells and HPs of the day).

Actually no, I remember my first system I used in college, it was a next system, dude that thing blew away any top end PC at the time 6 times over, of course it was 6 times the price too, 30k vs 5k. This was when the 486 66mhz was top of the line chip and that's the pc I got a month before going to college.

Ahh... NeXT. Those 68040/68060 CPUs were no joke. Fun fact, John Carmack developed the Doom engine on a 68040 25MHz based NeXT cube.

give me some of what are you smoking, clearly is very good! :mad:

I can smell the skunkiness of it from here. Puff puff pass, mofo :D
 
Maybe one day we'll have a competitive ARM desktop CPU that runs on Windows. Once we do, it will be the end of Intel's monopoly. I see more potential in that happening than AMD ever making a comeback.

Problem: legacy compatibility. Virtualisation/emulation would hurt the performance a lot.
 
No matter how you slice it, even if you don't believe in the "rats leaving a sinking ship" theory, this is one more thing that AMD doesn't need.
 
Seriously man, how many companies have the billions of cash lined up to design state of the art ISA's that are not x86 but would be performance competitive with an i5 or an i7? Add to that the additional billions required to incent Microsoft and every other software developer to start making additional versions of software for this ISA. The list would be pretty small. Even if you recycled an existing one, you still need software for it. And you are not going to out margin Intel at any price if you outsource your production. And if you don't then caugh up another $4B for your own state of the art fab.

Bottom line, if AMD goes then Intel is all there is for a long, long time.
 
Software isn't an issue if you're on a non-Windows platform.

Sometimes we don't see direct challenges per se, but a movement to a different technology. E.g., consider just how much of a person's daily computing has moved to iOS and Android and thus ARM rather than Wintel.

What's ironic is that Intel is really pushing this ultrabook nonsense pretty hard, but that's an area where they could actually have competition, e.g. Chromebooks with ARM processors -- as opposed to areas where they have no competition (desktops).
 
VIA still has a license. Odds are if AMD goes belly up, VIA could start to give Intel a run for their money sooner than later.

As far as embedded x86, DM&P and ZF Micro could also take up the slack.

And then there is IBM, who has been working very closely with nVidia on NVLink with their Power 9 system.

Not to mention, maybe it's about time to end x86 and transition away from it.
 
Back
Top