Former AMD chip architect says it was wrong to can Arm project

Amid the renewed interest in Arm-based servers, it is easy to forget that one company with experience in building server platforms actually brought to market its own Arm-based processor before apparently losing interest: AMD. Now it has emerged that Jim Keller, a key architect who worked on Arm development at AMD, reckons the …

[Gene Cash]

Low bus factor

So it looks like this had a low "bus factor" - I have a feeling Keller was the major driver and major repository of ARM knowledge in the company, so when he left, they didn't have much choice other than to cancel it.

I wonder if his leaving is a underlying cause of why AMD hasn't been such a CPU powerhouse recently... just a SWAG.

[cornetman]

Re: Low bus factor

> I wonder if his leaving is a underlying cause of why AMD hasn't been such a CPU powerhouse recently... just a SWAG.

Not sure where you've been for the last 3-4 years. AMD are destroying Intel in various spaces.

My suspicion is that the secret to Lisa Su's success with AMD have been due to the following factors:

1. Laser focus on important tech areas that advance their portfolio in meaningful ways. The dropping of ARM might be related to avoid stretching their R&D resources too thin. AMD are a big company but they are still dwarfed by the likes of Intel. They surely cannot be as diverse as Intel. I also think this is the reason in part why their Radeon group has been so slow initially. I know that lots of people were really disappointed with early offerings on the graphical side.

2. Slow yet merciless execution with a clear 5-10 year plan.

3. Key strategic partnerships with key players. Keeping their foot in the door with Microsoft and Sony must have been a great money spinner for AMD, getting them through the lean periods during key development phases.

AMD have a lot of loyal fans due to their underdog status. I try to choose AMD over Intel because I hate the shady practices of Intel in the past and I think that honest, steady, hard-working face resonates with a lot of consumers. It certainly does with me.

[martinusher]

Re: Low bus factor

ARM also has a big "Me Too" factor -- everyone and their dog is developing it.

I figure if they're going ARM-like it will probably be with RISC-V. Even that's got a bit of me too in it.

[chasil]

AMD SPARC

The assertion of this article is that a common low-level RISC could be shared between an ARM and AMD64 implementation.

AMD has actually done this already, between the 29k and the K5.

Details on AMD SPARC:

"Based on the seminal Berkeley RISC, the 29k added a number of significant improvements. They were, for a time, the most popular RISC chips on the market.. Berkeley RISC-derived designs is the concept of register windows... In the original Berkeley design, SPARC, and i960, the windows were fixed in size.... It was here that the 29000 differed from these earlier designs, using a variable window size."

https://en.wikipedia.org/wiki/AMD_Am29000

K5 reused previous RISC elements:

"The K5 was based upon an internal highly parallel Am29000 RISC processor architecture with an x86 decoding front-end."

https://en.wikipedia.org/wiki/AMD_K5

So... here we go again?

[samoanbiscuit]

Re: Low bus factor

I think if AMD’s x86 offering weren’t well received, AMD would have continued down the ARM path. But with Zen selling like hotcakes, especially Zen 3 and onwards, AMD must have decided to stay in their x86 lane, where competiton is basically Intel and VIA, rather than enter the ARM fray, where competition is everyone in the tech space that can afford a CPU design team.

I think AMD will pivot to ARM or some other architecture if ARM or RISC-V manage to do to x86 what x86 did to MIPS, POWER, Alpha, SPARC, etc. But right now? They’re making maximum revenue off of x86 while it still retains advantages in the datacentre.

[Anonymous Coward]

Re: Low bus factor

what x86 did to MIPS, POWER, Alpha, SPARC, etc

In the desktop space maybe, but PowerPC is still one of the biggest players in the embedded space. If anything it's ARM that is pushing it out of that area now.

[andrewmm]

AMD / XILINX

And now , AMD have bought Xilinx

that a have a good few chips with combinations of AMD CPUs, Memory and programable logic in them

[elsergiovolador]

Re: AMD / XILINX

Xilinx makes quite a few CPLD and FPGA chips. Hope AMD won't force them to shelve them.

[eldakka]

Re: AMD / XILINX

> Xilinx makes quite a few CPLD and FPGA chips. Hope AMD won't force them to shelve them.

AMD bought Xilinx for their FPGA and other similar assets. While it's always possible they'll sqaunder it - many companies have bought other companies and squandered the takeover - since that was the purpose of the takeover, I see it highly unlikely they'll force them to shelve those products. The form those products take may change (embedded into server systems etc.), but the technology should remain.

[talk_is_cheap]

But why bother?

If AMD can advance their x86 based core design at the same speed as the ARM core design, there is little value in splitting their R&D budget and paying third-party additional IP fees. We already know that AMD is planning x86 cores with different levels of functionality and the only thing that seems unclear is how (or if) AMD plans to mix core types on a single CPU.

[William Higinbotham]

And AMD Moves On....

https://www.tomshardware.com/news/from-opteron-to-milan-crusher-supercomputer-comes-online-with-amd-cpus-and-gpus

[david 12]

Which reminds me...

All (??) of MS compilers (pre .NET) were based on the same engine: c, pascal, basic, FORTRAN, COBOL etc. Much to my disappointment they dropped most of the language front ends to concentrate on where the money was, and it seems to have worked for them.

AMD's concentration of the '86 family seems to have worked for them at present. What Might Have Been if they had spent resources on an ARM front end is one of those open questions which will never be answered.

[Cooe]

This is such a crap take... -_- ... AMD was on freaking life support in 2016... They didn't have the money to launch & support two ENTIRELY DIFFERENT server CPU platforms! They BARELY had the money to get the familiar x86 based Zen 1 out the door, and even then it was initially held together with nothing but duct-tape & prayers!

I get why Jim Keller's upset. He's a chip design LEGEND who had a fully custom CPU core that he spent a ton of time & work designing canceled right after he left. Who wouldn't be upset at least somewhat in that situation?

But Lisa Wu made by FAR the right choice in making that call! Zen needed all hands & money on deck as a last ditch attempt to literally save the company. Diverting a bunch of their EXTREMELY limited resources to a HUGE & risky bet on the then totally unproven (in HPC compute) ARM platform could have EASILY been a death sentence for AMD as a whole...

[bazza]

Agreed.

It's possible that AMD had one shot at securing decent prices with TSMC. It's all very well having two fabulous chip designs, but when you're fabless and broke and you need a good deal on the best silicon process available, we'll it's easy to see why X64 won. TSMC likely would have said "no way" to making a product far from certain to sell. They must have taken some convincing that AMD was worth supplying at all.

Maybe now is the time though. AMD has since given TSMC a lot of work, so I suspect that they've gained some technical credit and belief there. Plus, the Arm landscape has moved onwards too.

[Ace2]

The key is the whole “right after he left.” If you want to have a say in what a company does, maybe don’t quit your job!

[John Brown (no body)]

Agreed. After all, by all accounts he's GOOD at what he does, but seems to have move around a lot. Is he driven to find new challenges? Seduced my better offers? A real shit to work with? Anyone have any insights?

[John Savard]

Odd

Given what an incredible success Ryzen has been for AMD - it basically saved the company - and that before Ryzen, AMD's resources were limited, for AMD to concentrate everything on Ryzen, and not spread itself too thin... would not seem to have been a mistake at all. It may have been a disappointment, and now may be the time for AMD to begin re-entering the field of ARM server chips, but for AMD to concentrate everything on saving itself through its core business was an eminently sensible decision, rather than a mistake.

[Anonymous Coward]

What a world it could have been

People have been throwing around whatifs here, so I'll add one. In a bizzrarro world when a former Intel guy DOESN'T call of AMD antitrust lawsuit, where AMD has enough money to keep its fingers in the ARM game, and still acquired Seamicro, and didn't kill it.

That means just as the rise of the age of Heartbleed and 31 flavors of Spectre and other side channel attacks, matching the explosion of cloud hosting, and virtualization based hosting no longer being execution safe. Seamicro made ultra-high density microservers, which in the hands of major players in cloud hosting could have offered site operators a tiny dedicated bare iron server with no other on device tenants. Arm tenants would have allowed them to sip even less power and reach higher densities.

Instead, all was canceled, and we collectively were all buying stuff off ecommerce sites with no guarantee that the SSL cert hadn't been filched by another tenant. Is this not the best of all possible worlds?