Hacker News Reader

Comment by samfisher83 | Hacker News Reader

Opteron was crushing it back in the day. I think Tech is just cyclical. It seemed like AMD was in a lull for a while which basically made intel not really compete. Hopefully this forces Intel to step its game up. Competition is always good for the consumer.

View on HN

Back during the Opteron era, their chips had good compute, but higher heat output and power consumption. I use to run dual AthlonXPs on their older dual-MP boards back in University. It's crazy we now have 8~16 cores on a single board; on consumer (or I guess prosumer/developer) targeted hardware.

In 2016, AMD was near bankruptcy, after many years of mediocre tech and bad decisions. At that time it looked like they were never going to compete with Intel again in any meaningful way, didn't have the resources to pull ahead, had a negative balance sheet, and the best hope was that they could survive on some cheaper x86 niches, on the grace of people wanting some x86 competition to exist.

Would it have been a lull if all of intels flaws and security issues were known at the time? Those go back many years but weren't discovered til fairly recently and AMD didn't have most of them.

Not all of them were known back then obviously, but there were definitely known timing-info-leak flaws with Pentium 4 hyper threading[1], its remarkable on that basis that the modern class of execution timing exploits has taken so long to arise, considering their approach and concept really isn't fundamentally any different.

[1] http://www.daemonology.net/papers/htt.pdf by none other than our resident cpercival!

Was AMD in a lull or just making longterm bets on technology?

I've only followed casually in the last decade+ but I was under the impression that the ATI merger was meant to support a long term bet on better chips by AMD and we're starting to see the fruit of that labor.

They hired Jim Keller to design the Zen architecture around 2012, and he left in 2015. Many people point to him for their current success. He joined Intel in 2018.

Afaik Keller mostly worked on K12, the so for unreleased ARM cpu. Mike Clark was the chief architect for Zen.

Is AMD getting ahead of Intel with a superior architecture, or due to the fact that its running on 7nm instead of Intel's 14nm?

Part chip design, part 7nm being unexpectedly good.

But the chiplet architecture, supported by fast Infinity Fabric interconnect is what puts AMD ahead. Intel will have to go same route to stay competitive. And now they are two steps behind.

Chiplets were for sure an amazing bet AMD made that’s paid off big time.

The smaller dies and modular setup even on the same die gives them so much flexibility in their binning and package integration and must keep defect rates much lower than Intel’s monolithic high core count chips.

And the 14nm I/O die means they can still keep buying chips from GloFo, whom they still have a contract with as part of the spinoff from a decade ago.

> But the chiplet architecture, supported by fast Infinity Fabric interconnect is what puts AMD ahead.

Exactly this. Funny Intel is no longer mocking this as dies "glued together".

Intel has some very impressive chiplet technologies in house, aka EMIB. It's surprising they haven't used it for anything but that weird 8809G yet, but I'm sure it'll be used in more products next year.

As a layperson, I never understood that merger. How does buying a graphics card company make for better chips in general?

Some advantages:

* Integrated graphics in their CPUs is important for lower-end systems and APU/SoCs. Intel had an integrated GPU at the time, AMD didn't.

* It allowed them to provide the CPU+GPU for all Xbox and Playstation consoles since the acquisition

* It might have allowed them better deals with chip factories because their volume increased

It was a bad bet on netbooks that didn't fully anticipate the dominance of smartphones.

Their low-end Ryzen APUs can game better than low-end Intel chips with their IGP.

At the time it seemed that GPUs were becoming developer friendly and productivity might approach CPU languages, but then it all plateaued and GPU languages & OS integration stayed clunky. And fragmented.

Also, most computers come with integrated graphics today. It's the no brainer choice for most uses.

Integrating graphics card into their chips is what Intel seems to want to do as well.

So it would appear AMD was ahead of the game.

I guess, at the time, the thinking was that GPUs were a core threat to the CPU business itself, since a lot of the buzz was around using the GPU to do CPU activities.

Yes, there was this thing called HSA(Heterogeneous System Architecture). Does anyone remember that now?

The PS4 is an example of that. It's an APU with GDDR5 acting as both system and video memory.

To be honest for a while I thought that was where everything but servers would be going. You can tune your OS and applications around the slower GDDR5 and it seemed a no brainer: instead of 8GB of system memory topping out and then 4GB of VRAM doing jack shit you could have 12GB that could be dynamically assigned to whatever you wanted!

Alas, that dream was never meant to be.

Edit: I posted about this a while a back and apparently Arch Linux has a hacky way to use your VRAM as system RAM. Good stuff.

Zen in many ways is still moving towards that model, The concept of an IO die that all cores hook into, along with all of PCI-E lanes, no need for complicated cross core linking. It would be relatively "easy" to introduce specialized accelerator cores or beefy GPU chips into this mix on a separate chip, and not massively complicate the CPU design, or massively bloat its cost as a specialist die.

Yes, parts of it now present in IOMMU(v2)

Intel has such a poor track record in recent years of expanding into new markets and hasn’t had to compete in their core market for so long that it’s questionable if they can.

we just got done saying that about AMD, so..

Reply on news.ycombinator.com