[mhh__]

Do you realize that Apple silicon is the result of decades of chip design experience? They didn't just decide to do it Q3 next year, one day.

Intel rested on their laurels because AMD almost wiped themselves out, which led to them getting complacent with the process. The actual designs were quite good, it took AMD a few years of better process and newer designs to properly dethrone intel (which intel have now reclaimed in the mid range, this will now yo-yo as it should always do)

[gwd]

> Intel rested on their laurels because AMD almost wiped themselves out

You mean, Intel almost illegally wiped AMD out using its monopoly position in the x86 market:

https://www.amd.com/en/corporate/antitrust-ruling

[mhh__]

Bulldozer was terrible

[dmm]

The meltdown/spectre mitigations have wiped out most of the gap between bulldozer chips and the contemporary Intel ones.

Did Bulldozer really lose or was the competition cheating?

[rayiner]

> Did Bulldozer really lose or was the competition cheating?

Bulldozer and it’s competition were both designed at a time where leaking information to another process on the same physical core was outside the threat profile CPU memory protection was designed to mitigate against. Intel was optimizing aggressively within the envelope of expectations at the time, which were upended by the rise of cloud computing.

[mhh__]

Big citation needed there.

Spectre mitigations shouldn't effect quite a lot of programs because there's nothing to mitigate within the same address space.

[chomp]

https://www.phoronix.com/scan.php?page=article&item=spectre-...

Worst case I see is approximately 33% performance penalty in aggregate for Xeon parts, with much worse performance in specific scenarios. Comparing the original Bulldozer benchmarks, this does close the gap like GP suggested.

Also, refresh your memory on how the mitigations work because there’s definitely an impact for most programs.

[mhh__]

Ok yes I was too black and white, but still a lot of programs should be OK. You need to harden web browsers differently to (say) a calculator.

Also bulldozer is almost definitely vulnerable to spectre.

[gwd]

Hard to innovate properly when your income stream is artificially restricted.

[netr0ute]

Then how did Ryzen work?

[neogodless]

Semi-custom (aka console) chips kept them afloat, and they executed well on Ryzen while Intel made modest year-over-year improvements. Doesn't mean they weren't working on a relatively tight R&D budget, though.

[pengaru]

https://en.wikipedia.org/wiki/AMD%E2%80%93Chinese_joint_vent...

https://www.anandtech.com/show/15493/hygon-dhyana-reviewed-c...

[mhh__]

Ryzen worked in part because it didn't try to be super-clever. Bulldozer was very opinionated about how computing would look (cheap cpu + big coprocessor) whereas Zen is much more practical

[thow-58d4e8b]

Large part, likely the majority, of Intel's CPU dominance had been a result of their "unfair" advantage in fabs, not in hardware design. Often quoted rule of thumb - 90%/10% - 90% of the progress in semiconductors is attributable to process shrinkage, 10% to improved the hardware design. TSMC contributed the lion's share in making AMD competitive again. Design-wise, chips from Intel, AMD and Apple are all incredibly impressive. Afaik, Apple pays the top dollar to get access to the highest-end TSMC process - any performance comparisons should keep that in mind

[greggsy]

I’d be very interested to know how much licensing costs might be, given the co-dependence there is between Intel and AMD in that space

[Macha]

My understanding is that everything was pretty much wrapped up with the quid pro quo settlement where Intel allows AMD an x86 license and AMD allows Intel a license the x64 extensions, and it also applies to extensions since then like AVX.

Here's the terms: https://www.sec.gov/Archives/edgar/data/2488/000119312509236...

I don't think any money actively changes hands on an ongoing basis for this, but when this was signed Intel did pay $1.25 billion and AMD agreed to drop anti-trust complaints against Intel.