← Back to context

Comment by rhelz

12 days ago

I have endless admiration for Lisa Su, but lets be honest, the reason AMD and Nvidia are so big today is that Intel has had amazingly bad management since about 2003.

They massively botched the 32-bit to 64-bit transition....I did some work for the first Itanium, and everybody knew even before it came out that it would never show a profit. And they were contractually obligated to HP to not make 64-bit versions of the x86....so we just had to sit there and watch while AMD beat us to 1 Gigahertz, and had the 64-bit x86 market to itself....

When they fired Pat Gelsinger, their doom was sealed. Thank God they hired him back, but now they are in the same position AMD and Nvidia used to be in: Intel just has to wait for Nvidia and AMD to have bad management for two straight decades....

You're talking about very old events. Intel made mistakes during 2000-2005 (which allowed the rise of Opteron) but they crushed it from 2006-2016. Then they had different problems from 2016-2024.

  • Intel went into Rest & Vest mode after Haswell (4000 series) in 2013 - they hardly improved until 2020 - 6 lost generations of relabeled CPUs like 13000 - 14000 series (lots of money paid for innovative new igpu marketing names like 520 and 620!). They found out that community college talent does not make good employees in the C-Suite ... The iris pro 5200 (2013) wasn't improved upon until 2020 !

    • > They found out that community college talent does not make good employees in the C-Suite

      This is new to me, what is this referring to?

    • This is probably wrong. The 10 nm problems were due to Intel being too ambitious, not "resting".

  • Thanks for making me feel old :-)

    There was a palpable change when Andy Grove retired. It was like the elves leaving Middle earth.

I don't mean to take away from Intel's underwhelming management.

But regardless, Keller's Athlon 64 or Zen are great competitors.

Likewise, CUDA is Nvidia's massive achievement. The growth strategy of that product (involving lots of free engineer hours given to clients on-site) deserves credit.

  • // I don't mean to take away from Intel's underwhelming management

    chuckle lets give full credit where credit is due :-)

    Athlon was an epochal chip. Here's the thing though---if you are a market leader, one who was as dominant as Intel was, it doesn't matter what the competition does, you have the power to keep dominating them by doing something even more epochal.

    That's why it can be so frustrating working for a #2 or #3 company....you are still expected to deliver epochal results like clockwork. But even if you do, your success is completely out of your hands. Bringing out epochal products doesn't get you ahead, it just lets you stay in the game. Kind of like the Red Queen in Alice in Wonderland. You have to run as fast as you can just to stay still.

    All you can do is try to stay in the game long enough until the #1 company makes a mistake. If #1 is dominate enough, they can make all kinds of mistakes and still stay on top, just by sheer market inertia. Intel was so dominate that it took DECADES of back-to-back mistakes to lose its dominate position.

    Intel flubbed the 32-64 bit transition. On the low end, it flubbed the desktop to mobile transition. On the high end, it flubbed the CPU-GPU transition.

    Intel could have kept its dominate position if it had only flubbed one of them. But from 2002 to 2022, Intel flubbed every single transition in the market.

    Its a measure of just how awesome Intel used to be that it took 20 years....but there's only so many of those that you can do back-to-back and still stay #1.

I began supporting AMD as my choice for a gaming CPU when they began trouncing Intel in terms of performance vs. total draw power with an attractive price point around 2016 or so.

Then, a wave of speculative execution vulnerabilities were discovered / disclosed, resulting in an even larger differential for performance and power use after the SPECTRE patches were applied.

Considering this, I'm not sure that it's fair to cast the successes of AMD as mere failures from Intel. Dr. Su is simply a brilliant engineer and business leader.

Well geez, Intel might have amazingly bad management but where on the scale do we put AMD not giving any fucks about challenging the CUDA monopoly for going on 10 years now?

Instead they put mediocre people on things like OpenCL which even university students forced to use the mess could tell was going nowhere.

  • AMD basically did not have any money from 2010 until 2020. They were teetering near death. No money for r&d no money for new features in their gpus no money no money no money. There excavator architecture was extremely foolish trying to support multiple instruction dispatches with only 1 ALU and FPU per core! This was corrected with Ryzen and the last 4 years they have been able to pay off their massive debts, especially with the 5000 series!

    Going forward I expect we will see much more innovation from them because now they have some spare cash to spend on real innovations hopefully in hardware AND software! Note that Intel is much worse than AMD in software!

I have endless admiration for Lisa Su, but lets be honest, the reason Nvidia is so big today is that AMD has had amazingly bad management.