5 ms·

Not a coincidence. Articles with the words CPU or Microprocessor in the title are getting a lot of attention on HN (including a 2016 article on CPU bugs).

phoe-krk 9y ago
Honestly, I find buzz like this to be extremely bad taste. It is trivially simple to post links like that on HN and wait for the "ha, Intel, this is how you should have been doing things!!!1". If this bug is really present in all chips since 1995, then where have all these people been for the last 22 years? If they're so competent about CPU design, why haven't we seen literally tens of Intel competitor startups pop up and succeed?
- aidos 9y ago
  Posting the article itself isn't in bad taste, though a comment like your example one would be, and would hopefully be downvoted. It's only natural that you get some articles like these posted. For many of us, we've read a few details about the bugs raised this week and thought to ourselves — "man, CPUs are so complicated, how do they even managed to make them work at all?" An article like this helps to answer some of the questions for those of us interested in finding out more.
- CthulhuOvermind 9y ago
  I can comment on this. Hardware is a long and tedious process. CPU design is an extra level of difficult on top. In my field, it is routine to hire people in late 40s/50s to make CPUs. I'm in my late 20s, and viewed as a weird specimen. In my career I've made a armv8 CPU and now work RV64. It takes a couple of years to get a guy from uni into working shape. And that's just to teach him how to do 1-2 tasks in his field. If verification, this usually is coverage specification and test writing. Combine this with a cpu project duration average of 5ish years and it's quickly evident that you won't beat someone with a startup. In my office today, in a team of 10, no one, other than me is under 40.
  - xemdetia 9y ago
    Something that always strikes me is how insanely complex it is to even experiment as an outsider to the university path. From my understanding a relatively simple 10+ year ago tech ASIC is still going to run like $10k USD for a shared die prototype from limited providers, compared to software, component level systems design, or even mechanical devices where you can get a local machine shop involved or at least do some parts of it yourself. Yes, FPGA's exist but they are not even relevant when you are talking about the skillset/engineering discipline to make the FPGA itself and actually producing the physical thing.
- pjc50 9y ago
  CPUs are a very hard market to break into, for reasons mostly unrelated to technology. It's like asking why there haven't been so many operating system competitors to Windows.
- Cthulhu_ 9y ago
  > why haven't we seen literally tens of Intel competitor startups pop up and succeed? ARM-based startups have conquered 95% of the smartphone market and significant chunks of TV's, set-top boxes, cars, tablets, etc since 1995, totaling at about 15 billion ARM-based chips a year (compared to Intel's 500ish million / year). That's pretty successful, right?
  - InitialLastName 9y ago
    ARM-based startups like Qualcomm, Broadcom, and Apple? Who are these ARM-based startups you speak of?
    - monocasa 9y ago
      A lot of Maxim's offerings, at least, are startups that they acquired. That seems to be a large chunk of their business model, acquire rather than put in all the R&D themselves. The specific chips I know of have ARM cores in them.
      - InitialLastName 9y ago
        I wouldn't call Maxim a startup, nor would I guess that their market reach approaches anything like the 95% figure above.
        monocasa 9y ago
        Maxim acquires startups, not is a startup themselves. A good chunk of their products are one offs from firms that won the startup lottery.
- JdeBP 9y ago
  Intel itself tried to do a better CPU design. It was called the Itanium. * https://news.ycombinator.com/item?id=16070020 https://news.ycombinator.com/item?id=16070020
  - hajile 9y ago
    You'd have a hard time convincing me of that. They had ALREADY tried the Itanium architecture with the i860 and knew that it was horrifically slow for almost all applications. https://en.wikipedia.org/wiki/Intel_i860 https://en.wikipedia.org/wiki/Intel_i860 I'd also look at the market of the time. Itanium was in the tech press everywhere. HP-RISC and Alpha were abandoned. MIPS moved into the low-power market. Resources for SPARC and POWER seem to have been cut. Everyone making the top-level decisions was convinced that Itanium and a sufficiently smart compiler were the solution (hint: the halting problem prevents general static analysis as Turing discovered in the 1930s). Everyone that is, except AMD and Intel. AMD had no alternatives, so they pulled the x64 rabbit out of their hat. Meanwhile, Intel hadn't slowed down their x86 work. When Itanium fell through (like history indicated it would), Intel was out some R&D and some marketing, but had effectively killed off all their RISC competition (it took a decade or more to catch up). Intel then proceeded to Monopolize AMD almost out of existence (and only paid a couple billion in return for their hundreds of billions in profit). It all works out way too conveniently for Intel.
    - fellellor 9y ago
      Itanium supposedly had a "better" 64 bit architecture, but required software creators to incorporate major changes in their design. Also they would have broken backward compatibility with 32 bit software. Since x86 was already popular, the market went along with the extended x86-64 architecture.
JdeBP 9y ago
Although the supposedly 2016 article was in fact last updated today, and discusses the recently disclosed problems.

phoe-krk 9y ago

Honestly, I find buzz like this to be extremely bad taste. It is trivially simple to post links like that on HN and wait for the "ha, Intel, this is how you should have been doing things!!!1". If this bug is really present in all chips since 1995, then where have all these people been for the last 22 years? If they're so competent about CPU design, why haven't we seen literally tens of Intel competitor startups pop up and succeed?

aidos 9y ago
Posting the article itself isn't in bad taste, though a comment like your example one would be, and would hopefully be downvoted. It's only natural that you get some articles like these posted. For many of us, we've read a few details about the bugs raised this week and thought to ourselves — "man, CPUs are so complicated, how do they even managed to make them work at all?" An article like this helps to answer some of the questions for those of us interested in finding out more.
CthulhuOvermind 9y ago
I can comment on this. Hardware is a long and tedious process. CPU design is an extra level of difficult on top. In my field, it is routine to hire people in late 40s/50s to make CPUs. I'm in my late 20s, and viewed as a weird specimen. In my career I've made a armv8 CPU and now work RV64. It takes a couple of years to get a guy from uni into working shape. And that's just to teach him how to do 1-2 tasks in his field. If verification, this usually is coverage specification and test writing. Combine this with a cpu project duration average of 5ish years and it's quickly evident that you won't beat someone with a startup. In my office today, in a team of 10, no one, other than me is under 40.
- xemdetia 9y ago
  Something that always strikes me is how insanely complex it is to even experiment as an outsider to the university path. From my understanding a relatively simple 10+ year ago tech ASIC is still going to run like $10k USD for a shared die prototype from limited providers, compared to software, component level systems design, or even mechanical devices where you can get a local machine shop involved or at least do some parts of it yourself. Yes, FPGA's exist but they are not even relevant when you are talking about the skillset/engineering discipline to make the FPGA itself and actually producing the physical thing.
pjc50 9y ago
CPUs are a very hard market to break into, for reasons mostly unrelated to technology. It's like asking why there haven't been so many operating system competitors to Windows.
Cthulhu_ 9y ago
> why haven't we seen literally tens of Intel competitor startups pop up and succeed? ARM-based startups have conquered 95% of the smartphone market and significant chunks of TV's, set-top boxes, cars, tablets, etc since 1995, totaling at about 15 billion ARM-based chips a year (compared to Intel's 500ish million / year). That's pretty successful, right?
- InitialLastName 9y ago
  ARM-based startups like Qualcomm, Broadcom, and Apple? Who are these ARM-based startups you speak of?
  - monocasa 9y ago
    A lot of Maxim's offerings, at least, are startups that they acquired. That seems to be a large chunk of their business model, acquire rather than put in all the R&D themselves. The specific chips I know of have ARM cores in them.
    - InitialLastName 9y ago
      I wouldn't call Maxim a startup, nor would I guess that their market reach approaches anything like the 95% figure above.
      - monocasa 9y ago
        Maxim acquires startups, not is a startup themselves. A good chunk of their products are one offs from firms that won the startup lottery.
JdeBP 9y ago
Intel itself tried to do a better CPU design. It was called the Itanium. * https://news.ycombinator.com/item?id=16070020 https://news.ycombinator.com/item?id=16070020
- hajile 9y ago
  You'd have a hard time convincing me of that. They had ALREADY tried the Itanium architecture with the i860 and knew that it was horrifically slow for almost all applications. https://en.wikipedia.org/wiki/Intel_i860 https://en.wikipedia.org/wiki/Intel_i860 I'd also look at the market of the time. Itanium was in the tech press everywhere. HP-RISC and Alpha were abandoned. MIPS moved into the low-power market. Resources for SPARC and POWER seem to have been cut. Everyone making the top-level decisions was convinced that Itanium and a sufficiently smart compiler were the solution (hint: the halting problem prevents general static analysis as Turing discovered in the 1930s). Everyone that is, except AMD and Intel. AMD had no alternatives, so they pulled the x64 rabbit out of their hat. Meanwhile, Intel hadn't slowed down their x86 work. When Itanium fell through (like history indicated it would), Intel was out some R&D and some marketing, but had effectively killed off all their RISC competition (it took a decade or more to catch up). Intel then proceeded to Monopolize AMD almost out of existence (and only paid a couple billion in return for their hundreds of billions in profit). It all works out way too conveniently for Intel.
  - fellellor 9y ago
    Itanium supposedly had a "better" 64 bit architecture, but required software creators to incorporate major changes in their design. Also they would have broken backward compatibility with 32 bit software. Since x86 was already popular, the market went along with the extended x86-64 architecture.

JdeBP 9y ago

Although the supposedly 2016 article was in fact last updated today, and discusses the recently disclosed problems.