juanrga (nospam.delete@this.juanrga.com) on August 10, 2014 5:34 am wrote:
> Evidently the x86 tax is more noticeable on small phone-like cores but the tax doesn't magically
> vanish for big cores (only reduces the amount by a factor of about 2x or 3x). This is the reason
> why a 90W ARM SoC is able to offer 80--90% of the performance of a Haswell 140W Xeon. It is not because
> those companies have alien engineers working in the microarchitecture. It is not because they rely
> on an advanced SOI foundry process beyond Intel bulk strengths. It is the ISA advantage.
>

And where is this magical 90W ARM server SOC? Oh right, it still doesn't exist. Claims of future performance should be taken with lots of salt.


> Intel didn't kill the old RISC guys because the "ISA doesn't matter". Intel killed them by a mere question
> of volume and basic market dynamics. Intel attacked all them with cheaper high-volume products and people
> started to buy the chip that provided a 80% of the performance at 50% cost. The small volume guys found they
> couldn't sustain RD costs at same rate than Intel could and were finally killed. More about this below.
>
No, Intel killed them in performance. Intel was able to offer the same or higher performance at a significantly lower price with a much much larger software base.

None of these things apply to ARM in the server space. The software base is non-existent. The performance is currently non-existent. And the price is currently non-existent, and really won't be less unless the future ARM server vendors want to basically kill themselves.


> The first figure in the papers plots the evolution of performance of the first vector computers
> against old RISC guys (MIPS, alpha, HP, SPARC,...) and against x86 (AMD/Intel). The Figure
> 2a compares vector vs commodity and 2b compares commodity vs ARM. The new processors always
> start from what you call a "crap baseline", but evolve a much faster rate (2--4x faster)
> and in few years caught the old processors in performance and replace them.
>
ARM has been around for almost 30 years. It retreated to the embedded/mobile space because it couldn't compete. Its not starting. Its a RISC processor. Like Sparc, Like Power, Like MIPS, Like Alpha. There is nothing new about ARM. Every 5 or 10 years there is a whole new wave of "this new thing that isn't x86 is going to overtake x86" and 5 years later its basically a footnote in history. The reality is that ISA at the end of the day is pretty marginal. Always has been, always will be. And certainly in the server space, ISA doesn't matter as far as performance. The main performance bottlenecks aren't ISA related and neither is a vast amount of the power.

>
> In fact, AMD also knows that ARM will win over x86 and this is why it is now an ARM license.
>
No, AMD needed a new market because they were getting slaughtered in their current one. They took a gamble with a new uArch and it came up pretty much goose egged. AMD's embracing of ARM says more about AMD than it does about x86.


> ARM is a winner because has (i) ISA advantage (efficiency), (ii) volume/cost advantage (commodity
> hardware), and (iii) ecosystem advantage (no monopoly, customization, and so).
>

ARM has zero ISA advantage in servers.
ARM has significant dis-advantages in volume/cost(there is none).
ARM has significant dis-advantages in ecosystem (there is none).

In fact, over the past couple of years, ARM has actually lost market share in the amount of server space hardware they did have: low end NAS. It has been taken away precisely by the products that the ARM fanbase likes to make fun of: Atom.