By: David Kanter (dkanter.delete@this.realworldtech.com), March 6, 2015 11:37 pm
Room: Moderated Discussions
Brett (ggtgp.delete@this.yahoo.com) on March 6, 2015 12:57 pm wrote:
> David Kanter (dkanter.delete@this.realworldtech.com) on March 6, 2015 12:42 am wrote:
> > > A57 is definitely more than 20% faster at the same clock, as you can easily conclude
> > > from various benchmarks (eg. Geekbench shows 40% single threaded gain overall
> > > in AArch32 mode between Galaxy S5 and Galaxy S6 at the same clock).
> >
> > That's a bogus comparison since Geekbench contains lots of AES
> > stuff, and ARM64 contains special instructions for AES.
>
> Are you accusing ARM of pulling an Intel, making sure that benchmarks are dominated
> by feature X right before Intel adds feature X to the instruction set. ;)
No, I'm saying that anyone making claims about A57 vs. A15 is being disingenious if they do not acknowledge the impact of highly workload specific instructions.
It's one thing for Oracle to include database accelerators, when they own the database and it's a fairly important workload.
It's a totally different thing for Intel, AMD, ARM, or whoever, to claim that AES acceleration via dedicated hardware will impact general workloads in a meaningful way.
> This tactic is old hat, blatantly obvious for over a decade. Happy
> to see that ARM is playing by the rules of the CPU market.
> > What's the speedup on GCC?
>
> GCC is the worst pointer chasing spaghetti on the planet. You have to brute force tweak hundreds
> of CPU details so as not to get burned by one of a hundred glass jaws that will cripple your
> performance. This takes billions of dollars that AMD does not have, much less tiny ARM.
That's the real world. Most code is terrible and optimizing it is hard. That's how CPU designers create value.
Anyone can optimize a loop that regularly accesses arrays. Dealing with branches, irregulat data structures, etc. is the interesting stuff.
David
> David Kanter (dkanter.delete@this.realworldtech.com) on March 6, 2015 12:42 am wrote:
> > > A57 is definitely more than 20% faster at the same clock, as you can easily conclude
> > > from various benchmarks (eg. Geekbench shows 40% single threaded gain overall
> > > in AArch32 mode between Galaxy S5 and Galaxy S6 at the same clock).
> >
> > That's a bogus comparison since Geekbench contains lots of AES
> > stuff, and ARM64 contains special instructions for AES.
>
> Are you accusing ARM of pulling an Intel, making sure that benchmarks are dominated
> by feature X right before Intel adds feature X to the instruction set. ;)
No, I'm saying that anyone making claims about A57 vs. A15 is being disingenious if they do not acknowledge the impact of highly workload specific instructions.
It's one thing for Oracle to include database accelerators, when they own the database and it's a fairly important workload.
It's a totally different thing for Intel, AMD, ARM, or whoever, to claim that AES acceleration via dedicated hardware will impact general workloads in a meaningful way.
> This tactic is old hat, blatantly obvious for over a decade. Happy
> to see that ARM is playing by the rules of the CPU market.
> > What's the speedup on GCC?
>
> GCC is the worst pointer chasing spaghetti on the planet. You have to brute force tweak hundreds
> of CPU details so as not to get burned by one of a hundred glass jaws that will cripple your
> performance. This takes billions of dollars that AMD does not have, much less tiny ARM.
That's the real world. Most code is terrible and optimizing it is hard. That's how CPU designers create value.
Anyone can optimize a loop that regularly accesses arrays. Dealing with branches, irregulat data structures, etc. is the interesting stuff.
David
