By: anon (spam.delete.delete@this.this.spam.com), October 13, 2018 4:17 am
Room: Moderated Discussions
Simon Farnsworth (simon.delete@this.farnz.org.uk) on October 13, 2018 4:52 am wrote:
> Maynard Handley (name99.delete@this.name99.org) on October 12, 2018 3:37 pm wrote:
> > Simon Farnsworth (simon.delete@this.farnz.org.uk) on October 12, 2018 11:51 am wrote:
> > > Maynard Handley (name99.delete@this.name99.org) on October 12, 2018 10:33 am wrote:
> >
> >
> > > > - Finally (and CF are very careful how they say this, but I think the message is obvious,
> > > > especially when you get to the comments) QC was, uh, not an optimal partner...
> > > > Essentially QC and Falkor showed the DESIRABILITY of an ARM64 throughput server, but not one from ARM.
> > >
> > > Well, it showed the desirability of a high thread count throughput server with just
> > > enough single threaded compute. ARM64 is a mirage here - as shown by them jumping
> > > to a high thread count Intel chip once Intel was willing to offer them one.
> > >
> > > Equally, they'll happily move to ARM64, MIPS, PowerPC, or any other ISA to get the chips they want
> > > - they've got no legacy hangups about staying with x86 if there's a good design out there.
> >
> > I'd agree with that phrasing.
> > One could assert that, once these baseline metrics are
> > met (performance parity in the ways CF demands) ARM64
> > PROBABLY has (or can more easily be made to have) other properties that CF might care about, things like
> > - better security (broadly defined, but could mean smaller attack surface, faster turnaround
> > to fix known HW bugs, HW root of trust that works + associated non-vulnerable BMC, ...) OR
> > - better memory capabilities in ways that CF wants but Intel won't offer for strategic reasons (like
> > more memory controllers, or some sort of wide memory support, or non-Optane NVMe support).
> > But CF hasn't asserted that these are important to them, so that would be very much a hypothetical.
> >
> >
> Of those, the only one that's a big issue is things that Intel won't offer for strategic reasons.
> Everything else is possible with Intel if you can convince Intel to care about your business.
>
> They're all possible with any ISA, too - including x86 and RISC-V.
>
> > > It's also worth maintaining and improving it as leverage against Intel - unless ARM vendors give
> > > up on the server market entirely, there's a very good chance that there will be another one along
> > > in the future that you can use to put Intel in their box and get a custom deal. For a big company,
> > > maintaining the warning to Intel that you can replace them if they get difficult has value.
> >
> > I'm sure some element of "warning Intel" is there, but I suspect it's much lower priority than many
> > people assume, substantially secondary to the technical issues. I obviously can't prove that, but I'd
> > note that CF has been substantially less interested in posting "Look at how fast AMD servers run compared
> > to Intel" than "look at how fast ARM64 servers run compared to Intel". If this were primarily about
> > veiled threats, I'd expect just as many of these AMD posts (which would, again, not have to be dishonest
> > or misleading in any way --- what AMD is selling IS a very good match for CF's needs).
>
> Intel already understands the AMD threat - and that's not an existential threat to Intel's business
> model. An ISA switch is a threat to Intel's business model; as they've already demonstrated
> in mobile, they struggle to win business if the legacy ISA in the space is not x86, because
> it's hard to produce a profitable chip that's enough better than the legacy ISA to motivate
> a switch. AMD, on the other hand, produces drop-in replacements for Intel, and that implies
> that Intel's chips are drop-in replacements for AMD if Intel produces a better option.
>
> Winning against ARM64, however, would require engineering effort - either Intel wins
> even though it's producing ARM64 chips, or it has to be good enough to convince a company
> that's gone ARM64 that it's worth the engineering effort of porting back to x86.
Well it's not like all hardware would be replaced within a day. They just need to keep a foot in the door. As long as they don't allow themselves to be completely replaced it's never a matter of porting back. So every year they manage to sell x86 CPUs to a customer is an extra year they have to come up with a better architecture should that customer start buying ARM64 exclusively next year.
If a few discounts mean that your oldest CPUs are still x86 by the time Intel is competitive again they never have to work against legacy ISA advantage.
So if they know that the next 2 years they won't have much to offer they'd probably be more willing to lower the price, but the CF situation isn't anywhere near that point yet. Lower clocked custom SKUs are easy, they'd do something similar to prevent CF from switching to AMD.
> Maynard Handley (name99.delete@this.name99.org) on October 12, 2018 3:37 pm wrote:
> > Simon Farnsworth (simon.delete@this.farnz.org.uk) on October 12, 2018 11:51 am wrote:
> > > Maynard Handley (name99.delete@this.name99.org) on October 12, 2018 10:33 am wrote:
> >
> >
> > > > - Finally (and CF are very careful how they say this, but I think the message is obvious,
> > > > especially when you get to the comments) QC was, uh, not an optimal partner...
> > > > Essentially QC and Falkor showed the DESIRABILITY of an ARM64 throughput server, but not one from ARM.
> > >
> > > Well, it showed the desirability of a high thread count throughput server with just
> > > enough single threaded compute. ARM64 is a mirage here - as shown by them jumping
> > > to a high thread count Intel chip once Intel was willing to offer them one.
> > >
> > > Equally, they'll happily move to ARM64, MIPS, PowerPC, or any other ISA to get the chips they want
> > > - they've got no legacy hangups about staying with x86 if there's a good design out there.
> >
> > I'd agree with that phrasing.
> > One could assert that, once these baseline metrics are
> > met (performance parity in the ways CF demands) ARM64
> > PROBABLY has (or can more easily be made to have) other properties that CF might care about, things like
> > - better security (broadly defined, but could mean smaller attack surface, faster turnaround
> > to fix known HW bugs, HW root of trust that works + associated non-vulnerable BMC, ...) OR
> > - better memory capabilities in ways that CF wants but Intel won't offer for strategic reasons (like
> > more memory controllers, or some sort of wide memory support, or non-Optane NVMe support).
> > But CF hasn't asserted that these are important to them, so that would be very much a hypothetical.
> >
> >
> Of those, the only one that's a big issue is things that Intel won't offer for strategic reasons.
> Everything else is possible with Intel if you can convince Intel to care about your business.
>
> They're all possible with any ISA, too - including x86 and RISC-V.
>
> > > It's also worth maintaining and improving it as leverage against Intel - unless ARM vendors give
> > > up on the server market entirely, there's a very good chance that there will be another one along
> > > in the future that you can use to put Intel in their box and get a custom deal. For a big company,
> > > maintaining the warning to Intel that you can replace them if they get difficult has value.
> >
> > I'm sure some element of "warning Intel" is there, but I suspect it's much lower priority than many
> > people assume, substantially secondary to the technical issues. I obviously can't prove that, but I'd
> > note that CF has been substantially less interested in posting "Look at how fast AMD servers run compared
> > to Intel" than "look at how fast ARM64 servers run compared to Intel". If this were primarily about
> > veiled threats, I'd expect just as many of these AMD posts (which would, again, not have to be dishonest
> > or misleading in any way --- what AMD is selling IS a very good match for CF's needs).
>
> Intel already understands the AMD threat - and that's not an existential threat to Intel's business
> model. An ISA switch is a threat to Intel's business model; as they've already demonstrated
> in mobile, they struggle to win business if the legacy ISA in the space is not x86, because
> it's hard to produce a profitable chip that's enough better than the legacy ISA to motivate
> a switch. AMD, on the other hand, produces drop-in replacements for Intel, and that implies
> that Intel's chips are drop-in replacements for AMD if Intel produces a better option.
>
> Winning against ARM64, however, would require engineering effort - either Intel wins
> even though it's producing ARM64 chips, or it has to be good enough to convince a company
> that's gone ARM64 that it's worth the engineering effort of porting back to x86.
Well it's not like all hardware would be replaced within a day. They just need to keep a foot in the door. As long as they don't allow themselves to be completely replaced it's never a matter of porting back. So every year they manage to sell x86 CPUs to a customer is an extra year they have to come up with a better architecture should that customer start buying ARM64 exclusively next year.
If a few discounts mean that your oldest CPUs are still x86 by the time Intel is competitive again they never have to work against legacy ISA advantage.
So if they know that the next 2 years they won't have much to offer they'd probably be more willing to lower the price, but the CF situation isn't anywhere near that point yet. Lower clocked custom SKUs are easy, they'd do something similar to prevent CF from switching to AMD.