By: dmcq (dmcq.delete@this.fano.co.uk), August 10, 2014 4:06 am
Room: Moderated Discussions
Michael S (already5chosen.delete@this.yahoo.com) on August 10, 2014 2:21 am wrote:
> Brett (ggtgp.delete@this.yahoo.com) on August 9, 2014 10:37 pm wrote:
> > Linus Torvalds (torvalds.delete@this.linux-foundation.org) on August 9, 2014 1:36 pm wrote:
> > > Brett (ggtgp.delete@this.yahoo.com) on August 9, 2014 11:51 am wrote:
> > > >
> > > > For those that are designing 40 watt ARM64 chips there are no legacy 32 bit apps to support
> > > > in the laptop/desktop/server space.
> > >
> > > Christ, you people.
> > >
> > > You can't have it both ways. You try to argue that "ISA matters a lot for decoding", but
> > > then at the same time you try to argue that "ISA doesn't matter at all for users".
> >
> > Not an ARM fan boy the way you assume.
>
> That's true. You are fan boy by proxy (apple).
>
> > I have in the past here called ARM32 crap,
>
> I like Thumb2 much better than I like Aarch64. It's a pity that Aarch64 is not derived from Thumb2.
Thumb2 is fine for embedded work but it is not a good basis for a fast server. And I would have to say the MIPS or PowerPC would also have been a better basis for a high performance server than extending the ARM 32 bit ISA. It is full of crap which is very good for embedded stuff but not for a server. ARM were right to abandon it.
And on the ISA decode business. The decode is very important. But there are other areas a good ISA can help with. Cutting down needing memory accesses, having a good modern cache coherency model supported well by the ISA. Dealing with interrupts quickly, supporting virtualization efficiently, simplifying the control flow. Consider even something as simple as the divide instruction - should one cause an interrupt for a divide by zero? Well it seems obvious one should but consider that it is an expensive instruction anyway and instructions round it can proceed in parallel and a test can be generated by a compiler easily. Why lumber it with the possibility of changing the control flow? And if it isn't in the arithmetic unit then it would be because one split it into micro-operations and just sticking in the equivalent instructions avoids the need for such split.
.....
> >
> > > Do you seriously believe that the alleged performance and power advantage (and yes, it very much is alleged
> > > - I would even go as far as call it "drug-induced hallucinations" - so far nobody has come even close to
> > > Intel in the space you claim is so ripe for ARM64 in either performance or price) are so massive that users
> > > should/would ignore the fact that the break in ISA also causes a lot of real and inevitable problems?
> >
> > In a few short years no one will care about ARM32, everything will be
> > 64bit, even the phones will be dumping compatibility with 32bit.
>
> It's possible. And it's a pity.
> But, I'd guess that 'few'='your current estimate of few' + 4.
It'll certainly be around for a long time in embedded controllers. I would guess that Apple will soon start emulating 32 bit ARM and remove it from their processors, if they don't do that in the A8 it'll be because it is a tick in Intel terms rather than a tock.
.....
> >
> > > ARM64 isn't that magical. MIPS and alpha were there before it with pretty similar "simple
> > > decoding". You're repeating arguments that didn't make sense the last time around, and that
> > > have been soundly disproven in that thing we call "the real world" (tm) or "the market".
> >
> > Good points.
> > You realize of course that 20 years ago the RISC camp was split
> > into a half dozen factions, none big enough to survive.
> > Sparc and Power will be dead in a few years,
>
> Now you are out of your depth. Even dead Itanium is going to last for at least 5-6
> years. Such live bodies as SPARC and esp. Power will be around for much longer.
>
> > both will convert to ARM64, just like AMD is.
> >
> > Hows that for a prediction.
>
> No good.
64 bit ARM is already making a rapid advance against MIPS and PowerPC in plans for new embedded controllers.
> > There will be a full spectrum of ARM64 implementations from low to as high as Intel can go.
> >
> > > I think ARM has a chance, but the arguments for it in this thread have been absolutely moronic. The
> > > performance claims about how much better ARM64 is over ARM32 seem to be based on Geekbench, for chrissake.
> > > Some of the other arguments have been about how relatively quickly ARM has improved, which is largely
> > > based on the fact that Cortex-A9 was a complete and utter disaster particularly from an uncore standpoint,
> > > so when people bandy about "40% improvement" and compare that to how Haswell didn't make as much of
> > > a difference, they are clearly not understanding how the ARM baseline was crap.
> >
> > Agreed. Old Arm chips are pretty crap.
>
> ARM Cortex-A9 inner core is not bad. I'd dare to say that for phones it is somewhat
> better than A15. Unfortunately it is coupled with brain damaged "modular" L2 cache.
> And 32B cache lines are not helping it's case in the DDR3-dominated world.
The ARM Cortex-A17 is far better than the A9 even though it is derived from it, a 60% improvement from better memory handling and branch prediction and a couple of other tweaks. It basically is what ARM should have had instead of the A15 for the high end smartphone market and if they have any sense they're currently making a 64 bit version of it.
> >
> > > Even now, people seem to think that Apple A7 is somehow a high-performance chip. It's not really
> > > all that impressive, and again, the whole "look at how great it is" seems to be based almost entirely
> > > on pure crap (geekbench). And absolutely none of that is relevant to the server market.
> > >
> > > No, if ARM has a chance, it's not because it will outperform intel server chips
> > > (I can pretty much guarantee it won't), it is because of other market forces.
There do seem to be a number of other forces in ARM's favor like companies just not wanting to be in thrall to Intel. I don't know how the Russian business about using ARM will go and Intel has had some problems in China. It is a pity but Snowden's revelations have made people a bit paranoid though I think it more likely China would put a spy into a chip they produced for their own citizens than that Intel would do it to them for the NSA.
> > You will be wrong, and sooner than you think possible. All it takes is for POWER to go ARM64.
> >
> > > For example, there are a lot of customers who want to make sure that they have alternatives,
> > > and are worried about the fact that Intel is so crushingly dominant. Those customers don't
> > > necessarily care about ARM at all, just go back a few months to look at the POWER8 motherboard
> > > news etc, but that "we really worry about a monoculture" is very much a real issue.
> > >
> > > Similarly, there are a lot of chip companies that want to get a part of the market, and if you don't have
> > > the resources of Intel, it's really hard to compete in the x86 space. Because that complexity may not
> > > be a huge performance issue in the end, but it does mean that there is a fairly high bar of entry.
> >
> > Another prediction:
> > In ten years AMD will not be making any x86 chips. And no one will care. Plenty of ARM64 choices.
>
> It's more likely that in ten years AMD will not exist as independent
> company. If it still exists it's going to make x86.
Their x86 business aint going to save them but I guess they might produce custom SoCs with x86 processors on. If they are independently producing servers in 10 years time it will be ARM ones or some successor chip, who's for The Mill anyone? ;-)
....
> > > So I'm dead serious when I say that anybody who talks about the alleged huge advantages of simple decoding,
> > > but then blithely ignores the advantages of backwards compatibility, is a f*cking moron.
> >
> > Yes backwards compatibility MATTERS for x86. For ARM64 servers it is merely a short term nuisance.
> > Yes the near complete lack of apps will hurt at first, but the
> > bandwagon will gather steam and nothing will stop it.
> >
> > > And there's a lot of those f*cking morons in this thread. Some of them double down on their stupidity
> > > by claiming that it's a big advantage that you can jettison all that legacy baggage.
> >
> > It is a small advantage to jettison all that legacy baggage, every little bit helps.
I get the feeling Linus is like someone who's had to use Microsoft Word and is terrified of changing to anything else because of all the work they've done coping with it.
> > > Linus "rant over" Torvalds
> >
> > Brett, who is apparently a f*cking moron. ;) ;) ;)
> >
>
Dmcq, who is also apparently a f*cking moron. ;) ;) ;)
> Brett (ggtgp.delete@this.yahoo.com) on August 9, 2014 10:37 pm wrote:
> > Linus Torvalds (torvalds.delete@this.linux-foundation.org) on August 9, 2014 1:36 pm wrote:
> > > Brett (ggtgp.delete@this.yahoo.com) on August 9, 2014 11:51 am wrote:
> > > >
> > > > For those that are designing 40 watt ARM64 chips there are no legacy 32 bit apps to support
> > > > in the laptop/desktop/server space.
> > >
> > > Christ, you people.
> > >
> > > You can't have it both ways. You try to argue that "ISA matters a lot for decoding", but
> > > then at the same time you try to argue that "ISA doesn't matter at all for users".
> >
> > Not an ARM fan boy the way you assume.
>
> That's true. You are fan boy by proxy (apple).
>
> > I have in the past here called ARM32 crap,
>
> I like Thumb2 much better than I like Aarch64. It's a pity that Aarch64 is not derived from Thumb2.
Thumb2 is fine for embedded work but it is not a good basis for a fast server. And I would have to say the MIPS or PowerPC would also have been a better basis for a high performance server than extending the ARM 32 bit ISA. It is full of crap which is very good for embedded stuff but not for a server. ARM were right to abandon it.
And on the ISA decode business. The decode is very important. But there are other areas a good ISA can help with. Cutting down needing memory accesses, having a good modern cache coherency model supported well by the ISA. Dealing with interrupts quickly, supporting virtualization efficiently, simplifying the control flow. Consider even something as simple as the divide instruction - should one cause an interrupt for a divide by zero? Well it seems obvious one should but consider that it is an expensive instruction anyway and instructions round it can proceed in parallel and a test can be generated by a compiler easily. Why lumber it with the possibility of changing the control flow? And if it isn't in the arithmetic unit then it would be because one split it into micro-operations and just sticking in the equivalent instructions avoids the need for such split.
.....
> >
> > > Do you seriously believe that the alleged performance and power advantage (and yes, it very much is alleged
> > > - I would even go as far as call it "drug-induced hallucinations" - so far nobody has come even close to
> > > Intel in the space you claim is so ripe for ARM64 in either performance or price) are so massive that users
> > > should/would ignore the fact that the break in ISA also causes a lot of real and inevitable problems?
> >
> > In a few short years no one will care about ARM32, everything will be
> > 64bit, even the phones will be dumping compatibility with 32bit.
>
> It's possible. And it's a pity.
> But, I'd guess that 'few'='your current estimate of few' + 4.
It'll certainly be around for a long time in embedded controllers. I would guess that Apple will soon start emulating 32 bit ARM and remove it from their processors, if they don't do that in the A8 it'll be because it is a tick in Intel terms rather than a tock.
.....
> >
> > > ARM64 isn't that magical. MIPS and alpha were there before it with pretty similar "simple
> > > decoding". You're repeating arguments that didn't make sense the last time around, and that
> > > have been soundly disproven in that thing we call "the real world" (tm) or "the market".
> >
> > Good points.
> > You realize of course that 20 years ago the RISC camp was split
> > into a half dozen factions, none big enough to survive.
> > Sparc and Power will be dead in a few years,
>
> Now you are out of your depth. Even dead Itanium is going to last for at least 5-6
> years. Such live bodies as SPARC and esp. Power will be around for much longer.
>
> > both will convert to ARM64, just like AMD is.
> >
> > Hows that for a prediction.
>
> No good.
64 bit ARM is already making a rapid advance against MIPS and PowerPC in plans for new embedded controllers.
> > There will be a full spectrum of ARM64 implementations from low to as high as Intel can go.
> >
> > > I think ARM has a chance, but the arguments for it in this thread have been absolutely moronic. The
> > > performance claims about how much better ARM64 is over ARM32 seem to be based on Geekbench, for chrissake.
> > > Some of the other arguments have been about how relatively quickly ARM has improved, which is largely
> > > based on the fact that Cortex-A9 was a complete and utter disaster particularly from an uncore standpoint,
> > > so when people bandy about "40% improvement" and compare that to how Haswell didn't make as much of
> > > a difference, they are clearly not understanding how the ARM baseline was crap.
> >
> > Agreed. Old Arm chips are pretty crap.
>
> ARM Cortex-A9 inner core is not bad. I'd dare to say that for phones it is somewhat
> better than A15. Unfortunately it is coupled with brain damaged "modular" L2 cache.
> And 32B cache lines are not helping it's case in the DDR3-dominated world.
The ARM Cortex-A17 is far better than the A9 even though it is derived from it, a 60% improvement from better memory handling and branch prediction and a couple of other tweaks. It basically is what ARM should have had instead of the A15 for the high end smartphone market and if they have any sense they're currently making a 64 bit version of it.
> >
> > > Even now, people seem to think that Apple A7 is somehow a high-performance chip. It's not really
> > > all that impressive, and again, the whole "look at how great it is" seems to be based almost entirely
> > > on pure crap (geekbench). And absolutely none of that is relevant to the server market.
> > >
> > > No, if ARM has a chance, it's not because it will outperform intel server chips
> > > (I can pretty much guarantee it won't), it is because of other market forces.
There do seem to be a number of other forces in ARM's favor like companies just not wanting to be in thrall to Intel. I don't know how the Russian business about using ARM will go and Intel has had some problems in China. It is a pity but Snowden's revelations have made people a bit paranoid though I think it more likely China would put a spy into a chip they produced for their own citizens than that Intel would do it to them for the NSA.
> > You will be wrong, and sooner than you think possible. All it takes is for POWER to go ARM64.
> >
> > > For example, there are a lot of customers who want to make sure that they have alternatives,
> > > and are worried about the fact that Intel is so crushingly dominant. Those customers don't
> > > necessarily care about ARM at all, just go back a few months to look at the POWER8 motherboard
> > > news etc, but that "we really worry about a monoculture" is very much a real issue.
> > >
> > > Similarly, there are a lot of chip companies that want to get a part of the market, and if you don't have
> > > the resources of Intel, it's really hard to compete in the x86 space. Because that complexity may not
> > > be a huge performance issue in the end, but it does mean that there is a fairly high bar of entry.
> >
> > Another prediction:
> > In ten years AMD will not be making any x86 chips. And no one will care. Plenty of ARM64 choices.
>
> It's more likely that in ten years AMD will not exist as independent
> company. If it still exists it's going to make x86.
Their x86 business aint going to save them but I guess they might produce custom SoCs with x86 processors on. If they are independently producing servers in 10 years time it will be ARM ones or some successor chip, who's for The Mill anyone? ;-)
....
> > > So I'm dead serious when I say that anybody who talks about the alleged huge advantages of simple decoding,
> > > but then blithely ignores the advantages of backwards compatibility, is a f*cking moron.
> >
> > Yes backwards compatibility MATTERS for x86. For ARM64 servers it is merely a short term nuisance.
> > Yes the near complete lack of apps will hurt at first, but the
> > bandwagon will gather steam and nothing will stop it.
> >
> > > And there's a lot of those f*cking morons in this thread. Some of them double down on their stupidity
> > > by claiming that it's a big advantage that you can jettison all that legacy baggage.
> >
> > It is a small advantage to jettison all that legacy baggage, every little bit helps.
I get the feeling Linus is like someone who's had to use Microsoft Word and is terrified of changing to anything else because of all the work they've done coping with it.
> > > Linus "rant over" Torvalds
> >
> > Brett, who is apparently a f*cking moron. ;) ;) ;)
> >
>
Dmcq, who is also apparently a f*cking moron. ;) ;) ;)