[HN Gopher] x86 is dead, long live x86
       ___________________________________________________________________
        
       x86 is dead, long live x86
        
       Author : harpratap
       Score  : 95 points
       Date   : 2023-07-31 14:05 UTC (8 hours ago)
        
 (HTM) web link (engineering.mercari.com)
 (TXT) w3m dump (engineering.mercari.com)
        
       | satanicpanic wrote:
       | [flagged]
        
       | retskrad wrote:
       | x86 lives on desktop now. Windows, Intel, AMD and Nvidia have
       | more or less thrown in the white flag and given the laptop market
       | to Apple.
        
         | cardanome wrote:
         | Apple products are only dominant the US. For the rest of the
         | world they are more of an luxury item.
         | 
         | That said, I am not terribly interested in ARM-based laptops
         | for now. Yes, they may be more energy efficient and all but
         | that hardly matters to me compared to just having the same x86
         | architecture I run on my desktop and servers. That sweet
         | binary-compatibility means less headache.
         | 
         | People underestimate the advantages that CPU architecture
         | monoculture gave us, though they are getting admittingly less
         | important year by year. Maybe one day I am going to run an ARM
         | laptop or even RISC-V.
        
           | mpweiher wrote:
           | > ... just having the same x86 architecture I run on my
           | desktop and servers.
           | 
           | Yep! For me, it's the same ARM64 architecture I run on my
           | desktop and servers[1]. :-)
           | 
           | Hetzner's offering is very competitive, cheaper than their
           | already rock-bottom x86 offerings. Then there's AWS Graviton,
           | Oracle with their free tier (not sure how expensive that gets
           | if you actually have to load it) and both Azure and Google
           | also have ARM offerings.
           | 
           | [1] https://blog.metaobject.com/2023/05/setting-up-hetzner-
           | arm-i...
        
           | agloe_dreams wrote:
           | A key element to Apple and ARM and why you are right outside
           | the us is this:
           | 
           | 1. Apple's chips are so far beyond everything else that it
           | makes obvious sense for Apple only. Snapdragon is at least
           | 30% slower single core while having worse Performance per
           | watt.
           | 
           | 2. Apple wasn't playing fair with their translation layer.
           | The Rosetta layer cheats a little because apple also made the
           | chip. The secret sauce is that the M1 has a hardware
           | compatibility mode (That technically breaks ARM spec) for x86
           | memory order that basically gives near 1:1 performance.
           | 
           | 3. Microsoft heavily botched their ARM rollout (again. Hello,
           | Windows RT). The translation layer on W10/11 is just bad, not
           | because of bad coding but just the technical limits of what
           | they were trying to do.
           | 
           | 4. Google is in a great place with ARM compatibility in
           | Chrome OS as the only consumer-facing apps are either built-
           | in or on the play store...which was designed for ARM in the
           | first place. Problem is that nobody will give them a good
           | chip for ChromeOS and the focus is on low end, so Mediatek is
           | just cruising.
           | 
           | ARM is a great arch...but right now it is Apple VS x86, not
           | ARM.
        
             | yyyk wrote:
             | >3. Microsoft heavily botched their ARM rollout (again.
             | Hello, Windows RT). The translation layer on W10/11 is just
             | bad, not because of bad coding but just the technical
             | limits of what they were trying to do.
             | 
             | It's not bad at all, however Intel patents heavily
             | restricted the initial implementation, to the point they
             | couldn't ship the 64bit emulation. Apple decided to wait
             | out the patents.
        
             | ngcc_hk wrote:
             | Unfair ... strange use of phrase. Innovative would be a
             | better use of word as nothing prepare others to do the same
             | ... it is good to have a vendor play differently not follow
             | the slow and dominated by arm player c.
             | 
             | The conclusion is on the ball but only current. M1 has been
             | a bit old now and if other arm vendor and the two major
             | architecture (ibm power for mainframe ... cannot be
             | counted) it is at least x86-32 plus x86-64 vs arm vs apple.
        
               | agloe_dreams wrote:
               | The wording of unfair is more of a fun use of the word.
               | Because Apple vertically integrated the whole product and
               | software supply chain, they are singularly able to do
               | things others cannot. In that context, if one is playing
               | "fair" on the terms of everyone else when they could do
               | more...they are losing.
        
             | astrange wrote:
             | > That technically breaks ARM spec
             | 
             | It does not (or it wouldn't be there) and they aren't the
             | only ARM vendor with TSO (Fujitsu also does it.)
        
               | monocasa wrote:
               | Optionally enabling it does break the ARM spec.
               | 
               | Their cores break the ARM spec in other ways too. Added
               | instructions for AMX, new guard privilege modes,
               | HCR_EL2.E2H can't be disabled, GPRs are clobbered on WFI,
               | etc.
        
         | MisterBastahrd wrote:
         | Unless Apple can deliver a sub-800 dollar product, they'll
         | never own the laptop market.
        
           | r00fus wrote:
           | Correct. And the fact is, Apple would rather not (as it comes
           | with government antitrust scrutiny) Instead they'll focus on
           | profit share.
        
         | oaiey wrote:
         | Sounds like a bubble thing. Sure x86 may go (as will arm one
         | day), but there is a very solid, healthy and huge Laptop market
         | outside than Apple.
        
         | aidenn0 wrote:
         | Just had a family reunion. The only Apple laptop was my wife's,
         | out of about a dozen.
         | 
         | Furthermore their market share has never been over 20% of units
         | shipped[1]
         | 
         | 1: https://www.statista.com/statistics/576473/united-states-
         | qua...
        
         | jimmaswell wrote:
         | Based on what? I would never buy an Apple laptop and no one I
         | know has one either. The Windows "gaming laptop" I got a few
         | years ago for game dev is perfectly adequate and has features
         | missing from Macbooks:
         | 
         | - ethernet port
         | 
         | - hdmi port
         | 
         | - multiple usb A and C ports
         | 
         | - solid keyboard
         | 
         | Same for my work laptop, again Windows/x86 and no one I know
         | with a work laptop is supplied a Mac either.
        
           | scarface_74 wrote:
           | > Based on what? I would never buy an Apple laptop and no one
           | I know has one either
           | 
           | Yes because no one that you know buys an Apple laptop that
           | must mean that Apple isn't selling any...
           | 
           | > Same for my work laptop, again Windows/x86 and no one I
           | know with a work laptop is supplied a Mac either.
           | 
           | I find the lack of self awareness...amusing.
           | 
           | Because no one you know uses a Mac laptop, that must mean no
           | one uses one.
           | 
           | Well my anecdote from working at the second largest employer
           | in the US that the vast majority of technical people prefer
           | Macs even though they can choose Macs or Windows.
        
             | debugnik wrote:
             | They're not arguing that Apple doesn't sell laptops,
             | they're refuting that "Windows, Intel, AMD and Nvidia have
             | [...] given the laptop market to Apple". Unless one thinks
             | the laptop market only exists in the US, this idea simply
             | can't match reality, not yet anyway.
        
               | scarface_74 wrote:
               | If they are arguing that, don't you think their anecdotal
               | "proof" is silly?
               | 
               | > I would never buy an Apple laptop and _no one I know
               | has one either_.
               | 
               | >no one I know with a work laptop is supplied a Mac
               | either.
               | 
               | I don't know anyone that uses WeChat. Wouldn't that be a
               | silly argument?
        
               | debugnik wrote:
               | Not if the anecdote contradicts an equally
               | unsubstantiated generalisation, honestly. If you're going
               | to complain about the quality of the discussion, do it
               | with both sides, not just the argument you side against.
        
               | scarface_74 wrote:
               | Well, it's really not that hard to search on Google for
               | "percent of laptops running ARM" and find something
               | slightly less silly than "no one I know owns a MacBook
               | nor does anyone I work with"
               | 
               | And the first link is
               | 
               | https://www.patentlyapple.com/2023/02/apple-currently-
               | owns-9...
               | 
               | It would be much more substantial than anecdotal what his
               | friends and coworkers do.
               | 
               | It's about as bad as the old Slashdot "do people still
               | watch TV? I haven't own a TV in 10 years"
        
               | debugnik wrote:
               | But why do you expect people to put any effort to answer
               | someone that didn't put any either? Subpar comments are
               | begging for subpar replies.
               | 
               | If you want to read a non-silly thread, I suggest looking
               | for a non-silly opener, this website has more than
               | average of those.
        
             | arp242 wrote:
             | "Apple isn't selling any laptops" wasn't what was said. The
             | original claim was x86 has "thrown in the white flag" and
             | that Apple dominates the laptop market, which is a pretty
             | ridiculous claim since x86 is still the vast majority of
             | laptop sales and Windows is still the dominant laptop OS.
        
               | havblue wrote:
               | I don't personally see MacBooks as a viable platform
               | either right now, but I would say that this looks like a
               | steady decline for windows x86 laptops until Chrome
               | laptops take over or an arm-based windows replacement
               | takes hold. "Hey this isn't a surrender, it's a war of
               | attrition for the next five years!!!"
        
               | arp242 wrote:
               | Given that 15-20 years ago almost all laptops were
               | x86/Windows, "decline" is pretty much the only direction
               | the platform could go in.
               | 
               | I don't expect either x86 or Windows to go away any time
               | soon - in general I think x86 has been somewhat unfairly
               | maligned, and for most applications (including laptops)
               | it works pretty well and is even the best available
               | choice. I would prefer to say "diversification" or "the
               | existence of actual competition" rather than "decline",
               | as that doesn't imply it's going away.
        
               | scarface_74 wrote:
               | > I don't personally see MacBooks as a viable platform
               | either right now
               | 
               | But yet Apple is selling millions. Someone must see it as
               | a viable experience.
        
               | scarface_74 wrote:
               | > Apple isn't selling any laptops" wasn't what was said
               | 
               | The parent poster said
               | 
               | > would never buy an Apple laptop and no one I know has
               | one either
               | 
               | What relevance then was there in the statement that the
               | original poster made?
               | 
               | > Apple dominates the laptop market, which is a pretty
               | ridiculous claim since x86 is still the vast majority of
               | laptop sales and Windows is still the dominant laptop OS.
               | 
               | Nor did I argue otherwise. I merely called out the
               | silliness of the statement that because the parent poster
               | "doesn't know anyone who owns one" or that because every
               | developer he knows uses a Windows laptop, that his
               | anecdotal experience has any relevance.
        
           | katbyte wrote:
           | MacBooks now post 2019 have a great keyboard, hdmi and
           | multiple usb c ports, as well sd card and headphone jack.
           | 
           | The only thing it is actually missing from your list is
           | Ethernet and usb a
        
           | gumby wrote:
           | That's what's so great about capitalism: I'm not interested
           | in any of those features on a laptop, so Apple is great for
           | me.* You do, and several other companies compete for your
           | business.
           | 
           | On a desktop (which these days isn't a tower) I do want some
           | of those features, and I get them.
           | 
           | * except kbd, and I find the apple kbds just fine -- I even
           | survived a couple of rounds of the infamous "butterfly" kbds.
           | But I know some people were unhappy and eventually Apple got
           | their act together.
        
           | glimshe wrote:
           | I would never buy an Apple laptop. I have one for work and I
           | hate it. Linux and Windows are by far more productive for me,
           | so all I need to do is keep getting better PC laptops. While
           | Mac hardware does have a slight edge on cost/benefit at that
           | price point, macOS is just too cumbersome for my needs.
           | 
           | EDIT: It's funny to see this downvoted. I'm literally
           | speaking of my particular use case and needs! Are there
           | people who believe Apples are better in _all_ scenarios?
        
             | adamc wrote:
             | The whole "click once on the app for it to wake up, click
             | again for it to know which window you wanted" makes using
             | Mac OS highly unpleasant on a daily basis. Yes, I'm on Mac
             | OS, and no, I don't like it.
        
               | cthalupa wrote:
               | As a daily Mac user for about 13 years now I have no idea
               | what phenomenon you're talking about.
               | 
               | For any program that is open, even if it hasn't been in
               | use for a while, I click it once, and it comes to the
               | forefront immediately.
               | 
               | For programs that are not open, I click it once, and then
               | if I wait for it to launch, it appears in the forefront.
               | If I immediately click back over to my web browser and
               | begin using it before the program launches, it does not
               | appear in the forefront... because I interacted with
               | another program after starting the launch.
        
               | Clamchop wrote:
               | They might be referring to how clicks don't register with
               | an app until it has focus (the first click being to focus
               | it).
               | 
               | On Windows you can, say, click once to pause on a video
               | playing in a browser window that doesn't have focus but
               | is visible. On macOS, it's two clicks, because you have
               | to focus the app first.
               | 
               | I don't care for the behavior but it's so minor to me
               | that it carries essentially no weight.
        
               | cthalupa wrote:
               | Ah! This does make sense. I mostly run things fullscreen
               | on my laptop and use Mission Control to swap windows so I
               | really only encounter this when my laptop is docked to a
               | larger display.
               | 
               | It doesn't particularly bother me one way or the other to
               | the point I didn't even think of this, despite also being
               | a daily Windows users on my desktop computer.
        
               | TillE wrote:
               | Use Mission Control for everything, it's great.
        
               | zdragnar wrote:
               | Except for how there's no stable sort order for windows,
               | even if you turn off the setting that's supposed to not
               | rearrange it all the time. Mission control was probably
               | one of the things I liked least about the Mac when I used
               | one for work... To the point that I ended up using
               | hammerspoon to create shortcuts to switch between my
               | applications directly rather than deal with it.
        
           | holbrad wrote:
           | Just chiming in to agree, I don't think we have a single mac
           | in our entire company (100's of people). It's all just
           | windows/x86 like every other business I've worked at.
        
           | cthalupa wrote:
           | MBP once again have HDMI ports and good keyboards. I wouldn't
           | mind a single USB-A port but I don't particularly miss it,
           | and 3 USB-Cs is plenty for my uses.
           | 
           | I haven't needed an ethernet port on a laptop since the mid
           | 2000s troubleshooting why my newly installed stack of Cisco
           | switches weren't working, and that sort of use cases is rare
           | enough that having a dongle seems fine.
           | 
           | Conversely, about 80% of the people I know use Apple laptops
           | for work.
        
           | ezfe wrote:
           | MacBook Pros have HDMI ports and many USB-C ports. I also
           | don't know any issue with MacBook keyboards except the
           | 2016-2019 model years.
           | 
           | I work for a large financial services company and receive a
           | MacBook Pro for my work.
           | 
           | My work laptop battery can handle all-day IntelliJ & web
           | browsing without charging, I've never heard the fan and the
           | laptop stays cool. It has a Geekbench multicore score of
           | 13,649.
           | 
           | My personal laptop also lasts that long and stays cool, and
           | doesn't even have a fan.
        
           | edgyquant wrote:
           | We live in different worlds. All engineers I know have Macs
           | and the ones that don't have Linux. The one guy on my team
           | who uses windows is a constant problem as he has to find work
           | arounds for every process we have/service we build.
        
             | oblak wrote:
             | 1. Build a pipeline with macOS in mind (because "All
             | engineers I know have Macs")
             | 
             | 2. Complain when the macOS-oriented process doesn't work
             | for everyone.
        
               | ezfe wrote:
               | I work on a split team, Windows and macOS. We do
               | straightforward Java work and most of the problems that
               | we have are related to installing command-line software
               | on Windows.
               | 
               | Installing Postgres, Redis, etc. on Windows is wildly
               | complicated compared to on macOS or Linux.
        
               | jimmaswell wrote:
               | Tried WSL?
        
               | WeylandYutani wrote:
               | Apple is doing what people hated Microsoft for. It's
               | amazing that a closed ecosystem is so welcomed by Apple
               | fans. But I suppose man has always been tribal.
        
               | Dylan16807 wrote:
               | Any closed ecosystem is annoying but I'm not going to
               | complain very hard while their desktop market share is
               | below 20-25%.
        
             | shortrounddev2 wrote:
             | Every company I've worked for provided a macbook to us, and
             | all my coworkers use a macbook. When I have the option, I
             | pick PC. My last company gave me a choice between mac or
             | PC, and I picked the thinkpad. My current company gave me a
             | macbook pro, but I opted to use my personal Windows
             | machines instead.
             | 
             | I'm the guy who has to come up with a workaround, but it's
             | never a problem for others. Usually I'm translating some
             | bash command into powershell, which takes maybe 2 minutes.
             | Everything else runs just fine, and I don't even need to
             | use WSL. Windows is a perfectly viable development
             | platform, if you give a shit about learning the differences
             | between a unix-like environment and windows instead of just
             | copping out and using WSL
        
             | hota_mazi wrote:
             | You probably live in the US and parent lives... anywhere
             | else in the world really.
        
             | redwall_hp wrote:
             | Yep. I work in an entire office of engineers with only Macs
             | (mostly ARM ones now) and I only have a Mac at home,
             | besides a very old ThinkPad running Ubuntu. (Aside from
             | FFXIV on MacOS, I've moved any gaming to PS5 and Switch.)
             | 
             | I haven't actively used Windows since 2008. It's such an
             | obnoxious OS to use, and I have never seen a Windows laptop
             | with quality in the same zip code as a Mac. There's always
             | a terrible touch pad, keyboard flex, plasticky trash fit
             | and finish, mediocre display or something that just ruins
             | the whole thing.
        
               | selectodude wrote:
               | I just received a brand new Thinkpad from work. I don't
               | even mind Windows but holy crap is that thing an absolute
               | chore to use. I don't understand how this is still an
               | issue while Apple has been putting out what amounts to
               | the perfect laptop for like 10+ years now.
        
               | innocenat wrote:
               | On the other hand, I moved from MacBook to ThinkPad and
               | holy crap it's just so much easier to use.
        
               | rejectfinite wrote:
               | >I don't even mind Windows but holy crap is that thing an
               | absolute chore to use.
               | 
               | What are your issues? I have a T490 that is a few years
               | old and its great for my use. But I keep it docked in to
               | a monitor and external mouse and keyboard 90% of the
               | time.
        
               | hota_mazi wrote:
               | > I haven't actively used Windows since 2008. It's such
               | an obnoxious OS to use
               | 
               | You might want to modernize your perception of it,
               | holding on to 15 year old views is not a very smart move
               | in the tech industry.
               | 
               | Windows 10 and 11 are a joy to use, and to develop. I
               | used a Mac for 15 years and switched to Windows to
               | develop 5 years ago, and now macOS looks extremely
               | antiquated and creaky for development.
        
               | Clamchop wrote:
               | Sure are a lot of extreme opinions on operating systems
               | when they all seem pretty damned similar to me on the
               | fundamentals.
               | 
               | Development work strikes me as maybe the least
               | distinguishing kind of task to compare them on. Very few
               | hard choices to make unless you're developing for Mac or
               | iOS.
        
               | shortrounddev2 wrote:
               | I like the aluminum bodies of macs, but I hate the flat
               | keyboards. In my view, no laptop brand has had a good
               | keyboard since the Thinkpad T420.
               | 
               | In terms of build quality, apple laptops are pretty good,
               | but I can't stand macOS and apple keeps fucking with
               | their ports.
               | 
               | My Inspiron 16 has a good glass screen and a pretty solid
               | aluminum body. It's got all the ports I want (HDMI, USB-A
               | 3.0, SD card reader, USB-C, and a 3.5mm headphone jack).
               | Personally, the macbook pro my company issued me is one
               | of my least favorite laptops I've ever used. I also had
               | compatibility or performance problems with x86 based
               | software when I got my first M1 macbook, I don't know if
               | they're improved anything since then
        
             | parineum wrote:
             | If your dev environment is meant to be multiplatform, it's
             | sounds a lot more like the other developers are checking in
             | things that don't work on anything but macs.
             | 
             | Your Linux devs are just used to dealing with it.
        
           | babypuncher wrote:
           | Macbooks have HDMI ports and solid keyboards.
           | 
           | I don't see ethernet ports making a comeback on thin laptops
           | anytime soon though. They are handy for LAN parties, so bulky
           | gaming laptops will keep shipping with them, but for everyone
           | else it makes more sense to leave that functionality to a
           | dock.
        
         | ngcc_hk wrote:
         | Strange and whilst I have 8 macOS and 3 ipad plus 5 iOS devices
         | I found in the train everyone use small dell notebook.
         | 
         | I guess like me I like macbook 12 and that is what you use for
         | non-desktop like notebook.
        
           | dc443 wrote:
           | I've been hoarding apple devices too. I have a 12 inch
           | macbook, recently brought it up to Ventura. But the blasted
           | thing won't hold a charge. The battery is fine over 90%
           | health but something is causing power drain while it sleeps
           | even when I try to hibernate it.
           | 
           | It's impossible to find a use case for the thing. I might
           | have to sell it.
        
         | o1y32 wrote:
         | A few programmers on HN think only they use laptops (and
         | specifically Macbooks). Not surprising.
         | 
         | If you just bother to open your eyes just a little bit wider
         | you would notice that there is a huge market for Chromebooks,
         | budget laptops, gaming laptops, mobile workstations, ultrabooks
         | for students, gamers, business people, bankers etc. New things
         | are happening every month. We are seeing more efficient laptops
         | from Intel and AMD. Companies like Framework are doing actual
         | innovations in the decade old laptop area. And there are
         | workflows that can only be done on Windows.
         | 
         | Your claim is completely unfounded.
        
         | kens wrote:
         | ARM has a lot more market share in people's minds than in
         | actual numbers. One research firm says that ARM has 15% of the
         | laptop market share in 2023, expected to increase to 25% by
         | 2027. (Surprisingly, Apple only has 90% of the ARM laptop
         | market.)
         | 
         | In the server market, just an estimated 8% of CPU shipments in
         | 2023 were ARM.
         | 
         | https://www.counterpointresearch.com/arm-based-pcs-to-nearly...
         | https://www.digitimes.com/news/a20230217VL209/amd-arm-digiti...
        
           | Andrex wrote:
           | I keep meaning to research this, and maybe this isn't the
           | right place, but how are things going for Linux executables?
           | Last I tried Linux on ARM it seemed like application-level
           | support for ARM was very spotty.
           | 
           | It doesn't seem like a Rosetta 2-like effort is making it
           | into mainline Linux anytime soon, if ever.
        
             | arp242 wrote:
             | For open source applications it's generally pretty good.
             | Closed source: not so much.
             | 
             | qemu-x86_64 is probably the closest there is to Rosetta; it
             | works fairly well and I think conceptually it's identical
             | or similar to Rosetta, but I don't really know what the
             | performance is like, and it's of course not as
             | integrated/automatic as Rosetta.
        
               | zekica wrote:
               | IMO, better options are box86/box64 or FEX.
        
               | Andrex wrote:
               | Hadn't heard of qemu (in this context), I'll take a look,
               | thank you!
        
             | nazgulsenpai wrote:
             | I suppose it depends on the distribution but most I've used
             | (Manjaro, Void, Alpine) are close to 1:1 with x86. Of
             | course third party applications have to compile for ARM if
             | they don't intend to distribute source code.
        
           | agloe_dreams wrote:
           | That is spot-on. The fact of the matter is that it is Apple
           | VS x86, not ARM. The M1/2 have dominant mindshare due to
           | being actually better.
           | 
           | On Windows, ARM products just suck. The product is just bad
           | because there is no reason to use it. The chips available are
           | worse than x86 and then the software issue is bad due to a
           | whole set of reasons that Microsoft can't change on their
           | own..\ Google has that last 10% probably, ChromeOS moves ARM
           | devices all day, every day. Just cheap ones.
        
       | netr0ute wrote:
       | Why do these kinds of articles essentially refuse to admit RISC-V
       | exists?
        
         | snvzz wrote:
         | The context is AWS and there's no AWS RISC-V instances yet.
         | 
         | But worry not, RISC-V is inevitable.
        
         | hedgehog wrote:
         | RISC-V is important, and I use it every day for work, but there
         | are no mainstream mature server platforms using it and that's
         | what this article is about.
        
         | harpratap wrote:
         | Cloud world is really slow. Imagine writing about Zen3 Milan in
         | Aug 2023 when AMD has already announced Zen4c Bergamo. Actually
         | we were "ahead of the curve" since we got access to T2D before
         | it was made publicly available in Tokyo region, and even then
         | it took several months to get enough capacity in Tokyo to fully
         | migrate our production Kubernetes cluster.
         | 
         | I really wish we could test out RISC-V SoCs from the likes of
         | tenstorrent, but it's a long journey
        
         | xcdzvyn wrote:
         | Because it essentially doesn't? At least when I tried looking a
         | few months ago, it was really hard to find any commercially
         | available RISC-V SoCs, let alone high performance ones. Sure
         | there's little hobbyist boards going for $200>, but that's
         | about it.
        
         | jsheard wrote:
         | Probably because the article is looking at silicon that's
         | actually available today, and big-iron RISC-V silicon akin to
         | Xeon/EPYC/Graviton/Ampere is more of a hypothetical at this
         | point.
        
       | harpratap wrote:
       | We did a migration from GCP's Intel based E2 instances to AMD's
       | T2D instances and saw huge 30% savings in overall compute! It is
       | similar amount of savings folks got from switching to AWS
       | Graviton instances, so looks like AMD might keep the x86 ISA
       | still alive
        
         | jeffbee wrote:
         | E2 is just extremely old. E2 are a mixed fleet that contain
         | CPUs as old as Haswell. Haswell launched over 10 years ago. It
         | makes sense that you get a better bang for your buck from using
         | something that isn't gravely obsolete. You should also keep
         | that in mind when benchmarking E2, since it's a grab-bag of
         | CPUs you need to control for the actual CPU type, either by
         | specifying the minimum one you require or assuming the worst.
        
           | harpratap wrote:
           | That is true. But in cloud our hands our tied, we cannot
           | really switch from one one generation to another so easily.
           | GCP has so far never launched a new chip while keeping the
           | price same or lower. We did same cost/perf analysis on newer
           | generation chips like Ice Lake from Intel and even Milan from
           | AMD in the form of N2D, but both are quiet expensive for the
           | performance uplift they provide. The unique thing about T2D
           | is that the price is competitive with 10 year old E2, which
           | has been the case only for ARM based CPUs like Graviton and
           | Ampere Altra (T2A from GCP)
        
             | jeffbee wrote:
             | I get what you mean. For the record, here's a GCP engineer
             | holding forth on price/performance ratios who also
             | concludes that T2D is optimal. https://medium.com/google-
             | cloud/google-cloud-platform-cpu-pe...
             | 
             | ETA: Since you are using Go and targeting a specific modern
             | CPU, you may also get a measurable benefit from setting
             | GOAMD64=v3, so the go compiler generates code using AVX2,
             | BMI2, LZCNT, etc.
        
               | harpratap wrote:
               | Thanks, that's a really insightful article!
               | 
               | > Since you are using Go and targeting a specific modern
               | CPU, you may also get a measurable benefit from setting
               | GOAMD64=v3
               | 
               | That's actually a long pending open issue in our backlog
               | :)
        
       | singhrac wrote:
       | As a separate data point, I briefly switched one of our servers
       | from an r6a.4xlarge (AMD Epyc) to a r6i.4xlarge (Intel Xeon) and
       | saw a 30% speedup in our number-heavy compute task. I would love
       | to find out why (MKL or AVX512? Do I need to recompile numpy?),
       | but for the time being it pays to stay on Xeon.
       | 
       | We eventually switched to m-instances since that fits our
       | compute/memory usage better when we're at limits.
        
         | re-thc wrote:
         | If it's AVX512, you'll be excited to know that m7a, currently
         | in preview has great support for this. AMD Epyc 4th gen i.e.
         | genoa is a lot faster.
        
       | amusingimpala75 wrote:
       | I would just like to point out that that is not how the saying
       | goes. When Queen Elizabeth died it would have been "the queen is
       | dead, long live the king", as the dead thing is the predecessor
       | and the living thing is the one that follows it
        
         | ihattendorf wrote:
         | https://en.wikipedia.org/wiki/The_king_is_dead,_long_live_th...
         | !
         | 
         | In this case, the article is implying that Intel (x86) is dead
         | to them and AMD (x86) is the successor. Whether Intel is dead
         | or not is up for debate (I doubt they are), but the saying is
         | used correctly.
        
         | harpratap wrote:
         | The dead thing is Intel (x86) and the successor is AMD (x86) as
         | opposed to ARM in our case. Isn't it correct?
        
           | re-thc wrote:
           | > Isn't it correct?
           | 
           | No. Intel isn't dead. They may be behind (for now) but
           | they're definitely catching up and have in a way on the
           | desktop.
           | 
           | It's not certain that Intel will die and AMD will for sure
           | win. Competition is great.
        
             | saghm wrote:
             | If we're going to be that pedantic, we might as well just
             | say that because Intel and AMD aren't living entities so
             | they can't be "dead". Or we could just accept that the
             | original meaning of the phrase has evolved and is flexible
             | enough to be used in situations where it's not absolutely
             | literal.
        
               | Teever wrote:
               | I just closed another post on HN because the top thread
               | descended into some bullshit semantics pissing contest so
               | I come here and see the same.
               | 
               | Is it just me or has anyone else noticed an increase in
               | arguing over the meaning of words on HN lately?
        
               | arcanemachiner wrote:
               | Nerds have been arguing online about irrelevant pedantry
               | as long as the Internet has been around. And this place
               | attracts some pretty hardcore nerds.
        
               | lcnPylGDnU4H9OF wrote:
               | I'd guess it's a timing thing. This site likes to
               | prioritize more highly the "current" discussions which
               | are generating upvotes as long as they're not also
               | generating a ton of downvotes.
        
               | kergonath wrote:
               | I've noticed that as well. It beggars belief that people
               | are seriously arguing about applying a centuries-old
               | foreign metaphors about kings to modern large
               | corporations. That's straight over Pedantic territory
               | into the Obnoxious Contrarian wilderness. Sigh...
        
               | coldtea wrote:
               | Sorry, but you've just noticed it?
               | 
               | Wars about semantics have been standard fare on geekdom
               | since the dawn of time, and on Hacker News since the dawn
               | of Arc.
               | 
               | What appears as "Obnoxious Contrarian wilderness" is
               | good-old hacker "well, actually" pendanticness (with some
               | sprinkling of on-the-spectrum focus on details and
               | semantics).
        
               | kergonath wrote:
               | > Sorry, but you've just noticed it?
               | 
               | Yes, I have noticed "an increase in arguing over the
               | meaning of words on HN lately". I also know how to
               | discuss the meaning of words and how they form the
               | context of a conversation, once taken together.
               | 
               | > Wars about semantics have been standard fare on geekdom
               | since the dawn of time, and on Hacker News since the dawn
               | of Arc.
               | 
               | Sure, and I enjoy very much some pedantry (and am not
               | above doing my bit every now and then), but that's not
               | the point. If someone wants to have a heated discussion
               | about how a centuries-old metaphor would apply to the
               | kings of France but not to a multinational company, then
               | sure, have at it, as long as you don't derail otherwise
               | useful threads. I will just opt out of spending my time
               | that way and instead get some interesting pedantry.
               | 
               | I must admit I am not really sure what image you want to
               | imply by "dawn of Arc", though it does go a bit with the
               | French king vibe. Would you care to explicit a little?
               | 
               | > What appears as "Obnoxious Contrarian wilderness" is
               | good-old hacker "well, actually" pendanticness (with some
               | sprinkling of on-the-spectrum focus on details and
               | semantics).
               | 
               | Pedantry and semantics are fine, and indeed one of the
               | reasons why I am here. But the point beyond that has to
               | be knowledge and enlightenment, otherwise it's just, yes,
               | obnoxious. There is also a thin line between a good "well
               | actually" and a nuisance (not even talking about bad
               | faith trolling).
               | 
               | Please note that I did not address any of the people in
               | question, because that would be stupid, rude, and
               | counter-productive. I am happy if they have their
               | arguments on their side. I was just replying to a fellow
               | commenter that they were not alone with that feeling.
               | There's no need to be rude about it.
        
               | dragonwriter wrote:
               | > I must admit I am not really sure what image you want
               | to imply by "dawn of Arc"
               | 
               | Arc is a Lisp dialect invented by PG whose main notable
               | use is that it is the implementation language for HN: the
               | "dawn of Arc" is thus a flowery way of saying something
               | like "the first concrete steps to the creation of HN".
        
               | kergonath wrote:
               | Interesting, I did not know that. Thanks!
        
               | coldtea wrote:
               | > _There's no need to be rude about it._
               | 
               | I'm sorry, then! (Was I tho? I thought I answered mostly
               | matter-of-factly)
        
               | coldtea wrote:
               | > _Is it just me or has anyone else noticed an increase
               | in arguing over the meaning of words on HN lately?_
               | 
               | Depends on what you mean by increase, arguing, and words.
        
               | throwaway742 wrote:
               | It's unironically one of the things I like about this
               | site.
        
             | GauntletWizard wrote:
             | Perhaps it would be better phrased as "'intel x86'" vs
             | "'AMD x86-64'". Specific sets of mandatory instructions and
             | extension parameters - But AMD's set "won" at some point in
             | the past, roughly corresponding to the launch of Intel's
             | "Core" microarchitecture in 2006 when they started
             | competing in the x86 space again
        
             | Clamchop wrote:
             | Weird. The topic was the phrase.
             | 
             | Mercari replaced Intel with AMD. One x86 out, another in.
             | Usage is correct if figurative.
             | 
             | Article does not claim Intel is dead. States that they are
             | "catching up". But for their nodes, Intel out ("dead"), AMD
             | in.
             | 
             | It's reasonable.
        
             | coldtea wrote:
             | Your point is about the factual statement (whether Intel
             | x86 is actually dead).
             | 
             | But this subthread (as started by amusingimpala75), and
             | your direct parent's question was about whether the
             | idiomatic usage of "(old) X is dead, long live (new) X" was
             | correct (given that the author believes AMD taking over
             | Intel is the factual case). That is, not about whether the
             | factual statement is correct or not, just whether it's
             | expressed well idiomatically.
        
         | coldtea wrote:
         | > _When Queen Elizabeth died it would have been "the queen is
         | dead, long live the king", as the dead thing is the predecessor
         | and the living thing is the one that follows it_
         | 
         | And if a freshly dead king is replaced by a new king, it's "the
         | king is dead, long live the king".
         | 
         | The predecessor and the living thing doesn't have to be of the
         | opposite sex, or use a different term:
         | 
         | https://en.wikipedia.org/wiki/The_king_is_dead,_long_live_th...
         | !
         | 
         | It's a very famous idiom.
        
       | WeylandYutani wrote:
       | "The macOS (OS X) version of this game does not work on macOS
       | Catalina (version 10.15) or later due to the removal of support
       | for 32-bit-only apps"
       | 
       | And that's why x86 is good.
        
         | rideontime wrote:
         | I'm failing to see the connection, could you please elaborate?
        
       | m4r1k wrote:
       | If you wanna have additional insights, head over my Medium ports
       | with SPEC CPU 2017 results. https://medium.com/google-
       | cloud/google-cloud-platform-cpu-pe...
        
       | tester756 wrote:
       | x86 would be dead if ARM ISA offered like at least 20% perf boost
       | or energy savings.
       | 
       | But since ISA doesn't imply perf. characteristics itself, then
       | x86 will be alive.
       | 
       | The hard part in changing hardware's stuff is getting software to
       | adjust.
        
       | slfnflctd wrote:
       | When I think about how long chips like the 6502 have still been
       | in active use (almost 50 years now), it is hard to conceive of a
       | world where there isn't a significant presence of x86 activity
       | for the rest of my life.
       | 
       | The majority of 'the market' may go elsewhere, but for a
       | gazillion reasons, x86 will not be disappearing for quite a
       | while. At this point it would honestly surprise me if we didn't
       | at least have high quality emulation available until the end of
       | the human race as we know it.
       | 
       | Sure, we've probably lost most of the software ever written on
       | it, but a whole lot of interesting artifacts from a key
       | transition point for our species still remain locked up in this
       | architecture.
        
         | riffic wrote:
         | A thing's future longevity can sometimes be predicted by how
         | long it's already been around.
        
           | briandw wrote:
           | https://en.wikipedia.org/wiki/Lindy_effect
        
           | _a_a_a_ wrote:
           | But not humans, regrettably.
        
             | wongarsu wrote:
             | It used to somewhat work for humans, until about the
             | mid-19th century (in Europe, other places vary). A newborn
             | was unlikely to reach an age of 5, a 5 year old had better
             | chances but might still die from illness in childhood, but
             | someone who reached an age of 20 could be expected to reach
             | an age of 40, and if you reached 40 you might well make it
             | to 80.
        
             | User23 wrote:
             | 99 year old humans are considerably more likely to live to
             | be 100 than 50 year old humans.
        
               | Dylan16807 wrote:
               | "future longevity" in this case means how many years are
               | left, not how many years total.
        
         | thefurdrake wrote:
         | x86 is now permanently a part of humanity. 1000 years from now,
         | when we've transcended our physical bodies and exist only as
         | streams of sentient data and energy traveling between the
         | stars, I 100% guarantee x86 will be detectable somewhere.
        
         | tracker1 wrote:
         | Given the new 128-core AMD server parts are on-par with ARM in
         | terms of power efficiency and capable of more raw compute, it
         | may even grow a bit.
         | 
         | I think there's lots of room for ARM, Risc-V and x86_64 in the
         | future. There's reasons to support any of them over the others.
         | And given how well developer tool are getting support across
         | them all, it may actually grow a lot. I think the down side is
         | a lot of the secondary compute accelerators, such as what intel
         | is pushing and what the various ARM and Risc-V implementations
         | include in practice.
         | 
         | The further from a common core you get, the more complex
         | porting or cross platform tooling gets. Even if for big gains
         | in some parts. For example, working on personal/hobby projects
         | in ARM boards that aren't RPi is sometimes an exercise in
         | frustration, with no mainline support at all.
        
           | mlindner wrote:
           | > I think the down side is a lot of the secondary compute
           | accelerators, such as what intel is pushing and what the
           | various ARM and Risc-V implementations include in practice.
           | 
           | I'm curious why this is a downside. The current trends in
           | computing is that we're long past the point of single
           | threaded compute. The first step of that was multi processor
           | and multi core and that'll continue with more and more
           | dedicated and specialized computing sub-processors. Energy
           | prices are more and more becoming a major determining factor
           | as is the area needed for cooling. By having more separated
           | subprocessors you get both efficiency and easier ability to
           | cool the parts.
        
         | Gordonjcp wrote:
         | Stuff like 6502s and Z80s are a bit like little single-cylinder
         | engines - the world will move onto all sorts of interesting new
         | places, but something somewhere will always be powered by a wee
         | Briggs & Stratton that starts first pull of the string, and
         | we'll be glad of it.
        
           | Dylan16807 wrote:
           | We'll always need tiny cores, but it's worth noting that
           | RISC-V can squeeze down to pretty small sizes and is so much
           | nicer to use. Notably you can go smaller than a 68k or 8086.
        
             | hedgehog wrote:
             | I suspect 32-bit RISC-V cores will become the minimum unit
             | of processor for new designs. Not a meaningful cost
             | increase over say a 6502 (or ARM) to build, but the
             | convenience of having mainstream compiler support and that
             | kind of thing does make a difference in the cost of
             | building a product.
        
               | mjevans wrote:
               | Less transistors does mean less power too. I could see
               | applications where that matters.
        
         | babypuncher wrote:
         | As we learned from Terminator 2, the machines that eventually
         | rise up to eradicate humanity will still be running some kind
         | of 6502 derivative.
        
         | sneed_chucker wrote:
         | Are 6502 chips still used? What's the application?
        
           | _whiteCaps_ wrote:
           | I don't know anything about physical 6502's but they've been
           | embedded in FPGAs where you need a small MCU in a larger
           | design. Same with Intel 8051's.
        
           | krylon wrote:
           | Tamagochis are running on 6502 compatible chips. Which makes
           | me think it is used in other toys, too.
        
             | tabtab wrote:
             | What are the top embedded usage chips and what is their
             | typical niche?
        
               | krylon wrote:
               | I wouldn't know. I just watched a talk a couple of years
               | back about reverse engineering tamagochis, and they
               | turned out to run on some 6502 compatible chip made by a
               | company that - IIRC - sells to toy makers mostly.
        
           | babypuncher wrote:
           | Skynet is due to start producing their T-800 line of
           | Terminators in 2026, which will use a 6502-derived CPU.
        
         | trashtester wrote:
         | .... until the end of the human race as we know it.
         | 
         | I think this is the critical part. If humanity (as we know it)
         | only lasts 10 more years, then sure x86 will still be around
         | somewhere.
         | 
         | If we last a million years, it will probably be gone long
         | before that. Even in a thousand years it's probably gone a long
         | time ago.
        
           | postmodest wrote:
           | I'm reminded of the Vernor Vinge novel where a character
           | hacking some fleet's automation hundreds of thousands of
           | years in the future casually mentions that the tech stack is
           | so old that the system time stamp is still the Unix Epoch
           | 
           | "And via a million million circuitous threads of inheritance,
           | many of the oldest programs still ran in the bowels of the
           | Qeng Ho system. Take the Traders' method of timekeeping. The
           | frame corrections were incredibly complex--and down at the
           | very bottom of it was a little program that ran a counter.
           | Second by second, the Qeng Ho counted from the instant that a
           | human had first set foot on Old Earth's moon. But if you
           | looked at it still more closely...the starting instant was
           | actually about fifteen million seconds later, the 0-second of
           | one of Humankind's first computer operating systems."
        
             | skywal_l wrote:
             | Part of the "Fire Upon the Deep" series of novels (that
             | particular reference is from the "A Deepness in the Sky").
        
           | tetha wrote:
           | I mean a thousand years are very hard to imagine, and how
           | many changes there are.
           | 
           | But the cynical operator in my head could just laugh. We as a
           | tech community are still running MS-DOS productively. Just
           | wait, someone will run the door controls of our first space
           | ships on some x86 chip. Or some similar system you just need,
           | but that never gets time to be updated properly. Just wait,
           | the new cruise liner spaceship of the milky way republic is
           | going to run some x86 emulator for their window control.
        
         | FirmwareBurner wrote:
         | _> chips like the 6502 have still been in active use (almost 50
         | years now)_
         | 
         | Also 8051 cores can still be found in modern products
        
           | pclmulqdq wrote:
           | 8051s are just now getting phased out as control cores of IP
           | blocks, which is pretty wild.
           | 
           | The 32-bit ARM and RISC-V cores are small enough and easier
           | to program.
        
         | Symmetry wrote:
         | Plausibly we're headed for a world where feature size decreases
         | stall out but manufacturing improvements continue to lower the
         | price of transistors over time. In a world like that throwing
         | in a few x86 cores even if the dominant ISA shifts might be
         | worth it from a backwards compatibility standpoint even if
         | other ISA become dominant.
         | 
         | There's lots of complications to address there (strict x86
         | memory ordering versus loose ARM ordering, for instance) but I
         | expect they're solvable.
        
           | tracker1 wrote:
           | IIRC, AMD's model for chiplets could already combine x86 and
           | ARM as an example, not sure if any such beasts are in
           | existence, but should be possible.
        
             | Symmetry wrote:
             | Yeah, I'm actually thinking something like a socket
             | integrator buying an 8 core chiplet from AMD or Intel to
             | connect via CXL or whatever to the ARM/RISC-V/GPU/Tensor
             | cores in the other chiplets from different manufacturers.
        
       | mobilio wrote:
       | I've recently switch from VPS with Intel to physical server with
       | latest AMD Zen4.
       | 
       | Single thread performance blow my mind with scores like 4000.
       | 
       | Without change a single line of code = performance was 10x than
       | before.
        
         | jsheard wrote:
         | 4000 in what benchmark? I don't doubt that bare-metal Zen4 is
         | fast but that number doesn't mean anything by itself.
        
           | aidenn0 wrote:
           | Yeah, I read that and just saw Vegeta saying "it's over 8000"
        
             | jeffmcjunkin wrote:
             | Ahem, that's _9000_.
        
               | aidenn0 wrote:
               | That's a mistranslation:
               | https://en.m.wikipedia.org/wiki/It%27s_Over_9000!
        
               | kingrazor wrote:
               | 9000 in the first few English dubs. 8000 in the manga and
               | DBZ Kai.
        
           | mobilio wrote:
           | I'm checking PassMark:
           | 
           | https://www.cpubenchmark.net/cpu.php?cpu=AMD+Ryzen+5+7600&id.
           | ..
        
         | papichulo2023 wrote:
         | Arent VPS cpus usually clocked pretty low? Also if you are
         | sharing the cpu with another intensive service, the cache will
         | miss a lot.
        
         | bombcar wrote:
         | It's really important to benchmark this stuff, because
         | depending on the workload, a cheap VPS is the way to go (you're
         | almost always idle, say) or getting some of the latest and
         | greatest hardware under your desk can quickly pay for itself.
        
         | distcs wrote:
         | > Single thread performance blow my mind with scores like 4000.
         | 
         | What exactly is this number 4000? What does it mean? Where can
         | I read more about this scoring system?
        
           | mobilio wrote:
           | I'm checking PassMark:
           | 
           | https://www.cpubenchmark.net/cpu.php?cpu=AMD+Ryzen+5+7600&id.
           | ..
        
           | c-hendricks wrote:
           | Assuming they mean Geekbench.
        
       ___________________________________________________________________
       (page generated 2023-07-31 23:02 UTC)