|
Lambert posted:Pretty sure you're wrong and Apple has a poo poo app store ecosystem. Have fun using your lovely free operating systems I guess??
|
#
?
Jul 23, 2019 00:59
|
|
|
|
| # ? May 30, 2024 13:46 |
|
|
BangersInMyKnickers posted:Have fun using your lovely free operating systems I guess?? lol @ the Apple fanboy trying to censure me. There have been plenty of complaints about the mac app store by developers in the past few years, those weren't exactly obscure. Some even pulled out, resorting to selling their apps on their own websites only.
|
|
#
?
Jul 23, 2019 01:00
|
|
|
Lambert posted:lol @ the Apple fanboy trying to censure me. well that's a great point I guess they should be like the other good app stores such as __________________ and ______________________.
|
|
#
?
Jul 23, 2019 01:02
|
|
|
lol at people thinking that if Apple moved to ARM they wouldn't just put a 20W TDP chip or something insane like that into their desktop. The A12X is already faster than a lot of X86 processors and it's only at 7W TDP.
|
|
#
?
Jul 23, 2019 01:04
|
|
|
BangersInMyKnickers posted:well that's a great point I guess they should be like the other good app stores such as __________________ and ______________________. fair enough.
|
|
#
?
Jul 23, 2019 01:07
|
|
|
ratbert90 posted:lol at people thinking that if Apple moved to ARM they wouldn't just put a 20W TDP chip or something insane like that into their desktop. The A12X is already faster than a lot of X86 processors and it's only at 7W TDP. 8 cores, of which only 4 cores can reach a clock rate* of 2.49GHz. Yes, that's "a lot" of x86 processors  *clock rate isn't everything, but come on.
|
|
#
?
Jul 23, 2019 01:16
|
|
|
ratbert90 posted:lol at people thinking that if Apple moved to ARM they wouldn't just put a 20W TDP chip or something insane like that into their desktop. The A12X is already faster than a lot of X86 processors and it's only at 7W TDP. I sincerely doubt that all the electrical and thermal overhead in x86 CPUs is all due to terrible inefficiencies all over the place.
|
|
#
?
Jul 23, 2019 01:18
|
|
|
iospace posted:8 cores, of which only 4 cores can reach a clock rate* of 2.49GHz. Yes, that's "a lot" of x86 processors iospace posted:8 cores, of which only 4 cores can reach a clock rate* of 2.49GHz. Yes, that's "a lot" of x86 processors A12 ipc is as good as if not better than intel and is fabbed on low power mobile processes yet still manages to embarrres intel processors If Apple wanted it could fab a big perf core but it's certainly not cheap so they just get intel to do it
|
|
#
?
Jul 23, 2019 01:21
|
|
|
Mobile core designs target lower clocks and efficiency but I was surprised to see this design running A72s at 4GHz. https://www.google.com/amp/s/wccftech.com/tsmcs-custom-built-octa-core-a72-chip-reaches-4ghz-at-1-20v/amp/
|
|
#
?
Jul 23, 2019 01:26
|
|
|
How can you compare IPC between two vastly different instruction sets?
|
|
#
?
Jul 23, 2019 01:26
|
|
|
I still doubt all of it. Neither Intel nor AMD are loving idiots. x86 is pretty much just the front-end, internally, they're basically just RISC CPUs like the ARM is. And no, the decoder isn't as power hungry as it keeps getting claimed.
|
|
#
?
Jul 23, 2019 01:34
|
|
|
How does ARM compare to 32/64 (x86?) bit processors? Are those just the weird soldered on mobile things like android's snapdragon or apple's A series chips on their phone/laptops? Would gaming eventually go to ARM or is that strictly for very basic computing at the moment?
|
|
#
?
Jul 23, 2019 01:35
|
|
|
buglord posted:How does ARM compare to 32/64 (x86?) bit processors? Are those just the weird soldered on mobile things like android's snapdragon or apple's A series chips on their phone/laptops? Would gaming eventually go to ARM or is that strictly for very basic computing at the moment? Currently yes, it's used almost exclusively in embedded/mobile applications. But ARM is just an instruction set, there's no reason that you couldn't build an ARM CPU that is just as high performance as a top of the line x86 CPU. It's just that nobody is making significant R&D effort in that direction because rewriting software to run on a different platform is nontrivial*. I've no doubt it'll happen eventually though *A lot of applications could probably get ported from x86_64 to ARM64 by just recompiling but then you have stuff that might directly check for and use SSE/AVX and other special instructions. hypervisors would also need some effort and those are the bedrock of modern datacenters. Then you have things that might not be necessarily incompatible but just assumed something that differs between the two platforms. My intuition as an engineer is that there's going to be a lot of random poo poo nobody ever thought about
|
|
#
?
Jul 23, 2019 01:48
|
|
|
VostokProgram posted:Currently yes, it's used almost exclusively in embedded/mobile applications. Arm servers have been popping up here and there for things that need moderate amounts of compute or can thread incredibly well (AES encryption for TLS concentrators and whatnot) and they have had some very good results for those use cases, the issue has been mostly compiler optimizations around x86 that need to be re-worked for arm
|
|
#
?
Jul 23, 2019 02:09
|
|
|
Going back to AMD, do I need an X570 mobo to get the memory compatibility improvements? The older chipsets are a lot cheaper but it almost isn't worth it if I have to fight it to get the RAM working at a higher speed. I just want to use 3200CL16 with no hassle.
|
|
#
?
Jul 23, 2019 02:28
|
|
|
For what it's worth, I have a cheaper MSI x470 with a 3900x and I just hit the XMP profile for 3200 with no issues.
|
|
#
?
Jul 23, 2019 02:35
|
|
|
Lambert posted:I'm all about Apple destroying the last remnants of their desktop user base, hope they go for it. It won’t. Apple has a ton of experience switching between architectures. They are already slow rolling in apps that are ARM based onto their desktop and laptop platforms. In addition they have parts of their machines interoperating between x86 and ARM and more and more functionality will fall under the ARM chips and less on the x86 cpu. Apple switching to ARM on their desktop class hardware will hurt Intel more than it hurts themselves.
|
|
#
?
Jul 23, 2019 02:51
|
|
|
Let’s also not forget that Apple has, for more than a few years now, the ‘bitcode’ compilation model. Which is their now required way you must submit apps to the App Store. And this compilation model makes no assumptions about the end target architecture. They can freely move an app between x86 or ARM on the backend when a user requests a download and the user is completely oblivious to it. There is no reason this kind of technology can’t be used to port desktop class software to ARM. No emulation required. They wouldn’t even need developers to do anything except compile to Apple’s bitcode instead of to x86 or ARM directly. xgalaxy fucked around with this message at 03:04 on Jul 23, 2019 |
|
#
?
Jul 23, 2019 03:01
|
|
|
My B450m mortar is running 3200cl16 fine.
|
|
#
?
Jul 23, 2019 03:28
|
|
|
Lambert posted:lol @ the Apple fanboy trying to censure me. True, but it's hard not to see the writing on the wall wrt Marzipan/Catalyst/'bitcode' and Apple's ongoing work to bridge Mac software development and development for iOS/iPad OS. Apple doesn't need a robust or even viable Mac app store to facilitate an architecture switch. They didn't have one when they switched from PPC to Intel (or from Motorola to PPC). The fact that they have arguably the gold standard in the iOS app store makes it much easier in some ways.
|
|
#
?
Jul 23, 2019 04:01
|
|
|
iospace posted:"ARM taking over x86" is the next "tablets are going to replace laptops" I'm not going to make any predictions about when any architecture will supplant x86. I will say that the major players who drive the CPU market through datacenter purchasing constantly evaluate new processors (and these days spin their own, though so far that has remained a special-purpose case). As soon as anything beats x86 on the cost/performance/density matrix (and I'm including power in "cost"), it will start replacing x86 CPUs as upgrades roll through the datacenters. It's about money, and nothing else. Paul MaudDib posted:And yeah you can nominally use ARM or RISC-V, if you don't mind throwing away all your software and rewriting everything. The only way I could see that working is if someone like Amazon subsidized the poo poo out of development and instance costs of a competitor. But realistically it's just not going to happen, legacy codebases are a kind of natural monopoly and it's just not cost-effective (or even possible) for a competitor to re-write everything. This is a red herring for a lot of use cases. Not all, but a lot. Like, maybe it would be a problem for Adobe, if they're still doing heavy assembler optimization. And it's a problem if your OS doesn't already run on other_arch, but pretty much everything that isn't Windows does. If your code is in Perl or Python or Ruby, or any such thing, you don't give a poo poo about the underlying machine. Your code is JITted to the language runtime's VM. If you're writing in C or C++, you've kept your code nice and portable, and you'll be fine as soon as the arch is added to your compiler (gcc, clang, LLVM), right? And just as an ultra-contrary example, if you're writing in Go, you don't even need a cross-compiler environment to produce binaries for any supported OS, on any supported arch. I've built binaries for deployment to Linux servers, on my Macbook running OS X (which, granted, is also x86_64), just by doing '$ GOOS=linux GOARCH=amd64 go build'. I could build that for my Raspberry Pi by switching it to 'GOARCH=arm64'.
|
|
#
?
Jul 23, 2019 04:48
|
|
|
I'm pretty excited about RISC-V, between Qualcomm, Google, and Western Digital investing pretty heavily, that's a lot of silicon sway. There have been some pretty major Linux advances in the platform recently, too. With all the simultaneous emulation/compatibility layer advances even in the last year, it's all pretty exciting. I think if there's any takeaway from AMD in the last few years, it's that momentum only goes so far. It's also more likely for RISC-V to take over than for US copyright law to get overhauled 
|
|
#
?
Jul 23, 2019 04:50
|
|
|
uhhhhahhhhohahhh posted:lot's of people are having similar issues. some people are hitting 60c while idling from it - but the main focus has been on monitoring programs causing it. messing with the power plans and disabling pbo fixes it for some people from what i've seen, but it seems bios and setup dependant. the AMD guy who posts on their subreddit has said it's normal and not worry about it but that theyre also looking into. he's supposed to be posting an update about it this week. Thanks, guess I'll look out for updates to the BIOS / chipset. I was this close to doing a completely fresh Windows install. I've not had any other issues but maybe I should anyway, it's been a while. I've been having high idle temperatures as well and was worried I'd hosed up the cooler installation somehow so glad to hear it's not that. Going to do a bit of a cable tidy when I have a spare hour as it won't hurt. Party Boat fucked around with this message at 05:49 on Jul 23, 2019 |
|
#
?
Jul 23, 2019 05:46
|
|
|
mdxi posted:If you're writing in C or C++, you've kept your code nice and portable lmao
|
|
#
?
Jul 23, 2019 07:20
|
|
|
mdxi posted:I'm not going to make any predictions about when any architecture will supplant x86. I will say that the major players who drive the CPU market through datacenter purchasing constantly evaluate new processors (and these days spin their own, though so far that has remained a special-purpose case). Lol
|
|
#
?
Jul 23, 2019 07:33
|
|
|
mdxi posted:And it's a problem if your OS doesn't already run on other_arch, but pretty much everything that isn't Windows does. Windows does run on ARM. You can buy laptops with Snapdragon CPUs that run Windows right now.
|
|
#
?
Jul 23, 2019 08:11
|
|
|
Lambert posted:Windows does run on ARM. You can buy laptops with Snapdragon CPUs that run Windows right now. It's not something that comes up in the era of x86 dominance, but Windows NT's architecture independence was a big deal in the 90s. You could get Windows for Alpha, PowerPC and MIPS up through NT 4, and 2000(I think), XP, and the Vista/7 era server releases supported Itanium. The aspects of Windows' design that enabled that didn't go away in Vista/7/etc, there just wasn't a non-x86 architecture worth running a desktop OS on during that time. Theris fucked around with this message at 12:12 on Jul 23, 2019 |
|
#
?
Jul 23, 2019 12:04
|
|
|
NewFatMike posted:I'm pretty excited about RISC-V, between Qualcomm, Google, and Western Digital investing pretty heavily, that's a lot of silicon sway. As someone who just got started in the RISC-V ISA, I'm...not so excited about it anymore. At least, speaking from the perspective of an ASIC team member trying to branch some hardware core IP off of the few existing cores that currently have RTL up on GitHub. With the current amount of documentation actually released to the public, as well as the behavior of certain automatic RTL generation tools, there's not as much modularity to customize and portability to FPGAs as has been implied, to say the least. There also isn't the sort of widespread community out there to ask for help at all, as it is with Linux. It's almost as if  you have to commit to joining the RISC-V Foundation to be in the in-crowd or pay up $$$$$$$ to be a SiFive/Andes/Codasip customer so they can do it for you...much like an ARM-license holding vendor wouldI get that the Foundation is trying to protect itself from exploitation, but the way the reality has played out is that it's still not really the sort of thing an average joe can innovate with on his own FPGA from Digi-Key. We've had to, uh, reverse-engineer a few things just to have an idea of what's inside the box inside the box, and what the box is made of, so to speak. edit: added the word "RISC-V" to ISA just to be clear even though it was implied by my quote! Sidesaddle Cavalry fucked around with this message at 12:32 on Jul 23, 2019 |
|
#
?
Jul 23, 2019 12:23
|
|
|
Lambert posted:Windows does run on ARM. You can buy laptops with Snapdragon CPUs that run Windows right now. They've already began executing some strategies to mitigate this. They've been getting in bed with open source projects, started integrating linux into win10, have some on the surface pro-consumer policies going on, aren't actively pursuing consumers who purchase grey market keys, are trying to make their office stack as portable as google docs, and have an x86 emulation layer on ARM. They're also diversifying big time. The current CEO is doing a pretty good job of preparing for the future and trying to rebuild their image, although win10's QA is an ongoing disaster for non-business users. Even for gaming, Valve is already "at war" with Microsoft over UWP and Microsoft's store. They've been actively developing compatibility layers for playing windows games on linux for quite a while now, and as developers migrate to fewer and fewer mainstream engines many games support it out of the box. A switch from x86 to another architecture is all a big company needs to wedge themselves into the desktop space. Khorne fucked around with this message at 00:14 on Jul 24, 2019 |
|
#
?
Jul 23, 2019 13:21
|
|
|
sth sth RISCailure-V The official mailing lists are, at best, worthy of the "idiots on social media" thread. The foundation seems to be a mediocrity committee, focusing in their vector ISAlure. Their virtualization ISA sketch had to be fixed by a guy posting on the mailing list, reminding them not to copy the broken v1 VM ISA ARM had to fix in their v2 VM extension after linux people told ARM to pull on their finger. They can't finalize their bit mangling extension, surely not useful for the embedded micro controller market they claim to target. They are gating stuff like crypto behind their craptastic vector ISA only the single PhD, and maybe his advisor, give a gently caress about. I'm not familiar with ARM's SVE but I think it shows you can do scalable SIMD, no need for going all hipster on Mr. Cray's 80s ISAs.
|
|
#
?
Jul 23, 2019 14:45
|
|
|
Sidesaddle Cavalry posted:As someone who just got started in the RISC-V ISA, I'm...not so excited about it anymore. At least, speaking from the perspective of an ASIC team member trying to branch some hardware core IP off of the few existing cores that currently have RTL up on GitHub. With the current amount of documentation actually released to the public, as well as the behavior of certain automatic RTL generation tools, there's not as much modularity to customize and portability to FPGAs as has been implied, to say the least. There also isn't the sort of widespread community out there to ask for help at all, as it is with Linux. It's almost as if That's somewhat disappointing. I guess it's SiFive's business model to help companies implement the ISA, so there's some perverse incentive to keep it somewhat opaque. How much of it do you think is teething issues from its relatively young stage? They just ratified the 1.0 spec a little while ago, right?
|
|
#
?
Jul 23, 2019 14:45
|
|
|
The Scottish dude is done with AMD and won't release any more (public) speculation videos: https://www.youtube.com/watch?v=kU5h0MYpmfg
|
|
#
?
Jul 23, 2019 16:19
|
|
|
Theris posted:It's not something that comes up in the era of x86 dominance, but Windows NT's architecture independence was a big deal in the 90s. You could get Windows for Alpha, PowerPC and MIPS up through NT 4 Well, those last two were around for about two seconds, the only one that had any sort of real success/sales was Alpha. It was certainly intended to be a big deal because back at the time nobody thought Intel could keep up with the RISCs, but in actually practice welp turns out they could.
|
|
#
?
Jul 23, 2019 16:30
|
|
|
Mr.Radar posted:The Scottish dude is done with AMD and won't release any more (public) speculation videos: And nothing of value was lost.
|
|
#
?
Jul 23, 2019 16:41
|
|
|
Hilariously he puts the blame for the inaccuracy of his "leaks" on AMD, for not immediately refuting them and for ~misleading~ marketing slides
|
|
#
?
Jul 23, 2019 16:48
|
|
|
He was the main 3900X 5ghz leaker
|
|
#
?
Jul 23, 2019 16:52
|
|
|
NewFatMike posted:That's somewhat disappointing. I guess it's SiFive's business model to help companies implement the ISA, so there's some perverse incentive to keep it somewhat opaque. Version 2.0 for the basic integer stuff, and the general use extensions (i.e. multiply and divide, floating point) plus compressed instructions. Side note: base integer set for the 32-bit 16-register embedded style isn't there yet at v.1.9. Vector math and SIMD are the unfrozen mediocrity as karoshi posted. And that's just been the programming side of things. I'd love to draw more conclusions about the way the UCB Rocket team has done things, but I still don't feel confident enough about the Chisel and Scala hardware model config infrastructure (as well as the TileLink interface) to comment more. It's just been so arcane and it certainly feels like right now that the the groups of people who can support it are already consolidated behind paywalls. Western Digital's SweRV core is also on GitHub and is a little more transparent in SystemVerilog, but their RTL maker has a bizarre behavior with flip-flops that makes pushing to an FPGA not ideal. Haven't investigated more as to why yet.
|
|
#
?
Jul 23, 2019 17:14
|
|
|
Mr.Radar posted:The Scottish dude is done with AMD and won't release any more (public) speculation videos: 41 loving minutes
|
|
#
?
Jul 23, 2019 18:01
|
|
|
alreet guys, i'll be going
|
|
#
?
Jul 23, 2019 18:04
|
|
|
|
| # ? May 30, 2024 13:46 |
|
|
It's ya boi, reeeet now.
|
|
#
?
Jul 23, 2019 18:17
|
|