What percentage of iMac buyers & Mini buyers choose those because it's the only option vs that's what they really wanted?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
New References to AMD Processors Discovered in macOS 10.15.4 Beta
- Thread starter MacRumors
- Start date
- Sort by reaction score
If they use a fab with a license they should be fine (though I haven’t seen the agreements). Still a few fabs out there (not TSMC, though, I believe).
Registered / ECC Ram is not a problem with Threadripper and the TRX40 chipset. While it's currently limited to 512GB using 8x DDR4-2933 LRDIMM ECC REGISTERED dimms. As soon as 256gb parts come out of sampling, 1TB will be possible.
[automerge]1581215944[/automerge]
That motherboard has been shipping for many months. Reading between the headlines... It was just certified by intel.
Most of what you're saying, I agree with, except for one nit:
Intel never forgot the importance of backwards compatibility-- they had hardware support for it.
Another key feature of the Itanium processor is its full support of the IA-32 instruction set
in hardware (see Figure A). This includes support for running a mix of IA-32 applications and
IA-64 applications on an IA-64 operating system, as well as IA-32 applications on an IA-32
operating system, in both uniprocessor and multiprocessor configurations. The IA-32 engine
makes use of the EPIC machine’s registers, caches, and execution resources.
My point to @RalfTheDog is that this was an example of Intel itself trying to support x86 on a VLIW architecture and it caused significant pain. Backwards compatibility consumed 30% of the die and was a significant source of delay in the release of the chip. Even with all of this hardware support, it ran x86 code at about the speed of a 486.
Eventually, Intel gave up and released the IA-32 Execution Layer, which sounds like hardware but was a software translation and JIT cross-compiler to bring IA-32 instructions into the IA-64 world. This was a similar approach to that used by Transmeta, another VLIW architecture on the market at the time that led to similar results. Still, the IA-32 EL at least (and rather embarassingly) turned out to be faster than dedicated hardware.
There's nothing magic about VLIW in accelerating legacy instruction sets. Even Intel couldn't do it well. Sure compilers are smarter now, but the VlIW problem is still hard and better compilers can just as easily transcode to ARMs native instruction set.
The reason AMD swooped in with x86-64 and somehow managed to make Intel look like fools while simultaneously saving them from doom isn't because Intel forgot backwards compatibility, it's because they forgot they had hungry competitors. Itanium failed because of Intel's arrogance in assuming the world would just fall in line because they declared IA-64 the future, failure to execute and years of delay and performance failures, and completely broken internal management and product development around the project that led to constantly rotating management and a culture of fear around raising concerns and admitting problems.
Itanium killed a few good architectures, DEC's Alpha most sadly, but AMD took a much less ambitious approach to 64 bit and won the support of Microsoft in the end.
And it wasn’t like we were planning this all along. We had a much more ambitious version of K8 that had been started, but had to be abandoned when a huge percentage of the design team quit. Fred Weber realized that given the small team we had, given that we had no license to Itanium’s architecture, and given that nobody would likely buy our own competing “clean paper” architecture if there was no software for it, we did pretty much the only thing we could, and extended x86 to 64 bits, played super-nice with Microsoft and tried to accommodate whatever their wish list was, split the chip design up for 8 or so physical design leads, and went to work.
5nm Technology - Taiwan Semiconductor Manufacturing Company Limited
TSMC’s 5nm (N5) Fin Field-Effect Transistor (FinFET) technology successfully entered volume production in the second quarter of 2020 and experienced a strong ramp in the second half of 2020.
It's not if, but what quarter in 2020. We already know Zen 3 is Q3 so put money that 5nm starts then.
[automerge]1581221744[/automerge]
Whether a motherboard has 12 LPDRAM DDR4 slots, or they put in 8 or 16 is up to the manufacturer. Intel's Mac boards with 1.5TB capacity are custom designed 12 slot memory motherboards created to Apple's specifications.
I don't think that would have worked. I remember reading something a long while back that Intel owns the x86 instruction set rights, and they license it to AMD. I believe somewhere in the licensing agreement it states that the license to build x86 CPU's will not transfer to any new owner, so if Apple bought AMD, they would not get that license. I don't know if this is or was true though.
The Ryzen 5 4500U can be used for Apple MacBook Air and completely fanless. The Ryzen 3 4300U is for the new entry level $999 12” Macbook.
Last edited:
Threadripper supports only U-DIMM. It doesn't even support both R-DIMM and LR-DIMM. You must be mistaken.
[automerge]1581237924[/automerge]
It cant. 15W is still high for Macbook Air. TDP for Intel Y is only 9W. Having 15W for MacBook Air will increase more heat and less performance.
Apple can request for lower TDP from AMD that shouldn't be a huge difference in performance and more reasonably priced than Intel.
Then what's the point of having AMD APU at 9W? I highly doubt it and MBA will have ARM chip instead.
If the next 3 months is not "anytime soon" then you are right.
If they live up to AMDs claims (which they did with their Desktop offerings last year), they are going past Intel now. With double the Cores.
It will be very hard for Intel to catch up if they don't get their act together. AMD is profiting from the aggressive TSMC release cycle. Which basically looks like this and they have never failed so far:
TSMC for Apple and some other Smartphones:
2019: 7nm+
2020: 5nm
2021: 5nm+
2022: 3nm
TSMC for AMD (once the capacity is freeing up):
2019: 7nm
2020: 7nm+
2021: 5nm
2022: 5nm+
2023: 3nm
Keep in mind that the Mobile CPUs from AMD are always a bit behind their Desktop offerings, by about 6-8 months.
Last edited:
they can be configured between 12-25W.
but TDP is not an official standard anyway and AMD and Intel measure them differently and they can't really be compared directly.
the ryzen 4000 u-series were designed for ultrathin laptops, so it shouldn't be a problem for someone like apple to implement them.
Edit: in their presentation they said 12-25W.. but their website says 10-25W... weird.
Last edited:
Apple breaking up with Intel is not as far stretched.
1. They already bought the smartphone modem department.
2. ThunderBolt is becoming a standard, aren't they putting the controller into t2-chip?
3. AMD is wiping the floor with Intel on everything but very low powered, but they might just start doing it as well
Mac Pro: Epyc is not that expensive, for the premium you pay for it Apple can afford going with that for better memory support.
Updating MacOS: It might mean redesigning some products, updating kernel a lot, redoing a lot of optimizations etc. MacOS does already runs on Ryzen (Hackintosh).
Everything else: AMD does have a very good offering ATM at all levels already or in the near future. As Mac is becoming a less important platform, Apple might just pull it off.
But... It might just be an Intel CPU with inbuilt AMD graphics or something like that.
I don't think Apple is going for ARM any time soon for Macs. I think the whole idea that ARM is the future is by extending the capabilities of the iOS platform and taking market shares from the low-end of Macs. But tbh, it hasn't been gaining as much ground as one might expect. Nearly all software is limited in its capabilites on iOS, not to mention worse I/O, multitasking, peak load etc.
1. They already bought the smartphone modem department.
2. ThunderBolt is becoming a standard, aren't they putting the controller into t2-chip?
3. AMD is wiping the floor with Intel on everything but very low powered, but they might just start doing it as well
Mac Pro: Epyc is not that expensive, for the premium you pay for it Apple can afford going with that for better memory support.
Updating MacOS: It might mean redesigning some products, updating kernel a lot, redoing a lot of optimizations etc. MacOS does already runs on Ryzen (Hackintosh).
Everything else: AMD does have a very good offering ATM at all levels already or in the near future. As Mac is becoming a less important platform, Apple might just pull it off.
But... It might just be an Intel CPU with inbuilt AMD graphics or something like that.
I don't think Apple is going for ARM any time soon for Macs. I think the whole idea that ARM is the future is by extending the capabilities of the iOS platform and taking market shares from the low-end of Macs. But tbh, it hasn't been gaining as much ground as one might expect. Nearly all software is limited in its capabilites on iOS, not to mention worse I/O, multitasking, peak load etc.
Yes it is that right:
And remember: Cavium uses the same ARM v8 architecture, as Graviton 1.
The post I was replying to stated that there are no ARM-based supercomputers...the link I posted stated "Cray to offer supercomputers with Fujitsu A64FX Arm processors from 2020"...so how is what I said incorrect?
The links you posted show that ARM doesn't compete core for core against other platforms and I never debated that! But it is factually wrong to say that no supercomputers exist based on the ARM platform when Cray (arguably one of the best know supercomputer manufacturers) is building them!
Threadripper is not compatible with RDIMM or LRDIMM ECC memory, only UDIMM ECC. If you want to use RDIMMs or LRDIMMs you have to use EPYC parts.
I'm inclined to agree in the short term - the message from the last WWDC with "iPadOS", catalyst and the emphasis on how super-powerful the new iPad Pros are seems to be that they're trying to push the iPad Pro into low-end Mac territory.
I just think that a hypothetical switch to ARM is more feasible than people think - given the Mac platform is already on its third radically different CPU architecture and second radically different OS, and those past changes happened when there was far more assembly language and CPU-dependent C code around.
...but I'm not holding my breath!
Thing is, I think a range of nice ARM-based personal computers running Unix (i.e. MacOS) would be an interesting development in its own right. There's a lot of interest in using ARM for servers etc. at the moment, but there's not a lot of widely-available, well supported consumer hardware beyond the Raspberry Pi (which is terrific, but lacks basics like SATA/PCIe and drives everything through a slightly shonky USB controller).
I don't know, maybe they have something in the works that will be adequate. They've certainly been kicking arse recently, and it's always great to have competition in the field. And with Intel getting into the GPU field as well I'm hopeful that we will soon see a *major* price/performance war in both the discrete and integrated CPU and GPU markets. That would be a very good thing for consumers. AMD CPUs in Apple hardware definitely makes a lot of sense right now. At the very least having those designs ready to go will allow Apple to choose whichever chips are best value for performance at the time.
Nobody has mentioned Apple's move to LLVM bitcode/IL which - if I'm not mistaken, would allow for most applications built in objective c or swift to be be compiled to bitcode - and from there to any one of several backed ISAs (PowerPC G5...)?
Shouldn't Apple be well positioned to work over both ARM and x86-64? And shouldn't that not be a factor for AMD in that both AMD and Intel apps are both compiled to x86-64? The only issue I've seen between the two (from 'ryzentosh' discussions) is that this falls down with apps like Docker Desktop that's tightly coupled to Intel's Hyper-V....
Shouldn't Apple be well positioned to work over both ARM and x86-64? And shouldn't that not be a factor for AMD in that both AMD and Intel apps are both compiled to x86-64? The only issue I've seen between the two (from 'ryzentosh' discussions) is that this falls down with apps like Docker Desktop that's tightly coupled to Intel's Hyper-V....
I'm aware of that but Y series can go lower. You know that Intel has both Y and U series right?
[automerge]1581274675[/automerge]
You still dont understand. Y series can go lower than 9W. This is not possible with APU at 15W