therunningvm

cloudguy said:

saarek said:

I hope this isn’t true.

I’m hoping for a true performance king that humiliates AMD & Intel and sets the bar in terms of performance. Primarily GPU based upgrades over the M1 wouldn’t be the big step up in the true “Pro” Macs over the current M1 that I was hoping for.

With all due respect why do you believe that this is even possible? As I have stated numerous times, the idea that ARM is inherently superior to x86 was wishful thinking. If it were true, ARM would have more than 3% of the server market. As I have also stated, most of the benchmarking was skewed: it only compared the M1 to the Intel chips that it replaced in macOS devices. Those were mostly 2 and 4 core "mobile" chips. They were also outdated chips: 9th and 10th gen. There were already 11th gen Intel chips on the market when the M1 Macs were introduced.

 Yet all the M1 crushes Intel who is now doomed! "benchmarks" ignored them just as they ignored how comparing 2 and 4 core "mobile" chips to a chip with 4 performance + 4 efficiency cores never made any sense. They just took Apple's claim - unsubstantiated by any data - that the M1 was faster than "80% of Windows PCs" and ran with it. Also, 4Q this year we are going to see 10nm big.LITTLE chips from Intel followed by 5nm Zen 4 chips - Athlon architecture, not big.LITTLE - chips from AMD. Intel hasn't started hyping their 12th gen chips yet, but AMD is claiming that their Zen 4 chips will have up to 40% performance gains over their current chips. 

Thinking that Apple was going to dominate Intel and AMD in PCs the way they dominate Qualcomm and Samsung in mobile never made any sense. Especially if it was based on the superiority of ARM because Qualcomm and Samsung make ARM chips too. Apple versus Qualcomm was "ARM CPU with laptop performance versus ARM CPU with embedded appliance performance." Fine. Apple versus Intel and AMD: ARM CPU with laptop performance versus x86 CPUs that power 97% of the world's servers. A totally different ballgame.

This has nothing to do with Isa or even big.little. One apple firestorm core matches or beats Ryzen 3 in SPECCint 2006, an industry standard benchmark. I'm not sure why you bring up big.little, considering it doesn't change the fact apple's core design is superior.

Oh sure, AMD will have zen 4 by the end of the year, but that's going to be just on paper. AMD still can't ship in any meaningful quantities, given their mobile chips barely compete in the high end of laptops. That's still dominated by intel. Both fail at sustaining clocks, let alone keep power draw at a reasonable level.

You also seem to believe that x86 being dominant in servers somehow means they're still superior to anything ARM, which is quite absurd. Go look at some of Ampere's server chips.

zimmie said:

therunningvm said:

zimmie said:

I wonder how they are able to load the kernel image at all. The tutorial should answer, but from the mention of pongoOS, I suspect they are exploiting some flaw in the secure boot process itself to load an arbitrary image and pass execution to it.

Reading through the commit for DART support, it looks like Apple did indeed design their own IOMMU core rather than using ARM's reference SMMU.

Apple opened up the tooling to load custom kernels on Apple Silicon.

When was that? Last I heard (end of November), they said you could boot any signed version of macOS, but that they would not support booting other operating systems directly on the hardware. They said everything had to run through the hypervisor framework.

Murvel said:

Hardly upsetting. Apple does this type of thing all the time.

keithw said:

I'm finding it extraordinary difficult to believe that Apple could replace the existing Xeon-based Mac Pro (or even iMac Pro) with comparable performance & expandability in only two years.  Low end, no problem at all.