Path: csiph.com!eternal-september.org!feeder.eternal-september.org!reader01.eternal-september.org!kreme.eternal-september.org!.POSTED!not-for-mail From: Lewis Newsgroups: comp.sys.mac.system Subject: Re: Explore the new system architectire of Apple Silicon Macs Date: Tue, 28 Jul 2020 00:31:56 -0000 (UTC) Organization: Miskatonic U Lines: 71 Message-ID: References: <210720202009353973%nospam@nospam.invalid> <230720200010573127%nospam@nospam.invalid> <230720201027217465%nospam@nospam.invalid> <%9pSG.19303$CM.5449@fx01.iad> <9YHSG.107292$I15.71812@fx36.iad> <_I1TG.117751$%p.22585@fx33.iad> <250720202316531185%nospam@nospam.invalid> <260720201459085729%nospam@nospam.invalid> <74mTG.53276$BL.26025@fx16.iad> <260720201949019297%nospam@nospam.invalid> <270720200009035410%nospam@nospam.invalid> Injection-Date: Tue, 28 Jul 2020 00:31:56 -0000 (UTC) Injection-Info: kreme.eternal-september.org; posting-host="19acfbaf95b8873b07a6356863bf642f"; logging-data="24965"; mail-complaints-to="abuse@eternal-september.org"; posting-account="U2FsdGVkX18OT1fvzl2lsecoK3ZBJnMQ" User-Agent: slrn/1.0.3 (Darwin) Cancel-Lock: sha1:+LTNIe0OIecJnz3ySTsVP8GPaDQ= X-Face: )^b5"R:T7U>9~:PEn3YkzMfW*[b1qKeU.fP9C8~8HpU9}lA&6`bH1z X-Clacks-Overhead: GNU Terry Pratchett Mail-Copies-To: nobody Xref: csiph.com comp.sys.mac.system:133818 In message JF Mezei wrote: > On 2020-07-27 00:09, nospam wrote: >>> So you agree there is commonality. >> >> nowhere near the level you claim. >> >> the instruction set is common. beyond that, very little. > So you no longer agree about shared core designs? No one ever agreed with that. You need to learn to read better. >> which means that chips designed specifically for macs will be much >> better than an a* series. > benchmarks for A13 show it being ob ar with Core i5. No one cares about that because the Macs will not be runnign an A13. > the Axx chips already competitive against Intel chips. The point is to improve, not hold steady. > The surprise is the difference between Rosetta2 translated binaries vs > natively compiles, The performance hit is greater than I expected. I > guess the translator can't really optimize the code soince it lacks > context of instruction dependancies. Making assumptions about what the Apple Silicon Macs will do based on teh DTK is nothing less than stupid. >> apple is not adding anything to the a* series. > You can beleive what you want. But Apple would be stupid to develop the > Mac chips totally separately since the product lines have so much in common. See, there is a difference between "totally separately" and "identical", a wide gulf of difference with many gradations. Yes, Apple Silicon will be based on ARM. Yes, Apple will have learned from their work with many different ARM and other chips that they have been designing. No, the Mac will not be running a version of the A-series SoCs. > MacPro currently configurable from 7 to 28 cores. So making an Axx chip > with 28 cores is not out of the question IF Apple decides to make a MacPro. Which is unlikely to happen this year. We will have a much better idea of what a Mac Pro refresh will look like once a couple of other Macs with Apple Silicon come out. > The MacPro also supports up to 4 fancy GPUs. How many you configure > changes the price of your computer. Will apple integrate 4 of its own > GPUs into the SoC for MacPro and not proce MacPros differently whether > you need 1 or 4 GPUs ? I would guess a future MacPro will have an integrated GPU and a discrete GPU and continue to support multiple GPUs. > "hardware acceleration" is marketing gobbledeegook. Apple,s > presentation made it clean the chips used on Macs will be similar to the > ones on iPhone/iPad with integrated Apple GPU. FCVO similar. As similar as the current MacPro chips are similar to the MacBook Air, perhaps. Similar is not the same. -- Adolescence is the period between childhood and adultery