Jump to content


Sneaky Snake

Member Since 28 Jul 2007
Offline Last Active Yesterday, 12:52 PM
***--

Posts I've Made

In Topic: AS ported games for M1-Chip

21 November 2020 - 10:33 AM

View PostHomy, on 20 November 2020 - 07:27 PM, said:

There are many reviews where they test Fortnite.

Have the tested the iOS version running on the M1 vs the macOS (x86) version?

In Topic: AS ported games for M1-Chip

20 November 2020 - 06:57 PM

View PostBen, on 20 November 2020 - 05:35 PM, said:

Has anyone tried some of the “real” games that have been ported to iOS on a M1 Mac? Thinking of things like X-Com, X-Com 2, Civ VI, etc...

Fortnite would also be an interesting game to try if it wasn't currently gone from the store due to Apple and Epic's legal feud. It has a 120 fps option on my iPad Pro. I don't care for the game much personally, but am very curious about it's performance. The iOS version could be benchmarked against the macOS native version.

In Topic: Apple's M1 chip - huge GPU gain over Iris graphics

20 November 2020 - 01:31 PM

I think you are vastly underselling how big of a deal performance per watt is.

Running complex jobs on battery power is probably it's least interesting benefit. Have a lower TDP/heat chip inside a mobile device like laptop allows for one of two things:
  • More performance in the same chassis/thermal design, or:
  • The same performance in a lighter, quieter, and cooler design
Also, the iMac and iMac Pro are pretty bad for cooling, so while they don't care about how many watts they are pulling from the wall due to them being a desktop, a cooler and more efficient chip will help them tremendously with staying at full performance while under sustained load. The Mac Pro is the only product that Apple sells that has a chassis that can completely disregard performance per watt.

Apple's M1 being so power efficient means that there is huge thermal headroom when you move to their more "Pro" focused chassis like the 16" MBP. My current 16" MBP is right at the limit of what the 96W power adapter can provide and also right at the limit of what its chassis can cool. I'm considering applying thermal pads to my VRM's to slightly improve cooling under extended loads where it still thermals throttles.

An M2, M1X or whatever they have planned for the 16" Macbook Pro would fix almost all of these issues. If they doubled the core counts on the CPU and GPU I would be looking at equal or better CPU performance as well as better GPU performance compared to my 5500M - all in the design that would give me way better battery life while on the road all day (I travel a lot for work between the 10 different sites that I manage for IT), and my laptop wouldn't be screaming like a banshee and losing 50% of its battery during a 90 min Teams call when I'm screen sharing. Also the video/audio editing work that I do on the side would also be improved.

AMD/Intel cannot do this - there simply isn't any power/thermal room to slap in more cores with their current designs. They have to improve their IPC performance to give a noticeable upgrade. Apple can simply slap in double or triple the cores compared to the M1 and still be way under my 16" MBP's thermal/power limit.

Intel finally has a decent IPC boost coming in 2021 with a 20% gain, but who knows how many years it is going to take them to do the next big jump after that one.

In Topic: Apple's M1 chip - huge GPU gain over Iris graphics

19 November 2020 - 08:16 PM

Fully agree with your assessment Tetsuya, except for the following:

Quote

Notice the mobile Ryzen's crushing it in multicore (and Tiger Lake beating it single-core at relatively equivalent clock speeds, as TL doesn't maintain its high boost very long at all).  Again, Apple could address this deficit in higher end-rigs by adding more cores, but all we can really comment on at this point is what there is.  And there are (deliberately) no Desktop parts shown on that list.  It basically gets taken out to bite the curb by any remotely fast modern X86-64 chip.  The 10600K @ 5ghz and the 5600X at its stock clocks (little point to OCing Ryzen because of Thermal Velocity Boost) absolutely demolish it.

While yes - those chips do demolish it; they are doing so with a substantially higher power usage, TDP, and core count. The M1 is 10W chip, that boosts up to around 20W it seems - and that 20W is for the entire SoC (RAM, IO, CPU, GPU, etc.). Compare that to the 4900H which boosts up to 54W and Tiger Lake which boosts up to 28W (while not factoring in the TDP/power required for the memory).

Single thread performance with the M2 likely won't go up much (will likely be 15-25% like you said), but multi-core performance is going to go up substantially - as evidenced by Apple planning to convert their entire ecosystem over to Apple Silicon. If Apple Silicon core counts were not going to 2x, 3x, or even up to 4x for the 16" MBP, iMac, iMac Pro, Mac Pro, etc. then Apple would not have those product lines scheduled for the next 18 months.

Further, there is nothing stopping Apple from integrating with higher end dedicated GPUs for products like the iMac Pro and Mac Pro. ARM and dedicated GPUs already co-exist in the datacenter world. I'm guessing that Apple isn't that interested in creating massive GPUs to drop into the ARM Mac Pro so I would bet that their partnership with AMD will continue for high end graphics.

Ultimately only Apple knows what their plans are, but I think they will definitely be competing with significantly more powerful (and higher TDP) chips in their Macs that are actually professional focused. We still need to see Apple Silicon designs for the following:
  • 13" MBP with 4 ports
  • 16" MBP
  • iMac
  • iMac Pro
  • Mac Pro
Basically their entire lineup focused on real professionals hasn't been updated yet. In the next 6-12 months I would be shocked if we don't see newer Apple M chips cranking out significantly higher performance then the M1.

In Topic: Apple's M1 chip - huge GPU gain over Iris graphics

18 November 2020 - 01:50 PM

View Postmacdude22, on 18 November 2020 - 11:30 AM, said:

Maybe we should start putting GeekBench #s in the sigs.

I will just put 99,999 for all of my scores. Yes my iPod Touch outperforms the Mac Pro by a 7500%.