So far, the ARM based Macs have been glorified iPhones with Lightning replaced with thunderbolt.
Today's iMac and ipad Pro announcements highlight how much IO
functionality is dependant on the CPU.
So far, the ARM based Macs have been glorified iPhones with Lightning replaced with thunderbolt. And today's announcement says more about
future of IOS devices than on how Apple intends to evolve the Mx
architecture to scale to the Mac Pro.
Apple has proven that it can create compelling raw core performance and
low power consumption. But it remains to be seen if it will scale its
chip capabilities to encompass fill range of computers, or whether it
intends to scale computers down to iDevice fixed config concept and
limited expansion.
I have to say that Apple timing when it announced the move couldn't have
been better. Intel is faltering, unable to move beyond 14nm while Apple
is already down to 5nm. But despite this huge technilogical advantage,
Apple still lags behind in what a XEON chip can do (IO, memory) so
something like Mac Pro is still out of range.
Only time will tell what Apple's plans are. Today didn't reveal anything
new.
You are, as always, an idiot. My M1 mini can run circles around your Mac
Pro is every single measure there is, and can do it for $799 with 8GB of
RAM.
You cannot buy a faster desktop or laptop from anyone, anywhere, for any amount of money.
On 2021-04-20 20:40, Lewis wrote:
You are, as always, an idiot. My M1 mini can run circles around your Mac
Pro is every single measure there is, and can do it for $799 with 8GB of
RAM.
Do you work for Trump Org?
You would fit well within their charlattan environment of lying to
make sales.
You cannot buy a faster desktop or laptop from anyone, anywhere, for any
amount of money.
A computer is more than just the speed of a core. But of course,
because that doesn't fit your charlattan agenda, you convieniently avoid
that discussion.
Also not clear to me how video rendering works with regards to having 4
slow cores. The fast cores finish their task rapidly, but video
re-assembly has to wait for the slow cores to finish.
On 2021-04-20 20:40, Lewis wrote:
You are, as always, an idiot. My M1 mini can run circles around your Mac
Pro is every single measure there is, and can do it for $799 with 8GB of
RAM.
Do you work for Trump Org? You would fit well within their charlattan environment of lying to make sales.
You cannot buy a faster desktop or laptop from anyone, anywhere, for any
amount of money.
A computer is more than just the speed of a core. But of course,
because that doesn't fit your charlattan agenda, you convieniently avoid
that discussion.
Also not clear to me how video rendering works with regards to having 4
slow cores. The fast cores finish their task rapidly, but video
re-assembly has to wait for the slow cores to finish. So the results for
fast cores retained in memory (while thet start work on the next 4
slices of work) until the slow cores are done and the sequential segment
of movie can then be assembled and written to disk.
Please provide anything that suggests video rendering works remotely as
you describe on the Mx. Remotely. Slightest whiff at all.
On 2021-04-21 08:20, Alan Browne wrote:
Please provide anything that suggests video rendering works remotely as
you describe on the Mx. Remotely. Slightest whiff at all.
Each core is given a frame range to render.
(If you look at how After
Effects works with the xpressions, each frame can be rendered
independantly because there they aarchitected it such that the logic in
each frame is not dependant on the result of the preceeding frame. All values are calculated based on the frame number/time code, and its
position between 2 keyframes. (so if a mask changes shape between frames
100 and 200, frame 137 calculates the mask shape based on fixed values
at frames 100 and 200 and not based on what frame 136 had calculated.
The results from each task is received by the "master" who assembles the output in the right order. So when you have a core that is late, it
need sto hold on to the results from other cores before being able to assemble the strem in the right order. (meanwhile, the faster cores have
been given other tasks).
With equal cores, this is generally fairly minimal. But when you have
half cores that are deliberately slow and half that are very fast, you
may end up having to hold on to results in RAM a lot longer waiting for missing gaps in the stream expected from a slow core.
This is why the Apple apologist claims that M1 requires less RAM are
bogus. Rendering and compressing video streams takes as much if not
more memory becayse you bneed bigger buffers to hold on to fast results
while waiting from slow results.
Final Cut Pro likely has adjusted its logic to deal with core asymetry,
but not sure that software like ffmpeg, handbrake etc have. They still
work fine since any multi threaded software needs to handle results not coming back in the right order, but that handling means keeping results
in RAM for longer and that means more ram, not less as apologists claim.
On 2021-04-21 08:20, Alan Browne wrote:
Please provide anything that suggests video rendering works remotely as
you describe on the Mx. Remotely. Slightest whiff at all.
Each core is given a fra....
Sorry. I wasn't clear. I didn't want Mezei theories. I wanted factual information fro an authoritative source. Not that I need it. It's very clear that the M1 performs very well at such tasks.
So please, go ahead and continue to support your ilk's claims that 16GB
is more than enough to render complex videos and use 3D software or
After Effects.
On 2021-04-21 08:20, Alan Browne wrote:
Please provide anything that suggests video rendering works remotely as
you describe on the Mx. Remotely. Slightest whiff at all.
Each core is given a frame range to render. (If you look at how After Effects works with the xpressions, each frame can be rendered
On 2021-04-21 17:32, Alan Browne wrote:
Sorry. I wasn't clear. I didn't want Mezei theories. I wanted factual
information fro an authoritative source. Not that I need it. It's very
clear that the M1 performs very well at such tasks.
I was responding to your ilk's arguments that M1 requires less RAM,
POinting out that when using multiple cores, the video processing loads
as much video as it can to spread between cores and needs to hold
results in RAM until the oiutput can be written out in sequence. So
slower cores require output fron faster cores be help in RAM longer
while waiting for missing frames from slower cores.
| Sysop: | Gate Keeper |
|---|---|
| Location: | Shelby, NC |
| Users: | 790 |
| Nodes: | 20 (0 / 20) |
| Uptime: | 41:36:41 |
| Calls: | 12,115 |
| Files: | 5,294 |
| Messages: | 564,934 |