Apple unveils M2 chipset with 18% faster CPU, 35% faster GPU compared to the M1
- c
- captain fokou
- a0t
- 10 Jun 2022
HaterzSuckkkkk, 10 Jun 2022Geekbench compute isn't reliable. It computes graphics... moreGeekbench is unreliable and very favorable to Apple products but even there the M1 ultra doesn't touch the 3060. Never mind the 30 fing 90
- c
- captain fokou
- a0t
- 10 Jun 2022
HaterzSuckkkkk, 10 Jun 2022"M1 ultra is slower than the 3050" No comment.... moreI wouldn't call blender 3.1 and V-ray gaming benchmarks. And geekbench is the only benchmark the M1 ultra surpasses the 3050 and approaches the 3060.
- K
- Kev
- 0Za
- 10 Jun 2022
rock, 08 Jun 2022Must be a a nice job defending a trillion dollar company lo... moreWhile it isn't useful information, Nvidia GPUs come from the same foundries as Exynos chips. While their architecture is vastly different, the crux is that Samsung have SoCs and GPUs that have ridiculous power inefficiency and heating issues. However, on a stationary PC GPU, that isn't an issue because most PCs use powerful fans and active water-cooling. So, the power consumption and overheating isn't an issue.
The M-chips by Apple are different. They're just bigger A-chips with higher clock speeds and just MORE of everything. So it's not a surprise that it runs very efficiently.
Just to prove my point, Qualcomm can't stop banging about how the SD 8 Gen 1+ is 30% more efficient than the regular Gen 1. I bet that most of it is because of the better fabbers at TSMC.
I would trust Apple's power efficiency claims. The chips are built on arguably the best architecture, and iPhones have proven themselves to be remarkably efficient next to Android phones. The performance gains are more dubious though. Intel does a perfectly good job at keeping up with the M1s, although they are held back by much worse thermals and much higher power consumption.
Let's not forget. Apple wanted their own silicon because their release dates didn't line up with Intel, making their launches incoherent and varying wildly in frequency. If they really wanted to make their own silicon to beat Intel, they would have made something new. Something that ran hotter but offered MUCH more performance than they now can. But what we've got now is essentially two A14s spliced together that power the 2020 Macbook Air.
- H
- HaterzSuckkkkk
- ybx
- 10 Jun 2022
captain fokou, 06 Jun 2022Your points are so stupid
1) You should watch the graphs a... more"M1 ultra is slower than the 3050"
No comment.
Geekbench compute sucks as a benchmark for GPU.
All the other benchmarks are for shadow of the tomb raider, and those are GAMING bechmarks, which the M1 sucks at.
- H
- HaterzSuckkkkk
- ybx
- 10 Jun 2022
captain fokou, 06 Jun 2022I will trust only numbers from reputable reviews. Apple is ... moreGeekbench compute isn't reliable. It computes graphics power in short bursts and doesn't measure the actual thing.
All the things that measure the M1 ultra's GPU are either GAMING benchmarks or UNRELIABLE benchmarks.
You call that reliable?
- ?
- Anonymous
- u5k
- 10 Jun 2022
rock, 08 Jun 2022Must be a a nice job defending a trillion dollar company lo... moreYou don't even understand what you're talking about do you. Nvidia GPU is notably more efficient in Blender than it is in the other tests because it relies on the hardware acceleration engine that is built separately from the CUDA cores. IT IS NOT A RAW GPU PERFORMANCE TEST.
Your claim is like saying Intel or AMD APUs are 5x less efficient than M1 because they are 5x slower when encoding or decoding vidoes, when in fact we are only looking at the number produced by M1's dedicated Media Engine. Exactly the same thing is happening here.
- r
- rock
- uvh
- 08 Jun 2022
Anonymous, 08 Jun 2022The fact he sees the total AC power draw instead of measuri... moreMust be a a nice job defending a trillion dollar company losing by a unimaginable percentage in gpu heavy works. M1 ultra takes in 5 times less power to give 5 times less performance.
- ?
- Anonymous
- 6wN
- 08 Jun 2022
Powerful hardware with joke software and many limits. That's the way Apple works.
- ?
- Anonymous
- AJj
- 08 Jun 2022
Anonymous, 08 Jun 2022The fact he sees the total AC power draw instead of measuri... moreHoping Apple to implement something similar to Optix in their M series chips. Then nVidia is gone for good.
- ?
- Anonymous
- AJj
- 08 Jun 2022
kek, 06 Jun 2022All that power and it seems the M2 still only supports 1 ex... moreGet something else, cause of you don't understand this is business and there are no free lunches.
- ?
- Anonymous
- TSR
- 08 Jun 2022
Hemedans, 07 Jun 20221650 is Cheaper than M2 and its not for people who want low... moreHow much watt difference?
Flagship EV car is rapid charging at 150kW-200kW now.
15W-200W?
Meh.. so weak.
- ?
- Anonymous
- u5k
- 08 Jun 2022
tim, 07 Jun 2022https://www.youtube.com/watch?v=vEkjupEC6gwThe fact he sees the total AC power draw instead of measuring how much each of CPU and GPU consumes power instead of using Powermetrics suggests how little he knows about the validity of his entire test process. Mixing a x86 only app like V-Ray (which the guy in the video admits himself) into a comparison really muddles the point of his argument too.
Blender is one application RTX3000 series GPUs have a distinct speed advantage over M1 series due to the inclusion of dedicated OptiX ray tracing acceleration engine, somewhat similar to M1 having a media engine for fast video encoding and decoding. This test isn't anything to do with the raw performance of GPU. This is also why in benchmarks like GFXBench they perform comparably to each other in terms of FPS, although M1 Ultra draws significantly less power for the same workload.
- ?
- Anonymous
- vxx
- 08 Jun 2022
tim, 07 Jun 2022https://www.youtube.com/watch?v=vEkjupEC6gwBlender 3.1 added a Metal support but the app itself is still in x86 binary, so the translation takes away a huge chunk of performance in the process.
- t
- tim
- uvh
- 07 Jun 2022
Anonymous, 07 Jun 2022You have no evidence whatsoever to support your hypothetica... morehttps://www.youtube.com/watch?v=vEkjupEC6gw
- c
- captain fokou
- atL
- 07 Jun 2022
Anonymous, 07 Jun 2022Why did NVIDIA create Total Garbage Puke trash like 1060,16... moreOK, I know it may be hard to understand but some people can't pay 350 for a gpu or do not need to.
- c
- captain fokou
- atL
- 07 Jun 2022
Anonymous, 07 Jun 2022Look at Intel graphs for 12th gen Core introduction. Core 1... moreWho do I believe? How about neither of these companies and I don't understand what this has to do with my comment
- ?
- Anonymous
- vxx
- 07 Jun 2022
Anonymous, 07 Jun 2022Lol. The new xps 13 plus with i7 1260P destroyed even the M... moreM1 Max has a solid 20% lead in multi threaded performance over 1260P. Nice try, but Intel still lags far behind AMD and Apple :)
- ?
- Anonymous
- vxx
- 07 Jun 2022
Anonymous, 07 Jun 2022not true. m1 ultra is 5 times slower using 5 times less pow... moreYou have no evidence whatsoever to support your hypothetical numbers, why should we ever believe you?
- ?
- Anonymous
- 6wN
- 07 Jun 2022
007bond, 07 Jun 2022Entry level laptop of apple starting at Maxed out price of ... moreWindows is miles better.