3090 Founders Edition is $1500 at Best Buy. Even scalper price is ~$1800 but nice try.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
M1 Ultra Doesn't Beat Out Nvidia's RTX 3090 GPU Despite Apple's Charts
- Thread starter MacRumors
- Start date
- Sort by reaction score
3090 Founders Edition is $1500 at Best Buy. Even scalper price is ~$1800 but nice try.
Apple Silicon’s first generation includes M1, M1 Pro, M1 Max, and M1 Ultra. All of these M1 chips scale up from the same fabrication process used on the A14 chip, and apparently they all share the same CPU & GPU core design (or at least nearly identical, from what I’ve read).
The second generation AS desktop will be M2 (based on A15 chip), third gen will be called M3, etc.
TheYayAreaLiving 🎗️
Suspended
They should have let Craig Federighi introduce the Mac Studio. Maybe this wouldn’t have happened. He is magical. 🪄
Am I the only one that thinks this chart is obviously talking about performance per watt and not raw performance? This chart is not misleading at all. Its simply stating M1 Ultra is much more efficient
Good luck finding a founders edition.
MSI NVIDIA GeForce RTX 3090 GAMING X TRIO 24G 24GB GDDR6X PCI Express 4.0 Graphics Card RTX 3090 GAMING X TRIO 24G - Best Buy
Shop MSI NVIDIA GeForce RTX 3090 GAMING X TRIO 24G 24GB GDDR6X PCI Express 4.0 Graphics Card products at Best Buy. Find low everyday prices and buy online for delivery or in-store pick-up. Price Match Guarantee.
www.bestbuy.com
This is why i won’t buy a Mac desktop. Windows (essential for my work) is just much better in this regard. But still love the m1 max laptop for personal use.
It’s still a software game imo. For macs it’s anemic and will be much worse when rosetta goes poof. Can’t use one for work. This is where apples focus needs to be but we won’t see this out of Cook.
It’s still a software game imo. For macs it’s anemic and will be much worse when rosetta goes poof. Can’t use one for work. This is where apples focus needs to be but we won’t see this out of Cook.
It's the problem with trying to use a scaled up mobile architecture for desktop class computing. Think of the RAM scenario. It doesn't matter if your 128G of RAM is fast, or if your CPU is fast. If your 300G+ in-memory workload is in swap then these boxes will perform worse than the 2019 Mac Pro.
There’s some speculation out there that the next MacOS update will vastly improve the M1 Ultra’s performance. That makes sense, and we’ll know in a couple months.
Apple graph is BS anyway. M1 Ultra 64GPU can't even hang and it's ~180W.
We use gaming cards for pro apps, as the apps are built using gaming engines [unreal engine specifically]. A gaming GPU performs far better than a quadro for this.
We also use CAD software that would benefit from the quadro, but the reality is the RTX cards perform very well in this area too, hence always using RTX cards in PC [3080ti at present].
We have a mac studio ultra on order. If the GPU performance is as the Verge have suggested, it will be boxed back up for return within a day. It wont take me long to test out a few designs to see where it is at.
I am actually now thinking however, that this is the end of macs in our studio. A bit ironic that the Mac Studio killed the macs in the studio............
Be sure to report back to this thread.
Very unlikely considering M1 Max had poor hashrate, worse efficiency and high cost compared to $400 3060ti Founders Edition GPU.
Ethereum
3060ti FE tuned 60.6 MH/s @ 115.6W = 0.524 MH/s per W
M1 Max GPU 10.3 MH/s @ 24W = 0.429 MH/s per W
M1 Max GPU+CPU 20MH/s @ 62W = 0.323 MH/s per W
M1 Max CPU 10.4 MH/s @ 40W = 0.260 MH/s per W
I have 2 3090's and neither card comes close to 500 watts. Usually in the 200's rendering, with occasional peaks to 320-330watts. You can even undervolt them a bit and get them down to 260-270 watts max.
Now the stupid uber overclocked cards are available that do get close to 500 watts, but they perform only 5% better than stock on average which is crap.
To achieve that performance with the RTX3090, you need a computer system case with (my guess) 650 watts of power available--totally ridiculous. And you still have the issue of communicating between the CPU cores and the GPU card over the PCI-e bus.
But then, they couldn’t rage post. MacRumors doesn’t wants objective coverage, it wants page views!
With tons of microstutter to boot. That's just average FPS, I bet the 99%tile numbers would be horrendous if you could reliably measure them in MacOS. MacOS ports also don't have the eye candy or draw distance that the PC titles usually have also which also makes their performance seem better than they really are. I had hopes my M1Max MBP16 would be decent in gaming, but it's worse than my 2019 MBP 16 with most gaming titles. Just runs quieter with more stuttering.
I really don't understand what Apple was thinking when I saw this graph. They specifically state it has 21 TFLOPS which I know are not apples to apples, but 21 TFLOPS on the M1 Ultra (which is hugely impressive and essentially two PS5's) was never going to come close to an RTX 3090 at around 36 TFLOPS.
Is it amazing and extremely fast? Yes. Does it wipe the floor in terms of efficiency? Yes. Can it match an RTX 3090's performance? No.
Is it amazing and extremely fast? Yes. Does it wipe the floor in terms of efficiency? Yes. Can it match an RTX 3090's performance? No.
Stuttering isn't as bad as crashing but doubt any of Apple's sponsored reviewers are going to report that.
https://gizmodo.com/review-apples-mac-studio-is-the-ultimate-mac-desktop-1848662076
"It ran Total War: Warhammer II at an average of 108 fps with the graphics set to Ultra, falling behind the Omen with its RTX 3090 (167 fps). I then entered the sewers of Metro: Exodus, which ran well above 60fps at 4K Ultra, averaging around 80-90fps. This sounds great, but in reality, both of these games crashed multiple times in my testing."
Along the way, these 'experts' at the Verge forgot how to read graphs. Apple's graph is a performance per wattage (power). Their marketing from the beginning has always been about power efficiency. This article and Verge 'review' on 'performance' should be ahem...reviewed.
We really need to see native M1 triple A games running on Metal to see what this is truly capable of.