Surprise: Mac Studio with M1 Ultra Can’t Beat NVIDIA GeForce RTX 3090, Contrary to Apple’s Claims

[Brian_B]

Of all 3, I think Nvidia would be most cabable of doing this relatively quickly, but I think all 3 could do it in at least the equivalent development time of a console. All of Apples advantage, plus faster everything else if needed would most likely be the result.

I'd say... of all 3, nVidia has the least amount of experience developing a product for third party specifications - even the Switch uses an off-the-shelf SOC, not a custom package.

And nVidia will be limited by an ARM license or some other open standard (RISC V?), something they have done, but don't have a great deal of experience customizing either. Apple's product is also ARM, but they've been through more than a dozen iterations and refinements on theirs by now, and AMD/Intel both have x86 in addition to ARM (and RISC and anything else open) has options to use.

If I were betting and those 3 are the options, I'd put my money on AMD to make the most marketable product. nVidia might be able to make the fastest, but I think they would also attach the least attractive commercial terms to it. I'd put Intel almost neck and neck with nVidia, for about the same reasons. But I don't know that any of them would beat Apple, honestly - I think they have already won the "Design something for Adobe" contest. That power envelope delta is pretty big.

 

[Grimlakin]

I wouldn't be surprised if Apple had a financial stake in Adobe to be honest. That tends to lock down partnerships without the need for explicit contracts. Reports contradict but Apple owns between 15 and 20 percent of adobe.

 

[Burticus]

Did everyone see the size of that thing? That's no moon... it's Apple's new cpu. 3x bigger than a Ryzen cpu.

 

[Uvilla]

I'd say... of all 3, nVidia has the least amount of experience developing a product for third party specifications - even the Switch uses an off-the-shelf SOC, not a custom package.

And nVidia will be limited by an ARM license or some other open standard (RISC V?), something they have done, but don't have a great deal of experience customizing either. Apple's product is also ARM, but they've been through more than a dozen iterations and refinements on theirs by now, and AMD/Intel both have x86 in addition to ARM (and RISC and anything else open) has options to use.

If I were betting and those 3 are the options, I'd put my money on AMD to make the most marketable product. nVidia might be able to make the fastest, but I think they would also attach the least attractive commercial terms to it. I'd put Intel almost neck and neck with nVidia, for about the same reasons. But I don't know that any of them would beat Apple, honestly - I think they have already won the "Design something for Adobe" contest. That power envelope delta is pretty big.

In a way, Im surprised none of the 3 (AMD, Nvidia and Intel) hasn't decided to go all in on all manner of productivity acceleration ( whatever codecs, anything, everything) include it ALL in all future cards. Have a line of professional cards that would have applicable hardware differences (reliable clock speeds, stable memory, big silent heatsinks/with reliable fans whatever) but instead of pricing said cards at a million bucks price it close to game cards but sell it with a service/license contract. The benefit of said service should/ could be quite significant for a business. I think going all in for everyone not only gives you a competitive edge, it pushes computing forward for everyone too. Plus said service contract is just basically paid driver debugging for everyone else kind of self maintains . Business gets everything stable, and problems addressed actively, everyone else gets in part a trickle down from solutions of the problems actively addressed.

 

[LazyGamer]

In a way, Im surprised none of the 3 (AMD, Nvidia and Intel) hasn't decided to go all in on all manner of productivity acceleration ( whatever codecs, anything, everything) include it ALL in all future cards. Have a line of professional cards that would have applicable hardware differences (reliable clock speeds, stable memory, big silent heatsinks/with reliable fans whatever) but instead of pricing said cards at a million bucks price it close to game cards but sell it with a service/license contract. The benefit of said service should/ could be quite significant for a business. I think going all in for everyone not only gives you a competitive edge, it pushes computing forward for everyone too. Plus said service contract is just basically paid driver debugging for everyone else kind of self maintains . Business gets everything stable, and problems addressed actively, everyone else gets in part a trickle down from solutions of the problems actively addressed.

Both AMD and Nvidia ship 'creator' drivers for consumer cards. Beyond that, FirePro and Quadro cards are exactly that in hardware - if such stability is needed.

But in terms of an Apple competitor, Nvidia is as close as you're going to get. Between CUDA, which has similar adoption in the PC space (it's the default, despite being Nvidia-specific), to NVENC being broadly supported and high-quality, content creators and graphics professionals that aren't using Apple are best off with an Nvidia GPU.

To wit, when the RTX 3090 was introduced, content creation - and specifically, 3D modeling - was a core target application. Yes, the RTX 3090 is the fastest GPU in the lineup, but it's also the one most focused on content creators.

(I'll also say out of personal experience - and this is a sample of one anecdote - I wound up swapping out an RX460 4GB for a slower 1050Ti 4GB in my wife's desktop because Photoshop kept crashing with the AMD GPU...)