Nvidia and Apple have a long way to go before they will work together. Also, Nvidia is so chip constrained now they can barely ship products. That is why a 3070 with a retail price of $450 runs $1500 if you want one in 2021.
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.
Aspects where Intel Macs still outperform the M1 models?
- Thread starter jent
- Start date
-
- Tags
- apple silicon intel macs
- Sort by reaction score
here (non usa) , got stock nvdia a lot. I got old r9 amd but macos cannot support
. My rx 570 broke ?If you are saying you can get a RTX 3070 for the $400 please send me a link.here (non usa) , got stock nvdia a lot. I got old r9 amd but macos cannot support
Support for external cards is something they will likely have to add for Mac Pros if there ever is to be a Mxx-based Mac Pro.
Why? Apple's perf-per watt is better than Nvidia, and Mac Pro are very expensive machines. If Apple wants, they could build an absolute monster of a GPU for the Mac Pro, making eGPUs obsolete.
Depends on what you want to do with your MacPro. I do Machine Learning and that means I need access to the GPU. More specifically, now it means I need access to Nvidia's ML libraries that only run on Nvidia GPU, and can't use AMD because of lack of libraries. Mxx GPU support has started to appear. But I wonder how well they will compare to Nvidia in terms of support of the wider ML community which has a lot invested in Nvidia.
Also, In the graphics space is it likely that Apple will build external cards that would plug into the AS architecture? One of the major performance features of the M1 is the shared memory fabric between the process, memory, and GPU. Can this be preserved? And what would an Apple Mxx graphics card cost compared to Nvidia, AMD, or other current graphics manufacturer's product?
Last edited:
A couple of these (with the exception of the Mac mini) were always an issue with the systems that the M1 replaced. There was always two ports, 16GB of RAM max on the intel variants of these systems.
And BTW, the Mac mini has two controllers, just like the previous Mac mini - except fewer ports. So you can get the OWC thunderbolt dock.
Apple ARM chips have a much better performance per watt and they just run way cooler. You can run sustained, intensive workloads on the ARM chip that would cause the Intel chip to overheat unless it were throttle or the jet engine fan engaged afterburners.
We've been conditioned to expect noisy fans when CPUs run at speed for an enduring period. That myth was totally blown out of existence by the Apple ARM chips. Until you realize the difference in normal usage then it doesn't sink home how much of a game changer this is.
Note if you hesitate with ARM Macs because some random Intel Mac thing is 'better' then stick to Intel Macs until you feel enough comfort with them.
We've been conditioned to expect noisy fans when CPUs run at speed for an enduring period. That myth was totally blown out of existence by the Apple ARM chips. Until you realize the difference in normal usage then it doesn't sink home how much of a game changer this is.
Note if you hesitate with ARM Macs because some random Intel Mac thing is 'better' then stick to Intel Macs until you feel enough comfort with them.
If you are invested into CUDA ecosystem I am afraid Macs are just not a good option…
Good question. Apple did build a new ML library that is apparently flexible enough to be a backend to tensorflow but I suppose you are talking about wider CUDA software. Maybe someone will try to implement CUDA on Metal or something like that, but I’m wondering about feature parity.
I don’t see much possibility here. eGPUs are incredibly niche (no way of making a profit) and they break the unified memory model. Maybe one could remedy it by using some sort of super-fast bus with some sort of coherent cache technology, but that kind of sounds like science fiction to me. Not to mention that such tech would massively raise the price of both the GPU and the host computer…
I don't know why people expect so much from a MacBook Air. It's never been designed as a performance machine. The M1 is amazing but people need to understand that the M1 isn't about making the Air a performance machine, it's really about making it run cooler and sustain long battery life. After all the general audience buying an Air are students, people who write papers and do common stuff such as read emails and web surf. It's not meant for sustained transcoding, running CAD programs or for constant transferring of files such as RAW images. This thread about what Intel can do better than the M1 is silly when Apple is not selling the Air as a high-end performance machine.
On the topic of the original post: Intel Macs are much better space heaters!
I have an unheated basement with my office. My i7-10700 Windows system actually runs really cool. My Late 2009 iMac and 2008 Dell XPS Studio will put out a lot of heat.
Many people on this forum would suggest that everyone should dump whatever Intel Mac they are currently using and buy an M1 MBA or if they insist on a desktop, the M1 Mac mini. The M1 MBP is considered by most to not be worth the money. However, some people are sticking with their Intel Macs for now and the purpose of this thread is to explain why someone might do that.
There is no doubt that the M1 MBA is a great light weight laptop and incredible value for money. Apple had good reasons to deliver the machines it did first. The M1 is probably not that different to the A14X that will power the new iPad Pro so it was the easiest and least risky to develop first. Also by targeting their volume market, they are able to grow the installed base quickly.
That volume also pushes software developers to more quickly do native ports or port from 32-bits to 64-bits. The faster they get software ported, the faster they can drop Intel systems and, eventually, support.
I think a lot of reviewers got a little caught up in the moment. That being said, for many, the Air is the most accessible Mac. Starting at $999, it's still expensive for many people. But the fact that the Air can handle considerably more substantial workloads than it has in the past is quite impressive. People who travel a lot (or will resume traveling a lot) can potentially downsize and save space and weight by using the Air when that may not have been a viable option before.
this is malaysia one.. https://shopee.com.my/MSI-GeForce-R...aphic-Card-(G3070V2XC)-i.412162701.9233898966. myr 800 equi 192 dollar. for me i rather buy at the shop.. maybe if i have free time check at real shop it is the real price.. a lot also price 400 maybe it's new and those cheap one maybe from bitminer re-used. By da way , what the average price for that thing your place ?
Last edited:
When you can get them they are very expensive and always from a questionable source. Here is an example on Amazon. $1700 USD. Maybe I need to make a trip to KL?
Amazon.com: 2021 Newst GeForce RTX 3070 XC3 Ultra Gaming, 8GB GDDR6, iCX3 Cooling, ARGB LED, Metal Backplate PCI Express 4.0,1 x HDMI 2.1 3 x DisplayPort 1.4a w/ AllyFlex HDMI Cable: Computers & Accessories
Same for my 9900K deskside. It never gets above 34C and the 2070s GPUs never get about 40c even in 2 day ML training sessions. All air-cooled by fans tuning very slow and silent.
i see in the shoppe i see same spec https://shopee.com.my/EVGA-RTX-3070...5-3755-KR-)-READY-STOCK-i.99616108.9904058438
shoppee - 1,496.56
amazon - $1,749.00
facebook market place - 1,521.19
Fuhh 200 dollar diff same item.
Now covid era , it totally hard to go anywhere without quarantine 14 days x 2. Try find other country nearest to your place. I try to check newegg( no stock), micro center (no stock). Unsure want to ? or ?.
For me , i more prefer you to go the evga,msi website or any other so you can book them and ask them politely it's for work environment, they may entertain quicker.
When you put space towards additional heat sink mass and airflow, yes, you can get things going quiet. But the total heat produced is still going to be similar for similar loads on same-model chips (assuming no throttling in either case). It just might not seem that way when you put your hand near the vents to feel it, due to the heat being more dispersed when it comes out of larger vents/etc.
My 5600X gaming rig with a 2080S stays pretty quiet, but it took work getting the airflow right to optimize it for noise. Since I’m using a smaller (<20L) case, the exhaust is still quite warm. Airflow and case choice are huge when it comes to getting a quiet system.
Apple’s push to go thin and appliance-like in their designs doesn’t help on the noise front because they need faster airflow through the smaller volume to dump the same amount of heat, but that’s a good reason why they are very interested in power consumption to cut down on the waste heat in the first place. It’s not like they seem terribly interested in building big coolers.
My CPU normally runs at 2-20% (20% during the day for my trading programs) and it's a 65 Watt CPU. It's never seen 100% CPU though it can get up to 70% starting up a program but that's just for a few seconds. So I got a low-power desktop chip that isn't used heavily so typical power consumption is low.
I'd love to have a 5900x and I am shopping for one but they are essentially unicorns right now. Getting a GPU would be even more difficult and RAM is starting to get difficult to find. So for me, it's a race between 5900x availability and Apple M1X. I do have a massive heat sink on the CPU. I am running one of my trading programs right now and CPU temps are average 29-30 degrees. GPU is running at 45 degrees but that's what you get with a four-year-old GPU. Hard to get newer GPUs on smaller process nodes these days that aren't overkill. The GPU is just driving 3x4k monitors - I actually don't do anything GPU-intensive.
If I got an M1X desktop, I'd still have to run ATP on a Windows system.
Well eventually they're going to have to as Apple will drop support for Intel Macs at some point and perhaps earlier than expected as they will want more adoption of AS Macs.
Most? Most who? A few people on a forum or YouTube reviewers? Please step into reality that the majority of Apple's real customers are not forum or tech social media junkies. Some or even at a stretch many, not most. You'd need to show proof of that statement.
The thread title is about aspects of Intel Macs outperforming M1 Macs.
I've heard this from time to time here but you have to understand that MacOS is not the same slimmed down and dumbed down system as iOS. It takes a lot more to run MacOS.
I hope you're not referring to forum or social media junkies because they're certainly nowhere near Apple's volume market. Sadly this forum would love to believe they make up the majority of Apple's customers and their needs and wants should cover everyone else's.
I recently switched from a 10-core 64GB iMac Pro to a 16GB Mac Mini M1. For my mostly-Java software development workflow, the Mini M1 has been significantly faster and more responsive experience across the board. Builds of our large Java8-based platform are 30% faster on average. And, the user experience is much snappier. I sold the iMac Pro a week or so ago after running both side-by-side for a couple weeks.
Amazing...
Amazing...
It should have Intel and AMD worried and should have software developers considering a port.
Absolutely. Thanks to the efforts of the open source community and Homebrew, everything in my dev toolchain is completely ARM native. Only a few non-development-related stragglers running in Rosetta right now, like Dropbox.
Speaking of Dropbox... that was my biggest pain point in converting. The client continues to be a massive RAM and CPU hog. So, I had to significantly scale back the number of files managed there, as well as shift completely to Smart Sync, in order to keep it under control.
Other than that, the 16GB RAM ceiling has not been a problem at all. Memory Pressure rarely goes above 35% on my Mini, and the system feels extremely snappy, despite running all the same apps and tools that I used to run on my 64GB iMac Pro.