Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Yeah we have plenty of servers and access to nvidias DGX systems but it’s just frustrating I always have to VNC instead of firing up some quick tests on my local computer. The 9980HK is plenty fast so it’s worth it in that regard but still.


I have a RTX 2080 and GTX 1070 in my desktop system (windows/Linux) and still use the cloud instance more often. It is faster and does not sound like a 747 taking off for 1/2 hour (or maybe it does, in the datacenter) while it trains my models.

And, a Macbook Pro will never be a great ML machine regardless of GPUs. It cannot dissipate the heat and the puny 96 W Power supply limits how hard the GPU can be run.
 
There aren't any dual core options from Intel for 10th gen and beyond. Unless they use a dual core ARM CPU, it's basically impossible for them to continue with it moving forward. And I'm sure they would have loved to have quad core in current MBAs if Intel had delivered.
The current i5-Y they use is semi-custom (617 graphics) so it's not completely out the realm of possibility they ask Intel for a custom 10th gen dual core Y chip (probably in the form of taking quad core CPUs from the reject bin where 1-2 of the cores don't work and just turning them off). It would be a bit dumb though as it would leave the MBA getting smoked by competing machines.
 
No. Mojave and Catalina have no nvidia driver support.

Unfortunately, at this moment if you need CUDA support then you shouldn’t buy a Mac. and yeah that’s really sad.
You can do it in boot camp with an eGPU
[automerge]1577570547[/automerge]
I have a RTX 2080 and GTX 1070 in my desktop system (windows/Linux) and still use the cloud instance more often. It is faster and does not sound like a 747 taking off for 1/2 hour (or maybe it does, in the datacenter) while it trains my models.

And, a Macbook Pro will never be a great ML machine regardless of GPUs. It cannot dissipate the heat and the puny 96 W Power supply limits how hard the GPU can be run.

right I get that laptop GPUs suck. Probably gonna get/build a workstation (maybe Mac Pro) and put Linux with nvidia gpus
 
right I get that laptop GPUs suck. Probably gonna get/build a workstation (maybe Mac Pro) and put Linux with nvidia gpus

Please post how that works out if you go that way. I have an i9900K system with 64 GB or memory and several TB of NVMe SSD plus the 2 Nvidia Graphics cards. I still find it better to go to the cloud. I can spin up a container instance in a few minutes and throw a lot of hardware at the problem.

I am curious to see how your MacPro experience works out. I can always another system. :eek:
 
Yeah I know. I purchased a mbp for a change since I have a gaming desktop. I wanted to get into coding, web dev and design. So I got this! I barely game anymore tbh.

Well then, you get command line Unix straight out of the box. That's worth quite a bit if you want to get into Web dev. You can get those on a Windows machine too, but it's more hassle and when you're doing Web dev with all of the frameworks and libraries that are exploding in number, you really want someone else to deal with maintaining the tool foundation so you can concentrate on the bigger stuff.
 
Please post how that works out if you go that way. I have an i9900K system with 64 GB or memory and several TB of NVMe SSD plus the 2 Nvidia Graphics cards. I still find it better to go to the cloud. I can spin up a container instance in a few minutes and throw a lot of hardware at the problem.

I am curious to see how your MacPro experience works out. I can always another system. :eek:

Huh really? When you say cloud do you mean like AWS? At my institution we have access to Linux boxes with 32 cores 256gb of ram with like 4 titans rtxs or a Quadro 8000 (I think) and stuff runs decently well. We’re actually in the process of switching our code to make use of GPUs but I can update with how the Mac Pro works in my workflow
It’s just like my opinion man, but if the solution to something on a Mac is to install Windows on it, then what are we even doing here? Just buy a thinkpad and be done with it.

actually it won’t work because most cuda stuff I do is on a Linux distro anyway. And everything we do is Unix-based so almost no one uses a windows computer where I work. Furthermore I’m not a huge fan of it so it’s not really an option.
 
Huh really? When you say cloud do you mean like AWS? At my institution we have access to Linux boxes with 32 cores 256gb of ram with like 4 titans rtxs or a Quadro 8000 (I think) and stuff runs decently well. We’re actually in the process of switching our code to make use of GPUs but I can update with how the Mac Pro works in my workflow


actually it won’t work because most cuda stuff I do is on a Linux distro anyway. And everything we do is Unix-based so almost no one uses a windows computer where I work. Furthermore I’m not a huge fan of it so it’s not really an option.

then how is boot camp an option!?

also you can install Linux on any PC...
I mention thinkpad as an arbitrary example. Obv once you go outside Apple hardware your freedom is plentiful

[automerge]1577580824[/automerge]
Cuz a ThinkPad can’t run an eGPU.
This is 100% BS. Yes they can.

they also are available with internal nVidia dGPU
 
Last edited:
it's basically equal to nvidia 1660 mobile with a little extra vRAM. 2070 max q has 8 gb of vRAM but it is about 2x more powerful and consumes about 2x the power.
 
  • Like
Reactions: khalitzy
It's been like this for 20 years. GPU drivers are much better on Windows, they are updated twice a month, they aren't controlled by your computer maker.
Yes, i fondly remember updating my nvidia drivers every few weeks on the massive tower PC i built from the free Opteron AMD gave me one year in lieu of a holiday bonus. I was pissed that I only got a CPU instead of real cash that year, but excited to build a rig with a chip I designed.

Of course, what really got me pissed was constant reboots on Windows due to bad GPU drivers, that NVIDIA kept messing up, and which required constant attention, updating, and parameter twiddling just to keep the machine running.

Know what? Maybe my memories aren’t so fond after all.

(P.S.: yes, even though i worked at AMD i used NVidia cards. I started at AMD long before they bought ATI, and once they bought ATI the two teams were completely separate for a long time.)
 
There aren't any dual core options from Intel for 10th gen and beyond. Unless they use a dual core ARM CPU, it's basically impossible for them to continue with it moving forward. And I'm sure they would have loved to have quad core in current MBAs if Intel had delivered.

There are still some in the 10th generation category, though they are not given a typical 10th gen name.
Although highly unlikely, I wouldn't put it past Apple to slap some under clocked/powered 35 watt desktop CPU in an Air, JUST to prevent loss of sales of a single 13" MBPs.

Intel Pentium G64002/43.8 GHz3.8 GHzN/ATBD4 MB65WTBD
Intel Pentium G6400T2/43.2 GHz3.2 GHzN/ATBD4 MB35WTBD
Intel Celeron G59002/23.2 GHz3.2 GHzN/ATBD2 MB65WTBD
Intel Celeron G5900T2/23.0 GHz3.0 GHzN/ATBD2 MB35WTBD
 
My Thinkpad has T2000 with 4gig ram. It's basically a workstation version of 1650ti with slightly less performance. It's pretty nice to have in the form factor of Thinpad P1
 
There are still some in the 10th generation category, though they are not given a typical 10th gen name.
Although highly unlikely, I wouldn't put it past Apple to slap some under clocked/powered 35 watt desktop CPU in an Air, JUST to prevent loss of sales of a single 13" MBPs.

Intel Pentium G64002/43.8 GHz3.8 GHzN/ATBD4 MB65WTBD
Intel Pentium G6400T2/43.2 GHz3.2 GHzN/ATBD4 MB35WTBD
Intel Celeron G59002/23.2 GHz3.2 GHzN/ATBD2 MB65WTBD
Intel Celeron G5900T2/23.0 GHz3.0 GHzN/ATBD2 MB35WTBD

You must be joking. The Air is designed for 7W chips. There's no "under clocked/powered 35 watt desktop CPU" that would fit in the Air. I can only assume you're trolling.
 
Huh really? When you say cloud do you mean like AWS? At my institution we have access to Linux boxes with 32 cores 256gb of ram with like 4 titans rtxs or a Quadro 8000 (I think) and stuff runs decently well. We’re actually in the process of switching our code to make use of GPUs but I can update with how the Mac Pro works in my workflow

Yes, by Cloud I mean AWS, Google Cloud Service (GCS), or Microsoft Azure. I have things running on all 3 platforms. I use Nvidia GPUs on all three for acceleration via TensorFlow and PyTorch, and also Tensor Processing Units (TPUs) on GCS.
 
Last edited:
You must be joking. The Air is designed for 7W chips. There's no "under clocked/powered 35 watt desktop CPU" that would fit in the Air. I can only assume you're trolling.
My apologies.

They will go with either an i3-10110Y @ 5.5 Watt or i3-10110U @ 15 Watt - both are 2 core/4 threads that Apple will happily use, rather than losing the sale of a single 13" MBP.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.