Agreed but which laptop line to move to? Lenovo and Alienware have heat issues. Razer has durability issue and poor customer services, etc.
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.
What if Apple does not release a new MBP at WWDC in June?
- Thread starter hajime
- Start date
- Sort by reaction score
Agreed but which laptop line to move to? Lenovo and Alienware have heat issues. Razer has durability issue and poor customer services, etc.
Yeah that's the issue. Overall.. I like MacOS, I like the MBP shell (also love the keyboard) and the customer service in my experience has been great, and way better than other companies. My friend just bought a Thinkpad and the UHD screen was outstanding. It's getting harder to justify the cost of the MBP unless you can expense it or money is no issue. I bought a 2016 MBP on launch day because I just wanted it and like others, had waited a long time for Apple to release a new product. I liked it but after a while it was just too slow and I sold it. For 3k, I want it to fly.
I wouldn't compare Apple laptops to Alienware or Razer. There's just too many differences and yes the customer service thing will get you.
And they still will not have Cuda support. Sigh. Like it or not CUDA more than the chips are what make it easier for harried application developers to make use of the GPU. And no developer want to code one way for one chip and a different way for another chip.
With Thinkpad and Alienware, we could upgrade the RAM, SSD and change battery easily. They also have 32GB and 64GB options at lower prices. Sounds good. I have horrible experience with heat and crazy fan with my MBP 2010 right now. Don't want to spend another 3K to get a new computer with these issues.
I guess we are at a transition period. MBP is getting worse and worse but Windows laptops are getting better and better. Which is still better?
People seem to be talking about moving to Windows based laptops, what is Windows like now tho? I’ve been using a Mac since my first in 2011 and I find MacOS really good, that’s not to say I haven’t been tempted by other devices because I have, for some reason just lately I’ve been tempted by the Surface yet in the past I’ve been against those type of devices..... it’s very weird 
MacOS seems to have the edge tho and depending on what updates they show on Monday it might even have more of an edge over Windows.
MacOS seems to have the edge tho and depending on what updates they show on Monday it might even have more of an edge over Windows.
People seem to be talking about moving to Windows based laptops, what is Windows like now tho? I’ve been using a Mac since my first in 2011 and I find MacOS really good, that’s not to say I haven’t been tempted by other devices because I have, for some reason just lately I’ve been tempted by the Surface yet in the past I’ve been against those type of devices..... it’s very weird
MacOS seems to have the edge tho and depending on what updates they show on Monday it might even have more of an edge over Windows.
Not sure what MacOS edge you are talking about. Just received my base model Razerblade Stealth. So far so good. Leaves my Macbook Air in the dust, AND it cost less than the current (outdated) MB Air. Windows X is fine. Like anything new takes some time to get used to.
I say to all the doubters dont be afraid try something new. You may be surprised.
I find MacOS doesn’t crash, I don’t have to worry about viruses (yes I know that Mac’s can get them but I never have) it’s faster to load, doesn’t seem to slow down over time and so on. What worries me is if I were to try out a windows machine again and I don’t like it I’m stuck, for example if I try the Surface which I would need to spend around £1000 or more and I don’t like it.
My MacBook Pro 2011 has served me very well over the years and i’ve had no issues, no crashing, no need to do any maintenance or anything.
And yet those same developers chose to support Nvidia's monopoly instead of forcing them to properly support open-source tools..
As a developer you use the tools available. Where was open CL in 2012/13 when CUDA became dominant? Still in committees being argued about.
Now Open CL may be better, but it may be too late as companies like Microsoft, Google, Amazon, and others are providing very cheap (free in some cases) GPU compute cycles online. Many in AI/ML community are moving that way since all you need is a browser to access this power.
Maybe graphics programs like those from Adobe will do a better job support OpenCL.
But I read that DIY PCs are still faster than those rental ones and in the long term, better to get one’s own machine. Also no need to worry about continuous charging due to Forgotten to log off.
Not faster that is for sure. I have a couple of 10xx Nvidia cards in my machine and still push to my Google Cloud account for heavy lifting.
But you are right about charges. Make sure you code correctly and launch as late as you can and terminate compute sessions as soon as you can.
OpenCL was pretty much there. I mean, OpenCL 1.2 spec was released in 2011, and OpenCL 2.0 spec in 2013. However, Nvidia pretty much ignored it and used its market share advantage to push CUDA instead. Sure, CUDA had superior development tools, but I'd argue that the entire community would be better off if Nvidia would invest at least 1/3 of its CUDA vigor into OpenCL instead. It took them until 2015 to actually release a driver With OpenCL 1.2 support. They still(!!!) don't support OpenCL 2.0
Mo matter how you look at it, its just a monopolist taking advantage of its position and the community falling for their trap. The fragmentation we have now is not least the responsibility of the community.
The difference with home built machines you get to choose the components, and at times those components may be faster then the standard stock offering. Of course you lose the ability to call someone for support of something isn't working right with home built machines. You're left to your own devices. I've done this in the past, and it can be a fun process, and as long as you know the advantages and disadvanges it can be rewarding.
CUDA was there and worked. A spec is a piece of paper. If you have a job to do today, you use what works. That has happened over and over in engineering and software development.
But, as I said earlier, the need for OpenCL or CUDA may be becoming less and less for the ML/AI community.
We do not need 64 bit floating accuracy to train models. We do not even need 16-bit accuracy since we are predicting probability. So a traditional GPU card is overkill.
As a result you see people like Google building their own hardware, the Tensor Processing Unit (TPU) that are using 8-bits. Even running a a lowly 800MHz they are fast enough because they are simple integer based math processors. Because what we need is the ability to do many operations in parallel and processors like these are design to do. TPUs can do hundreds of thousands of matrix operations in a single cycle. Just the perfect thing to do calculations when data is all the pixels in an image.
And now this sort of power is available with a change of a couple lines to run on the cloud instead of on your macbook. So the whole OpenCL/CUDA debate on a dedicated system is becoming less and less important.
Last edited:
I know, I have this kind of chip in my phone
I didn't look at the APIs, but how configurable is this? Can you run custom algorithm on this or is it just predefined stuff + your data? There are still a lot of people doing research etc.
Hello, as WWDC is approaching and there does not seem to have any good evidence on the upcoming of new MBP, what suggestions do you have if Apple does not release new MBP with improved keyboard on Monday? Will the possible release be either Monday or another three months? Will it happen later in the week? As mentioned earlier today, I cannot find a Windows laptop line that can replace the MBP yet. Usually they have heat/noisy fan, reliability, customer services issues, etc. I also cannot wait for another three months as all my work have been on hold for 2.5 years already. I will suffer big lost if I continue to wait. After application of new thermal paste, my MBP 2010 still runs very hot and sometimes I still have crazy fan spinning issues. It also cannot drive 4K display. I have to buy a new computer the next few days.
Last edited:
I know, I have this kind of chip in my phone
I didn't look at the APIs, but how configurable is this? Can you run custom algorithm on this or is it just predefined stuff + your data? There are still a lot of people doing research etc.
First, I want your cell phone!
Second, you can definitely do it on your own NN design. For Google you can use TensorFlow to build your network. TensorFlow comes with implementations of all of the usual layers (Fully connected, Convolution, Pool, LSTM, etc.) So you can use them to build your network.
Or if you are creating the Leman layer, can also write your own layers and use them. But from what I understand this not a trivial task. Unfortunately, that is beyond my expertise.
BTW, I misspoke when I said that you just change the config file and correct my post. You also need to call a method to get you connected and call the TPU variant of the Estimator. Usually you use an (If TPU) .... to make the switch.
See https://www.tensorflow.org/programmers_guide/using_tpu
I know Microsoft has their project Brainwave, https://www.microsoft.com/en-us/research/blog/microsoft-unveils-project-brainwave/ that is their dedicate NN processor.
I don't have high expectations for Microsoft's project. What did they mean by "massive FPGA infrastructure"?
I've used FPGA's in data acquisition and wasn't impressed
There is actually a bunch of newer phones that come with dedicated hardware for AI processing. iPhone X (NeuralEngine), latest Android phones... I do not know the details of these chips, but I assume they are also rather simple, fast limited-precision vector ALUs specialised on matrix operation.
Ah, thats cool. We'll look into it. We started dabbling with some deep learning for some of our research projects lately, this could be helpful.
I have heard that type of comment before.
[doublepost=1527885191][/doublepost]
Interesting. I have never thought to look into the details of my Samsung S8's processor. Maybe I too already have this type of chip. Thanks.
Last edited:
Heck even something as simple as window management is much nicer in Windows with the snap functions and minimising to the app icon with roll-over previews...
And what if you have a 2009 or 2010 MacBook Pro? hahahaha
I'm in that scenario. My 2010, honestly, still works fine, although maybe with Sierra it was a bit better.
This machine is still fine for light loads, like my typical usage. Only issue is the battery-life and its weight. That's why for me, and my light workload (although I'm still considering to learn to edit video and such things), I'd be fine with an 13" MacBook. Passively cooled, and with at least two USB Type C.
Anyway, what I don't want honestly is a f****** Touch Bar on my Mac. So maybe it would be a good idea to purchase a 13" MacBook Pro without TouchBar and then, cancel the purchase if they don't remove them from the Apple Store.