I don't think I'm going to buy 13.3" rMBP until they have dedicated GPU's

[linjac321]

I'm going to keep my macbook air until 13.3" rMBP has a dedicated gpu in it. Clearly too many people and I have seen this happen in the Apple store where everything is set up to perfection where there are too many ui issues with integrated graphics. Even though intel is making great progress in power, there vram is just to low powered to handle pixel pushing screens. vram is critically important to how many pixels a gpu can push instantaneously.

The memory of dedicated and integrated are just that far apart and will never get close to dedicated power because integrated uses system memory whereas dedicated has their own memory. System memory has and always will be 2-3 generations behind dedicated memory. It has been like that for years and i don't see that changing in the near future. They need to at least put a 840m in there. 920m in the future maybe.

I would consider the 15 inch but its to big for me to be mobile after using my 13.3" MBA for so long.

 

[maflynn]

You may be waiting a long time, i.e., forever. I don't think apple is going to put a dGPU in the 13" MBP, for a variety of reasons.
1. Heat dissipation, the smaller enclosure with one fan will be an issue.
2. Product differentiation between the 15" and 13"
3. Apple is seemingly moving away from the dGPU. Just look at the fact, that only the high end 15" model as a dGPU now.
4. Apple's track record with dGPU - the 2008, 2010, 2011 models all have failing dGPUs.

In light of those items, I don't think we'll be seeing any dGPU in the 13" model.

 

[linjac321]

You may be waiting a long time, i.e., forever. I don't think apple is going to put a dGPU in the 13" MBP, for a variety of reasons.
1. Heat dissipation, the smaller enclosure with one fan will be an issue.
2. Product differentiation between the 15" and 13"
3. Apple is seemingly moving away from the dGPU. Just look at the fact, that only the high end 15" model as a dGPU now.
4. Apple's track record with dGPU - the 2008, 2010, 2011 models all have failing dGPUs.

In light of those items, I don't think we'll be seeing any dGPU in the 13" model.

Then this might might be my last mac, i will probably have to get a windows machine flip it to linux. As long as i can get 9-10 hours, I'm good. probably use xfc or something. You talk about product differentiation, and you are right, but after looking at the air and the pro benchmarks, the only thing different between those is the retina screen that is clearly not mature yet.

I hope they don't pull out the dedicated on the 15", that is like the only alternative for me to stay on a mac. they probably won't but if they do, i am done with apple's idiocy.

 

[Radiating]

Ok. You're making a huge assumption that it is impossible to make an interface that works correctly without a dedicated GPU.

I can tell you as a fact that this is a ridiculous assumption and completely false. The rMBP 13.3 has never had any issues for me with the latest model. The first generation was sluggish, but after that there are literally no issues. Apple doesn't set up their store machines in some magical way to make them run faster. In fact they run slower than in the real world due to constantly resetting applications.

It sounds like you just put together a paranoid conspiracy theory.

My advice would be to actually take home a MacBook Retina 13.3 and try it out yourself. You have two weeks to put it through any tests you want.

 

[maflynn]

Then this might might be my last mac

That's your decision, as you need to find a machine that best fits your needs.

I hope they don't pull out the dedicated on the 15", that is like the only alternative for me to stay on a mac. they probably won't but if they do, i am done with apple's idiocy.

My next MBP will be iGPU only. Firstly the Broadwell iGPU is supposed to be much faster then the Iris Pro, which is one of Intel's best iGPU and most of my needs don't require it.

I disagree with you that its Apple's idiocy, most people don't need one, and if the broadwell gpu, can drive a 4k display there's even less reason to have one.

 

[leman]

Even though intel is making great progress in power, there vram is just to low powered to handle pixel pushing screens. vram is critically important to how many pixels a gpu can push instantaneously.

I would say that you are wrong, but what you write simply just makes no sense. You are just spitting out terminology and have no idea what it means.

The memory of dedicated and integrated are just that far apart and will never get close to dedicated power because integrated uses system memory whereas dedicated has their own memory.

Future is in systems where memory is fully shared between the GPU and CPU. The issue is not system memory per se, but the slow speed of the current system memory. Future RAM technology (which is already sampled today) will offer 10x of current DDR3 speeds with a point-to-point connections, making dedicated video RAM completely obsoleted.

System memory has and always will be 2-3 generations behind dedicated memory.

Again, complete nonsense. Please inform yourself about the difference between GDDR and DDR etc. before making such nonsensical claims.

 

[linjac321]

That's your decision, as you need to find a machine that best fits your needs.
My next MBP will be iGPU only. Firstly the Broadwell iGPU is supposed to be much faster then the Iris Pro, which is one of Intel's best iGPU and most of my needs don't require it.

I don't mind iGPU, but the video memory is bullocks. If iGPU can catch up there, i would cry about a dedicated. But sadly at this point in time, theres nothing. As for broadwells increase in gpu performance, thats what they said the 4000 and similarly the 5000 from the 4000, but looks like not much has changed from the ui lag issue of the 13.3". As for the issues for video cards, they have been using nvidia, i know and radeons have issues. I typically avoid and stuff just because they are not aw that good unless its the desktop grade radeons. That red ring of death in the xbox 360 was also due to the same issue.

I am really anal about stutters. I really don't get any with my MBA and i am using it with an external display. 1080p+mba screen is way less than rmbp so thats why i don't have any issues with lag. not even scroll lag on the verge which everybody uses as a benchmark to look for stutters.

I disagree with you that most people don't need one. People who do need one, buy one with dedicated. Especially in the entertainment industry. Musicians, video editors, etc. They don't buy mac pro, they buy 15" dedicated, because they can take their work with them and stuff.

----------

I would say that you are wrong, but what you write simply just makes no sense. You are just spitting out terminology and have no idea what it means.



Future is in systems where memory is fully shared between the GPU and CPU. The issue is not system memory per se, but the slow speed of the current system memory. Future RAM technology (which is already sampled today) will offer 10x of current DDR3 speeds with a point-to-point connections, making dedicated video RAM completely obsoleted.



Again, complete nonsense. Please inform yourself about the difference between GDDR and DDR etc. before making such nonsensical claims.

Again 10x is overestimated numbers, until they have proven through benchmarks. There is no proof that. Companies over exaggerate these numbers to get people to purchase. Just like broadwells "UP TO" 40% faster gpu. Annand tech proved there wasn't a great deal of difference between the 4000 and the 5000 in gpu speed with the macbook air. average increase was about 15-20%. Whereas intel marketed it "UP TO" 50-60% over ivy bridge.

GDDR and DDR are the same technology, the difference is that GDDR is just dedicated to the graphics card and can't be shared with any other component. On top of that shared system memory has a crap load of latency, compared to dedicated graphics memory. hence the stutters you may get with rMBP igpus.

"GDDR or graphics double data rate memory refers to memory specifically designed for use on graphics cards" straight from wikipedia. look things up before stating that i am misinforming people.

btw GDDR6 is going to start emerging soon while ddr4 is just coming out and will take 2-3 years until all new systems will have them.

Based on what i have seen with apple products, they usually do the "just enough" technique to keep their margins high. the just enough to keep the customer happy. They don't really cater towards power users except for the macbook 15" dedicated and they are still pushing it to the "just enough" standard of theirs.

 

[leman]

Again 10x is overestimated numbers, until they have proven through benchmarks. There is no proof that.

There is also no proof that a dGPU is better at desktop compositing than an integrated one. *shrug*. I am talking about maximal theoretical memory bandwidth. Technologies like HMC allowed for 160GB/s in early samples, by time they actually hit the bigger market they will be much faster.

Annand tech proved there wasn't a great deal of difference between the 4000 and the 5000 in gpu speed with the macbook air. average increase was about 15-20%. Whereas intel marketed it "UP TO" 50-60% over ivy bridge.

The "UP TO 60%" was in regards to Iris Pro. The new MBA's GPU performs better with reduced TDB, nothing wrong with that. Sure, a bigger improvement in the GPU would be nice.

GDDR and DDR are the same technology, the difference is that GDDR is just dedicated to the graphics card and can't be shared with any other component. On top of that shared system memory has a crap load of latency, compared to dedicated graphics memory. hence the stutters you may get with rMBP igpus.

DDR3 has lower latency than GDDR5 (which, in your terms is the one with 'crap load of latency'). GDDR5 is faster for GPUs simply because it is optimised for parallel access. In reality, it will be faster or slower than DDR3 depending on the algorithm.

"GDDR or graphics double data rate memory refers to memory specifically designed for use on graphics cards" straight from wikipedia. look things up before stating that i am misinforming people.

They don't really cater towards power users

This is entirely correct and there is nothing wrong with that.

 

[maflynn]

I disagree with you that most people don't need one. People who do need one, buy one with dedicated. Especially in the entertainment industry. Musicians, video editors, etc. They don't buy mac pro, they buy 15" dedicated, because they can take their work with them and stuff.

Why would musicians need a dedicated gpu, when most of their work is not visual but auditory?

Apple's focus is been more to the consumer, who is more interested in Facebook, emails, and office apps. Yes, I'll not deny a lot professionals use the MBP, but that doesn't mean Apple going to focus their time and effort on them. We've seen them move away from pros in a number of ways.

While you may disagree, and that's your right, I don't see most consumers wanting or caring about a discrete GPU.

 

[crawler1975]

I'm going to keep my macbook air until 13.3" rMBP has a dedicated gpu in it. Clearly too many people and I have seen this happen in the Apple store where everything is set up to perfection where there are too many ui issues with integrated graphics. Even though intel is making great progress in power, there vram is just to low powered to handle pixel pushing screens. vram is critically important to how many pixels a gpu can push instantaneously.

The memory of dedicated and integrated are just that far apart and will never get close to dedicated power because integrated uses system memory whereas dedicated has their own memory. System memory has and always will be 2-3 generations behind dedicated memory. It has been like that for years and i don't see that changing in the near future. They need to at least put a 840m in there. 920m in the future maybe.

I would consider the 15 inch but its to big for me to be mobile after using my 13.3" MBA for so long.

I agree with most posters here. You'll be waiting for a very long long time for that to happen ... Apple clearly differentiates their laptops. If you want dedicated graphics then you will need to get the high end 15" rMBP. Apple's bottom line is profit - it is absurd for them to create a 13.3 rMBP with dGPU, knowing that they have the 15" already... Obviously, they don't want their high end product to take a hit. Just my opinion

 

[leman]

Why would musicians need a dedicated gpu, when most of their work is not visual but auditory?

I didn't write that quote But I can kind of answer it — I can imagine that some software that musicians use for their work can benefit from GPU acceleration. Whether the software actually uses GPU acceleration — no idea.

 

[iKrivetko]

They've never had those, and likely never will, sorry to burst your bubble.

 

[linjac321]

yea, i made a mistake higher ddr versions are higher in latency than the lower ones. I was writing that in the heat of the moment lol.

**also musicians, no musicians use dedicated. again i wrote it in heat of the moment.

But most programs in production use it. (I'm talking to you adobe) have high resource usage in memory and they need dedicated video card.

I am just saying that these integrated cards are not ready and won't be ready for retina resolutions at least for the next 2-3 years, if apple really wants to put retina on a 13.3" they need dedicated, at least for now.

I agree, apple wants to build for the majority not the minority. but for now, users that i know using a 13.3" rMBP are experiencing some sort of lag issues on both windows and OS X, due to the igpu. 15" igpu users are also feeling it. I even tested it at the apple store and its the same issue.

Im not totally against igpu. igpus are just not ready for retina resolutions at this point in time.

i like the fact that igpus are getting faster, its the shared memory that is causing it to be slow and even when ddr4 comes out on MacBooks, it will be the same situation until, the have dedicated memory available for the igpu.

I just want a rMBP 13.3" that doesn't experience such issues.

 

[Woochoo]

You may be waiting a long time, i.e., forever. I don't think apple is going to put a dGPU in the 13" MBP, for a variety of reasons.
1. Heat dissipation, the smaller enclosure with one fan will be an issue.
2. Product differentiation between the 15" and 13"
3. Apple is seemingly moving away from the dGPU. Just look at the fact, that only the high end 15" model as a dGPU now.
4. Apple's track record with dGPU - the 2008, 2010, 2011 models all have failing dGPUs.

In light of those items, I don't think we'll be seeing any dGPU in the 13" model.

Add 5. iGPU increases the benfeits margin, since it's included in the own CPU (despite CPU being maybe a bit more expensive than it used to be).

To the musicians one, some of them are not only musicians but also artists, so I guess some of them will need the GPU for visuals (Max+Jitter, Quartz, etc). But I guess an Iris Pro would do the trick with that since that kind of visuals are not exactly like playing Bioshock Infinite in Ultra.

 

[linjac321]

I didn't write that quote But I can kind of answer it — I can imagine that some software that musicians use for their work can benefit from GPU acceleration. Whether the software actually uses GPU acceleration — no idea.

no, gpu acceleration only is towards graphical works or certain computational computing something like CUDA

 

[Samuelsan2001]

The Lag issue again

The lag issue seems to be software related and due to the way that OSX scales and uses transparancy etc. The iGPU is perfectly capable of running the retina screen and indeed the few High def windows machines oyt there run fine on lesser graphics cards than the iris some with higher pixel density than the rMBP. If you have an issue with something please at least find out what is causing the issue rather than making ridiculous sweeping statements.

 

[linjac321]

i think we are going a little off topic here. I created this post because the fluidity of today's 13.3" rMBP have a poor ui experience due to the igpu. The point i was trying to make is that right now, it needs a dedicated video card. but in the next 2-3 years or whenever intel creates a proper dedicated memory for the gpu is when apple should go back to igpu. The fact that i saw it in the apple store on several different models confirms my suspicions.

----------

The lag issue seems to be software related and due to the way that OSX scales and uses transparancy etc. The iGPU is perfectly capable of running the retina screen and indeed the few High def windows machines oyt there run fine on lesser graphics cards than the iris some with higher pixel density than the rMBP. If you have an issue with something please at least find out what is causing the issue rather than making ridiculous sweeping statements.

These are not ridiculous statements. I have seen countless times people with this system, even the i7 versions, lag. sometimes really severely. I have seen it in both windows and OS X. Windows systems with integrated graphics are reporting similar issues too. Its not only macs.

 

[leman]

no, gpu acceleration only is towards graphical works or certain computational computing something like CUDA

First: your sentence makes no sense. Second: its obvious that I was talking about OpenCL/CUDA in context of audio processing.

 

[linjac321]

First: your sentence makes no sense. Second: its obvious that I was talking about OpenCL/CUDA in context of audio processing.

what doesn't make sense about that sentence? I literally said anything process that is graphical (i.e. photoshop, premiere, video encoding, etc.) takes advantage of gpu acceleration. Then, there was my mention about CUDA. I don't think audio processing takes advantage of gpu acceleration from what i have looked at.

----------

Is there anyway to close this thread? I think its gotten a little out of hand. I just wanted to explain why i won't buy rMBP at least for now, and its become a debate about dedicated and integrated gpus and such.

And for those who think that I'm bashing against igpu. i am not. Im just stating its not ready yet at this point in time. But it is getting there and will eventually get there to the point 840m and less dedicated video cards will be irrelevant.

 

[leman]

what doesn't make sense about that sentence?

The grammar. I literally can't parse that sentence. I can only guess what you were trying to say.

I literally said anything process that is graphical (i.e. photoshop, premiere, video encoding, etc.) takes advantage of gpu acceleration. Then, there was my mention about CUDA. I don't think audio processing takes advantage of gpu acceleration from what i have looked at.

You are massively misinformed. The traditional role of the GPU is to speed up application drawing. It has very little to do with image and video processing workflows which have always been done on the CPU. It is only in recent times that the GPUs became flexible and user-programmable which allows image and video processors to use OpenCL/CUDA to provide GPU acceleration for their algorithms. It does not matter much for the GPU whether data it is working on is video or audio, if you program it accordingly.

 

[linjac321]

The grammar. I literally can't parse that sentence. I can only guess what you were trying to say.



You are massively misinformed. The traditional role of the GPU is to speed up application drawing. It has very little to do with image and video processing workflows which have always been done on the CPU. It is only in recent times that the GPUs became flexible and user-programmable which allows image and video processors to use OpenCL/CUDA to provide GPU acceleration for their algorithms. It does not matter much for the GPU whether data it is working on is video or audio, if you program it accordingly.

are you saying if you can program it correctly, a gpu can process cpu tasks?

 

[leman]

are you saying if you can program it correctly, a gpu can process cpu tasks?

That is a gross oversimplification, but in essence, yes. That is what CUDA/OpenCL is all about. Without them, the only use of GPUs would be for gaming and 3D CAD (and it was, until recently).

 

[ee4life]

That is a gross oversimplification, but in essence, yes. That is what CUDA/OpenCL is all about. Without them, the only use of GPUs would be for gaming and 3D CAD (and it was, until recently).

I would add that to further educate oneself about this topic, please research GPGPU. In essence, if your problem fits certain criteria, offloading it to the GPU can yield massive performance improvements. As leman has stated, whether it will accelerate computation has nothing to do with what the data is used for. It is dependent on the operations that will be performed with the data

 

[mcarling]

I'm going to keep my macbook air until 13.3" rMBP has a dedicated gpu in it.

Every 13.3" rMBP ever produced has a dedicated GPU integrated onto the same die with the CPU. They have never had a discrete GPU and they never will because the discrete GPU market is nearly dead, for good reasons.

 

[yjchua95]

I don't mind iGPU, but the video memory is bullocks. If iGPU can catch up there, i would cry about a dedicated. But sadly at this point in time, theres nothing. As for broadwells increase in gpu performance, thats what they said the 4000 and similarly the 5000 from the 4000, but looks like not much has changed from the ui lag issue of the 13.3". As for the issues for video cards, they have been using nvidia, i know and radeons have issues. I typically avoid and stuff just because they are not aw that good unless its the desktop grade radeons. That red ring of death in the xbox 360 was also due to the same issue.

I don't see much of the lag that you and a few others keep complaining about.

Here's a video of my late-2013 rMBP 13" (2.8GHz i7, 16GB RAM, 512GB SSD): https://www.dropbox.com/s/2yq9f5z2uq1t7jy/IMG_6220.MOV?dl=0