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.
Nvidia Maxwell released
- Thread starter leman
- Start date
- Sort by reaction score
So what? It doesn't relate neither to Macbooks, nor to nVidia Maxwell GPUs (870m is Kepler one).
p.s. 850m and 860m are full 640-core GM107 with GDDR5 memory! so excited about it!
Not sure if this was already mentioned here, but Anand has the list of 8XXM GPUs:
http://www.anandtech.com/show/7834/nvidia-geforce-800m-lineup-battery-boost
The 850M will be Maxwell, and 22% faster than the 760M - with a 750M like TDP I assume.
Looks quite impressive, in particular the comparison with Iris Pro on page 4.
http://www.anandtech.com/show/7834/nvidia-geforce-800m-lineup-battery-boost
The 850M will be Maxwell, and 22% faster than the 760M - with a 750M like TDP I assume.
Looks quite impressive, in particular the comparison with Iris Pro on page 4.
TDP will actually be 5-7 TDP lower than the current 750M. Which is why the 860M actually becomes viable. But it won't happen
It would be awesome. Anand say the 860m should provide near enough the same performance as the desktop 750Ti. That would be epic.
I see lots of Nvidia marketing department PowerPoint slides, but no actual benchmarks. Marketing department muppets tend to be notoriously inaccurate and everything they say shouldn't just be taken with a pinch of salt, but a fistful.
With the desktop 750 and 750TI they let the chips get benchmarked right off the bat, now with these things we're going to have to wait until next month and this has me slightly worried. Are they intentionally trying to delay actual benchmarks so that they can freely boast more than what they actually should?
With the desktop 750 and 750TI they let the chips get benchmarked right off the bat, now with these things we're going to have to wait until next month and this has me slightly worried. Are they intentionally trying to delay actual benchmarks so that they can freely boast more than what they actually should?
Nvidia waiting until Microsoft announces DirectX12 at GDC on March 20th since they are going to be announcing that the upcoming Maxwell fully supports DirectX12. They'll also be showing more of Maxwell at the conference in late March, which they're going to release Maxwell mid/high end cards.
Where did you get all of that? never heard of anything similar. Maxwell will support DX12? Higher end Maxwell GPUs will be presented in the end of March? wtf?
Where did you get all of that? never heard of anything similar. Maxwell will support DX12? Higher end Maxwell GPUs will be presented in the end of March? wtf?
Oh you know
Yes I do believe Maxwell will support DX12, but regarding the Maxwell mid-high Maxwell GPU's they may just be announcing or leading into their official announcement later on. I never said anything about shipping immediately.
I don't think benchmarks will have any surprises. The spec of the mobile versions is known, the worst-case performance can be easily estimated from what we know about the 750 Ti. And - benchmarks of mobile cards often come later, because you actually need a laptop to benchmark them. Making a laptop usually takes longer than building a reference graphics card...
Yes, laptops are only starting to show up around now. There are only a handful on the market (as of right now), and even with some of the 860M ones, there seem to be two versions; one of which is a Keplar rebrand.
http://www.notebookcheck.com/Test-Schenker-XMG-P304-Clevo-W230SS-Notebook.112827.0.html
860m is about equal to 770m. Massive jump there.
The fundamental problem with your logic is that you assume absolute performance is one of the most important (if not the most important) criteria for Apple when making these decisions. There's quite a bit of evidence to suggest that it isn't. Is it more likely than not that we'll see one more round of dGPUs? Meh, maybe. Highly likely? I don't think so.
Even if this is true, the gap between the Iris Pro 2 and the Maxwell GPU will be higher than between the Iris Pro and the current dGPU...
Moore's Law and the last few decades of year on year increases in performance seem to back that up. Apple has already managed huge improvements to battery life, laptop size and weight etc. Even with these new GPU's and CPU's we should see a good jump in battery life plus a good improvement in performance. I don't see what else Apple would be aiming for?
'Here's the new 2014 rMBP, we haven't improved performance instead we erm, um, eh...' What exactly are you suggesting they would do? These are expected, fairly simple upgrades that should offer good improvements.
Which is why I think there's a good chance Apple will not use Maxwell in the first place. It's great for marketing to show benchmarks vs your previous generation. There are no benchmarks vs the product you never created...
The "other thing" Apple would be aiming for is increased profitability and simplified architecture. You're focused on the metrics of performance. From a business angle, that's only one dimension among many. Leaving graphics as-is and just doing a minor processor spec bumpwhich is what they did a year agois a very viable option.
A 40% performance jump in Iris Pro 2 would put it past the 750M. And remember this is at a FAR lower TDP than any discrete solution can produce.
Intel does not need to have their iGPU be performing even with the latest and greatest. All they need to do is be "in the game" - and performing like Iris Pro and its successors do is exactly that. From that point the far lower TDP and power needs makes the offering compelling.
Intel does not need to have their iGPU be performing even with the latest and greatest. All they need to do is be "in the game" - and performing like Iris Pro and its successors do is exactly that. From that point the far lower TDP and power needs makes the offering compelling.
Doing maths, a 40% increase in performance will give 3500 pts in 3dMark11. Which will be way ahead of GT750M.
The problem is - Nvidia Geforce 830-840M will be at the same level of performance. 860M will be in his own league of performance. Looking by the numbers and performance levels it will be between GTX770M and 680M. Which is pretty mind blowing.
And looking at competition - not putting at least GTX850M in Broadwell MBP will be just laughing at customers, in their faces.
That's one interpretation. Another is simply that they deem Broadwell's integrated graphics to be "good enough," which is a perfectly reasonable statement.
Indeed. It has to be remembered how critical TDP and power consumption are for mobile parts. The 860M is by itself a 45W part minimum - if they use the 4GB version it is 75W. Then you add the Broadwell CPU TDP to that which is expected to be about 47W. So the CPU-dGPU combination comes in at 92W which has big implications for cooling and battery life.
Meanwhile Iris Pro / Iris Pro 2 sails in at 47W for everything.
Remember, the whole case for integrated GPUs has always been the much lower power and thermal requirements. And in the past the tradeoff was dreadful performance. All Intel needs to do with the Iris Pro is keep its performance in the proverbial ballpark of middle range dGPUs and the thermal and power considerations make the rest of their case. And their case is not to replace all discrete GPUs even in high end gaming rigs - they simply are making the case for not bothering with the lower end dGPUs and many middle level ones and just using them instead.
Last edited:
Apple is not bound to this. Mac Pro GPUs have 109W of TDP and are made by Apple. Apple is licensing GPUs from AMD.
Guess in what market Nvidia just went? Yep, licensing GPUS.
In other words. If Apple will want - they will lower the TDP and increase the GPU clocks - the same what they did in DXXX from Mac Pro, and what they did to GT650M in retina Macbook Pro.
Everything points out that this is how Apple will be looking at GPUs.
This is MacBook Pro not Mac Pro. Desktops don't have the same TDP requirement. The only Change might be offering a 13 with the i7Q and Pro.
Thats is just wrong logic.
How can CPU+GPU be at 47W and
CPU alone is 47W? Remember when 650M/750M is running the integrated GPU is sitting idle.
Case in point, a 2012 15" macbook pro with 45W Ivy Bridge processor and 650M Nvidia graphics card uses 77.2W at full load:
http://www.notebookcheck.net/Review-Apple-MacBook-Pro-15-Retina-2-3-GHz-Mid-2012.78959.0.html
A 2013 13" Macbook pro with a 28W CPU/GPU uses 60.6W at full load
http://www.notebookcheck.net/Review-Apple-MacBook-Pro-Retina-13-Late-2013-Notebook.105035.0.html
Looking at theoretical:
15" 2012 Macbook Pro
Theoretical: 45W CPU + 45W GPU = 90W
Reality: 77.2W total including display, SSD etc.
13" 2013 Macbook Pro
Theoretical: 28W CPU/GPU
Reality: 60.6W total including display, SSD etc.
This either tells us the combined CPU+GPU is massively exceeding the TDP of the chip and/or
When the the machine with discrete graphics is running anything GPU intensive, the combined CPU is running nowhere near TDP as the integrated GPU is idle.