It could be worst... Apple could take example on the defunct Sega of America (Saturn/Dreamcast fiasco)
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.
2016 nMP
- Thread starter ManuelGomes
- Start date
-
- Tags
- nmp
- Sort by reaction score
- Status
It could be worst... Apple could take example on the defunct Sega of America (Saturn/Dreamcast fiasco)
Amd used to have reasonable good fp64 performance, now it's a joke ever seems it's faster to emulate fp64 than use its native double precision
People believe that ATI cards get 0 TFLOPs on the huge number of CUDA applications - and they're right.
(And don't bring up one example (or a few) of a cross-compiled app as a counter....)
Osborne Effect doesn't require the company to go out of business. The definition is about pre-announcing something that craters your sales.
You are right, Aiden. The question is, how much longer proprietary software will bring advantage for Nvidia. Looks like tide has changed lately dramatically.
So far all it looks like they are truly threatened by what AMD is doing lately. And they have to be, cause software is bringing what their hardware lacks. But software can catch up fast. Is this the same case for hardware?
That is why I believe Apple did the right thing with picking AMD. Apple is working on their own Proprietary solution, but one that can work on ANY hardware. Best hardware will always win then. Future however is long. Tide can once again change.
So far all it looks like they are truly threatened by what AMD is doing lately. And they have to be, cause software is bringing what their hardware lacks. But software can catch up fast. Is this the same case for hardware?
That is why I believe Apple did the right thing with picking AMD. Apple is working on their own Proprietary solution, but one that can work on ANY hardware. Best hardware will always win then. Future however is long. Tide can once again change.
Last edited:
I didn't want to comment but I read on an Hispanic blog about Apple that due the thermal issues with D700 and slow opencl grew Apple reconsidered to include nVidia GPU at least on the 2016 nMP (this is something to take with an grain of salt) also AMD delays maybe pushing back nVidia, I don't don't use either cuda or opencl as to decide on hardware based on its availability but it's undeniably user's cuda preference.
Time will tell. If Nvidia will come with great hardware, I will welcome it on Apple ecosystem again. BTW, GTC starts in few minutes
.
Nvidia keynote starting now (9AM PDT) at:Time will tell. If Nvidia will come with great hardware, I will welcome it on Apple ecosystem again. BTW, GTC starts in few minutes
http://www.ustream.tv/channel/fWbQyaEMfbh
or
http://www.gputechconf.com/
[doublepost=1459872443][/doublepost]
Mostly true - based on the definition of "crater". If Apple sold zero MP6,1s it would have next to no effect on their sales figures.
https://devblogs.nvidia.com/parallelforall/inside-pascal/
Tesla is a 300 Watts chip. Still, 5304 GFLOPs is pretty amazing figure. HBM2 with ECC.
Tesla is a 300 Watts chip. Still, 5304 GFLOPs is pretty amazing figure. HBM2 with ECC.
Last edited:
3584 CUDA cores? That was absolutely unexpected. With 300W of TDP, and HBM2 on 16 nm node?
And I was called as AMD fanboy when pointed that Pascal will not be as power efficient as Maxwell was...
But I can understand that TDP comes from extremely high boost clock. With that core clock, It will be BEAST.
And I was called as AMD fanboy when pointed that Pascal will not be as power efficient as Maxwell was...
But I can understand that TDP comes from extremely high boost clock. With that core clock, It will be BEAST.
So one Tesla chip is going to cost around USD $3800...? Just made a quick assumption according from DGX-1 $129.000 price.
Update: There's only eight Tesla units in DGX-1.. but that still doesn't tell how many chips are included in one Tesla unit..
Update: There's only eight Tesla units in DGX-1.. but that still doesn't tell how many chips are included in one Tesla unit..
Last edited:
$4K to $5K is the typical starting price for all of the Teslas, and has been for years.
On a new node, that delivers 70% reduction in power consumption accompanied with HBM2 which uses much less power than GDDR5.
And wouldn't have any software to run.nMP Late 2016 with a couple Power CPUs and a couple of Pascal GPUs linked through NVLINK would sound great, but that would cost you an arm and a leg, or both each, and then some
Too bad AMD lost the chance to first "announce" a HBM2 GPU. They co-developed the tech, I believe they should have had the need to show it first. Maybe they are still first one out the door...
Maybe Apple will go back to Power, who knows?! Hardly, but...
Said it before, and I understand people like the compatibility, but to me x86 would die already. I know, it's well optimized, over time, but it's still a lot of patches over 30+ year old tech, patched from the get go and along the times, to maintain compatibility with software, that for the same reason also lacked real improvements.
How is it possible that you look inside the Windows folder and still see there files that are decades old?
Maybe now that different architectures get attention, and software is portable between platforms, things will change.
Or not...
Said it before, and I understand people like the compatibility, but to me x86 would die already. I know, it's well optimized, over time, but it's still a lot of patches over 30+ year old tech, patched from the get go and along the times, to maintain compatibility with software, that for the same reason also lacked real improvements.
How is it possible that you look inside the Windows folder and still see there files that are decades old?
Maybe now that different architectures get attention, and software is portable between platforms, things will change.
Or not...
Intel/HP owns an much better cpu architecture than arm and powerX, Itanium ia64 but it's semi abandon by Intel and hp barely builds servers on this, Itanium updated with the new tech on Xeon could beat both power and arm.
Mago, I know. Itanium was my hope for the future until it wasn't. I don't expect it to come back though, no one seems to care about it anymore, unfortunately.
I'd love to see Itanium based Xeons, on the latest node, they would rock for sure.
Too bad it didn't survive other interests, it was indeed a much better, future proof, arch.
Maybe it will come back, but MS lost interest, Apple won't port OS X for sure, not just for just for a possible nMP based on it. I guess that would make us both happy
And now comes Aiden saying something smart about it...
I'd love to see Itanium based Xeons, on the latest node, they would rock for sure.
Too bad it didn't survive other interests, it was indeed a much better, future proof, arch.
Maybe it will come back, but MS lost interest, Apple won't port OS X for sure, not just for just for a possible nMP based on it. I guess that would make us both happy
And now comes Aiden saying something smart about it...
- Status