Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Model S AP 2.0 picture ( 50 pictures).

This site may earn commission on affiliate links.
No it isn't. You can look at the picture of the cooling unit NVIDIA demos with the unit, it is as big as the system itself, if not bigger: (And notice the external tubes connecting the cooling to the processing unit? (i.e. not self-contained)
nvidia_drive_px_2-100636040-large.jpg


I can't imagine that Elon is going to want those two fans whirring under the dash making a lot of noise, so they are probably going to pipe the cooling to somewhere else.

That is a big unit. It looks like it is the size of about 6 soda cans when the Drive PX 2 is combined with the cooling unit...

That is exactly what I described. It's a closed loop system, i.e. you don't ever access the fluid within, doesn't need refilling or anything else. That's the same kind of set up you'd see in a liquid cooled desktop setup. It is larger than the usual GPU radiator, but it's also having to work in a confined space, possibly without a good flow of outside air even and they don't have the size constraints of a desktop case so that can provide a larger and better radiator. Either way, it is still exactly as I was describing. Just google CPU or GPU liquid coolers if you don't believe me, it would be easier than trying to argue the point...

Chris
 
That is exactly what I described. It's a closed loop system, i.e. you don't ever access the fluid within, doesn't need refilling or anything else. That's the same kind of set up you'd see in a liquid cooled desktop setup. It is larger than the usual GPU radiator, but it's also having to work in a confined space, possibly without a good flow of outside air even and they don't have the size constraints of a desktop case so that can provide a larger and better radiator. Either way, it is still exactly as I was describing. Just google CPU or GPU liquid coolers if you don't believe me, it would be easier than trying to argue the point...

Chris
Its not just a glorified GPU. Instead of goggling what liquid cooled means....it would be better to watch the NVIDIA video. This unit is the opposite of a GPU in a computer. Liquid cooling is not even relevant to me.

This unit captures video and sensor data This unit captures the video of 8 cameras and 16 sensors in real time. Then uses all 24 inputs to determine object recognition (recognizes whether an object is a sign or a car or a person or something irrelevant.), object proximity, character recognition on street signs., IN REAL TIME for each input and keep up with all of that data from millisecond to millisecond. Not only does it do all of that it also keeps memory of what all of the objects are all doing each millisecond and then determines their direction and possible future movements in order to react without error.
Not only does it do all of what we just talked about it also sends live instructions to functions such as braking / acceleration / steering and such without error.

The human brain is amazing in that it can do this by only feeding it potato chips and orange juice every other day. The human brain has even been known to get bored doing all of this at the same time. LOL

That's why I say to everyone. This is a lot for the human brain to do all at once so STOP TEXTING AND DRIVING.

 
  • Funny
Reactions: darker123
Its not just a glorified GPU. Instead of goggling what liquid cooled means....it would be better to watch the NVIDIA video. This unit is the opposite of a GPU in a computer. Liquid cooling is not even relevant to me.

This unit captures video and sensor data This unit captures the video of 8 cameras and 16 sensors in real time. Then uses all 24 inputs to determine object recognition (recognizes whether an object is a sign or a car or a person or something irrelevant.), object proximity, character recognition on street signs., IN REAL TIME for each input and keep up with all of that data from millisecond to millisecond. Not only does it do all of that it also keeps memory of what all of the objects are all doing each millisecond and then determines their direction and possible future movements in order to react without error.
Not only does it do all of what we just talked about it also sends live instructions to functions such as braking / acceleration / steering and such without error.

The human brain is amazing in that it can do this by only feeding it potato chips and orange juice every other day. The human brain has even been known to get bored doing all of this at the same time. LOL

That's why I say to everyone. This is a lot for the human brain to do all at once so STOP TEXTING AND DRIVING.


Goodness, where you determined it is the "opposite" I'll never know. The PX 2 board is based on exactly the same Pascal GPU chips found on the current line of NVIDIA graphics cards. It's not magic, I know it sounds so very mystical that modern hardware can do these kinds of things but it's really not. I'm not even sure why we are arguing about this, or why the need exists to make the hardware out to be more than it really is. It's great stuff don't get me wrong but it's not magic or way ahead of its time. It's current technology and what sets Tesla apart is that they are using current technology instead of 3-5 year old stuff like most car companies. I'd recommend a little research in the future or at least figure out why and what you are trying to argue about. I just asked if the cameras in the fender forced a redesigned fender and for some reason the crazies come out of the woodwork trying to make this board into the second coming or something.

Nvidia Drive PX 2 Uses Integrated and Discrete Pascal GPU Cores - 24 DL TOPS, 8 TFLOPs and Up To 4GB GDDR5 [Updated]
 
Goodness, where you determined it is the "opposite" I'll never know. The PX 2 board is based on exactly the same Pascal GPU chips found on the current line of NVIDIA graphics cards. It's not magic, I know it sounds so very mystical that modern hardware can do these kinds of things but it's really not. I'm not even sure why we are arguing about this, or why the need exists to make the hardware out to be more than it really is. It's great stuff don't get me wrong but it's not magic or way ahead of its time. It's current technology and what sets Tesla apart is that they are using current technology instead of 3-5 year old stuff like most car companies. I'd recommend a little research in the future or at least figure out why and what you are trying to argue about. I just asked if the cameras in the fender forced a redesigned fender and for some reason the crazies come out of the woodwork trying to make this board into the second coming or something.

Nvidia Drive PX 2 Uses Integrated and Discrete Pascal GPU Cores - 24 DL TOPS, 8 TFLOPs and Up To 4GB GDDR5 [Updated]
When I say opposite I'm quoting the CEO of Nividia himself.

The GPU of a computer is designed to process data from its processors and OUTPUT video to a screen.
Data creates Video

The GPU of AutoPilot is designed to receive video and sensor INPUT and process data for its processors.
Video creates Data


I'm not arguing about anything.
 
Last edited:
  • Informative
Reactions: MorrisonHiker
It appears that the Computer PX2 is different than what's known as Drive PX2. Only 10 Watts of power consumption? What? WOW.

Some Computer super duper GPU's sometimes require new cyclopean power supplies.

The new single-processor configuration of DRIVE PX 2 for AutoCruise functions —which include highway automated driving and HD mapping—consumes just 10 watts of power. Plus, it enables vehicles to use deep neural networks to process data from multiple cameras and sensors.

DRIVE PX 2 can understand in real-time what's happening around the vehicle, precisely locate itself on an HD map, and plan a safe path forward. It's the world's most advanced self-driving car platform—combining deep learning, sensor fusion, and surround vision to change the driving experience. - See more at: Autonomous Car Development Platform from NVIDIA DRIVE PX2
 
When I say opposite I'm quoting the CEO of Nividia himself.

The GPU of a computer is designed to process data from its processors and OUTPUT video to a screen.
Data creates Video

The GPU of AutoPilot is designed to receive video and sensor INPUT and process data for its processors.
Video creates Data


I'm not arguing about anything.

Traditionally yes, GPU's just display data, however that last several generations have been utilized for "compute" functions as well. For instance, photoshop uses the compute functions to complete certain types of calculations many many times faster then a general CPU can, games use GPU's to perform the physics calculations, SETI@home uses it to analyze radio telescope signals in the hopes of finding alien civilizations. GPU's haven't just display graphics to the screen for a long time. They are excellent at performing many parallel instructions at once. The higher end Pascal chips have 3500+ "compute units" and each one of those have 8 or so "cores" but they are simple compared to a traditional CPU core. Luckily, signal processing (like from the radar or ultrasonics) and image processing/recognition from the cameras are things which GPU's excel. Likewise almost every deep learning server farm incorporates 100's if not 1000's of those same types of GPU's to process the info uploaded. That's the reason the traditional GPU companies are expanding into the automous driving world with their processors, it's a great growth opportunity for them and their processors are uniquely suited for it.

Chris
 
  • Like
Reactions: justg0
Traditionally yes, GPU's just display data, however that last several generations have been utilized for "compute" functions as well. For instance, photoshop uses the compute functions to complete certain types of calculations many many times faster then a general CPU can, games use GPU's to perform the physics calculations, SETI@home uses it to analyze radio telescope signals in the hopes of finding alien civilizations. GPU's haven't just display graphics to the screen for a long time. They are excellent at performing many parallel instructions at once. The higher end Pascal chips have 3500+ "compute units" and each one of those have 8 or so "cores" but they are simple compared to a traditional CPU core. Luckily, signal processing (like from the radar or ultrasonics) and image processing/recognition from the cameras are things which GPU's excel. Likewise almost every deep learning server farm incorporates 100's if not 1000's of those same types of GPU's to process the info uploaded. That's the reason the traditional GPU companies are expanding into the automous driving world with their processors, it's a great growth opportunity for them and their processors are uniquely suited for it.

Chris
The entire platform of Tesla is non-traditional. Actually the NVIDIA videos and quotes I've been making are old news. How old does news have to be to now be considered in the traditional category.

There is a saying: Old news to new people is considered new news.

Volvo has had this NVIDIA technology on their ICE cars for years now. Its new news to TESLA, however its been around for a while.

Tesla AP has even improved on the NVIDIA videos as I understand it. I suppose they will keep us posted.
 
Last edited:
I haven't looked, do any other manufacturers offer an option for a dash cam from the factory?

Yes. GM introduced the PDR option on the Corvette in 2014. It also works to monitor valet usage.

Performance Data Recorder: Corvette Technology | Chevrolet

It gets great reviews as well.

GM turned to Cosworth to help develop the new PDR, and the system pairs a high-definition 720p video camera with GPS tracking and vehicle telemetry. When joined with the free Cosworth Toolbox telemetry software, the PDR becomes one of the most powerful driver-improvement tools out there.

Check the option box on the 2015 Corvette order sheet, and GM will install your PDR system at the factory. The unit tucks up near the rear-view mirror, and records the driver's point of view through the windshield. It also captures in-car audio.

The 2015 Corvette's dash cam is insanely powerful

Hopefully one day Tesla will catch up with GM. ;)