What would an nvidia desktop cpu be like?

>>57915047
chicken

It'd burst into flames before we found out.

>>57915047
it wouldn't work
but since Nvidia would tell everyone the they do work, it's going to have the best benchmark results in the world.

cloacked because it will never happen since intel and amd will rather die before letting nvidia do anything lol

It would be ARM

it would be lit

I hope x86 dies

File: nvidia desktop solution.jpg (142KB, 612x612px) Image search: [Google]

142KB, 612x612px

>>57915047

>>57915047
It would be absolutely blazin' like the nVidia GT 420

>>57915374
this is AMD

>>57916564
>ARM

This. And it might actually happen given Microsoft's recent announcement.

Would literally burst into flames.

>>57916740
Amd didn't make the Fermi 480 and 590 though

>>57915047
It would be absolutely amazeballs

>>57916914
but made bulldozer

File: House_Fire_www_normancope_com.jpg (2MB, 3072x2048px) Image search: [Google]

2MB, 3072x2048px

>>57915047
Probably something like this

>>57915047
Would rival Intel in terms of performance but it'll be 2x more expensive.

File: IMG_1144.jpg (2MB, 5184x3456px) Image search: [Google]

2MB, 5184x3456px

>>57915047
You can get the JETSON board and use it as a desktop.

>>57916952
Which was hot because glofo's shitty node leaked like a motherfuck.

>>57917371
This. Tegra on a desktop? It'd be worse than Fermi.

>>57917794
nothing wrong with fermi

>>57916577
Underrated

GARBAGE it would be GARBAGE and inturn it would turn the GPUs into garbage as well they ONLY need to focus on the GPUS

File: Mad-K6.png (110KB, 307x277px) Image search: [Google]

110KB, 307x277px

>>57917851
They make CPU+GPU combos in the Tegra series.

>>57917909
and they are garbage they are why the nintendo switch is going to SUCK

>>57917981
Except for being expensive as fuck. What is it that suck about them?

>>57918015
they are slow they run hot the drink up battery like its water they are noly capable of 1080p max

they suck

>>57918038
They seem to be engineered to specific use-cases.
Where high resolution display-output is not the main focus.
And what do you mean they run hot. Most of what I have been able to find about them states very little power-consumption. In the <20W range for the entire SoC and surrounding chips.

So they are slow in what way? From the datasheets they appear to have quite some power, and that they are CUDA enabled gives room utilise a lot of parallel computing.

Remember that OP asked for Desktop CPUs. Battery is only a problem if you run on UPS.

>>57918125
THEY ARE SHIT JUST FACE IT THE ONLY THING ANYONES NEEDS OR WANTS ARE FULL SIZED DESKTOPS WITH MOUSE AND KEYBOARD

>tegra chips were ahead of its time. They didn't really take off because the socs didnt come with an integrated modem like the Qualcomm ones which drove the costs up.
TDP was bit on the high side for phones too. So we only saw them in tablets. They also deliver very late to the market

>>57915047
3.5 cores

>>57918015
They always get beat by Qualcomm and Apple's offerings.

>>57918752
what the fuck? when did i enter economics studies? why offerings instead of saying chips?

>>57918875
Because it's not as simple as "Tegra X is faster than Snapdragon 820"

You also need to factor in LTE integration, power consumption, manufacturing costs, etc. A smart phone maker cares about performance but they care more about their SoC provider being able to make something that does everything well, not just render a lot of pixels.

>>57918930
ok? its still a chip though why say offerings like we are in a board meeting 4chan is no place for professionalism

>>57918534
underrated

>>57918979
>being this much of an aspie about terminology that everyone understands anyway.

Bro u r the 1 b-ing overly-professional

>>57916601
>wanting BEST architecture to die
why tho my friendo why

File: nothing but inane comments.jpg (127KB, 500x500px) Image search: [Google]

127KB, 500x500px

>>57919448
I think you forget where you are.

>overpriced
>bloated and telemetry ridden drivers
>wood screws
>3.5gb
>coil whine
>20% reduced performance after a year

It would definitely be an experience

>>57918289
>So we only saw them in tablets
Even then the high power use and low performance relative to competing options still limited their use. Remember that shield disaster? I almost felt bad for nvidia, they couldn't get anyone to use their POS so they used it themselves and even then no one would buy it.

File: 1425083600921.png (181KB, 280x273px) Image search: [Google]

181KB, 280x273px

>>57915047
Hot as if Pentium 4 was made by AMD

File: spp[1].jpg (107KB, 600x392px) Image search: [Google]

107KB, 600x392px

'member nForce chipsets

File: speccy3.png (43KB, 953x623px) Image search: [Google]

43KB, 953x623px

>>57922537
>~40°C in idle

jesus no

>>57918038
>They are slow
Except they're as fast as Qualcomm's offerings from the same era. Tegra 4 is as fast as the 801 etc.
>They run hot
No not really
>They drink up battery
They're just as power efficient
>They are only capable of 1080p
Really? Because my Nvidia Shield portable can output 4k @ 30Hz, and I'm pretty sure the Nvidia Shield Console can do 4k @ 60Hz.

>>57915047
The Way Its Meant To Be Processed™

>>57916768
Unlikely. The JIT emulation of X86 code to ARM exists so that Microsoft can preemptively corner the ARM server market as it emerges. If it ever emerges, its a pretty big gamble. They're showing that new platforms will potentially not be lacking any legacy support while the ecosystem for ARM software builds.

No one has ever demanded an ARM desktop, no one aside from internet autists anyway. Manufacturers go to where the money is, and so far the only proven market for ARM parts is low margin high volume mobile and embedded.


>>57917545
It was IBM's process, GloFo just did a terrible job with it, and they were really late to market.

>>57918038
nvidia ARM processor has performance far above anything the other vendors are selling. To a point where it might be capable of competing with low end Intel Haswell.. if nvidia had the patents to use x86 and x86-64.

>>57919448
>x86
>best
Not really, it's more like the most well-optimized architecture with the most widespread software dependancy

>>57922707
They never will, unless AMD is divested upon its death. There's no way NVidia could ever get their hands on its graphics division.

>>57915047
>ITT: speculations and circle jerking of buttmad AMDrones.

Tbh it's rather amusing seeing /g/ latching to shit while convincing everyone else it's a bar of gold.

>>57915047
it would probably go out horribly for the first few years, causing massive damage to reputation. this is because nvidia would have to create its own cpu technology without causing on patent infringements on intel and amd technology.

same goes for intel not making graphics cards, that's why intel has made agreements with amd to use amd graphics cores for their future igpu's.

Tegra X1

Its buggy.

>CPU is marketed to have 4mb of cache but has 3.5mb instead

>>57922796
Of course they never will. Intel will not give them the x86 license. Well not never, Intel might be willing to license it at some point. ARM is spreading like cancer.

>>57922846
Nvidia already has their own CPU, it's an ARM CPU. It could be converted to an x86 CPU as well, since it is like a transmeta processor that uses instruction translation or code morphing as transmeta would call it.

>>57922594
I want nForce to return.

>>57922946
Pretty sure that because of antitrust Intel will always have at least one licensee of x86 to compete with.

>>57915047
arm

>>57923002
Yea but I really don't see AMD going belly up any time soon. I'm assuming they can stay around for a while, even if they only offer token competition.

>>57922707
>nvidia ARM processor has performance far above anything the other vendors are selling
Bwahahaha no. It's dogshit to the point where no one would touch it. Then nvidia tried to sell their own hardware and that was so bad that consumers wouldn't touch it. I'm amazed that they haven't given up.

File: 1478649985955.png (387KB, 2560x1440px) Image search: [Google]

387KB, 2560x1440px

>>57916768

they already making the product MS is just reacting to new platforms not the other way around

>>57923060
Basically this.
They never really delivered on the performance they were promising and Qualcomm has always been faster.

>>57916577
ayy

>>57923060
They just started selling their mobile SoCs to markets where power consumption doesn't matter.

>>57922707
ahahahahahahahahaahahahahahahahhahahahaahahahahahahahahahahahahahahahahahhahahahahahahahahahahahahahahhahaahahahahahahahhahaahahahhahahahahahahahahahahahahaahahahahhahahaahahahhaahahhahahahahahahahaahhahahahaha


aaaaaaaahahahahhahahahahahhahahahahahahhahahahahahaahahahahahhaha

>>57915047
a fire hazard

>>57922707

too bad the if amd or via are sold they lose their license for x86 as per existing agreements

amd would have had its revoked for squiring ati but intel needed the amd64 extension

>>57923140
acquiring fuck you mind control

File: 1WvCwUvY2jI.jpg (50KB, 453x604px) Image search: [Google]

50KB, 453x604px

>>57915047

worse than intel jewery I can tell you this much

>>57923140
There is so much wrong in this post that it is downright hilarious. I suggest you go look this stuff up instead of believing the stuff on /g/.

Part of the antitrust court cases between AMD and Intel was that Intel must come with a damn good reason to deny an x86 license to any third party that would buy AMD. As for the whole acquiring ATI thing, I have no idea where you got that or what you're smoking. Do you have any of it left? Would you mind sharing?

>>57923060
Seems like the performance is really good to me. It's not consistent, but it does rely on translation of instructions which makes it perform poorly if you don't feed it with enough recurring instructions.

It's a really interesting CPU. I'm almost more interested in seeing what nvidia can do with it than I am interested in seeing what Zen is like.

>>57923252
>It's a really interesting CPU
It is, though that's mostly because of the whole transmeta background. The rest of it is pretty boring.

It's a curiosity. It underperforms, draws too much power, spits out too much heat and is lacking in any substance other than nvidias graphics technology but it's interesting. It's just a shame that interesting wasn't enough to keep transmeta alive and it probably won't be enough to keep the Tegra going.

>>57923216
lmaoing @ your life

>>57915047
Overpriced.

>>57916952
>>57917545
Hate to be the barer of bad news, but my i7 gets way hotter than my 8150 did on the same cooler. Just saiyan.

>>57923345
I think nvidia can keep it going and keep iterating on it. They have the money to do it. They kind of have to too, if they want to stay in the SoC market. If they were to continue using the same ARM CPU designs as everyone else combined with their GPU, who'd buy their product? They offer nothing except the GPU and it's not that good.

They did also manage to sell their CPU to Nintendo. So it certainly ain't dead yet. I think in video games, that processor will not underperform. That's exactly the kind of work load it's going to be good at. Lots of repeating instructions.

File: TEGRA STRONK.jpg (91KB, 680x384px) Image search: [Google]

91KB, 680x384px

>>57923448
>They offer nothing except the GPU and it's not that good.
But that's exactly what they're doing now. The current CPU is terribad, by switching to generic ARM cores and tweaking them they could officially upgrade their standing to "thoroughly mediocre."

Even their launches pretty much admit this with the entire focus being the graphics tech.

>>57923517
That's the point though. Maybe the CPU doesn't cut it for phones and tablets right now, but it has something different to offer the market. If they used generic ARM cores what do they offer, exactly, that Qualcomm does not? The answer is fuck all.

There is a higher risk in going with their own CPU design, but at least there is a possibility that they'll actually stay in the market instead of getting muscled out. You don't need like 100 versions of the exact same shit.

>>57923605
>but it has something different to offer the market.
Such as? The scrapped potential to run x86 code some time in the nebulous future when Intel decides to let others in on the action?
>You don't need like 100 versions of the exact same shit.
You don't need to reinvent the wheel either, especially if your version comes with a flat tyre.

>>57923669
Exceptionally high performance if repeating instructions are used. The performance is only bad if the instructions that the CPU is executing do not repeat much, resulting in the CPU executing mostly ARM code instead of having the optimizer spit out already pre-made instruction packages.

And they aren't reinventing the wheel. Nobody was buying nvidia products on the market. They stopped selling the wheel that nobody wanted and now they are on the market with robotic legs. If you want to draw that analogy to a conclusion.
You never know, someone might want robotic legs.

>>57917450
will it run Linux?

>>57923787
>quad-core ARM Cortex-A15

of course

>>57923787

nVidia have a specialised Ubuntu version for the board.
But you can install anything that can run on ARM.

File: cf6.png (340KB, 506x658px) Image search: [Google]

340KB, 506x658px

>>57915047
Workworks™

File: workworks™.png (119KB, 640x201px) Image search: [Google]

119KB, 640x201px

>>57924007

>>57916577
kek

>>57915047

It would be like the cell and everything would work out and be perfectly optimized and it would be the best thing that ever happened and and and

File: first result for 'pc on fire'.jpg (149KB, 1920x1080px) Image search: [Google]

149KB, 1920x1080px

leaked pic of prototype nvidia desktop PC
it's said that it reaches the boiling point of peak performance

>>57924352
and what

>>57924380
>woodscrews

>>57924390

intel would have to make good cpu's again