NVIDIA Tegra Architecture

I was making a remark about how Nvidia has been using copy-pasted ARM designs, which has lead to lackluster products.

lackluster in what ? surely not in CPU performance where nvidia was always at the forefront of SoCs for phones/tablets (first dual core A9, first quad core A9, first quad core A15)
now, if we talk GPU, its another story...
 
lackluster in what ? surely not in CPU performance where nvidia was always at the forefront of SoCs for phones/tablets (first dual core A9, first quad core A9, first quad core A15)
now, if we talk GPU, its another story...

First quad core A15? You can buy a phone with a quad core A15 from Samsung, but you can't buy any quad core A15 product that has an NVIDIA SoC. That to me means Samsung was first. For Tegra 2 it was a little unfortunately that their dual core A9 implementation left a little to be desired. Apart from that, NVIDIA has been pushing CPU performance quite a bit.
 
First quad core A15? You can buy a phone with a quad core A15 from Samsung, but you can't buy any quad core A15 product that has an NVIDIA SoC. That to me means Samsung was first. For Tegra 2 it was a little unfortunately that their dual core A9 implementation left a little to be desired. Apart from that, NVIDIA has been pushing CPU performance quite a bit.
for quad core A15, it all depends where you live, SHIELD will be available this week in all main US retailers (june 27th). Galaxy S4 is with S600 in US, not Exynos...
 
First quad core A15? You can buy a phone with a quad core A15 from Samsung, but you can't buy any quad core A15 product that has an NVIDIA SoC. That to me means Samsung was first. For Tegra 2 it was a little unfortunately that their dual core A9 implementation left a little to be desired. Apart from that, NVIDIA has been pushing CPU performance quite a bit.

I think NVIDIA's strategy is, basically, to rush to market with whatever the new thing is. With Tegra 2 it was dual-A9, with Tegra 3 it was quad-A9, now it's quad-A15 but they're a bit late.

In practice, Tegra chips were always pretty bad compared to their competitors, but they were also first, so for a few months it didn't matter. It's not necessarily a bad strategy when you don't have the same level of expertise and resources as your competitors (Samsung, Qualcomm) but it does require you to successfully get your chip to market first.
 
for quad core A15, it all depends where you live

If we were to only judge based on where you live then we could probably find countries were NVIDIA never was first with either dual or quad core A9. In other words, going by your logic, NVIDIA never was first.
 
So who is the licensing customer? Intel, Apple, or Samsung would seem the most likely, and I doubt Nvidia would bother to announce it unless they already had a customer...
 
Intel seems unlikely. Besides, Intel already has a patent licencing agreement with NVIDIA and is already putting quite some effort into their own in house designed GPUs.
As for Samsung and Apple. Of those two, I think Samsung is the more likely to strike a deal with NVIDIA.
 
If we were to only judge based on where you live then we could probably find countries were NVIDIA never was first with either dual or quad core A9. In other words, going by your logic, NVIDIA never was first.
funny how you turn the slide to always put nvidia down :rolleyes:
it needs a lot of bad faith to not agree that nvidia was first dual core A9 soc, first quad core A9 soc, and even if T4 was not shipped first, it's still at cpu wave front tech (1st or 2nd, depending on how you see it...)
 
Intel seems unlikely. Besides, Intel already has a patent licencing agreement with NVIDIA and is already putting quite some effort into their own in house designed GPUs.
As for Samsung and Apple. Of those two, I think Samsung is the more likely to strike a deal with NVIDIA.

IMHO Samsung looks like the most evident target.
Intel is second in my mind. They already have licence IP agreement, it will not take too much more to licence the all thing...
 
very interesting concept unveiled today based on T4:

HP announces Slate 21 AIO, 21.5-inch Android tablet with Tegra 4

hpaio2.jpg


http://www.engadget.com/2013/06/24/hp-launches-slate-21-aio/#continued
 
I don't see Intel choosing mobile Kepler over Gen for tablet/netbook/nettop Atom SoCs, which is what they would doing. It would make sense if they needed better DirectX and OpenGL support for phone SoCs, but so long as they're making those Android-only that doesn't really make sense.

Any existing license agreements shouldn't really matter.

In my mind Apple is more likely than Samsung - they're less of a direct threat to Tegra and they're more stable in their licensing decisions (they've used IMG since the first iPhone while Samsung flip-flopped between their own GPU, IMG, ARM, and IMG again).

Apple could also have bigger aspirations for rolling out proper ARM laptops than Samsung does. If they want an ARM OS X platform they'll want proper OpenGL and nVidia is probably in the best position to provide that. Even in phones and tablets Apple is in a better position to get people to leverage better API support or nVidia-only extensions by virtue of the walled garden.
 
Apple? No chance.
I could only see some chinese/korean manufacturer like huawei or even LG using it for some obscure chipset.

Qualcomm have adreno..everyone uses that apart from apple..who make their own designs around industry leading img tech..who they also have shares in.
Intel will also continue to use img tech untill they can get gen down low enough.
 
funny how you turn the slide to always put nvidia down :rolleyes:
it needs a lot of bad faith to not agree that nvidia was first dual core A9 soc, first quad core A9 soc, and even if T4 was not shipped first, it's still at cpu wave front tech (1st or 2nd, depending on how you see it...)

Hmm, is anyone else planning to make an A15-based SoC for phones and tablets?
 
Hmm, is anyone else planning to make an A15-based SoC for phones and tablets?
As far as T4 goes I know of HP SlateBook x2 and Asus Transformer Pad Infinity.

Edit: completely misunderstood your question, sorry and thanks to Exophase :)
 
Last edited by a moderator:
Hmm, is anyone else planning to make an A15-based SoC for phones and tablets?

You mean outside of Exynos 5 SoCs and Tegra 4, right? Renesas announced one and OMAP5 still sort of exists, but it's anyone's guess if either will make it in a tablet ever (phone is almost a definite no). AFAIK MediaTek is also alleged to have an A15/A7 big.LITTLE SoC in the works and that would almost certainly make it in tablets. They and the higher tier Chinese SoC makers like Rockchip really have little choice but to embrace A15 unless they want to ride out A9 and A7 until A12 is available.
 
You mean outside of Exynos 5 SoCs and Tegra 4, right? Renesas announced one and OMAP5 still sort of exists, but it's anyone's guess if either will make it in a tablet ever (phone is almost a definite no). AFAIK MediaTek is also alleged to have an A15/A7 big.LITTLE SoC in the works and that would almost certainly make it in tablets. They and the higher tier Chinese SoC makers like Rockchip really have little choice but to embrace A15 unless they want to ride out A9 and A7 until A12 is available.

Yes, that's what I meant.

Apart from Samsung and NVIDIA, nobody seems to be very vocal about A15, which makes me wonder whether they might actually be waiting for A57 instead, since apparently it will be ready for commercial release in mid-2014.
 
I don't know about you gentlemen, but I'd personally prefer a quad Swift, Krait or A9r4 nowadays in a smartphone then A15.
 
I don't know about you gentlemen, but I'd personally prefer a quad Swift, Krait or A9r4 nowadays in a smartphone then A15.
+1
I tried a Chinese 10" IPS FHD tablet featuring Rockchip 3188 (quad core A9 at 1.8GHz) and it was very responsive under Android 4.2. Even smother than my Nexus4.
Tegra4i with A9 R4 at 2GHz should fly and keep consumption reasonable.
 
Back
Top