NVIDIA Tegra Architecture

Lacking integration on high volume phone parts seems like a poor outcome of the Icera acquisition. Yeah maybe they do still need a baseband processor for their automotive parts but why does it need to be their own? It's hard to believe the volumes there justify this.

Maybe this market has special needs for modems that other suppliers aren't meeting..
 
NVIDIA doesn't need to have Icera baseband modem tech for cars (for instance, the A3 is using a Tegra infotainment processor with a Qualcomm Gobi modem), but they have a greater chance of success with Icera in the automotive space than in the smartphone space. One way or another, I don't expect to see a baseband modem from them integrated on SoC die for a long time to come.
 
but they have a greater chance of success with Icera in the automotive space than in the smartphone space.

You may be correct here, but I still think the Icera acquisition will have ended up a net loss used only for this purpose, and the shareholders won't be too happy about it. It's questionable if it even justifies the R&D to keep making new products just for this market. Maybe we'll see a generation of chips using existing tech and then that'll be it.
 
"So we still need the baseband ... if only for automotive."

Sounds like they might be giving up on tablets too.

Seems you missed this from the article:

NVIDIA does plan to continue developing its own standalone baseband chips (useful for high-end phones, tablets, and -- the star of Tegra's show -- automotive).
The external modem can and will be used in high-end phones, tablets and automotive. Just because they highlighted automotive does not exclude the others.
 
Can, yes. Will? We'll see. What's Icera i500 being used in now?

It's at least plausible that they can sell them in the automotive space where a well tested reference design and support has a lot of value and there's not a lot of pressure to get the best performance, power efficiency, or cheapest part cost.
 
You may be correct here, but I still think the Icera acquisition will have ended up a net loss used only for this purpose, and the shareholders won't be too happy about it. It's questionable if it even justifies the R&D to keep making new products just for this market. Maybe we'll see a generation of chips using existing tech and then that'll be it.

To be fair, NVIDIA only paid $367M for Icera. Should it be a complete waste, it wouldn't be a very large one.
 
To be fair, NVIDIA only paid $367M for Icera. Should it be a complete waste, it wouldn't be a very large one.

That was the initial cost. Add in the cost of R&D to integrate the modem into the Tegra 4i and then add in the lost revenue for delaying the Tegra 4 to pull forward the Tegra 4i.

I really can't understand why it took a full year to qualify the Nvidia modem in the Tegra 4i when others like QualComm and MediaTek seem to release new versions within months.
 
That was the initial cost. Add in the cost of R&D to integrate the modem into the Tegra 4i and then add in the lost revenue for delaying the Tegra 4 to pull forward the Tegra 4i.

Fair point.

I really can't understand why it took a full year to qualify the Nvidia modem in the Tegra 4i when others like QualComm and MediaTek seem to release new versions within months.

I would imagine that Qualcomm et al. simply have a design pipeline: when a modem is released, the next one is already undergoing qualification. That's an effective way of doing things but it can't help you with the first iteration.
 
Intel has not been doing better with Infineon.

True, but Intel is a bottomless pit of money, revenue and R&D power. Plus, they use alien tech (fabs about 1.5 years ahead of everyone else).
If Intel hasn't done things fast enough in the ultra-mobile department, it's just because their priorities are somewhere else.

nVidia OTOH is highly dependent on Tegra's success. Discrete GPUs in x86 machines will stop making sense eventually (<10 years IMO) and I doubt they can live off HPC alone.
And I seriously doubt that nVidia's intentions with buying Icera were to make discrete baseband processors.
 
I really can't understand why it took a full year to qualify the Nvidia modem in the Tegra 4i when others like QualComm and MediaTek seem to reletase new versions within months.

NVIDIA had to prove that their i500 modem was robust and reliable for voice and data, and that is probably why they did the full qualification/certification testing. Note that many phones (including iPhones) don't go through the same rigorous testing on the baseband modem (which is understandable because they are mostly using proven technology from Qualcomm).

Note that it often takes 9-12 months between first silicon and commercial availability, irrespective of baseband modem certification, so Tegra 4i never really had any hope of mass production in 2013 in the first place.
 
The Tegra 4i is the first and LAST Nvidia integrated LTE modem SOC.

Wow..that's quite surprising..I would have thought they'd already be in the middle of developing the next version by now. I was expecting a A53 based version from them. Maybe the market is just too crowded and they don't see the returns.
Nvidia's external modem will continue to be available.
Did they ever sell significant quantities of the external modem? I doubt the business case is justified if they plan to only develop external modems in future.
Wow, T4i must have tanked really hard if nVidia is giving up on integrated modems right after their first try.

Yea..one reason I think is that T4i was just too late. And given the abundance of quad A7 based solutions in the market..that must have had an impact on pricing and demand as well.
For major SoC suppliers, who do we have with integrated baseband processor? Mediatek, Qualcomm Broadcom and Samsung (still lacking LTE, though)?
Samsung? Which of their SoC's have an integrated baseband?
An SoC with baseband modem integrated on die is ideal for cost-sensitive products that require connectivity. Since NVIDIA is focusing on high end ultra mobile products, it doesn't make sense for them to integrate a baseband modem. NVIDIA will have quicker time to market and higher application processor perf. for a given SoC die size by foregoing baseband modem integration

Are you saying Tegra 4i was targeted at the high end? And your second point is very debatable as well..given the success of MSM8974.
Obviously Qualcomm has a very strong baseband modem product roadmap. Note that most people change their phone every 1-2 years and will have newer baseband tech with each upgrade cycle, so having a software-defined modem is not a huge benefit there. But in the automotive space, most people don't change their car for at least 5 years, so having a software-defined modem may make some sense there
You have a point but how often do cars receive software updates? Maybe it will change in future but I right now I dont see much benefit to it.
Lacking integration on high volume phone parts seems like a poor outcome of the Icera acquisition. Yeah maybe they do still need a baseband processor for their automotive parts but why does it need to be their own? It's hard to believe the volumes there justify this.

Maybe this market has special needs for modems that other suppliers aren't meeting..

I completely agree with you. At the time it seemed that the major selling point of the Icera acquisition was to integrate the baseband on die to compete with the likes of Qualcomm. If they are now only going to make external modem's, I don't see the business case.
That was the initial cost. Add in the cost of R&D to integrate the modem into the Tegra 4i and then add in the lost revenue for delaying the Tegra 4 to pull forward the Tegra 4i.

Not counting the money spent on the external modem. But regarding your second point, wasn't it the other way around? Didn't they delay the release of 4i to pull forward Tegra 4?
I really can't understand why it took a full year to qualify the Nvidia modem in the Tegra 4i when others like QualComm and MediaTek seem to release new versions within months.
Well I dont think Qualcomm or Mediatek do it in just a few months, but they don't take as long as Nvidia has taken at least. Though given Qualcomm's vast experience, I wouldn't expect any less from them.
 
But regarding your second point, wasn't it the other way around? Didn't they delay the release of 4i to pull forward Tegra 4?
No it has been clearly stated in numerous Conference Calls that the Tegra 4 was delayed so that the Tegra 4i could be pulled forward thus creating the problem of the Tegra 4 missing key design wins because of the delay.

It turns out that was a major mistake because of the missed designs and also that the integrated modem is now dead for Nvidia.
 
I have heard some rumours as well. Take this with a grain of salt but apparently Google has chosen Tegra K1 for the next Nexus device. This is the Denver based K1. Now ordinarily I wouldn't have believed this myself, but after Apple moved to 64 bit last year, rumours were that Google was also working on a 64 bit version of Android. And it would seem likely that they'd reveal it at the next Google I/O conference in June. Now they'd have to have a device ready at the launch right? And which is the only 64 bit SoC that would be available in that timeframe? So yeah..I thought about it and it kinda makes sense to me..but lets wait and see what happens.
 
TK1 Denver would be a great fit for a Nexus console coming in Q4 2014, IMO. Google has supposedly been reworking on Nexus Q, so this could be it.

Likewise, the Nexus 10 is also in dire need of a refresh.
 
I have heard some rumours as well. Take this with a grain of salt but apparently Google has chosen Tegra K1 for the next Nexus device. This is the Denver based K1. Now ordinarily I wouldn't have believed this myself, but after Apple moved to 64 bit last year, rumours were that Google was also working on a 64 bit version of Android. And it would seem likely that they'd reveal it at the next Google I/O conference in June. Now they'd have to have a device ready at the launch right? And which is the only 64 bit SoC that would be available in that timeframe? So yeah..I thought about it and it kinda makes sense to me..but lets wait and see what happens.

There's no reason to believe that Tegra K1-Denver will be ready before Cortex-A57 devices, nor does it seem likely that Google would pick a chip without integrated LTE for a Nexus device.
 
Wasn't the LTE modem separate from the app SoC in Nexus 7 2013 tablet?

Hmm, it's Qualcomm, so I just assumed LTE was integrated, but now that you mention it, I don't know for sure. This page says the modem is a Qualcomm MDM9215M. Does that mean it's definitely discrete or could it be the name of the modem block in an SoC?

I also wonder why Google would make such a choice. I guess it means they can use the same app SoC and PCB in all versions and just leave a few components out of the LTE-less one. But it makes the LTE version sub-optimal.

Still, Tegra K1 + discrete LTE modem seems like an expensive proposition for a Nexus device, certainly compared to the S600 in the current model.

Then again, if the rumors are true that they're adding an inch and making a Nexus 8 this time…
 
Back
Top