Nvidia Tegra

New video codecs, performance improvements? I thought that it was still the same tensilica xtensa core...
It's definitely Tensilica Xtensa-based but that doesn't mean much since that's a highly configurable core. Snapdragon1 is probably also using XTensa for 720p (or maybe not and it was really only Freescale or STM who had licensed that last generation) but its configuration must surely be quite different to achieve 1080p Encode and the much higher bitrates. You could argue it's still incremental, and that'd be fair, but my point is I fail to see how Snapdragon3 could be a more fundamental change. I'm not sure whether it supports CABAC/High Profile (I think it does) but beyond that there's not much they're likely to add besides dual-stream support, which basically just requires copy-pasting hardware or doubling the clock speed (and adding some more dedicated RAM if it's the latter).

Isn't the GPU based on quad-pipeline z430? Snapdragon2 is supposed to be 4x faster than the snapdragon1 so it would seem so.
Yeah, it's z430 vs z460. One key factor that hasn't been discussed much is that while z430 was basically limited to one pipeline, z460 is more flexible and (before being acquired by Qualcomm) a company that'd have really insisted for a different pipeline count (or even a different ALU:TEX ratio or whatever) could have gotten it. It's not quite like SGX where any given IP is a very optimized but inflexible solution; here I've heard from two different people that their RTL is apparently flexible enough to change these things as required even without a Mali400 or SGX543-like core concept.

I don't know about Snapdragon2's fillrate exactly (triangle performance is massively higher though) but the most likely option remains a 4-pipeline solution at slightly higher clock speeds than Snapdragon1. For Snapdragon3, I have no idea but I'd assume a newer revision of z460 (or something like it) with 8 pipelines would make sense. Or maybe not. Sorry, not very useful I know ;)

Are my assumptions wrong?
Not really as I implied above - just there are clear architectural changes in these blocks from SD1 to SD2, whereas SD3 *might* be simply stronger through brute force. Other important changes in Snapdragon2 include things like a probably implemented/used power island for audio decode (ala Tegra and OMAP4) and apparently better clock speed optimizations for the CPU core. It's not surprising they could improve it after their first try, but remember it's not process related because there's little to no performance boost from 65LP to 45/40LP!

One other thing to keep in mind for both Tegra3 and Snapdragon3 is that 28LPG is a triple gate oxide process, which means they can use high performance/high leakage transistors *selectively* where they want to in order to achieve higher clock speeds everywhere and much higher clock speeds in places where they're mostly OK with higher leakage like the CPU core. If this is used properly, maybe we'll be pleasantly surprised by the clock speeds.

These Semis continue to be wildly, overly optimistic on their timetable for process transition, now for the 28nm generation.
How are they so wildly overly optimistic? TSMC wouldn't claim the first tape-outs will happen in Q1 (and more in Q2) 2010 if their customers weren't ready for it. This is not the 28HP process, where TSMC wouldn't be ready anyway. Remember: the first 45LP tape-out happened by Qualcomm in mid-2007. This is 2.5-3 years later! Handheld tape-outs always happen early and take a long time to come out in real products. OMAP3 taped-out in August 2006; it took nearly 3 years before the first phone based on it came out!
 
Sampling to volume production is hardly the routine transition implied by their projections.

The NEC N-01A and its OMAP3430 have been out since 2008, actually, in that particular case.
 
Why does it take so long from tape-out to product? What's the "weak" link here? And what's the best case scenario, let's say for someone like Apple who designs the chip (without 3G) and then builds and sells the end products? 18 months from tape-out to product?
 
To Arun:
What do you think this could mean for snapdragon2? It makes it more advanced but does it make it better?
Going back to snapdragon3. Do you think this time they will include on chip HDMI controller? Tegra, OMAP even new samsung SoC's(S5PC110 and S5PV210) have it
but not snapdragon. It would be cheaper than add a separate HDMI controller.
Possible specs for snapdragon3(purely my speculations):
CPU - probably dual core scorpion CPU at 1,5-2Ghz(hopefully they are going to do some architectural improvements that will make it more competitive with cortex A9)
GPU - roughly 300M poly/sec and 4G pixels/sec
Video - 1080p High Profile Decode, 1080p Encode(720p HP Encode) possible dual stream support for Blu-Ray.
Other things I hope will end up in snapdragon3 - LP-DDR2, 1080p/i HDMI support(what's the point of having 1080p playback on a device that won't have HDMI output).

I don't think those specs are far from truth considering that they seem to be constantly upgrading their scorpion CPU and that GPU got 4x boost from SD1 to SD2 rest of the specs seems logical considering when it is supposed to be released. HDMI should become a must have feature on a MID/PocketPC/Smartbook, LP-DDR2 support will guarantee great performance and will be crucial for seamless 1080p playback.

Why does it take so long from tape-out to product? What's the "weak" link here? And what's the best case scenario, let's say for someone like Apple who designs the chip (without 3G) and then builds and sells the end products? 18 months from tape-out to product?

IMO it takes less time cause they design the chip when designing the whole itself so they can do two things at the same time.
The whole phone designing process takes about 18-24 months. But when you have previous generation device you can just update some parts and the rest is left intact. That is why apple can release new devices every year without any delays.
If they would develop new device it would require more time and could even get delayed(ex. apple tablet).
 
Sampling to volume production is hardly the routine transition implied by their projections.

The NEC N-01A and its OMAP3430 have been out since 2008, actually, in that particular case.

That handset was presumed to be Omap3430 turned out to be NEC Medity 2 which seems a reletively unknown implementation of SGX.
 
HDMI should become a must have feature on a MID/PocketPC/Smartbook, LP-DDR2 support will guarantee great performance and will be crucial for seamless 1080p playback.
I'm not so sure that HDMI (at least the connector) will win in the smartphone/MID market longterm. MHL looks very promising and I guess things like the rumored Micro-DisplayPort and maybe mobile Light Peak aren't too far out. But HDMI (Type D connector) will be the first, no question there.
 
Last edited by a moderator:
I'm not so sure that HDMI (at least the connector) will win in the smartphone/MID market longterm. MHL looks very promising and I guess things like the rumored Micro-DisplayPort and maybe mobile Light Peak aren't too far out. But HDMI (Type D connector) will be the first, no question there.

In longterm MHL will probably win due to smaller size more appropriate for mobile devices but considering that the specs aren't finalized and the work has only just begun it is still at least 2-3 if not more before we see it in devices. Link with some informations.
For now mini-HDMI is the only choice.
 
To Arun:
What do you think this could mean for snapdragon2? It makes it more advanced but does it make it better?
That's nice! Quick Logic is an intriguing company, I should really get around to looking a bit more into them. However in this case it looks more like a cost reduction than a truly new capability, I think you could already do this with more expensive discrete components. It's also clearly aimed at the lower-end chipsets at least as much as Snapdragon.

Going back to snapdragon3. Do you think this time they will include on chip HDMI controller?
I definitely hope so, but I have zero insider info. Obviously if HDMI doesn't become very relevant to phones because of Zune-like Docks maybe using MHL as Mike said, then it's not much of a cost reduction when that extra integrated component is only useful to a small number of your customers! TBH, personally I'd prefer not to have to use a dock one way or another.

CPU - probably dual core scorpion CPU at 1,5-2Ghz(hopefully they are going to do some architectural improvements that will make it more competitive with cortex A9)
I have no idea about architecture or number of cores, but clock speeds should increase quite a bit thanks to the triple gate oxide in TSMC's 28LPG process.

Other things I hope will end up in snapdragon3 - LP-DDR2
Good question, does Snapdragon2 support LP-DDR2? SoCs from NVIDIA, Samsung, and TI in the same timeframe do; however that doesn't mean much. Also in netbooks, for cost reasons DDR3 makes more sense than LPDDR2; active power at 1.35v isn't too different, the only catch is that idle power is significantly worse which can affect music playback time significantly.

Why does it take so long from tape-out to product? What's the "weak" link here? And what's the best case scenario, let's say for someone like Apple who designs the chip (without 3G) and then builds and sells the end products? 18 months from tape-out to product?
The weak link between the final respin (sometimes 'sampling' doesn't mean the design is truly finalized, especially for wireless chips like Bluetooth/WiFi I think) and end-product availability is usually the OEM. Doing more of the SW work for them helps though, as does having very similar hardware interfaces compared to previous chips. Ideally you'd do nearly all the work yourself but this only makes sense for customers that don't want to customize the SW much; the one company which has been mind-blowingly successful at doing that is Mediatek (in the Chinese market).

I'm not so sure that HDMI (at least the connector) will win in the smartphone/MID market longterm
I'm far from convinced by anything but HDMI in the next few years, unless maybe it's dock-centric like MHL (and then it depends whether customers actually like that approach or not). I also think your average customer will realize the potential usage of HDMI better than DisplayPort or Light Peak so there is always an inherent advantage there, but we'll see.
 
I definitely hope so, but I have zero insider info. Obviously if HDMI doesn't become very relevant to phones because of Zune-like Docks maybe using MHL as Mike said, then it's not much of a cost reduction when that extra integrated component is only useful to a small number of your customers! TBH, personally I'd prefer not to have to use a dock one way or another.
MHL can be used without any kind of dock if the bridge chip gets integrated into the cable itself. It is possible but would make the cable a little bit larger but it wouldn't require any kind of dock so its the best way and if I understand correctly MHL can be integrated into microUSB connector.
It would be best solution for mobile devices but only if someone will do MHL->hdmi cable with MHL bridge integrated to make the docking station redundant.
It is always easier to take cable with you than docking station. Besides cable is always cheaper than a station.
So I fully understand why you would prefer to use cable.
 
The dock is a big profit opportunity. Who knows what the connect rate is. If it's low, then it means people don't bother connecting the device to the big screens.

It's going to be about costs, isn't it, in terms of royalties, in terms of how much the silicon to support these connectors cost.


On another note, any ideas on the market share of all these different ARM SoCs? Sounds like there's a likely consolidation of this market.
 
In longterm MHL will probably win due to smaller size more appropriate for mobile devices but considering that the specs aren't finalized and the work has only just begun it is still at least 2-3 if not more before we see it in devices. Link with some informations.
For now mini-HDMI is the only choice.
Don't you mean Micro-HDMI?

MHL can be used without any kind of dock if the bridge chip gets integrated into the cable itself. It is possible but would make the cable a little bit larger but it wouldn't require any kind of dock so its the best way and if I understand correctly MHL can be integrated into microUSB connector.
It would be best solution for mobile devices but only if someone will do MHL->hdmi cable with MHL bridge integrated to make the docking station redundant.
It is always easier to take cable with you than docking station. Besides cable is always cheaper than a station.
So I fully understand why you would prefer to use cable.
I haven't found an english article about it yet but german heise.de reports that there are already MHL-capable TVs out there. The Silicon Image HDMI chip they use already has the MHL-Receiver integrated (Samsung etc.). For these TVs you would only need a passive USB-HDMI adapter cable. The dock is only necessary for non-MHL capable HDMI-ports.

http://translate.google.com/transla...Porti-798289.html&sl=de&tl=en&history_state0=
 
http://www.pcper.com/comments.php?nid=7876

Practically-speaking, the TS would be on a par with the Wii in terms of its processing power and graphical grunt. The new unit is set to retain many features of its immediate predecessor too – notably a DSi-style built-in camera and a higher-resolution touch-screen interface. Full backwards-compatibility is also on the cards, we’re assured.

Nintendo has sold over 100 million Nintendo DS systems and if NVIDIA does indeed power the next revision of it with a Tegra-based processor, that is a huge boost to the company's bottom line and could put the Tegra brand on its way to "capturing 50% of nVidia's revenue within the next couple of years."
 
That's a huge coup for Nvidia if they've actually landed the contract. Would be quite a strange situation if their handheld is more capable than their home console.
 
Somebody correct me if I'm wrong, but isn't the Tegra roughly the same last generation visuals for portable devices? It's probably Dreamcast or worse.
 
Somebody correct me if I'm wrong, but isn't the Tegra roughly the same last generation visuals for portable devices? It's probably Dreamcast or worse.

Compared to the ARM9 and ARM7 cores in the current Nintendo DSi, Tegra is much faster clock-for-clock, even in its first generation ARM11-based iteration.
This without even mentioning clockspeeds (DSi's ARM9 goes up to 133MHz, apparently, while Tegra can hit speeds closer to 1GHz).
 
Yeah, compared to the current one, but I'm talking about brain_stew suggestion that it would be comparable to the main console. I seriously doubt it'll comparable to the Gamecube.
 
Yeah, compared to the current one, but I'm talking about brain_stew suggestion that it would be comparable to the main console. I seriously doubt it'll comparable to the Gamecube.

If they go for Tegra II, it should be substantially faster.
Going with the current one is an unknown quantity at this time. Remember that the general purpose CPU component and architecture both play a key role here, due to the tight integration of these closed console platforms.
Comparing a really old, crippled PowerPC G3 CPU with a current ARM11 or ARM Cortex A8, based purely on hardware specs, it really isn't an easy task, but i'd suggest the ARM ones are much faster by now.
 
Back
Top