Qualcomm shows working MSM8x60 at CES.

[metafor]

Design win != sales success and that goes for anything out there on sale.

Well, the topic of discussion was about why MSM8x60 didn't have more design wins; and primarily it's because it was late to market.

 

[Erinyes]

Design win != sales success and that goes for anything out there on sale.

Tell that to Nvidia!

But back on topic, Sony Ericsson hasnt yet announced any dual core phones yet. Given that they went with an all Qualcomm lineup for their new phones, quite possible they'll stick to them for the dual core phones

 

[Ailuros]

Well, the topic of discussion was about why MSM8x60 didn't have more design wins; and primarily it's because it was late to market.

Obviously; but does your gut feeling tell you that Qualcomm will sell more or less units all together than Tegra2? That's exactly the reasoning I had in the back of my mind with my design wins note.

 

[Deleted member 13524]

Here's the (first?) smartphone with a MSM8660 @ 1.5GHz:

Vega Nº5

It packs a 5-inch screen (engadget now calls these "tabletphones") and a disappointing 800*480 resolution.

 

[metafor]

Obviously; but does your gut feeling tell you that Qualcomm will sell more or less units all together than Tegra2? That's exactly the reasoning I had in the back of my mind with my design wins note.

8x60? I don't know. The market seems to be flooded by 7x30/8x55 devices more than anything else. The Droid series have been pretty good sellers in the past and I don't see that stopping. I don't know if 8x60 devices will outsell them.

 

[Arun]

Erineyes: Sony Ericsson will definitely sell ST-Ericsson U8500/U9500-based Android phones. I don't know if it's an exclusive relationship.

metafor/Ailuros: Well, I'd expect T30 to make more money for NV than T20 by Q1 2012, so I'm not sure it matters either way. What you should really compare IMO is the total amount of revenue (ideally gross profit and excluding the baseband part) for dual-core-and-above phones. Of course, that kind of data is even harder to get access to...

 

[metafor]

Well, I'd expect T30 to make more money for NV than T20 by Q1 2012, so I'm not sure it matters either way. What you should really compare IMO is the total amount of revenue (ideally gross profit and excluding the baseband part) for dual-core-and-above phones. Of course, that kind of data is even harder to get access to...

I don't know about by Q1. Unless tablet sales really pick up, it won't be until smartphones are released en mass with Tegra 3 that you'll see revenue rise from that chip and I don't think that'll happen before Q2. There may be a few initial designs released but if history is any indication, it takes a few months for the masses to catch on.

There's also the fact that T30 for phones will likely compete with 8960 coming out crosses fingers end of this year.

 

[Erinyes]

Erineyes: Sony Ericsson will definitely sell ST-Ericsson U8500/U9500-based Android phones. I don't know if it's an exclusive relationship.

metafor/Ailuros: Well, I'd expect T30 to make more money for NV than T20 by Q1 2012, so I'm not sure it matters either way. What you should really compare IMO is the total amount of revenue (ideally gross profit and excluding the baseband part) for dual-core-and-above phones. Of course, that kind of data is even harder to get access to...

IIRC U8500 started sampling quite late compared to the competition so its still a while off right? Btw did they end up using any ST Ericsson chips in their previous android phones? I only remember them using QSD8250 in the Vivaz/X10 and MSM 7227 in the Mini/X8.

Do you really think T30 will be selling in great volumes by then? It wont even show up in smartphones until late Q4 2011 or early Q1 2012. There is a strong possibility of it again being the reference platform for Ice cream sandwich though.

.There's also the fact that T30 for phones will likely compete with 8960 coming out crosses fingers end of this year.

I think you're being just a wee bit too optimistic here :smile: 8960 was scheduled to start sampling in Q2 this year, with availability in the middle of next year. OMAP 5 with Cortex A15 is scheduled for H2 2012 and i think its slated to be the first Cortex A15 SoC. Im sure Arun can correct me here though

 

[Arun]

I don't know about by Q1. Unless tablet sales really pick up, it won't be until smartphones are released en mass with Tegra 3 that you'll see revenue rise from that chip and I don't think that'll happen before Q2. There may be a few initial designs released but if history is any indication, it takes a few months for the masses to catch on.

Well, a few points:
1) Availability lags behind revenue, so Q1 revenue means Q2 availability
2) While Tegra 2 will keep selling in phones for a while, I expect most tablets to switch to Tegra 3 quite rapidly, with Tegra 2 tablet revenue mostly gone by Q1.
3) Tegra 3 ASPs are higher than Tegra 2.

So yeah, I cheated a bit, kill me Also I still think it's far from impossible that they somehow manage to release one or two phones in November/December 2011. We'll see what they say at Computex and whether WiFi tablets are still on schedule for late August - if they are, then maybe, otherwise no way.

There's also the fact that T30 for phones will likely compete with 8960 coming out crosses fingers end of this year.

Wait, 'coming' meaning 'in shipping devices'? It's only sampling this quarter, it's on a less mature process, and it loses at least 2-3 months via baseband certification. If MSM8960 manages to be available in phones at roughly the same time as T30, I'd be VERY impressed. Unless you mean 'commercial sampling and ready for mass production' rather than actually available. I certainly don't expect the first MSM8960 phones before June 2011 at the earliest (i.e. 6 months after T30) with real volumes in Q4.

IIRC U8500 started sampling quite late compared to the competition so its still a while off right?

Actually, it started sampling relatively early, but it's definitely suffered from various small delays. At this point I'd expect the first real products to be available in late Q3.

 

[metafor]

I think you're being just a wee bit too optimistic here :smile: 8960 was scheduled to start sampling in Q2 this year, with availability in the middle of next year. OMAP 5 with Cortex A15 is scheduled for H2 2012 and i think its slated to be the first Cortex A15 SoC. Im sure Arun can correct me here though

It's still very much an open question of when 8960 will make it into commercial devices but Q1 of 2012 isn't out of the question at this point. Of course, that depends on how much delay the OEM's take to push it through to the carriers.

 

[Arun]

It's still very much an open question of when 8960 will make it into commercial devices but Q1 of 2012 isn't out of the question at this point. Of course, that depends on how much delay the OEM's take to push it through to the carriers.

Well, I suppose a best-case scenario of 3 months behind T30's best-case scenario is still very good Let's see how they both deliver.

 

[metafor]

Well, a few points:
1) Availability lags behind revenue, so Q1 revenue means Q2 availability
2) While Tegra 2 will keep selling in phones for a while, I expect most tablets to switch to Tegra 3 quite rapidly, with Tegra 2 tablet revenue mostly gone by Q1.
3) Tegra 3 ASPs are higher than Tegra 2.

So yeah, I cheated a bit, kill me Also I still think it's far from impossible that they somehow manage to release one or two phones in November/December 2011. We'll see what they say at Computex and whether WiFi tablets are still on schedule for late August - if they are, then maybe, otherwise no way.

Well yes, that's kinda the point. It'll definitely have a 1-2 hot selling phones by January-Feb of next year but usually the volumes don't pick up until you have at least one or two different phones on each carrier, which takes a while to get to. IMO, we're only seeing that with Tegra 2 phones now.

Wait, 'coming' meaning 'in shipping devices'?

Oh hell no.

It's only sampling this quarter, it's on a less mature process, and it loses at least 2-3 months via baseband certification. If MSM8960 manages to be available in phones at roughly the same time as T30, I'd be VERY impressed. Unless you mean 'commercial sampling and ready for mass production' rather than actually available. I certainly don't expect the first MSM8960 phones before June 2011 at the earliest (i.e. 6 months after T30) with real volumes in Q4.

I don't know about 6 months but it'll definitely lag T30 devices by some number of months. Again, that's all in flux right now depending on the OEM's. But the chip should be available for mass market by Q1 of 2012. With the usual finger-crossings of course.

 

[Arun]

[...]
I don't know about 6 months but it'll definitely lag T30 devices by some number of months. Again, that's all in flux right now depending on the OEM's. But the chip should be available for mass market by Q1 of 2012. With the usual finger-crossings of course.

Fair enough Now I can't help but wonder if that's the first 28LP chip. If it is, then I wouldn't be surprised if 28HP beat 28LP to mass production, which would be slightly ironic...

 

[metafor]

Fair enough Now I can't help but wonder if that's the first 28LP chip. If it is, then I wouldn't be surprised if 28HP beat 28LP to mass production, which would be slightly ironic...

HP as in metal gates? Which chip has been announced with that?

 

[Arun]

HP as in metal gates? Which chip has been announced with that?

Some FPGAs, but more importantly PC GPUs: http://forum.beyond3d.com/showpost.php?p=1546398&postcount=109

It's a close thing, but if that happens and MSM8960 is indeed the first 28LP chip to enter mass production, it looks like 28LP won't be in mass production noticeably before (if not after) 28HP.

 

[metafor]

Some FPGAs, but more importantly PC GPUs: http://forum.beyond3d.com/showpost.php?p=1546398&postcount=109

It's a close thing, but if that happens and MSM8960 is indeed the first 28LP chip to enter mass production, it looks like 28LP won't be in mass production noticeably before (if not after) 28HP.

Interesting. Although I'd have to say that may not be metal gates. So far as I've heard, TSMC is nowhere near ready for mass market with the HKMG process though I could be wrong. Perhaps a bulk silicon 28HP?

 

[Arun]

Interesting. Although I'd have to say that may not be metal gates. So far as I've heard, TSMC is nowhere near ready for mass market with the HKMG process though I could be wrong. Perhaps a bulk silicon 28HP?

I'm very very confident it's HKMG. Here's part of the Q1 conference transcript from the 28th of April 2011:

Now I, let me talk about our advanced technologies. First, on 28 nanometers, we completed a very important milestone. We completed full technology qualifications for both oxynitrite, which is our 28 LP and high-k metal gate, which is our 28 HP. Now the significance of this milestone is that the way is now cleared to volume production on the 28 LP and 28 HP, both the oxynitrite and the high-k metal gate versions.

[...]

Zhou Ying – BNP

Hi, good evening. Just want to talk about the 28 nano. Could you get us some idea about what percentage of 28 nano right now is decide on high-k metal gate and what percentage on polygate?

Morris Chang

I don’t know whether we have that number, do we?

Lora Ho

At this point we have already shipped 28 nanometers to our customers and our customers have already started shipping in both high-k metal gate. So we also have other – in polygate.

Morris Chang

We have that number, it is two. But we don’t have that number here tonight.

Zhou Ying – BNP

Okay. Actually not exact the number, but maybe some idea, is it polygate over 50% of the total sell off?

Morris Chang

I would say, just taking a guess, I would say the rate of majority is being in high-k low gates.

Then there's also this in the Q4 CC:

Mehdi Hosseini - Susquehanna International

Dr. Chang, could you help us understand how should we think about the mix of 40-nanometer and 28-nanometer as we exit this year, like in Q4 timeframe? And also, how many tape-outs do you actually have for 28-nanometer that will be volume ramped in the second half?

Morris Chang

We plan to have around 2% or 3% of out total revenue in the forth quarter that will be 28-nanometer, 2% or 3% percent in the fourth quarter. The 40-nanometer is still increasing. And in the fourth quarter, it's 21% of revenue in the last quarter. And so by fourth quarter of 2011, I think that that percentage will be higher. But I do not care to give the number now.

Another part of the question, the tape-outs volume ramping. The tape-outs of the 28-nanaometer, they will start to ramp in the second half, starting in the third quarter and then more in the fourth quarter. But the real momentum we believe will be in the next year.

There's also an older quote saying they expected 2% of revenue in Q4 2011 to be for 28nm High-K, but I can't find it anymore, and old forecasts aren't very reliable anyway.

---

BTW, we talked briefly some time ago about the fact OMAP5's Cortex-A15 reached 2GHz on a 28nm SiON process. Guess how much it can reach on TSMC's 28HPM High-K process?

We have the 28 HPM technology that has gained enthusiastic adoption for tablets, smartphones and embedded SoC applications. The 28 HPM has 3-gigahertz ARM CPU for multi-core application processes. The 28 HPM has low SRAM standby leakage and VCCmin for several applications.
[...]
TSM 28 HPM is very well positioned. We are the industry leading 3-gigahertz operating frequency and the best performance power ratio for low power processes. Customers are adopting our 28 HPM enthusiastically. We stand to benefit from the popularity of these mobile devices.

Obviously that's going to be the top rate for tablets when running a single core and/or not thermally limited, but you get the idea. Pretty nice stuff if they deliver.

 

[metafor]

I'm very very confident it's HKMG. Here's part of the Q1 conference transcript from the 28th of April 2011:

Then there's also this in the Q4 CC:

There's also an older quote saying they expected 2% of revenue in Q4 2011 to be for 28nm High-K, but I can't find it anymore, and old forecasts aren't very reliable anyway.

Now that is interesting. Although I thought others were working on 28LP chips as well? Have none of them scheduled theirs for Q4 2011 mass production? I haven't kept up on the relative maturity of 28HP from TSMC, I guess they've made more progress than I thought.

BTW, we talked briefly some time ago about the fact OMAP5's Cortex-A15 reached 2GHz on a 28nm SiON process. Guess how much it can reach on TSMC's 28HPM High-K process?

Obviously that's going to be the top rate for tablets when running a single core and/or not thermally limited, but you get the idea. Pretty nice stuff if they deliver.

About what I'd expect given TI's history of ramping clockspeed. But I'd wager such a device may not be suitable for a tablet given the A15's uarch. But time will tell. TI's very good at implementing hard-throttling.

 

[Arun]

Now that is interesting. Although I thought others were working on 28LP chips as well? Have none of them scheduled theirs for Q4 2011 mass production? I haven't kept up on the relative maturity of 28HP from TSMC, I guess they've made more progress than I thought.

There's Altera which will use both 28LP SiON and 28HP(L?) High-K, but Xilink is using 28HPL High-K exclusively. As for handheld chips, someone else might be working on 28nm SiON besides Qualcomm and Texas Instruments, but NVIDIA/ST-Ericsson/Broadcom are definitely High-K exclusive. Strangely enough, even CSR has indicated they are more interested in 28nm High-K than SiON even for their RF chips. And there's not as much of an ecosystem around 28LP so smaller players are unlikely to even consider it.

I wouldn't be surprised if Atheros was more interested in 28LP... either way, they probably will have to be now that Qualcomm acquired them

About what I'd expect given TI's history of ramping clockspeed. But I'd wager such a device may not be suitable for a tablet given the A15's uarch. But time will tell. TI's very good at implementing hard-throttling.

Hmm? I'm not sure I understand - what I meant is that Cortex-A15 reaches 2GHz on UMC's 28LP SiON process and up to 3GHz on TSMC's 28HPM High-K process. There is no way OMAP5430 will go up to 3GHz as it's definitely on SiON. I'm also not sure what you mean by "such a device may not be suitable for a tablet given the A15's uarch" - do you mean the TDP would be too high? I don't think so for a dual-core, and even for a quad-core with aggressive thermal throttling it shouldn't be a problem. 3GHz peak for a single core (with a separate voltage regulator for the other 3 cores which could then run up to 1.5-2GHz) should definitely help for web browsing.

 

[metafor]

There's Altera which will use both 28LP SiON and 28HP(L?) High-K, but Xilink is using 28HPL High-K exclusively. As for handheld chips, someone else might be working on 28nm SiON besides Qualcomm and Texas Instruments, but NVIDIA/ST-Ericsson/Broadcom are definitely High-K exclusive. Strangely enough, even CSR has indicated they are more interested in 28nm High-K than SiON even for their RF chips. And there's not as much of an ecosystem around 28LP so smaller players are unlikely to even consider it.

I would've expected at least Broadcom and ST to have some products aimed at 28LP. There's a definite trade-off when going to HP that is unsuitable for handhelds.

I wouldn't be surprised if Atheros was more interested in 28LP... either way, they probably will have to be now that Qualcomm acquired them

I believe they're going to continue to operate as they do today though some of the IP may find itself into 28nm MSM's.

Hmm? I'm not sure I understand - what I meant is that Cortex-A15 reaches 2GHz on UMC's 28LP SiON process and up to 3GHz on TSMC's 28HPM High-K process. There is no way OMAP5430 will go up to 3GHz as it's definitely on SiON. I'm also not sure what you mean by "such a device may not be suitable for a tablet given the A15's uarch" - do you mean the TDP would be too high? I don't think so for a dual-core, and even for a quad-core with aggressive thermal throttling it shouldn't be a problem. 3GHz peak for a single core (with a separate voltage regulator for the other 3 cores which could then run up to 1.5-2GHz) should definitely help for web browsing.

TDP for a 3GHz A15 on 28HP would likely be too high for a lot of current tablet form factors. But like I said, aggressive throttling would help in that regard but the device would almost never be able to sustain that kind of peak frequency. There's also the problem that the PMIC would have to be severely ramped up to provide that kind of peak current as well. A similar problem existed at 65nm for a 1GHz Scorpion and that was a single-core on a bulk, low-current process. 28HP isn't that big of an improvement power-wise.