Krait's knocking . . .

[Bateluer]

http://www.anandtech.com/show/4940/qualcomm-new-snapdragon-s4-msm8960-krait-architecture

Keeping in mind the sheer amount of compute potential of the Adreno 22x family, it only makes sense that driver improvements could unlock a lot of performance. Qualcomm expects the 225 to be 50% faster than the outgoing 220

The sheet numbers on this platform are impressive, should easily dominate everything currently out and coming out for the next year. Have to see what they can deliver though. And when devices show up on the market with them.

 

[Oyeve]

Last saw him on Tatooine, he didn't look so good.

 

[mnewsham]

Last saw him on Tatooine, he didn't look so good.

lol'd

 

[Mopetar]

I imagine that the next Apple chip will probably be at least comparable and released in a similar timeframe and I imagine Nvidia will eventually release a die shrink of the Tegra 3 in some form or another. It'll be a big step up, but I don't know if it will be any bigger of a step than other competitors will be able to make.

 

[Bateluer]

I imagine that the next Apple chip will probably be at least comparable and released in a similar timeframe and I imagine Nvidia will eventually release a die shrink of the Tegra 3 in some form or another. It'll be a big step up, but I don't know if it will be any bigger of a step than other competitors will be able to make.

Not sure what the time table is for Apple's next chip, A6, but I don't think its going to be out in the same time frame as Tegra 3 and Krait. Having said that, I expect a big jump between 5 and A6, obviously.

Krait seems to have the numbers on its side, depends on when they can ship it to vendors, and how quickly those vendors can deliver a polished product. With phones, they have to dick around with carriers, at least, in the US. For a tablet, get the FCC approval and get it on the shelves at retail/e-tail stores. I want to see some product execution here, on the same level that nvidia currently has with most of their video cards. Announce the GPU on Day 1, parts available for purchase on Day 3.

 

[Mopetar]

Not sure what the time table is for Apple's next chip, A6, but I don't think its going to be out in the same time frame as Tegra 3 and Krait.

The Krait is listed as 1H 2012. The Tegra 3 will be out Q4 2011 in actual devices, but it's anyone's guess when or if Nvidia will release a die shrink for it. Assuming Apple keeps their early April release schedule for the iPad, the A6 will be out early Q2, well within the same range of time as the Krait.

 

[smartpatrol]

What's up with Tegra 2/Tegra 3 using single-channel memory? Also, I'm not sure if the chart is right, I think Snapdragon S3 also use only single-channel. Wikipedia lists it as having only single channel memory bus, and the HP Touchpad definitely uses only single-channel memory.

Anyway, is this going to be detrimental to performance? Tegra 3 sounds amazing except for its memory configuration.

 

[podspi]

Are modern ARM designs really more powerful than Atom? I'm typing this on an N270 and it is kind of blowing my mind right now the next telephone I purchase may be more powerful than this netbook. Although, as others have pointed out, compute is only one part of the equation. ARM SoCs usually have anemic I/O, and I didn't really see anything that stood out with Krait w.r.t. other ARM SoCs (with the exception of Tegra).

 

[Fire&Blood]

AFAIK, OMAP4 was the first with dual channel to make into a phone with the Optimus 3D.

All next gen smartphone SoC's are scheduled to land sometime in 2012.

Tegra 3 isn't entirely a next gen part and it's as late as the Tegra 2 was, a year. Nvidia's CEO said on several occasions that it would launch Q1 2011. Here is a link to an Engadget article from September last year: http://www.engadget.com/2010/09/21/nvidia-ceo-tegra-3-almost-done-tegra-4-on-the-way-expect-a-ne/

Anand's articles repeated several times that Nvidia claims they still can't saturate the single channel bus it and they like it's power savings over DC.

Because of the delay, Nvidia is basically competing with a A9 against upcoming A15 cores and their revisions. It's hard to guess how much more can be squeezed out of A9 with their 5 core solution and how well their GPU will perform. We have seen benchmarks but that's something Nvidia has always been known to manipulate. Tegra 3 will likely be the first quad core CPU to appear in a smartphone. They announced the 28nm Kal-El+ for smartphones, the sooner they can deliver it, more market share they can grab.

They may not look good on paper but I wouldn't underestimate them. Their GPU expertise is only rivaled by ARM and IMG. Tegra3 could put up a good fight against the first Krait parts with Adreno 2xx. Qualcomm's MSM8974 with Adreno 320 and TI's OMA5 should be superior but that's coming even later.

 

[destrekor]

Not sure what the time table is for Apple's next chip, A6, but I don't think its going to be out in the same time frame as Tegra 3 and Krait. Having said that, I expect a big jump between 5 and A6, obviously.

Krait seems to have the numbers on its side, depends on when they can ship it to vendors, and how quickly those vendors can deliver a polished product. With phones, they have to dick around with carriers, at least, in the US. For a tablet, get the FCC approval and get it on the shelves at retail/e-tail stores. I want to see some product execution here, on the same level that nvidia currently has with most of their video cards. Announce the GPU on Day 1, parts available for purchase on Day 3.

A6 is going to be whatever and whenever Apple so desires to based on what technology they want and when it's available, but obviously it's up to them finding a manufacturer to produce it. Samsung is not a likely candidate to build their next SoC, but that still leaves powerhouses willing and eager to put forth their best products when it comes to the sheer volume of potential sales: TI, Qualcomm, Nvidia are all itching to make a ton of money.

Of course, there's also the factor of having a spec sheet in hand and an order placed with a projected ship date that turns out to be unrealistic.

This year has been a mess in terms of delays. Let's hope the next generation fares better.

 

[Mopetar]

A6 is going to be whatever and whenever Apple so desires to based on what technology they want and when it's available, but obviously it's up to them finding a manufacturer to produce it.

TSMC has already been selected as the fab for the A6. Beyond that not a whole lot is known about the chip.

 

[Fire&Blood]

Quoted the most interesting bits from the article:

The key is this: other than TI's OMAP 5 in the second half of 2012 and Qualcomm's Krait, no one else has announced plans to release a new microarchitecture in the near term. Furthermore, if we only look at the first half of next year, Qualcomm is the only company that's focused on significantly improving per-core performance through a new architecture. Everyone else is either scaling up in core count (NVIDIA) or clock speeds. As we've seen in the PC industry however, generational performance gaps are hard to overcome - even with more cores or frequency.

Note that courtesy of the wider front-end and OoO execution engine, Krait should be a higher performance architecture than Intel's Atom. That's right, you'll be able to get better performance than some of the very first Centrino notebooks in your smartphones come 2012.

At launch Krait will run 25% faster than most A9s on the market today, a gap that will only grow as Qualcomm introduces subsequent versions of the core. It's not unreasonable to expect a 30 - 50% gain in performance over existing smartphone designs.

Quite a few optimistic claims throughout the article, something to look forward too. Once this makes it into products, we'll have a significant leap over existing tech but this where the US market complexity becomes a problem.

The problem for the USA consumer is matching these SoC's with our preferred phone makers, then with our preferred OS's and finally, with our preferred carrier. 5 out 6 new desirable phones get eliminated in this process of selection. Even when we get lucky with that, there still are considerable delays we have to endure in order to get the phone we want.

I hope the OEM's are aware of it and maybe new chipsets can help with the lack of GSM uniformity in the US. The day may never come but I'll keep hoping future AT&T and T-Mobile phones are compatible with each network, that would make for a wide selection of phones to choose form.

 

[Bateluer]

The problem for the USA consumer is matching these SoC's with our preferred phone makers, then with our preferred OS's and finally, with our preferred carrier. 5 out 6 new desirable phones get eliminated in this process of selection. Even when we get lucky with that, there still are considerable delays we have to endure in order to get the phone we want.

I hope the OEM's are aware of it and maybe new chipsets can help with the lack of GSM uniformity in the US. The day may never come but I'll keep hoping future AT&T and T-Mobile phones are compatible with each network, that would make for a wide selection of phones to choose form.

Agreed. Been saying this for a while now too. If you're making a wifi tablet, I expect products to make it to market before Qualcomm/TI/Nvidia/Samsung begin shipping their next gen parts.

 

[Mopetar]

I still think it's pretty pointless to buy a tablet or phone based on the SoC inside of it. Either it's a good product and the experience is worthwhile, or it isn't. An SoC probably won't make or break the device.

 

[Aikouka]

I think that if this SoC's LTE capabilities are better than the current LTE implementations (multi-chip), then I could see carriers wanting to use it. I mean, if you start talking to me about LTE in a phone, my first question is going to be, "how bad is the battery life?"

 

[Mopetar]

I think that if this SoC's LTE capabilities are better than the current LTE implementations (multi-chip), then I could see carriers wanting to use it.

I'd imagine that they won't be any worse considering Qualcomm is one of the major baseband manufacturers. The 28 nm process should also help reduce the battery drain.

 

[Aikouka]

I'd imagine that they won't be any worse considering Qualcomm is one of the major baseband manufacturers. The 28 nm process should also help reduce the battery drain.

Yeah, but Qualcomm also makes the current crop of LTE modems that we get, and they aren't exactly sippers of juice. While I know that die shrinks help reduce power consumption, I'm kind of wary to think that it will necessarily be enough.

I wonder if it will also require software (the mobile OS) to be smart with it and possibly treat it kind of like a faster CPU. What I mean by that is the typical saying is that "a faster CPU may use more power, but it gets its workload done faster" could apply to the 4G technologies as well. Essentially, the modem kicks into high gear to download your Cute Kitten YouTube video, and then quickly drops down into some low power state.