Someone give me a few reasons why Apple has not purchased Qualcomm yet?
Absolutely terrible move. Why would Apple want Qualcomm? There a. Ore than a few companies that would be important for them to buy, but this one isn't it.
Yeah TI's OMAP was "never one of the big players and couldn't get up with the competition".
That's not what I said. Do you even understand what quotation marks mean?
What I said was that TI are a small player in the chip making industry. They achieved a lot of volume with Nokia but never had the resources of Qualcomm or Intel. Puzzlingly, you've then listed a series of dud products that never sold in significant volumes to back up your argument.
There's an ever better way to look at Geekbench. If someone posts a single test of a device (often seen in forums) there will be a little box on the top right of the page. Clicking this brings up a graphic with points to plot up to 2,000 of the same device and their scores.
If you look at the Note 4 the scores are all over the place. If you look at any Apple device, all the results fit into a little tiny bar with all the dots clustered around it.
It's hilarious how inconsistent Android devices are when benchmarked and how consistent Apple devices are. Also funny to see the "loaded" Samsung scores where a few devices score way higher than normal, and how these tests are the ones always referred to in forums as to Note 4 performance (instead of the average score).
There's a problem with benchmarking which might explain some of the variability of the Android benchmarks. If you use Geekbench, for example, you will notice that they advise not running anything else at the same time. You should turn your phone off completely after making sure that everything that turns on when you turn the phone back on will remain off.
Due to Android using a Desktop multitasking model, many apps remain on when you turn the phone completely off, and back on again. While many Android geeks claim to be knowledgable, I fint that not to be so. They know a few minor picky things, but don't understand the system.
Because of that, most Android phones will have some apps running in the background which slow down various aspects of the benchmarks. It's much easier to control this in an iOS device. Once you turn the phone off completely, and turn it back on again, this problem doesn't exist, so the scores are much closer, just bearing the usual slight deviations in manufacturing tolerances of the components, of about 5%, or so.
I don't buy that 80% figure. If you look at the gazillion sub $100 Android devices being sold, I never see a Qualcomm processor inside. It'a always something like Mediatek. I'd say Qualcomm has almost 100% of high-end Android devices.
I don't care whether you buy it or not. Those are the numbers the industry says is correct.
I don't buy that 80% figure. If you look at the gazillion sub $100 Android devices being sold, I never see a Qualcomm processor inside. It'a always something like Mediatek. I'd say Qualcomm has almost 100% of high-end Android devices.
While it's hard to find these articles when you want them, a bit of searching has shown this one, though it doesn't have the numbers, it gives the idea. If I can have the time to look more, I will. I'll either add it here, or do another post.
What, did you think you were reading a scientific journal or something? This is a tech rumor website.
He's not "citing himself." He's providing links to his (and others) previous articles that explain events in greater detail.
He's often an egotistical ass, but I don't think his linked articles exemplify this.
In any case, researchers often cite their own work in scientific papers. Often they are the leading authorities on the subject or they are building on previous work they have done. It's not at all frowned upon (if the work is good.) On the contrary, it's common practice.
I found the citations and break-out quotations to be highly exaggerating of the situation actually described in the linked articles.
For example "the company's own Exynos chips were plagued by serious design defects" - that simply isn't the case. It had a weaker GPU than the A8 (and not much weaker at the same resolution). If it was plagued you wouldn't expect it to ship successfully in millions of devices.
Historically Samsung hasn't used its Exynos chips because they only had a certain amount of them to use, to they targetted them at certain markets (especially ones that like the number 8). Luckily one of its major customers has diversified its suppliers...
I'm thinking that the X-1 will never see a smartphone. It looks like a good fit for a car; less so for tablets, though I expect a Shield and maybe a few others to offer the X-1. It's quite possible that Nvidia will disappear from mobile devices entirely. It's not cost effective in low volumes and almost certainly not as power efficient as it needs to be.
It's showing really good numbers, better than the A8X. For a smartphone? Well, more likely for a tablet. But a point is that Nvidia, like Apple, Qualcomm, Marvel and one or two others, has an architecture license, not just a design license as Samsung does. The Snapdragons, for example, are Qualcomm's take on ARM, just as Apple's "A" series are.
The fact that they are having problems with the 810 isn't good, but they've already come out with several lower rated 64 bit chips that are being used. No doubt, they will solve these heating problems at some point. Those here who are getting frantic about their problems really need to sit back and take a breath. This is just one processor. Qualcomm will fix it, and a couple of years from now, everyone will forget those problems.
I remember way back when Intel released a chip that had some severe math problems that would show up at some of the worst times. Intel had to recall a lot of chips. At the time, Apple was still on the 68xxxx platform, and everybody was breaking out the champagne. I see that happening here as well.
It also puts a dent in its ability to put money in R&D, since margins on the cheaper chips are much less (less money), even though the volume is very high. They will be hurt by that for sure, though saying they'll be run out of the market is a bit ridiculous.
Apple is doing the same thing Intel has done for a long time, starving the competition by taking the top of the market. Intel right now is producing less and less of the world wide processors (yes, I'm counting mobile), but because they own the mid to top market, they're sitting pretty compared to AMD and even Qualcom. I'm betting that this will provide an overture for Intel in the top of the Smart Phone market eventually .
Qualcomm is, by far, the biggest mobile processor manufacturer, and the same for their radio chips, which include a processor and memory, in addition to the radio(s). That position is in no danger of being eroded in any significant way by this. Every processor company over the years has had boner chips. This is theirs. They will figure it out, and the problem will be over, so this year, it will make a bit of a dent. But next year, it will seem as though it never happened.
Apple, of course only makes high end medium level phones, and high end phones. You can bet they've been fooling around with lower cost models in their labs. Some day, I expect to see at least a $349 model, and possibly even a $299 model. Those could double their marketshare.
But for now, most phones sell for much less than what Apple charges. Some people her talk about Marvel and others, but so far, most of their chips go into feature phones, and very cheap smartphones. Just a very small number aspire higher, though that will change somewhat, in the future.
What absolute bullshit. Even without Samsung, there's more than enough Android manufacturers around to keep Qualcomm's APs profitable. Qualcomm gets paid whether the manufacturer makes a profit or not. Around 350 million non-Samsung Android devices were sold last year and that doesn't even include non-Android devices that also use Snapdragon. That's a healthy market for Qualcomm.
Before you post such a ridiculous and uninformed post you should really educate yourself on the subject matter. First, most of the android phones you mention, other than Samsung, are from cheap ass Chinese manufacturers, many of whom have NOT been paying Qualcomm as they should, or when they do, they don't report all their sales. How would we know that? Qualcomm reported it in their security filings. (And the overwhelming majority of those android devices you mention, including Samsung's, are cheap phones NOT using Snapdragon.)
Secondly, DED's point is specific to Snapdragon CPU/baseband combo chip - if not enough companies buy it in quantity they won't be able to afford to continue development and production due to high unit costs. This is basic business economics.
I found the citations and break-out quotations to be highly exaggerating of the situation actually described in the linked articles.
For example "the company's own Exynos chips were plagued by serious design defects" - that simply isn't the case. It had a weaker GPU than the A8 (and not much weaker at the same resolution). If it was plagued you wouldn't expect it to ship successfully in millions of devices.
Historically Samsung hasn't used its Exynos chips because they only had a certain amount of them to use, to they targetted them at certain markets (especially ones that like the number 8). Luckily one of its major customers has diversified its suppliers...
Historically, Samsung doesn't use its own chips in a number of markets, because their performance lags Qualcomm's chips at the same level. Samsung has tried to be a major supplier of their chip designs, which are really just standard ARM designs, but hasn't made any headway. So the only major company who uses their chip fabs has been Apple. Once Apple began to break away with their architecture license, something that Samsung doesn't have, for some odd reason, they pulled away in performance. Samsung is such a large company, that if they really want to, they could build another major fab. But they can't sell their chips, so they don't do it.
Early this year, when they reported sales and income for their chip division, it was way down, because Apple had pulled away for the new A8 series. If Samsung produced good chips, at competitive prices, they could have filled that major gap with third party orders, or more of their own phones, but they couldn't. That why they have been fighting to get Apple's business back. Ironic, isn't it, that they can't sell their own chips, or even use them in all of their own devices, because of price and performance, but can make Apple's in the same fab?
You can always spot the idiots posting. One is when they post an Antutu benchmark, the laughing stock of the benchmark world. Then you go to Anand and find the iPhone 5S from a year ago beats the Note 4 in 6 out of 9 CPU benchmarks. Utterly pathetic for Ssmsungs latest & greatest.
The other way to spot an idiot is when they bring up shipping estimated from IDC.
So what? The article is comparing the Snapdragon 810 against an as-yet-unknown Exynos chip as options for the S6. The fact that the Apple-only A7/A8/A8X is faster is neither here nor there - it's not an option for the S6!
The article is saying that this is terrible news for the S6, because the Exynos chips are slower than the Snapdragon chips. However when it comes to the latest generation, we are talking about octa-core A57/A53 big.LITTLE SoCs - Qualcomm's next generation 64-bit core is not in this chip (it may debut in the 820 on 14nm late this year).
So it really comes down to whether the Adreno GPU in the 810 is going to be faster than the as-yet-unknown-but-presumably-Mali GPU in the Exynos.
Now I personally think that Apple's route of having few, more powerful cores is more sensible than the current rush to 8 cores in the Android market, but as an Android OEM Samsung has to choose from the high end SoCs it has available to use. Indeed it is probably time they actually committed to using their own in-house design.
It's showing really good numbers, better than the A8X. For a smartphone? Well, more likely for a tablet. But a point is that Nvidia, like Apple, Qualcomm, Marvel and one or two others, has an architecture license, not just a design license as Samsung does. The Snapdragons, for example, are Qualcomm's take on ARM, just as Apple's "A" series are.
The fact that they are having problems with the 810 isn't good, but they've already come out with several lower rated 64 bit chips that are being used. No doubt, they will solve these heating problems at some point. Those here who are getting frantic about their problems really need to sit back and take a breath. This is just one processor. Qualcomm will fix it, and a couple of years from now, everyone will forget those problems.
I remember way back when Intel released a chip that had some severe math problems that would show up at some of the worst times. Intel had to recall a lot of chips. At the time, Apple was still on the 68xxxx platform, and everybody was breaking out the champagne. I see that happening here as well.
Not good folks!
It is showing great performance numbers, but not in a shipping product, and given the release this spring, it will almost certainly be trumped in almost every metric by the A9 / A9X in September. What is known is that Nvidia has over promised in the past and I expect the same here; hence why I don't suspect that the X1 it will get many, if any, design wins in tablets. Even the recent K1 Denver has seen few design wins, and the most notable, the Nexus 9, is uniformly considered an unsuitable tablet for it's cost.
Kudos to Nvidia for having a design license, but, I'm not seeing any progress in mobile for that.
Historically, Samsung doesn't use its own chips in a number of markets, because their performance lags Qualcomm's chips at the same level. Samsung has tried to be a major supplier of their chip designs, which are really just standard ARM designs, but hasn't made any headway. So the only major company who uses their chip fabs has been Apple. Once Apple began to break away with their architecture license, something that Samsung doesn't have, for some odd reason, they pulled away in performance. Samsung is such a large company, that if they really want to, they could build another major fab. But they can't sell their chips, so they don't do it.
Early this year, when they reported sales and income for their chip division, it was way down, because Apple had pulled away for the new A8 series. If Samsung produced good chips, at competitive prices, they could have filled that major gap with third party orders, or more of their own phones, but they couldn't. That why they have been fighting to get Apple's business back. Ironic, isn't it, that they can't sell their own chips, or even use them in all of their own devices, because of price and performance, but can make Apple's in the same fab?
I'm still looking for the benchmarks of this unreleased Exynos SoC versus the unreleased Snapdragon 810 to confirm that the Exynos' performance is lagging, or indeed that the existing Exynos SoCs lack their competitor Snapdragons.
It's obvious that the integrated baseband in Qualcomm's chips makes them far more attractive for the price. I expect there is a lot of aid at the design level too - the entire package. Samsung are probably quite poor at this aspect, so they don't sell the chips.
This is why NVIDIA actually bought a baseband company, and tried to do their own chip - the Tegra 4i. It failed, because it was late to market. Since then they've changed their target markets and don't need integrated baseband, so that's now separate.
Samsung's foundry business is separate from their application processor business. Maybe not as separate as their giant mega-ship construction business (that they outperform Apple in considerably!). Additionally it appears that Samsung's 14nm process is working, and TSMC's 16nm process is not, so products out later this year will surely show more of a foundry disparity. Guess who's now using Samsung as a foundry later this year? Qualcomm.
While it's hard to find these articles when you want them, a bit of searching has shown this one, though it doesn't have the numbers, it gives the idea. If I can have the time to look more, I will. I'll either add it here, or do another post.
"Although none of the major chip analysts have actual numbers, due to the difficulty of tracking so many phones on the market, all you have to do is look at the list of phones in the Wikipedia entry for Snapdragon, Qualcomm's brand of ARM-based processors, to see how ubiquitous Qualcomm is."
You have sources with actual concrete numbers? Or something newer (that article is 18 months old).
Edited: Here's a link showing Mediatek alone shipped 220 million processors in 2013 and expects that number to grow significantly in 2014.
There's a problem with benchmarking which might explain some of the variability of the Android benchmarks. If you use Geekbench, for example, you will notice that they advise not running anything else at the same time. You should turn your phone off completely after making sure that everything that turns on when you turn the phone back on will remain off.
Due to Android using a Desktop multitasking model, many apps remain on when you turn the phone completely off, and back on again. While many Android geeks claim to be knowledgable, I fint that not to be so. They know a few minor picky things, but don't understand the system.
Because of that, most Android phones will have some apps running in the background which slow down various aspects of the benchmarks. It's much easier to control this in an iOS device. Once you turn the phone off completely, and turn it back on again, this problem doesn't exist, so the scores are much closer, just bearing the usual slight deviations in manufacturing tolerances of the components, of about 5%, or so.
You're really reaching there... the variability is beyond insane, not just small. Your telling me that all those Iphones were all perfectly calibrated for the tests (those IOS users are so disciplined...) and you have to baby the little Android phones to make them work as well?
Were are YOU getting your information about how IOS versus Android users ran these test? Hypothesis is not reality, is it?
Isn't it possible that multi-tasking/memory management is just broken down and you see the result of that... Hey, my scenario is as worth while as yours.
So what? The article is comparing the Snapdragon 810 against an as-yet-unknown Exynos chip as options for the S6. The fact that the Apple-only A7/A8/A8X is faster is neither here nor there - it's not an option for the S6!
The article is saying that this is terrible news for the S6, because the Exynos chips are slower than the Snapdragon chips. However when it comes to the latest generation, we are talking about octa-core A57/A53 big.LITTLE SoCs - Qualcomm's next generation 64-bit core is not in this chip (it may debut in the 820 on 14nm late this year).
So it really comes down to whether the Adreno GPU in the 810 is going to be faster than the as-yet-unknown-but-presumably-Mali GPU in the Exynos.
Now I personally think that Apple's route of having few, more powerful cores is more sensible than the current rush to 8 cores in the Android market, but as an Android OEM Samsung has to choose from the high end SoCs it has available to use. Indeed it is probably time they actually committed to using their own in-house design.
You're right about the technical statements you make.
But, if performance falls off because of this, it's very possible that Samsung will lose sales to Apple in markets where performance matters. If so, then this is a problem. No matter what people say, Apple's chips do compete, indirectly, with everyone else's chips. A sale to Apple is a lost sale to Qualcomm, as well as the manufacturer using their, or anyone else's, chips. They know this very well, which is why, after initial poo pooing of Apple's 64 bit designs, everyone is rushing to produce them. Remember that Apple sold 191.3 million phones last year. That's 191.3 million SoCs that Qualcomm and it's competitors didn't sell. Apple is expected to sell a lot more phones this year, taking even more sales away.
But it takes a couple of years to produce something so different, even longer, when considering lead times for new technologies. Remember, this is why Intel has their tick tock schedule. So one year, new process technology, with relatively minor chip design updates. Then the second year, major new IP. That's because the new process required a chip design to be redone for the new process itself.
What's remarkable was that Apple was able to do the tick tock at the same time for the A7. Now, for the A8 and A8X as well. This is throwing everyone off course. It's forcing everyone to do the same. Qualcomm has gotten stuck. Intel is doing it with the Atom. Nvidia with the Tegra X-1.
But Apple now has the experience and the lead, because no one expected it. The advantage of being first, is that you're not on a real competitive schedule. If it takes another year, that's fine, as there's no competition for it on the horizon. But once it's out, everyone else has to scramble, or the public will think they're behind technologically. No one wants that. So errors are being made in the rush. Oddly, Qualcomm's other, lower end, 64 bit chips haven't had any real problems. But the bleeding edge isn't so lucky.
I suspect that next year, we're more likely to see a Krait based 64 bit high end chip from them. They just haven't had the time.
They seemed to be hinting they would have something in 2014. Did that not happen? At all? In any market with any device so they could say, "We did it!"?
It is showing great performance numbers, but not in a shipping product, and given the release this spring, it will almost certainly be trumped in almost every metric by the A9 / A9X in September. What is known is that Nvidia has over promised in the past and I expect the same here; hence why I don't suspect that the X1 it will get many, if any, design wins in tablets. Even the recent K1 Denver has seen few design wins, and the most notable, the Nexus 9, is uniformly considered an unsuitable tablet for it's cost.
Kudos to Nvidia for having a design license, but, I'm not seeing any progress in mobile for that.
Well yes. Nvidia has a history of giving us numbers which actual production, in actual devices, can't match. But still, even if every spec isn't quite there, it looks to be a very competitive chip. I just hope this inspires Apple too.
Well yes. Nvidia has a history of giving us numbers which actual production, in actual devices, can't match. But still, even if every spec isn't quite there, it looks to be a very competitive chip. I just hope this inspires Apple too.
Competitive in performance, but power efficiency is undetermined, hence my skepticism. I suspect that Apple isn't inspired by anything other than the level of the GPU, and even then, probably not worried about the tradeoffs that Nvidia had to make to get that performance.
Comments
Absolutely terrible move. Why would Apple want Qualcomm? There a. Ore than a few companies that would be important for them to buy, but this one isn't it.
That's not what I said. Do you even understand what quotation marks mean?
What I said was that TI are a small player in the chip making industry. They achieved a lot of volume with Nokia but never had the resources of Qualcomm or Intel. Puzzlingly, you've then listed a series of dud products that never sold in significant volumes to back up your argument.
There's a problem with benchmarking which might explain some of the variability of the Android benchmarks. If you use Geekbench, for example, you will notice that they advise not running anything else at the same time. You should turn your phone off completely after making sure that everything that turns on when you turn the phone back on will remain off.
Due to Android using a Desktop multitasking model, many apps remain on when you turn the phone completely off, and back on again. While many Android geeks claim to be knowledgable, I fint that not to be so. They know a few minor picky things, but don't understand the system.
Because of that, most Android phones will have some apps running in the background which slow down various aspects of the benchmarks. It's much easier to control this in an iOS device. Once you turn the phone off completely, and turn it back on again, this problem doesn't exist, so the scores are much closer, just bearing the usual slight deviations in manufacturing tolerances of the components, of about 5%, or so.
I don't care whether you buy it or not. Those are the numbers the industry says is correct.
While it's hard to find these articles when you want them, a bit of searching has shown this one, though it doesn't have the numbers, it gives the idea. If I can have the time to look more, I will. I'll either add it here, or do another post.
http://www.itworld.com/article/2706472/mobile/it-s-not-an-arm-market--it-s-a-qualcomm-market.html
What, did you think you were reading a scientific journal or something? This is a tech rumor website.
He's not "citing himself." He's providing links to his (and others) previous articles that explain events in greater detail.
He's often an egotistical ass, but I don't think his linked articles exemplify this.
In any case, researchers often cite their own work in scientific papers. Often they are the leading authorities on the subject or they are building on previous work they have done. It's not at all frowned upon (if the work is good.) On the contrary, it's common practice.
I found the citations and break-out quotations to be highly exaggerating of the situation actually described in the linked articles.
For example "the company's own Exynos chips were plagued by serious design defects" - that simply isn't the case. It had a weaker GPU than the A8 (and not much weaker at the same resolution). If it was plagued you wouldn't expect it to ship successfully in millions of devices.
Historically Samsung hasn't used its Exynos chips because they only had a certain amount of them to use, to they targetted them at certain markets (especially ones that like the number 8). Luckily one of its major customers has diversified its suppliers...
It's showing really good numbers, better than the A8X. For a smartphone? Well, more likely for a tablet. But a point is that Nvidia, like Apple, Qualcomm, Marvel and one or two others, has an architecture license, not just a design license as Samsung does. The Snapdragons, for example, are Qualcomm's take on ARM, just as Apple's "A" series are.
The fact that they are having problems with the 810 isn't good, but they've already come out with several lower rated 64 bit chips that are being used. No doubt, they will solve these heating problems at some point. Those here who are getting frantic about their problems really need to sit back and take a breath. This is just one processor. Qualcomm will fix it, and a couple of years from now, everyone will forget those problems.
I remember way back when Intel released a chip that had some severe math problems that would show up at some of the worst times. Intel had to recall a lot of chips. At the time, Apple was still on the 68xxxx platform, and everybody was breaking out the champagne. I see that happening here as well.
Not good folks!
Qualcomm is, by far, the biggest mobile processor manufacturer, and the same for their radio chips, which include a processor and memory, in addition to the radio(s). That position is in no danger of being eroded in any significant way by this. Every processor company over the years has had boner chips. This is theirs. They will figure it out, and the problem will be over, so this year, it will make a bit of a dent. But next year, it will seem as though it never happened.
Apple, of course only makes high end medium level phones, and high end phones. You can bet they've been fooling around with lower cost models in their labs. Some day, I expect to see at least a $349 model, and possibly even a $299 model. Those could double their marketshare.
But for now, most phones sell for much less than what Apple charges. Some people her talk about Marvel and others, but so far, most of their chips go into feature phones, and very cheap smartphones. Just a very small number aspire higher, though that will change somewhat, in the future.
Before you post such a ridiculous and uninformed post you should really educate yourself on the subject matter. First, most of the android phones you mention, other than Samsung, are from cheap ass Chinese manufacturers, many of whom have NOT been paying Qualcomm as they should, or when they do, they don't report all their sales. How would we know that? Qualcomm reported it in their security filings. (And the overwhelming majority of those android devices you mention, including Samsung's, are cheap phones NOT using Snapdragon.)
Secondly, DED's point is specific to Snapdragon CPU/baseband combo chip - if not enough companies buy it in quantity they won't be able to afford to continue development and production due to high unit costs. This is basic business economics.
Historically, Samsung doesn't use its own chips in a number of markets, because their performance lags Qualcomm's chips at the same level. Samsung has tried to be a major supplier of their chip designs, which are really just standard ARM designs, but hasn't made any headway. So the only major company who uses their chip fabs has been Apple. Once Apple began to break away with their architecture license, something that Samsung doesn't have, for some odd reason, they pulled away in performance. Samsung is such a large company, that if they really want to, they could build another major fab. But they can't sell their chips, so they don't do it.
Early this year, when they reported sales and income for their chip division, it was way down, because Apple had pulled away for the new A8 series. If Samsung produced good chips, at competitive prices, they could have filled that major gap with third party orders, or more of their own phones, but they couldn't. That why they have been fighting to get Apple's business back. Ironic, isn't it, that they can't sell their own chips, or even use them in all of their own devices, because of price and performance, but can make Apple's in the same fab?
You can always spot the idiots posting. One is when they post an Antutu benchmark, the laughing stock of the benchmark world. Then you go to Anand and find the iPhone 5S from a year ago beats the Note 4 in 6 out of 9 CPU benchmarks. Utterly pathetic for Ssmsungs latest & greatest.
The other way to spot an idiot is when they bring up shipping estimated from IDC.
So what? The article is comparing the Snapdragon 810 against an as-yet-unknown Exynos chip as options for the S6. The fact that the Apple-only A7/A8/A8X is faster is neither here nor there - it's not an option for the S6!
The article is saying that this is terrible news for the S6, because the Exynos chips are slower than the Snapdragon chips. However when it comes to the latest generation, we are talking about octa-core A57/A53 big.LITTLE SoCs - Qualcomm's next generation 64-bit core is not in this chip (it may debut in the 820 on 14nm late this year).
So it really comes down to whether the Adreno GPU in the 810 is going to be faster than the as-yet-unknown-but-presumably-Mali GPU in the Exynos.
Now I personally think that Apple's route of having few, more powerful cores is more sensible than the current rush to 8 cores in the Android market, but as an Android OEM Samsung has to choose from the high end SoCs it has available to use. Indeed it is probably time they actually committed to using their own in-house design.
It's showing really good numbers, better than the A8X. For a smartphone? Well, more likely for a tablet. But a point is that Nvidia, like Apple, Qualcomm, Marvel and one or two others, has an architecture license, not just a design license as Samsung does. The Snapdragons, for example, are Qualcomm's take on ARM, just as Apple's "A" series are.
The fact that they are having problems with the 810 isn't good, but they've already come out with several lower rated 64 bit chips that are being used. No doubt, they will solve these heating problems at some point. Those here who are getting frantic about their problems really need to sit back and take a breath. This is just one processor. Qualcomm will fix it, and a couple of years from now, everyone will forget those problems.
I remember way back when Intel released a chip that had some severe math problems that would show up at some of the worst times. Intel had to recall a lot of chips. At the time, Apple was still on the 68xxxx platform, and everybody was breaking out the champagne. I see that happening here as well.
Not good folks!
It is showing great performance numbers, but not in a shipping product, and given the release this spring, it will almost certainly be trumped in almost every metric by the A9 / A9X in September. What is known is that Nvidia has over promised in the past and I expect the same here; hence why I don't suspect that the X1 it will get many, if any, design wins in tablets. Even the recent K1 Denver has seen few design wins, and the most notable, the Nexus 9, is uniformly considered an unsuitable tablet for it's cost.
Kudos to Nvidia for having a design license, but, I'm not seeing any progress in mobile for that.
Historically, Samsung doesn't use its own chips in a number of markets, because their performance lags Qualcomm's chips at the same level. Samsung has tried to be a major supplier of their chip designs, which are really just standard ARM designs, but hasn't made any headway. So the only major company who uses their chip fabs has been Apple. Once Apple began to break away with their architecture license, something that Samsung doesn't have, for some odd reason, they pulled away in performance. Samsung is such a large company, that if they really want to, they could build another major fab. But they can't sell their chips, so they don't do it.
Early this year, when they reported sales and income for their chip division, it was way down, because Apple had pulled away for the new A8 series. If Samsung produced good chips, at competitive prices, they could have filled that major gap with third party orders, or more of their own phones, but they couldn't. That why they have been fighting to get Apple's business back. Ironic, isn't it, that they can't sell their own chips, or even use them in all of their own devices, because of price and performance, but can make Apple's in the same fab?
I'm still looking for the benchmarks of this unreleased Exynos SoC versus the unreleased Snapdragon 810 to confirm that the Exynos' performance is lagging, or indeed that the existing Exynos SoCs lack their competitor Snapdragons.
It's obvious that the integrated baseband in Qualcomm's chips makes them far more attractive for the price. I expect there is a lot of aid at the design level too - the entire package. Samsung are probably quite poor at this aspect, so they don't sell the chips.
This is why NVIDIA actually bought a baseband company, and tried to do their own chip - the Tegra 4i. It failed, because it was late to market. Since then they've changed their target markets and don't need integrated baseband, so that's now separate.
Samsung's foundry business is separate from their application processor business. Maybe not as separate as their giant mega-ship construction business (that they outperform Apple in considerably!). Additionally it appears that Samsung's 14nm process is working, and TSMC's 16nm process is not, so products out later this year will surely show more of a foundry disparity. Guess who's now using Samsung as a foundry later this year? Qualcomm.
While it's hard to find these articles when you want them, a bit of searching has shown this one, though it doesn't have the numbers, it gives the idea. If I can have the time to look more, I will. I'll either add it here, or do another post.
http://www.itworld.com/article/2706472/mobile/it-s-not-an-arm-market--it-s-a-qualcomm-market.html
From the article you linked:
"Although none of the major chip analysts have actual numbers, due to the difficulty of tracking so many phones on the market, all you have to do is look at the list of phones in the Wikipedia entry for Snapdragon, Qualcomm's brand of ARM-based processors, to see how ubiquitous Qualcomm is."
You have sources with actual concrete numbers? Or something newer (that article is 18 months old).
Edited: Here's a link showing Mediatek alone shipped 220 million processors in 2013 and expects that number to grow significantly in 2014.
http://global.ofweek.com/news/Sources-MediaTek-smartphone-solution-shipments-to-grow-robustly-in-2014-7158
There's a problem with benchmarking which might explain some of the variability of the Android benchmarks. If you use Geekbench, for example, you will notice that they advise not running anything else at the same time. You should turn your phone off completely after making sure that everything that turns on when you turn the phone back on will remain off.
Due to Android using a Desktop multitasking model, many apps remain on when you turn the phone completely off, and back on again. While many Android geeks claim to be knowledgable, I fint that not to be so. They know a few minor picky things, but don't understand the system.
Because of that, most Android phones will have some apps running in the background which slow down various aspects of the benchmarks. It's much easier to control this in an iOS device. Once you turn the phone off completely, and turn it back on again, this problem doesn't exist, so the scores are much closer, just bearing the usual slight deviations in manufacturing tolerances of the components, of about 5%, or so.
You're really reaching there... the variability is beyond insane, not just small. Your telling me that all those Iphones were all perfectly calibrated for the tests (those IOS users are so disciplined...) and you have to baby the little Android phones to make them work as well?
Were are YOU getting your information about how IOS versus Android users ran these test? Hypothesis is not reality, is it?
Isn't it possible that multi-tasking/memory management is just broken down and you see the result of that... Hey, my scenario is as worth while as yours.
You're right about the technical statements you make.
But, if performance falls off because of this, it's very possible that Samsung will lose sales to Apple in markets where performance matters. If so, then this is a problem. No matter what people say, Apple's chips do compete, indirectly, with everyone else's chips. A sale to Apple is a lost sale to Qualcomm, as well as the manufacturer using their, or anyone else's, chips. They know this very well, which is why, after initial poo pooing of Apple's 64 bit designs, everyone is rushing to produce them. Remember that Apple sold 191.3 million phones last year. That's 191.3 million SoCs that Qualcomm and it's competitors didn't sell. Apple is expected to sell a lot more phones this year, taking even more sales away.
But it takes a couple of years to produce something so different, even longer, when considering lead times for new technologies. Remember, this is why Intel has their tick tock schedule. So one year, new process technology, with relatively minor chip design updates. Then the second year, major new IP. That's because the new process required a chip design to be redone for the new process itself.
What's remarkable was that Apple was able to do the tick tock at the same time for the A7. Now, for the A8 and A8X as well. This is throwing everyone off course. It's forcing everyone to do the same. Qualcomm has gotten stuck. Intel is doing it with the Atom. Nvidia with the Tegra X-1.
But Apple now has the experience and the lead, because no one expected it. The advantage of being first, is that you're not on a real competitive schedule. If it takes another year, that's fine, as there's no competition for it on the horizon. But once it's out, everyone else has to scramble, or the public will think they're behind technologically. No one wants that. So errors are being made in the rush. Oddly, Qualcomm's other, lower end, 64 bit chips haven't had any real problems. But the bleeding edge isn't so lucky.
I suspect that next year, we're more likely to see a Krait based 64 bit high end chip from them. They just haven't had the time.
I'm making fun of the fact that Samsung was so quick to say "64-bit mobile CPU? Me too!!!" the same day Apple dropped the mic.
Well yes. Nvidia has a history of giving us numbers which actual production, in actual devices, can't match. But still, even if every spec isn't quite there, it looks to be a very competitive chip. I just hope this inspires Apple too.
I guess it would not "be cool" for Samsung to ask Apple if they could buy some A-series chips that they (Samsung) make.
Well yes. Nvidia has a history of giving us numbers which actual production, in actual devices, can't match. But still, even if every spec isn't quite there, it looks to be a very competitive chip. I just hope this inspires Apple too.
Competitive in performance, but power efficiency is undetermined, hence my skepticism. I suspect that Apple isn't inspired by anything other than the level of the GPU, and even then, probably not worried about the tradeoffs that Nvidia had to make to get that performance.