poke01
Diamond Member
- Mar 8, 2022
- 4,836
- 6,163
- 106
I thought the ISA doesn’t matter, which one is it? Does it only matter to suit your biases.I also don't think that rewriting an RTOS for a baseband in x86 is trivial either.
I thought the ISA doesn’t matter, which one is it? Does it only matter to suit your biases.I also don't think that rewriting an RTOS for a baseband in x86 is trivial either.
The reference was to Intel's failure at basebands, not Apple's current implementation of the standards.That's an old article from back when Apple was buying modems from Intel and thus had no control over what ISA was used in the baseband. There's no way they're using x86 in the C1, especially since they basically had to scrap everything Intel did and start over.
Getting a low power, compact physical implementation of a baseband is also VERY hard and has to bake in various analog signal processing steps entirely in silicon and calls for special doping.The hardware is easy (relatively speaking) it is the software that's a nightmare. That's because for an industry that runs on ITU "standards" like LTE, 5G, etc. there is a massive amount of grey area and wiggle room in those standards and different base stations vendors and different cellular providers all have their own wrinkles that a phone must properly handle. Qualcomm has been in it from the start so they have been adding checks and special cases to their baseband code since forever, leaving an Everest sized hill to climb for a competitor starting from scratch. That's probably why Samsung has never taken their modem worldwide and instead goes to Qualcomm in large regions despite having their own SoC line and even their own fabs the whole time they've been making smartphones.
I wonder how much of that cruft you can be rid of if you drop support for 2G & 3G. Yes I know they are still used outside of the US in many places, but much of that support is simply because many areas have wide deployment of IoT devices like alarms and utility meters on older tech that would be expensive to replace. I have to wonder how much area there is in the world that's covered ONLY by 2G or 3G but is without LTE or 5G coverage, so that a LTE/5G only phone would not work. I guess a lot more than I would have assumed, otherwise Apple wouldn't have gone to all that extra effort to support those outdated standards.
I think it's more the other way around --you pick your words and examples to try to portray mine as bias. (The subject was Intel's failure..)I thought the ISA doesn’t matter, which one is it? Does it only matter to suit your biases.
Well, it’s not surprising that Intel chooses the least efficient way to design. It’s evident in their CPUs as well.I think it's more the other way around --you pick your words and examples to try to portray mine as bias. (The subject was Intel's failure..)
User facing big cores don't have as much decoding complexity relative to caches, branch predictors, reorder buffers etc., microcontrollers do. Having to rewrite an RTOS in a different ISA rather than using / modifying an existing one is clearly going to be complex as well.
It seems like the X-mont core team is doing better than the X-cove team from recent benchmarks and efficiency comparisons. Maybe that team should just do all the cores...Well, it’s not surprising that Intel chooses the least efficient way to design. It’s evident in their CPUs as well.
It’s quite evident your beef with Apple is cause of that Qualcomm licence lawsuit which Apple started because they were cheap.It's nothing new for Apple to manipulate other suppliers to file lawsuits that benefit themselves.
They left cause Apple said no to servers. Yes, Gerard was sued and later it was dropped. Apple also sued Rivos and settled. These are intimidation tactics from a non-biased point of view. I don’t believe it’s the right approach either.Those Nuvia engineers didn't want to be at Apple and GWIII was attacked by the company following their departure.
Meanwhile, you have a CEO at Qualcomm that mislead customers with the X Elite. They painted a cosy picture on how good the X Elite was and how it more efficient than a M2 Max( it wasn’t).Tim Cook has publicly said that they shouldn't follow the rules and that they should write them --even if it means robbing others blind of their work. They don't care about others' valid claims to IP be it systems or devs.
When you own a basket of these smart phone supplier companies and want to analyze them objectively, you have to read between the lines of public statements and dig through patterns of behavior and motivation.It’s quite evident your beef with Apple is cause of that Qualcomm licence lawsuit which Apple started because they were cheap.
They left cause Apple said no to servers. Yes, Gerard was sued and later it was dropped. Apple also sued Rivos and settled. These are intimidation tactics from a non-biased point of view. I don’t believe it’s the right approach either.
Meanwhile, you have a CEO at Qualcomm that mislead customers with the X Elite. They painted a cosy picture on how good the X Elite was and how it more efficient than a M2 Max( it wasn’t).
I’ll end my point here, when you say one company does all this bad business and yet not a peep of Qualcomms bad business practices is proof of bias.
It’s good to be critical and I agree Apple is the bigger crook but that is cause Apple is a just a bigger company. It’s corporate greed and pleasing shareholders whatever means necessary.When you own a basket of these smart phone supplier companies and want to analyze them objectively, you have to read between the lines of public statements and dig through patterns of behavior and motivation.
Apple really does stand out to me in terms of the scope of their malfeasance --Qualcomm, no saint at all, is a small-time amateur crook by comparison (I believe they bought out all the viable envelope tracking competitors just to seal up the market). The press rarely looks deeper when it comes to Apple's public statements when they commit their malfeasance and just swallows them hook, line, and sinker, basking in the glow of their warm feelings for iPhones and marketing. Suppliers get their arm twisted and have to comply or risk losing the bulk of their revenues. Ask shareholders of GTAT (a viable solar panel manufacturer stuck w/ a terrible lease structure from Apple for sapphire ovens that ultimately failed to produce viable phone glass) or Dialog Semi (threatened with the dropping of their power management devices from iPhones when Apple had no alternative supplier lined up and forced to sell their IP for pennies on the dollar after what that announcement did to share prices) what they think of Apple and you might get a very different sentiment.
Maybe this do as I please because I'm big attitude is good for shareholders for now (hey, the zeitgeist now is: Trump or maybe Xi if you look east...), but there is ample evidence for the DoJ's anti-trust enforcement and a flat footedness for innovation in their focus on only 70 million units per quarter products and in-house everything --this worries me as a shareholder.
Getting a low power, compact physical implementation of a baseband is also VERY hard and has to bake in various analog signal processing steps entirely in silicon and calls for special doping.
I don't think Apple has much to benefit from Qualcomm winning since it is the prime driver of high-end devices in the Android ecosystem --ARM didnt' disclose its terms with Apple during discovery but it is likely peanuts compared to what Qualcomm pays as it is a cornerstone investor as well as an important contributor to the ISA, much more so than say Google.It’s good to be critical and I agree Apple is the bigger crook but that is cause Apple is a just a bigger company. It’s corporate greed and pleasing shareholders whatever means necessary.
That being said, I just find no relationship between QC Vs ARM and Apple here. If anything Apple would Qualcomm to win cause ARM was being petty with its contracts.
But facts are likely no one influenced ARM and it acted alone and Qualcomms own statements adhere to that belief. It’s good that Qualcomm is suing them ARM again, it’s better for the platform and ARM doesn’t pull these sorts of tricks again.
You're right for LTE and 5G on, so my bad, I'm only human. Was recalling the older 3G modems which did integrated some of the RF frontend components and required a fluorine implant.The baseband is all digital. The RF stuff is handled in a separate transceiver chip. That's how Apple does it, that's how Qualcomm does it, that's how I assume everyone else does it.
If that’s the case Google, Nvidia and Mediatek also have a vested interest against Qualcomm especially since they have PC/server intentions.I agree that it is possible that ARM just took a page out of the Apple playbook (very flimsy but very public case), but Apple has quite a bit more to gain from the blow Android would suffer if Qualcomm had lost. They have had a vested interest in keeping key designers from leaving for competing startups/rivals.
They certainly do -- but the beef between Qualcomm and Apple is special. It's rare that a smaller company survives a mauling by Apple who typically expects suppliers to supplicate for the "privilege" of working with them. The reason I talk about Qualcomm is because most people either haven't heard of or remembers the mauling that occurred at GTAT, Imagination, etc. They aren't really around anymore to talk about or are a husk of their former selves after a bad Apple deal.If that’s the case Google, Nvidia and Mediatek also have a vested interest against Qualcomm especially since they have PC/server intentions.
But generally ARM winning this case would have bad for the ecosystem.
