Qualcomm, HTC, chipsets and features: An Insider Q + A

Posted on Wednesday, Feb 13, 2008 by Malatesta
 

cartoon_detective.gif qualcomm.jpg

With the recent spate of Qualcomm info (they just showed a whole new lineup of next-gen chips, including an improved version of aGPS called…ready for it… gpsOneXTRA, I suppose tech companies have to pay for vowels) and frustration with HTC over the “missing driver controversy”, the community is wondering just what is going on here?

Was HTC just cheaping out (something yours truly has even said, ahem)? Is Qualcomm manipulating the numbers or do they have faulty chips? And what does this all mean for the future of WM devices?

Read on for some off-the-record information from someone "in the know" about what is going on with Qualcomm and HTC. The answers are quite fascinating…

At issue here are a few questions that frequently come up in our forums:

  • Why is aGPS not frequently enabled?
  • Who’s at fault in the HTC/Qualcomm driver class-action lawsuit?
  • Why don’t we have ~5mp cameras on a lot of devices, especially CDMA?

Surely there are even more tech ponderings that can be asked and maybe in the future we’ll update this piece to reflect those, but for now lets see what our tipster has to say on these issues. For what I hope are somewhat obvious reason, their identity needs to remain anonymous.

(While the technical information is accurate, the questions and answers are paraphrased by the author for the flow of the article.)

Q: Why do a lot of overseas GSM phones have massive 5MP cameras, yet their sister (CDMA) versions seem to have a 2.1MP limit? e.g. Samsung SGH-F490 vs. Sprint M800.

A:Currently, the costs of 5MP sensors are somewhat prohibitive. But in addition, current Qualcomm chips lack certain features like jpeg hardware and the system can be over-taxed from transferring the sensor-data into memory for post-processing. In essence, multitasking with the OS, with an active radio and using the camera may be too much for the processor so the resolution of the sensor may need to be reduced..

Q: What’s the deal with aGPS (GPSOne) both on the earlier MSM-6500 platform and the newer MSM-7500, specifically why was it never activated on the earlier devices and just who decides?

A: When it comes down to modern aGPS, it is strictly a money issue: carriers are not keen to pay for the drivers and the OEMs (HTC, Moto, etc.) are not going to foot the bill (Edit: Although it looks like Sprint and some other carriers are finally eating the cost on their new 2008 devices). On older devices, the aGPS on the MSM-6500 chips may not have been as efficient or accurate, since they were developed back in 2002, hence why they may not have been used.

And now for that million-dollar question: Who’s at fault in the HTC-Qualcomm debacle? Turns out it’s a lengthy answer and despite was some may want to believe, there really was no malicious intent from either side but the answer might point to a complicated system of pricing tiers, varying functions, promised performance gains not met and overall grayness where the only losers are the consumers…more on this issue after the Q/A.

Quick technical note: the ARM9 processors are found on the MSM-62/6500 chipsets (PDF!) like in the PPC-6700 and Treo 7xx series; the ARM11 processors are found on the MSM-72/7500 chipsets like the Sprint Mogul, HTC Touch, etc. where it handles the OS, graphics and programs. Also on that chipset, a secondary ARM9 processor handles the phone-radio aspect—many thought this dual-processor system would result in superior performance, but instead we have the HTC class-action suit. Interestingly, Qualcomm has recently pulled all of their data sheets for some reason. Hmm.

Q: HTC, Qualcomm and the missing drivers—where do we send the angry mob with torches?

A: Qualcomm has a tiered pricing policy with their chipsets—so although you bought the chip, you have not bought all the features. So you have to pay additional fees per phone to get things like aGPS, graphic acceleration, etc.
In the past, HTC had no problems when using the older MSM-6500 chips (ARM9 processors) without drivers hence their reluctance to pay for any or additional support with the new MSM-7500 chips (ARM11 processors), especially since the newer processors were advertised to match or outperform the older generation.
Unfortunately, Qualcomm’s ARM11 performance does not match their previous ARM9 processor and is therefore, not quite as advertised. To get the proper performance out of the ARM11, one has to have knowledge of the processor’s implementation and design, but since that processor is not publicly available; the solution requires cooperation and assistance. HTC in this instance does not have this knowledge and is therefore unable to directly fix the problem, so they are put in a tough situation as they already have millions of these devices sold but they don’t want to pay Qualcomm more than they have to.

Q: So Qualcomm sort of pulled a shell-game here, much like the recent CES + “Snapdragon” controversy?

A: The shell game comment is a bit strong but somewhat understandable. When Qualcomm sets out to make these processors, the marketing information comes out way before the final design and chip does, so while Qualcomm had the intention of outdoing themselves, in reality they fell behind the mark. Due to the nature of the industry, by the time these performance issues became knowledge, it was too late to fix as they had already moved on to development of other chipsets and processors.
No doubt Qualcomm behaves like a lot of companies and they don’t want to admit internally or publicly that their processors are underperforming, so this hampers any immediate resolution--the problem then becomes compounded as time goes on.
Because of this, HTC is put in an odd position as they are selling devices based on Qualcomm’s marketing and information, not on their own knowledge of how those processors work and are designed since they are not privy to that information. Qualcomm is not being as straightforward with them and in turn, HTC is trying to work around the issue by trying to fix or enhance their software, even though they are unaware that it is not really their fault. It is in HTC’s best interest to of course try and fix this issue right away and if they know what they need to know perhaps they would foot the bill for those drivers, but unfortunately all the butt covering at all levels is preventing certain people who need to know from getting the job done

So there you go folks—judge accordingly I suppose. In hindsight, I have to revise my “HTC was cheap” line and instead sort of paint them as an unwitting victim who has been put in a tough situation by Qualcomm. Along the same line, Qualcomm didn’t exactly out right deceive anyone, but they seemed to have slipped up a bit on the MSM-72/7500 platform quite a bit.

The latest Qualcomm chipsets are promising everything under the hood: Wifi, WiMax 3g, 4g, BT 2.1, FM radio, VOIP, video out, aGPS,/GPS, 2D + 3D video acceleration, mobile TV, ad nauseum. But the caveat should be obvious by now: you are not entitled to all of those features and it is up to the OEM and carriers to pick and choose what they want, a la carte style.

In a way, this makes sense as it certainly allows device manufactures and carriers to offer a wide range of devices with varying functionality and pricing, from low to high.

But it also means that as the target audience, you are at the mercy of those OEMs, carriers and Qualcomm to sign up and pay for those features (it certainly doesn’t help when the promised performance of new-gen chips fall below their predecessors!). This sort of begs the question: what about Android devices? They too will have to pass through the labyrinth of carrier testing, OEM development and yes, paying Qualcomm for drivers to unlock those magical and prized features. Will that decentralized and somewhat chaotic software model work when we have these ongoing issues now?

So what do you think? Post your thoughts and comments below. And if you have some questions of your own about GSM/CDMA chipsets and functionality, ask away and maybe we’ll do a sequel article.

And a special thanks to our tech insider for all the info!

 
 

Comments

As an owner of a Sprint HTC Mogul, I've been following this issue.
If what this "Insider Q&A" is true then HTC certainly has a legal leg to stand on and call out Qualcomm to produced the drivers for free.

Summary:
HTC is given information of Qualcomm chips that can achieve certain performance (not even touching the driver issue).
Qualcomm has problems with the chip performing under par.
HTC produces units under expectation of certain performance based on Qualcomm's claims.
Units produced are now "crippled" because of under performance.

My conclusion:
It would be only right for Qualcomm to produce drivers that would allow the chips to achieve their stated performance.
I don't have direct contact with the insider, but what he/she is saying rings very true to me.

@scsl: I don't think a lawsuid would happen. My feeling is still that, while interesting to see inside the business, none of this really is going to change the fact that sometimes chip features don't get utilized on the devices that they're on. I don't think HTC was advertising more than they sold -- it was a realization after the fact that these chips could possibly have video acceleration. Or at least that's how I remember this all shaking out.
@Dieter:
I'm not saying a lawsuit should happen either between HTC/Qualcomm or class action/HTC. It could be easily resolved if Qualcomm stepped up and admit their ARM11 chips aren't up to par without workaround at the driver levels and make things right by providing that information to HTC.

There are two distinct issues here:
1. Chipset assisted video acceleration
2. Actual usability without appropriate video drivers

Most of the HTC Mogul users are NOT looking for "video acceleration" drivers which utilize on-board ATI Imageon hardware. They are looking for HTC to provide a video driver that does not slow the machine to a crawl.

If I am reading this Q&A right, the differences between the ARM9 and ARM11 was different enough that HTC's prior solutions targeting the ARM9 chips did not translate well to the new ARM11 chipset AND that Qualcomm knew about performance problems with the ARM11, needs some kind of work around at the driver level, but will not provide them unless HTC forks over a bunch of money for those driver level work arounds.
@Malatesta:

First of all, thanks for an informative article. This is the kind of information I want from WMExperts, not the one shot "look at this!" blurbs so common here and elsewhere like at Engadget and Gizmodo.

The true "victim" in all this is the end consumer, typically an enthusiast or user with the funds to pony up for a high end HTC WM device. HTC and Qualcomm can finger point all they want, the fact is they both traded on performance claims for their respective products, claims which unfortunately didn't fully pan out for their customers. We expect such behavior from the U.S. carrier cartel, it's unfortunate to learn that it extends to the handset manufacturers and their suppliers.

It would serve HTC and Qualcomm well to put aside the "it's your responsibility" debate and develop the drivers necessary to get the best performance from our phones, especially for those of us that buy unlocked HTC branded devices. Surely the development cost would be measured in the pennies per sold device.
@Malatesta:

First of all, thanks for an informative article. This is the kind of information I want from WMExperts, not the one shot "look at this!" blurbs so common here and elsewhere like at Engadget and Gizmodo.

lol, thanks. We're trying to gauge how "techy" these articles should be as a lot of my stuff veers on the very-techy level, plus it's hard to come by that kind of info often (though the more "off the record" the merrier for myself as I'm very interested in these carrier+OS+OEM+Qualcomm relationships).

I also agree with your conclusion 100%--some one should "take one for the team" but I'm not holding my breath--I think Qualcomm at this point has moved beyond the MSM-72/7500 technology and would rather forget it.

scsl: you're summary post is dead on and I probably should have made that clearer in the article: video acceleration technically is not needed, it's the fact that the ARM11 just performs worse than the ARM9. Presumably either the software fix and/or video acceleration would help ameliorate the current problems.
@pelona
Point taken. We're working on it (note I went looking for more writers!). On the other hand, I'm a fanboy and sometimes it's tough not to get excited. ;)

@everybody
I like this @soandso thing. It's not as cool as the full vBulletin quote, but it's a heckuva lot quicker. :D
Maybe this Qualcomm/HTC episode will mark the beginning of mobile device consumer 'CPU awareness' in a similar fashion to desktop/notebook devices.
This will force CPU vendors to behave like Intel and AMD where having performance and features (available) sells their product - rather than having clever tiered pricing policies that cause performance and features to be disabled for commercial reasons.

Maybe we shouldn't trust the specifications until devices get reviewed and part of that review should be a benchmark suite - just like PCs?

Reading between the lines - it seems like Qualcomm developed a bit of a 'lemon' and they are happy that they are moving the stock out of the factory.
Thanks to all for the very informative article.

@mal: you can word these things however you want... I'll read them either way. thank you for providing info and insight on the inter-relationships between the different parts of the mobile industry, it sheds alot of light on how/why things are the way the are, and perhaps how they could be better.

quick thing: Not positive on the Qualcomm/hardware mfr relationship, but does this not open the door to a 3rd party software vendor who will buy the 'fancy' drivers from Qcomm and re-sell them to users? Unless the MSM7500 is forever an under-performer, I'd line-up to sign-up to pay-up for performance enhancing drivers... you take visa? lol

You're saying you'll buy an $500-$800 dollar phone and then pay more for drivers to get it working the way it was advertised?

quick thing: Not positive on the Qualcomm/hardware mfr relationship, but does this not open the door to a 3rd party software vendor who will buy the 'fancy' drivers from Qcomm and re-sell them to users? Unless the MSM7500 is forever an under-performer, I'd line-up to sign-up to pay-up for performance enhancing drivers... you take visa? lol


Whoa, that might be a good direction for the people that were organizing to petition HTC to point their efforts at...
I must say.
This is one of the most informative and intelligent written articles I've ever read.

Truly enjoyable to read, even though the outcome of the driver issue is uncertain.

Thanks
@Dieter

I'm a fanboy too, ever since the HP LX95. (It died an unfortunate death in a hot tub during and impromptu demo...) Until a few years ago I owned virtually every iPAQ until the operating system on phones was good enough to be useful. So like you, I'm an enthusiast and will forgive the occasional outburst of delight.

My point is that posts like this as well as the WM focus differentiates WMExperts from the pack. Your Touch vs. iPhone video is another example. Though there were similar comparisons out there, no one did it as well. The video led me to discover WMExperts and played a factor in my purchase of a Touch Enhanced. (One question WMExperts didn't answer was whether or not the GSM Touch is quadband. Though there were plenty of testimonials on Howards Forums and XDA Developers that it is, there was no definitive information. Even the HTC representative in the partners area of the Microsoft booth at CES this year didn't know, and he personally used the Touch! The answer is that for all practical purposes the GSM Touch is quad band, though I found out only by going back to the TI spec sheets. That's the kind of thing I hope to find out from WM Experts.)

Sorry to get off topic...
I'd say it's the greed of Qualcomm. They set out to meet performance levels they promised, then FAILED!? HELLO? That alone should be reason enough for them to give a discount of sorts on the features HTC won't pay for.

I don't see a resolve to this issue anytime soon. Instead, when other chipsets start coming out, qualcomm will need to get competitive and this sort of horse poo will go away.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <img>
  • Lines and paragraphs break automatically.

More information about formatting options

temp ad

software ad

Smartphone Round Robin

New in the Forums

Follow WMExperts

 

Subscribe via RSS
   

Add to Google Reader or Homepage


What is RSS?

Subscribe to the WMExperts Store Newsletter:

 
 
Creating smartphone communities
Android Central - Android reviews, news and forums Crackberry - Blackberry news, reviews and community TiPb - iPhone news, accessory reviews & forums
Pre Central - Palm Pre Review, News and Community Treo Central - Treo & Centro News and Forums WMExperts - Windows Mobile Reviews & News