The Intel Atom Thread

[NTMBK]

Nice review of a Bay Trail PC on a stick: http://www.eurogamer.net/articles/digitalfoundry-2015-hannspree-micro-pc-review Performance is a bit weak from the sound of it (it's the cut down F-series chip), but it sounds perfect for Steam streaming.

 

[Arachnotronic]

How do you explain otherwise the 4.2bn losses in this business.?.

Research and Development.

 

[Shivansps]

For those that dont understand what Intel have done in the Tablet space,

OEM/ODMs could buy BayTrail-T for $4, yeap four Bucks. That is what the Contra-Revenue was all about.

BUT,

Intel would offer a rebate of up to $3 (perhaps more) if you would reach a target. For example (numbers are fictional) OEM/ODMs would get a rebate of $3 if they would buy 100.000 BayTrails-T.

That is very well illustrated in Intels Financial reports.
From Q4 2014 the Intel Mobile and Communications Group (MCG) reported a Net Revenue of -6Millions

For those that dont know, Net Revenue is the Revenue after any discounts/returns/rebates. That is if your total sales (revenue) is $1000 but your discounts/rebates are -$1006 your Net Revenue = 1000-1006 = -$6

And although Contra-Revenue was created in order for BayTrail-T to be price-competitive against ARM products (Android), it also had a huge impact in Windows Tablets as well. None in the industry would buy a $10-15-20 AMD APU for Windows tablets when BayTrail-T was sold at $1.

Contra-Revenue was non existant when AMD released Bobcat? What i do remember that, back then, it only keep AMD away from the cheaper options.

Ill more than happy to pay $200 instead of $150 for a AMD tablet if its really better than BT-T, but history tells me if its really better there whould be devices.
Look, the E1 Micro-6200T is 100% pure crap compared to any BT, the A4 Micro-6400T used in the HP Stream 14z it said to throttled back to 800mhz during load, thats BAD, really bad BT-T never goes below the base clock of 1.33ghz, at those clocks BT-T, even the slower one, its faster than the AMD Mullins whiout doubt, and while battery is great, is no better than BT-T either, and gaming? its about the same too, its better, but you are still locked to low resolutions and low details, nothing that really worth it, and i cant see the A10 mullin to be any better because it shares the same SDP.

And im not sure what kind of heatsink the HP14z is using, i known its passive but BT-T tables use a very thin aluminium sheet, i dont think the HP 14z is using anything like that.

 

[eRacer]

Research and Development.

While Intel spends plenty of money on R&D, revenues are supposed to more than make up for such expenses. Intel is basically giving away processors in order to compete with ARM in tablets and phones. The question is whether Intel sees long term profits in that market, or is Intel just making a preemptive strike in ARM territory to prevent/slow ARM competitors from eventually taking the fight to Intel in the lucrative x86 server/desktop/notebook markets.

 

[Arachnotronic]

While Intel spends plenty of money on R&D, revenues are supposed to more than make up for such expenses. Intel is basically giving away processors in order to compete with ARM in tablets and phones. The question is whether Intel sees long term profits in that market, or is Intel just making a preemptive strike in ARM territory to prevent/slow ARM competitors from eventually taking the fight to Intel in the lucrative x86 server/desktop/notebook markets.

Intel has stated that its long-term goal is to be profitable in mobile. You have to spend significant R&D ahead of revenue given the development timelines for SoCs/IP.

 

[Abwx]

Research and Development.

I guess that you understand nothing to financials..

RD is always included in a chip cost, even in 50$ HW pentiums, and suddenly a baytrail would require more RD/unity than a HW whole retail price.?.

Mediatek 20$ chips include all RD, as well as AMD s 30-40$ Beema but hey, Baytrail total cost is 90$...

You are living in denial, get used to it, Intel is paying OEMs to order and use their non competitive chips.

Contra-Revenue was non existant when AMD released Bobcat? What i do remember that, back then, it only keep AMD away from the cheaper options.

Ill more than happy to pay $200 instead of $150 for a AMD tablet if its really better than BT-T, but history tells me if its really better there whould be devices.

With 90$ total subside for each tablet using a Baytrail there is no way that the difference would be only 50$, more likely 150-180$ at retail level.

 

[witeken]

The mobile spending (that which wouldn't be spend if mobile didn't exist) is about a couple hundred million USDs per quarter.

 

[Arachnotronic]

I guess that you understand nothing to financials..

RD is always included in a chip cost, even in 50$ HW pentiums, and suddenly a baytrail would require more RD HW retail price.?.

R&D per unit is variable based on the number of units you sell (it is also typically not considered part of cost of goods sold, which is what determines gross profit margins). Also, the R&D spent today doesn't reflect R&D spent on products being sold today. There is a very significant lead time for chip development, and the vast majority of the R&D dollars Intel spent in mobile during 2014 were for products that won't be released for years.

 

[Nothingness]

I guess that you understand nothing to financials..

That was really funny :biggrin: You obviously don't know who Arachnotronic is even though you've been on these forums for years.

 

[Abwx]

R&D per unit is variable based on the number of units you sell (it is also typically not considered part of cost of goods sold, which is what determines gross profit margins). Also, the R&D spent today doesn't reflect R&D spent on products being sold today. There is a very significant lead time for chip development, and the vast majority of the R&D dollars Intel spent in mobile during 2014 were for products that won't be released for years.

And this is where you argument is right dismissing your sayings..

RD is quite low when amortized with +40 millions chips/year, and since they intent to throw another 44 millions in 2015...

Actualy what they did was to design and manufacture a chip whose break even point is in the 30$ range, and in this price everything is included, if they did sell it at this price they would had cashed 1.4bn with no net income, so they gave those chips wich induced a 1.4bn losses, the difference with the 4.2bn is the money given to OEMs, that is 2.8bn for 46 millions chips = 60$ subisde per ordered chip, plus the chip of course...

Notice that with this shenanigans they cleared 1.4bn revenues from eventual competitors, since 60-70% of the tablets and laptops using thoses subsided chips use Windows the impact on this market is roughly 1bn revenue where AMD was competent to take a significant share.

That was really funny :biggrin: You obviously don't know who Arachnotronic is even though you've been on these forums for years.

He s explaining us that Baytrail RD and manufacturing cost is 90$, that s with such a product that Intel was trying to compete with 15-20$ mediatek and other spreadtrum chips...

More seriously i think that advising to buy Intel shares and being neutral is a difficult exercise, other are less emotional, just look at Intel stock values, dont expect it to get back to the recent level unless it s propelled by buy backs, the equivalent of contra revenues but this time to prop up a deseperatly stagnating stock.

 

[AtenRa]

Operating Income = Gross Income - expenses like COGS, wages, R&D, Depreciation etc.

So, Intels 4.2B loss (Edit: for the MC Group) of Operating Income in 2014 is not R&D only.

 

[coercitiv]

R&D per unit is variable based on the number of units you sell (it is also typically not considered part of cost of goods sold, which is what determines gross profit margins).

In 2014 Intel Mobile and Communication group reported a Net Revenue of $202 Million, with roughly 46 million tablets shipped by year's end. It is my understanding that net revenue includes any contra revenue, but does not include research spending. (Edit: aside from anything added to the chip price itself)

If all of the above is true, what is the maximum possible value for net income per mobile chip sold by Intel in 2014?

 

[witeken]

Operating Income = Gross Income - expenses like COGS, wages, R&D, Depreciation etc.

So, Intels 4.2B loss of Operating Income in 2014 is not R&D only.

Correct, roughly 2B is R&D (and 1B CR).

 

[Enigmoid]

RD is quite low when amortized with +40 millions chips/year, and since they intent to throw another 44 millions in 2015...

Its quite significant for low cost chips.

If we assume it cost 250 mil for R&D and 44 million chips are sold that is $5.68 per chip, quite significant when talking about $20-30 chips. Even if 100 mil chips were sold that would still be $2.5, or nearly 10% of the cost.

 

[Abwx]

If all of the above is true, what is the maximum possible value for net income per mobile chip sold by Intel in 2014?

If we include the miserable 200 millions revenues this yield 4bn losses for 46 millions chips, that s -65$ net revenue per chip, not counting that they no more have the chip itself, to summarize they deliver the chip and get a 65$ bill from the OEM...

Its quite significant for low cost chips.

If we assume it cost 250 mil for R&D and 44 million chips are sold that is $5.68 per chip, quite significant when talking about $20-30 chips. Even if 100 mil chips were sold that would still be $2.5, or nearly 10% of the cost.

According to AMD a design from clean sheet cost about 300 millions $, so your number is half right given that they recycled parts like GPU from other designs, that said 5$/chip is reasonable but you ll admit that there are quite bigger amounts at stakes with thoses contra revenues, and that these are no more comparable with even a full RD budget for such a chip.

 

[coercitiv]

If we include the miserable 200 millions revenues this yield 4bn losses for 46 millions chips, that s -65$ net revenue per chip, not counting that they no more have the chip itself, to summarize they deliver the chip and get a 65$ bill from the OEM...

Let's leave research out of this, let's consider Intel chose to write that off as one giant bill to enter mobile market, and take a look at income mobile chips brought in 2014.

If I were to state Intel had a net revenue per mobile chip lower than $4.5, would that be correct?

 

[Arachnotronic]

And this is where you argument is right dismissing your sayings..

RD is quite low when amortized with +40 millions chips/year, and since they intent to throw another 44 millions in 2015...

Actualy what they did was to design and manufacture a chip whose break even point is in the 30$ range, and in this price everything is included, if they did sell it at this price they would had cashed 1.4bn with no net income, so they gave those chips wich induced a 1.4bn losses, the difference with the 4.2bn is the money given to OEMs, that is 2.8bn for 46 millions chips = 60$ subisde per ordered chip, plus the chip of course...

Notice that with this shenanigans they cleared 1.4bn revenues from eventual competitors, since 60-70% of the tablets and laptops using thoses subsided chips use Windows the impact on this market is roughly 1bn revenue where AMD was competent to take a significant share.




He s explaining us that Baytrail RD and manufacturing cost is 90$, that s with such a product that Intel was trying to compete with 15-20$ mediatek and other spreadtrum chips...

Abwx

Let me try to explain this a little bit better.

It is not generally useful to take the R&D expenditures of a business that is not yet at scale and divide it by the units that are sold to arrive at a cost structure for that particular product line. First of all, a lot of those R&D dollars (in fact, I'd wager the vast majority) are being spent on projects that have nothing to do with Bay Trail.

For example, those R&D expenditures include the entire pipeline of low-power mobile IP that Intel is developing (image signal processors, video encode/decode, low power GPU, low power CPU, modems, RF transceivers, software, etc.). That pipeline, if Intel is investing properly, is many product generations and years deep.

Now, you are correct that Intel is losing a lot of money in this division because they are not yet generating material revenue from the products that are out in the market today. Further, as has been noted, Intel is providing pretty hefty contra-revenue subsidies to make Bay Trail competitive from a platform bill of materials perspective.

If Intel is being truthful, then as Intel's new products roll out, the contra-revenue goes away. In fact, Intel is guiding to an $800 million reduction in the ~$4.2 billion mobile loss seen in 2014 in 2015, so obviously some progress is being made on the cost structure front. Also keep in mind that even as SoFIA/Cherry Trail roll out, there will be designs out in the market being sold with Bay Trail in them, which is why the contra-revenue -- while significantly reduce in 2015 -- will still persist.

Anyway, as I've mentioned before, the only real "impact" Bay Trail has had on AMD is that it took significant share in the low-end of the PC market against Kabini/Beema. Given that Intel now reports that ~80% of its Pentium/Celeron mix, and 20%+ of its notebook chips sold are Bay Trail-M based, and given how high Intel's profitability in the PC segment is, it is very unlikely that Intel is "giving away" these parts. These parts are winning designs on their own merits (by having the right combination of performance/power/cost structure), and given that this is the case, I don't see why AMD would have done any better in tablets even ex-Intel contra revenue.

 

[Khato]

It is not generally useful to take the R&D expenditures of a business that is not yet at scale and divide it by the units that are sold to arrive at a cost structure for that particular product line. First of all, a lot of those R&D dollars (in fact, I'd wager the vast majority) are being spent on projects that have nothing to do with Bay Trail.

For example, those R&D expenditures include the entire pipeline of low-power mobile IP that Intel is developing (image signal processors, video encode/decode, low power GPU, low power CPU, modems, RF transceivers, software, etc.). That pipeline, if Intel is investing properly, is many product generations and years deep.

Well said.

The other point that most aren't aware of is that the R&D cost for Baytrail, Cherrytrail, etc falls under MCG... but the revenue of the Celeron and Pentium branded versions is attributed entirely to PCCG - MCG only gets the Atom revenue. This is part of the reason for the merge as it makes MCG look at least slightly worse than it really is.

 

[AtenRa]

Anyway, as I've mentioned before, the only real "impact" Bay Trail has had on AMD is that it took significant share in the low-end of the PC market against Kabini/Beema. Given that Intel now reports that ~80% of its Pentium/Celeron mix, and 20%+ of its notebook chips sold are Bay Trail-M based, and given how high Intel's profitability in the PC segment is, it is very unlikely that Intel is "giving away" these parts. These parts are winning designs on their own merits (by having the right combination of performance/power/cost structure), and given that this is the case, I don't see why AMD would have done any better in tablets even ex-Intel contra revenue.

When an OEM/ODM purchase BayTrail-T chips, it will also raise its overall purchase targets and he will get a bigger discount/rebates for both Tablet, Mobile and Desktop chips from Intel.
So buying BayTrail-T chips it also affect mobile/desktop baytrail chip prices and that affect AMDs ability to price compete against BayTrail.

AMD had Tablet wins with BobCat chips (C50-Z01 and Z60) in 2011-2012. There were Windows AMD tablets with those APUs in 2011-2012 by ACER, MSI, Gigabyte and more. No matter how good in performance BayTrail-T was in late 2013, that was not the reason OEMs/ODMs stoped producing AMD Temash Tablets. And performance was certainly not the issue that kept Mullins from tablets either since it was faster than BayTrail in 2014.

It is simple guys, Contra-Revenue and Aggressive rebates killed any change AMD had in Tablets.

 

[Arachnotronic]

When an OEM/ODM purchase BayTrail-T chips, it will also raise its overall purchase targets and he will get a bigger discount/rebates for both Tablet, Mobile and Desktop chips from Intel.
So buying BayTrail-T chips it also affect mobile/desktop baytrail chip prices and that affect AMDs ability to price compete against BayTrail.

AMD had Tablet wins with BobCat chips (C50-Z01 and Z60) in 2011-2012. There were Windows AMD tablets with those APUs in 2011-2012 by ACER, MSI, Gigabyte and more. No matter how good in performance BayTrail-T was in late 2013, that was not the reason OEMs/ODMs stoped producing AMD Temash Tablets. And performance was certainly not the issue that kept Mullins from tablets either since it was faster than BayTrail in 2014.

It is simple guys, Contra-Revenue and Aggressive rebates killed any change AMD had in Tablets.

Mullins didn't even have an integrated image signal processor, FYI -- the Intel mobile SoCs, for all their faults, did. This is unacceptable for a truly "mobile" solution where the camera and its functionality are a big part of what the value proposition of the device.

 

[coercitiv]

Mullins didn't even have an integrated image signal processor, FYI -- the Intel mobile SoCs, for all their faults, did. This is unacceptable for a truly "mobile" solution where the camera and its functionality are a big part of what the value proposition of the device.

I can already imagine the BoM on that Mullins...

 

[Shivansps]

When an OEM/ODM purchase BayTrail-T chips, it will also raise its overall purchase targets and he will get a bigger discount/rebates for both Tablet, Mobile and Desktop chips from Intel.
So buying BayTrail-T chips it also affect mobile/desktop baytrail chip prices and that affect AMDs ability to price compete against BayTrail.

AMD had Tablet wins with BobCat chips (C50-Z01 and Z60) in 2011-2012. There were Windows AMD tablets with those APUs in 2011-2012 by ACER, MSI, Gigabyte and more. No matter how good in performance BayTrail-T was in late 2013, that was not the reason OEMs/ODMs stoped producing AMD Temash Tablets. And performance was certainly not the issue that kept Mullins from tablets either since it was faster than BayTrail in 2014.

It is simple guys, Contra-Revenue and Aggressive rebates killed any change AMD had in Tablets.

Its not faster, the HP Stream 14z with a quad mullin already goes down to 800 mhz and it sells for just $220, a 800mhz quad mullin its not faster than a 1.33ghz BT-T.

If HP can sell a quad mullin in a 14" $220 laptop tecnically there is nothing stopping them from selling a $300 AMD Mullin tablet, the Asus T100 is just over $300.

So as i was thinking ealier, the problem is the AMD SoC not able to keep up with BT on tablet form factor, there is no conspiracy here.

And as i said earlier, yes, Mullin have superior igp perf, thats no suprise, but is also no enoght.

BTW, the C and Z Bobcats where slow as hell, yes some oems decided to build windows tablets with them, but there was also no better option, There where also a very limited number of Atom tablets too.

 

[Roland00Address]

Guys lets keep two topics out of this thread. This thread is about

1) Intel baytrail/cherry trail performance

2) Intel baytrail and cherry trail products that are about to hit the market or reviews of products that did hit the market.


Things that have nothing to do with this thread.

A) Intel Contra Revenue

B) AMD in any shape or form, including the merits of Mullins, or how much Intel is screwing AMD by going after ARM and AMD is caught in the crosshairs.

 

[SPBHM]

Nice review of a Bay Trail PC on a stick: http://www.eurogamer.net/articles/digitalfoundry-2015-hannspree-micro-pc-review Performance is a bit weak from the sound of it (it's the cut down F-series chip), but it sounds perfect for Steam streaming.

http://images.eurogamer.net/2013/ar....jpg/EG11/resize/600x-1/quality/90/format/jpg

it's pretty cool but, I think it's probably going to have thermal related problems and I would trade some size for at least 4 usb + ethernet (like the raspberry pi), and the price is not that low as they mentioned, comparing to tablets.

 

[Shivansps]

This is very, very insteresting

Pipo W3F is a 2-in-1 Windows Bay Trail tablet for $190

Chinese tablet maker Pipo has introduced a new Windows tablet with a full HD display, an Intel Atom Bay Trail processor, and a keyboard cover that lets you use the system like a notebook.

Low-cost Windows tablets are a dime a dozen these days, but the Pipo Work W3F really doesn’t cost all that many dimes: It’s available from AliExpress for about $190 and up.

The tablet features an Intel Atom Z3735F quad-core processor, 2GB of RAM, a 1920 x 1200 pixel IPS display, and a 7800mAh battery.

It has two full-sized USB 3.0 ports, and a micro USB port as well as a micro SD card slot. There’s a 5MP camera on the back and a 2MP front-facing camera.

There’s a WiFi-only model with 32GB of storage and a 3G version that has 64GB of storage. Both ship with Windows 8.1 and Microsoft Office 365 software.