Intel Z3770 vs Snapdragon 800 in Kraken 1.1

[Nothingness]

Indeed, the majority of the idiotic investor crowd who doesn't care about the long-term implications would go crazy.

Most investors are certainly completely stupid and only live in their financial bubble without understanding the technical side of things.

But I know I'd sure be buying Intel stock in that situation. Intel will win any price war that it enters - their only concern is navigating it correctly to not come out of it as a monopoly for all the regulators to hound. Keep in mind that it's not only the ARM IP companies (Qualcomm, mediatek, NVIDIA, so on so such) that would feel the effect of a price war, but also their foundries. Really I'd be more concerned about the effect it can potentially have on the foundries - even if they don't reduce their pricing per wafer to help out they'll still have to deal with selling however many less wafers that Intel has gained. And with how much the foundries have been expanding lately in order to meet current/future demand it'd be rather bad for them if that demand suddenly evaporates.

Basically, it's not so much about making money for Intel as it would be depriving the competition of sales.

I surely hope this won't happen as that'd mean we would basically have a single provider from phones up to servers. This won't benefit anyone except investors. Do you really want that?

 

[piasabird]

When the I-phone came out maybe they said the same kinds of things about Apple I-phones. Who is laughing now? Have an open mind at least.

 

[BallaTheFeared]

When the I-phone came out maybe they said the same kinds of things about Apple I-phones. Who is laughing now? Have an open mind at least.

Android. :whiste:

 

[piasabird]

http://www.qualcomm.com/snapdragon/smartphones
Seems like lots of products use snapdragon processors of one kind or another.

 

[Khato]

I surely hope this won't happen as that'd mean we would basically have a single provider from phones up to servers. This won't benefit anyone except investors. Do you really want that?

Yes, but I have ulterior motives

Qualcomm has more money (zero debt, double the bank balance currently, and rising) and less dependence on chip manufacturing than Intel does.

I covered it in another thread a few weeks ago, but basically put Intel is now spending most of their profit on capex, ie building new fabs. The cost of fab building and process R&D rises while their profits drop.

That's why Qualcomm is making so much money compared to Intel. If Intel took 50% of Qualcomm's chip business, Qualcomm would barely flinch. If the opposite happened Intel would basically be finished.

Guess that's one way to look at it. Here's my perspective. Unlike Qualcomm who's mostly just sitting on all that cash Intel is investing their profits. Yes the cost of fab building and process R&D continues to increase with every new node, but profits are still keeping up with that pace thanks to the large gains in total revenue last year. (Yes revenues are down in comparison... but that's after how many straight quarters of increase, and going from $35B in 2009 to $43B in 2010 to $54B in 2011.) Intel still has quite a bit of buffer before it has to start slowing down in that regard.

As for what would happen to Qualcomm if Intel took 50% of their chip business... how would they not flinch? Unless you're restricting that statement to only Qualcomm's Snapdragon line? Qualcomm's chip business (reported in their earnings as QCT) is responsible for about three quarters of their revenue. And through their Infineon acquisition Intel has competitive offerings for the majority of Qualcomm's chip business.

Lastly, how do the foundries fit into your picture? Because it's not just Qualcomm vs Intel, it's Qualcomm + their foundries vs Intel. Qualcomm may well be sitting pretty and easily be able to mitigate the damage Intel can cause them directly (that's actually something I agree with) but the same can't be said for their foundry partner(s). How badly would TSMC be hurt if 50% of Qualcomm's orders suddenly disappeared and they had leading edge fabs sitting idle? It's the foundries that are Qualcomm's Achilles heel as they have to make the same kind of massive investments into infrastructure and research as Intel but without the huge cash flow/margins cushion.

 

[liahos1]

Qualcomm has more money (zero debt, double the bank balance currently, and rising) and less dependence on chip manufacturing than Intel does.

I covered it in another thread a few weeks ago, but basically put Intel is now spending most of their profit on capex, ie building new fabs. The cost of fab building and process R&D rises while their profits drop.

That's why Qualcomm is making so much money compared to Intel. If Intel took 50% of Qualcomm's chip business, Qualcomm would barely flinch. If the opposite happened Intel would basically be finished.

Yes, but I have ulterior motives



Guess that's one way to look at it. Here's my perspective. Unlike Qualcomm who's mostly just sitting on all that cash Intel is investing their profits. Yes the cost of fab building and process R&D continues to increase with every new node, but profits are still keeping up with that pace thanks to the large gains in total revenue last year. (Yes revenues are down in comparison... but that's after how many straight quarters of increase, and going from $35B in 2009 to $43B in 2010 to $54B in 2011.) Intel still has quite a bit of buffer before it has to start slowing down in that regard.

As for what would happen to Qualcomm if Intel took 50% of their chip business... how would they not flinch? Unless you're restricting that statement to only Qualcomm's Snapdragon line? Qualcomm's chip business (reported in their earnings as QCT) is responsible for about three quarters of their revenue. And through their Infineon acquisition Intel has competitive offerings for the majority of Qualcomm's chip business.

Lastly, how do the foundries fit into your picture? Because it's not just Qualcomm vs Intel, it's Qualcomm + their foundries vs Intel. Qualcomm may well be sitting pretty and easily be able to mitigate the damage Intel can cause them directly (that's actually something I agree with) but the same can't be said for their foundry partner(s). How badly would TSMC be hurt if 50% of Qualcomm's orders suddenly disappeared and they had leading edge fabs sitting idle? It's the foundries that are Qualcomm's Achilles heel as they have to make the same kind of massive investments into infrastructure and research as Intel but without the huge cash flow/margins cushion.

this is a crucial point many ppl miss with regard to the coming mobile wars. If QCOM/NVDA et all suffer and lose business to an increasingly aggressive intel, TSMC is not going to have the scale to continue capexing /R&Ding at the leading edge. they in turn will become the next UMC. A casualty of moore's law. In that event, qcom loses its QCT business but still has a great licensing business. But that business model slices both ways (just like intels). If you miss out on the next wave of innovation, or if without a QCT business you become a pure NPE, then the ability to license your ip becomes an increasingly litigious affair. I'm sure they dont look at the loss of 50% of their qct business and say no problem!

 

[SiliconWars]

Guess that's one way to look at it. Here's my perspective. Unlike Qualcomm who's mostly just sitting on all that cash Intel is investing their profits. Yes the cost of fab building and process R&D continues to increase with every new node, but profits are still keeping up with that pace thanks to the large gains in total revenue last year. (Yes revenues are down in comparison... but that's after how many straight quarters of increase, and going from $35B in 2009 to $43B in 2010 to $54B in 2011.) Intel still has quite a bit of buffer before it has to start slowing down in that regard.

Intel was brilliant between 2009 and 2011 for sure, but the problem is they can't afford slip-ups. Last year was a slip, this year is heading for another slip. The only constant is that capex must go up to maintain their fab advantage, and this year already they've cut it twice, while TSMC increased theirs twice.

As for what would happen to Qualcomm if Intel took 50% of their chip business... how would they not flinch? Unless you're restricting that statement to only Qualcomm's Snapdragon line? Qualcomm's chip business (reported in their earnings as QCT) is responsible for about three quarters of their revenue. And through their Infineon acquisition Intel has competitive offerings for the majority of Qualcomm's chip business.

Granted if Intel became competitive vs Qualcomm's non-MCM segments things would get interesting. They aren't making any inroads there though, when they bought Infineon it was always going to take time for this to work, and they are basically nowhere in terms of Qualcomm's QCT. The realistic target is the actual MCM segment, which isn't worth a whole lot in terms of profit, and takes a huge amount of chips to get there.

Lastly, how do the foundries fit into your picture? Because it's not just Qualcomm vs Intel, it's Qualcomm + their foundries vs Intel. Qualcomm may well be sitting pretty and easily be able to mitigate the damage Intel can cause them directly (that's actually something I agree with) but the same can't be said for their foundry partner(s). How badly would TSMC be hurt if 50% of Qualcomm's orders suddenly disappeared and they had leading edge fabs sitting idle? It's the foundries that are Qualcomm's Achilles heel as they have to make the same kind of massive investments into infrastructure and research as Intel but without the huge cash flow/margins cushion.

Well, why would Qualcomm allow that to happen? That's the whole reason why I believe Intel can't afford a price war. Intel is already on the ragged edge in terms of being able to afford fab capex.

Intel cut 1 - http://newsroom.intel.com/community...-reports-first-quarter-revenue-of-126-billion

Intel cut 2 - http://www.reuters.com/article/2013/07/17/intel-results-idUSL1N0FN20G20130717

TSMC increase 1 - http://www.digitimes.com/news/a20130814PD202.html

So in the last 6 months Intel has dropped capex by $2 billion while TSMC has raised it by $2 billion. In 6 months they have basically become equal in fab spending, both at ~ $11 billion.

If Qualcomm couldn't afford it, Apple could, or the 100's of other customers TSMC has. The fab R&D costs are spread evenly over a huge number of customers for TSMC, while Intel has to foot their own bill by themselves mostly. The small volume stuff they get from Altera is a drop in the ocean.

Think about the next $50 billion needed in fab capex. Apple could pay it all by themselves, Qualcomm probably could as well - this is why it's important to have money in the bank and to be making money as well. They need TSMC for this to work, so they can't and won't let TSMC fail under any circumstances. Intels cash is stagnating and they are spending most of what they make trying to maintain their fab advantage, but they are fighting a losing battle against costs.

 

[Nothingness]

Yes, but I have ulterior motives

You are an Intel investor! Booh :biggrin: Or worse, you are working for Intel :ninja:

As for what would happen to Qualcomm if Intel took 50% of their chip business... how would they not flinch? Unless you're restricting that statement to only Qualcomm's Snapdragon line? Qualcomm's chip business (reported in their earnings as QCT) is responsible for about three quarters of their revenue. And through their Infineon acquisition Intel has competitive offerings for the majority of Qualcomm's chip business.

Yes but 3/4 of the income is from licensing (QTL), which Intel has to pay for I guess

 

[Khato]

Intel was brilliant between 2009 and 2011 for sure, but the problem is they can't afford slip-ups. Last year was a slip, this year is heading for another slip. The only constant is that capex must go up to maintain their fab advantage, and this year already they've cut it twice, while TSMC increased theirs twice.

Pretty much all the more that needs to be said on this is that you're confusing process advantage with capacity. Intel's cuts have zero impact upon their advantage in process technology, they are purely reductions in how much capacity they are bringing up on 14nm initially. Meanwhile TSMC's increase in expenditure is to handle expected capacity demands a year or more out.

If Qualcomm couldn't afford it, Apple could, or the 100's of other customers TSMC has. The fab R&D costs are spread evenly over a huge number of customers for TSMC, while Intel has to foot their own bill by themselves mostly. The small volume stuff they get from Altera is a drop in the ocean.

Think about the next $50 billion needed in fab capex. Apple could pay it all by themselves, Qualcomm probably could as well - this is why it's important to have money in the bank and to be making money as well. They need TSMC for this to work, so they can't and won't let TSMC fail under any circumstances. Intels cash is stagnating and they are spending most of what they make trying to maintain their fab advantage, but they are fighting a losing battle against costs.

Oh, there's no question that they have the ability to help out TSMC, but will they? Because what you're basically advocating there is for Qualcomm/Apple/all of TSMC's customers to pay more in order to keep Moore's law alive at TSMC. But that is only financially feasible if there's adequate volume to support it.

I guess basically we're arguing two sides of the same coin. Your impression being that Intel's traditional marketshare is stagnating at a rate that won't allow it to continue the level of investment necessary. Whereas mine is that Intel can keep up their current model without making inroads in the mobile segment for at least another four years... and if they use their manufacturing might to their advantage, foregoing traditional margins in order to capture a sizable portion of the mobile market it will instead be TSMC/the other foundries that are suddenly in trouble. Sure their customers could keep them on life support, but at the cost of draining their own cash reserves with a bleak outlook on actually making a difference.

 

[Abwx]

Qualcomm probably could as well - this is why it's important to have money in the bank and to be making money as well. They need TSMC for this to work,

I thought it was common knowledge :

Qualcomm, a world-known provider of chips for mobile gadgets, has entered into a contract with GlobalFoundries company with the intention to develop chips based on 28 nm technology.

 

[Khato]

Yes but 3/4 of the income is from licensing (QTL), which Intel has to pay for I guess

Quite true - and I guess the numbers are actually that 2/3 of their revenue comes from QCT, but it's only responsible for 1/3 of their EBT. It's unfortunate that there's no good breakdown about what their licensing revenue is actually from, aka no indication as to how stable a source it's going to be in the coming years.

 

[SiliconWars]

Pretty much all the more that needs to be said on this is that you're confusing process advantage with capacity. Intel's cuts have zero impact upon their advantage in process technology, they are purely reductions in how much capacity they are bringing up on 14nm initially. Meanwhile TSMC's increase in expenditure is to handle expected capacity demands a year or more out.

Intel is cutting capacity because of lower demand. TSMC is doing the opposite.

Oh, there's no question that they have the ability to help out TSMC, but will they? Because what you're basically advocating there is for Qualcomm/Apple/all of TSMC's customers to pay more in order to keep Moore's law alive at TSMC. But that is only financially feasible if there's adequate volume to support it.

Why wouldn't there be adequate volume?

I guess basically we're arguing two sides of the same coin. Your impression being that Intel's traditional marketshare is stagnating at a rate that won't allow it to continue the level of investment necessary. Whereas mine is that Intel can keep up their current model without making inroads in the mobile segment for at least another four years... and if they use their manufacturing might to their advantage, foregoing traditional margins in order to capture a sizable portion of the mobile market it will instead be TSMC/the other foundries that are suddenly in trouble. Sure their customers could keep them on life support, but at the cost of draining their own cash reserves with a bleak outlook on actually making a difference.

Companies like Qualcomm and Apple already have such a huge bank balance lead on Intel, it's basically a forgone conclusion. One day a fab will cost $50 billion to build and tool (possibly at 7nm?). Apple could already build two of them right now, and by then probably six or seven. Intel will never have $50 billion in the bank, and will never have anything near the profits needed to get there in the same timeframe.

You need to make money and save money as well. Intel is spending all their money on trying to maintain a process advantage that isn't bringing them the knockout victory they need.

Qualcomm spends basically nothing on fabs, and they just keep on banking more money. If they ever had to build their own they'll have enough money to do it many times over, at the cutting edge. It's only a matter of time.

 

[Khato]

Why wouldn't there be adequate volume?

As per the hypothetical where Intel aggressively pursues mobile market share at the cost of high margins every chip that Intel sells is one that isn't sold by the competition. And as you say, TSMC is increasing their available capacity - that's being done in anticipation of future demand. If demand comes up short/volume isn't as expected the cost per wafer on a node has to increase accordingly. Basically, I don't see how the independent foundries can't be considered to be in a more precarious position than Intel.

 

[SiliconWars]

As per the hypothetical where Intel aggressively pursues mobile market share at the cost of high margins every chip that Intel sells is one that isn't sold by the competition. And as you say, TSMC is increasing their available capacity - that's being done in anticipation of future demand. If demand comes up short/volume isn't as expected the cost per wafer on a node has to increase accordingly. Basically, I don't see how the independent foundries can't be considered to be in a more precarious position than Intel.

I thought you'd say that, but surely you can see that it's a far more precarious situation for Intel trying to defend their classic segments share from the ARM guys? That's what I meant about the 50% loss. Even a 20% server loss for Intel would be enough to wipe out half of their yearly profit, which would mean capex needed to be halved again. This could happen in two to three years with the A53.

Any damage Intel can do to the ARM crowd can largely be absorbed by weight of numbers, but Intel cannot afford to lose anything they already have.

The ARM guys have too much in hand, and too much can go wrong for Intel. It's a time game, and Intel is running out of time. It's like they need to beat everyone or lose, all or nothing.

 

[ViRGE]

I'm only going to say this once: get back on the topic of the Intel Z3770 vs. the Snapdragon 800 on the Kraken benchmark. This is NOT a discussion of business practices, and any further posts on that subject in this thread will be infracted and removed.

-ViRGE

 

[mikk]

http://wccftech.com/intel-atom-z3000-series-bay-trail-socs-performance-lineup-22nm-silvermont-cores/

 

[Gideon]

http://wccftech.com/intel-atom-z3000-series-bay-trail-socs-performance-lineup-22nm-silvermont-cores/

Whoa, this confirms 4 EUs @ UP TO 677 Mhz (prolly turbo). While Dx11 feature-set is cool and all, the GPU will surely be the chips weakest point.

 

[Homeles]

Does it really matter though? Anand was tossing around the idea of iPad4-level graphics, which is enough to drive a 2048x1536 resolution display. I may be missing something, but graphics don't seem like a relatively critical component.

 

[NTMBK]

Whoa, this confirms 4 EUs @ UP TO 677 Mhz (prolly turbo). While Dx11 feature-set is cool and all, the GPU will surely be the chips weakest point.

Yup. I find it pretty funny how their slides happily compare CPU performance to the latest ARM parts, but only compare GPU performance to last-gen stuff:

 

[Sweepr]

Does it really matter though? Anand was tossing around the idea of iPad4-level graphics, which is enough to drive a 2048x1536 resolution display. I may be missing something, but graphics don't seem like a relatively critical component.

Snapdragon 600-like graphics performance or even better would be quite good, far from a complete disaster even if Snapdragon 800's Adreno 330 is faster. I'd personally give up graphics performance (not interest in hardcore Android gaming and even Adreno 330-like performance would be too slow for Windows games ported from X360/PS3) for better CPU performance (productivity) and battery life.
14nm process (more EUs) and Gen 8 / Gen 9 graphics will probably improve gfx performance a lot, but that will be in 2014.

 

[seitur]

That GPU is kinda disappointing, but I will wait with opinion until it releases.

btw. Does Intel does any architectural performance advancements with it's GPUs with each generation? Of course aside of making them able to to process higher DX and OGL.
What I mean does 4 Haswell EUs would be stronger than 4 Ivy Bridge EUs. etc

Lack of more in-depth info about Intel GPU tech is irritating.

 

[Arachnotronic]

Blowout CPU performance, solid GPU, probably really good power consumption - yep, looks like a winner.

SiliconWars, weren't you saying that a 60% performance advantage in CPU wasn't likely? Well, there are your SPECInt results...of course, unless that benchmark - like Kraken - isn't representative.

All we need is a Geekbench 3 result, and then you can go about "cleaning up"...

 

[SiliconWars]

Blowout CPU performance, solid GPU, probably really good power consumption - yep, looks like a winner.

SiliconWars, weren't you saying that a 60% performance advantage in CPU wasn't likely? Well, there are your SPECInt results...of course, unless that benchmark - like Kraken - isn't representative.

It's representative of SPECint, yes. Notice they didn't use SPECfp?

Notice that after using S800 and Tegra 4 for SPECint, they then used the S4 and Tegra 3 for the other benchmarks (including Sysmark? lol)

All we need is a Geekbench 3 result, and then you can go about "cleaning up"...

I guess that's because you don't like the Geekbench 2 results we already have?

 

[mikk]

btw. Does Intel does any architectural performance advancements with it's GPUs with each generation? Of course aside of making them able to to process higher DX and OGL.
What I mean does 4 Haswell EUs would be stronger than 4 Ivy Bridge EUs. etc

Yes they do. Even Gen7.5 had some enhancements over Gen7. I expect much bigger architectural improvements in Gen8.

I guess that's because you don't like the Geekbench 2 results we already have?

That is because FP results are bugged in Geekbench 2. No surprise that you didn't know though.

 

[liahos1]

isnt this going to be pretty straightforward within the next two years or am i missing something.

Based on AMAT/LAM comments it doesnt seem like the industry is going to be adding a ton of 20nm and 20nm/finfet capacity next 12 months.

Intel's has the biggest transistor budget and will have widening budget to by 2h of next year even accounting for the minimal volumes we're going to get at 20nm from TSMC

If i look at the kraken, and IDF comparison slides, they already have superior cpu performance and at S600 or better GPU performance.

If ur getting your cpu muarch iterated next year along with potentially two gpu muarch iterations i dont see how this isnt intel's game to win from a performance, efficiency and cost point of view??

I am not a tech person, i'm an investor but the sheer amount of negativity blasted towards intel is baffling?