[GTO]

Interesting how Intel used to indirectly bully Nvidia before, and how Nvidia has grown:

Now that Dell is sending Intel more money per PC shipped, the PC maker now has three options: 1) it can raise prices, thereby making its offerings less competitive in a cut-throat PC marketplace where anyone with a screwdriver and a credit card can start a PC assembly shop, 2) it can eat the cost increase and see its own margins suffer (and its stock price get hammered), or 3) it can start squeezing the rest of the component makers that provide its PC parts (i.e. the GPU, the soundcard, the motherboard, etc.) to lower their prices and their margins, in order to make up the difference.

Not surprisingly, Dell and the rest of the PC vendors got into the habit of choosing option 3. Because there are multiple GPU vendors in the market, Dell could go to Nvidia and ATI and play them off against each other, forcing them to offer lower prices in order to secure a spot in a Dell PC. And likewise with other component makers.

This, then, was the basic mechanism by which Intel was able to “steal” margin from everyone else inside the PC box, especially the GPU makers like Nvidia and ATI.

Source

[itisravi]

I wonder how much of having its own fabrication plants has slowed down Intel (unlike Nvidia which is fabless and outsources manufacturing) from spending on R&D, in the sense that setting up a fab lab is a huge financial investment and it must keep churning out the current architecture models to recover the cost and make profits.

[asbshyam]

Originally Posted by V.Narayan

"Marketing Myopia" by Theodore Levitt. This is a blind siding caused by the frailness of human thinking influenced by super success with one formula. ..

Can't agree more, the arrogance and over confidence with which they pushed Intel Atom onto the Android Platform. Intel thought they would dictate the market like they did on other x86 platforms.

This was also at a time (around 2012) when all the mobile players were just marking their spots in the mobile market - Samsung was still making chips for Apple and it's own Exynos lineup, Snapdragon was just breaking into the market. Intel with genuine effort could have reigned supreme.

The same thing happened with the Ryzen uprising, Intel was selling 6 core CPUs as HEDT and anything beyond was god level. And then came AMD, 6 core is now the new quad core and has a lot of entry level options.

It is now time for Intel to roll up it's sleeves and follow the market.

[samirotiv]

Originally Posted by GTO

I can't for the life of me figure out how or why Intel missed out on the smartphone revolution (just like their longtime partner Microsoft).

Can any BHPian explain why in layman terms?

It's the fundamental difference between CISC and RISC architectures.

Essentially, computers are like onions - every layer builds on top of an inner layer. The innermost layer from a computer organization standpoint is the CPU. They provide an instruction set, which allows the next outer layer to build on top of it.

There are two schools of thought here - one is to make the CPU complex, and to provide a larger instruction set, with several complex instructions. This simplifies the outer layers of the onion. This is called CISC - Complex Instruction Set Computing.

The other is to make the CPU as simple as possible, and having fewer and simpler instructions - which moves the complexity to the outer layers of the onion. This is called RISC - Reduced Instruction Set Computing.

With this in mind, the more complex a chip is, the area (read: transistors) it requires. It is difficult to make a tiny CISC CPU.

Intel's only successful architecture is x86 - which was extended by AMD to x86-64. It is a CISC architecture. On the other hand, ARM is a RISC implementation. It's one of the reasons why Intel has missed the bus, and ARM has made it to most mobile devices.

[zensure]

Originally Posted by Sutripta

The first commercially available RISC processor IIRC was the Intel i960. A commercial failure. Apparently still used by ISRO!

Sutripta

BTW, i960 had a predecessor - i860, which was also a RISC. I had used it extensively for parallel computing in early '90s. It had bugs and while too fast for integer operations, not so fast for floating point operations.

Re the power consumption part - it is bad if the chip consumes more power for the same amount of compute power. Higher power consumption of intel chips is one of the main reason Apple is moving away from Intel. More power consumption means need a bigger battery and complex engineering to cool the logic board.

Suresh

[Sutripta]

Originally Posted by zensure

BTW, i960 had a predecessor - i860, which was also a RISC.

Strange. I thought the i960 preceded the i860. But going by sequentially increasing numbers guess you are right.

Sutripta

[zensure]

Originally Posted by Sutripta

Strange. I thought the i960 preceded the i860. But going by sequentially increasing numbers guess you are right.

Sutripta

Apologies Sutripta. I checked again and you were right. i960 came first and supposedly improved i860 came later. I too was taken in by the number sequencing. I stand corrected now.

Suresh

[Sutripta]

Originally Posted by zensure

BTW, i960 had a predecessor - i860, which was also a RISC. I had used it extensively for parallel computing in early '90s.

Not the thread for this, but would be interested in knowing more. esp the software part which was supposed to leverage parallel architectures to the full.

The 90s - heady days! Today's guys don't know what they have missed!

Any exposure to things like Transputers?

Sutripta

[ZenMaster]

Originally Posted by GTO

I can't for the life of me figure out how or why Intel missed out on the smartphone revolution (just like their longtime partner Microsoft).

Can any BHPian explain why in layman terms?

Its basically just like Nokia, they rested on their laurels, milking their Cash cow and failed to innovate.

Intel got a double blow by ARM jumping from smartphones to computers and they also got blindsided by AMD which came with all guns blazing and then some with their RYZEN processors which basically knocked intel off their feet.

ARM revolutionised 21st century technology and Intel just didn't capitalise it. By the time they put their foot in the race, competitors where miles ahead and were generations ahead in development. They tried shoehorning their watered down versions of their powerful but inefficient chips into phones and tablets but to no avail. They just couldn't run well without a Fan basically.

They did find success later in Ultrabooks and Surface like products but that is just a transition phase and they couldn't bring themselves to accept it and all the while Apple, Qualcomm and Samsung pushed themselves into the next gen by using and developing their processors and devices on ARM chip design for powering their future devices.


They also didn't think that AMD could do what it did. Put their head down and develop a brand new architecture ( i.e. ZEN architecture which is like the "Core" architecture is to intel.) which took advantage of their Integrated GPU for high speed processing as I understand. Experts might know whether its right or not. We can hear computer scientists raving about GPU based processing which is must faster and is used in Big data etc.(This is why Nvidia and AMD is surging ahead)
They then went to undercut Intel heavily on pricing.

Everyone who used the Ryzen processors became immediate fans. The 3rd and 4th gneration RYZEN processors destroyed intel on 90% of the performances benchmarks and blasted in out of the water in terms of VFM pricing.
Esecially in Laptops even though the effect in less in India because of taxes and horrible pricing by sellers.

The newly announced Playstation and Xbox both use AMD processors.

Many large data Centres now run on AMD processor because they basically provide the same or better performance of much much cheaper.

Intel never thought their clients would even consider AMD especially in laptops & gaming systems and now they are.

They were caught sleeping, thinking that nobody could steal their share of the Pie.

[shyaGO]

Its not entirely trues that Intel did not enjoy success in mobile market. When we talk of cell phones, there is apps processor and there is baseband processing (3G/4G etc). Fruit company still use Intel solution for baseband for last few generation until Intel decided to exit 5G business. This was aftermath Qcom and Apple settling their lawsuits. Eventually Intel 4G/5G division was bought by the fruit company.
With Mobile Eye acquisition by Intel in 2017 - showdown between Intel and nVdia is now playing closer to Team-BHP turf

[zensure]

Originally Posted by Sutripta

Not the thread for this, but would be interested in knowing more. esp the software part which was supposed to leverage parallel architectures to the full.

The 90s - heady days! Today's guys don't know what they have missed!

Any exposure to things like Transputers?

Sutripta

No, I don't have exposure to Transputers. I think it was dying when I started computing and focus had shifted towards RISC by then.

The 90s - indeed heady days. Not sure how many had heard of Cray.. All pervasive Cray started dying in late 80s to early 90s. IBM POWER chip started gaining prominence. All these type of high end machines were denied to India by US. There were efforts to build some powerful computers from first principles and I was involved in one such effort involving i860 and MIMD parallel computing. Served our purpose for a while. Programming in C using sockets and transporting data from one machine to another and do the computing. It was primitive parallel computing and finding an alternative to the high end computing machines. Heady days indeed. Can't think that these days computing power is taken for granted! We have come a long long way indeed.

Coming back to NVIDIA and GPU - I think it is a lot of hype and hope at present. GPUs need more power and the RAM used is normally a generation higher than the regular RAM in the same machine/laptop. At the moment it provides more bang for the buck, but not necessarily at the same power efficient level. And these are quite expensive and hence niche products at the moment. Need to see what would be right balance between CPU and GPU.

Suresh

[Sutripta]

^^^
I'm sure everyone here has heard of Cray. But how many have heard of Hillis's Connection Machine, or Transputers or ... The dead ends of the computation evolution. Shame that these are not taught.

Sutripta

[NiInJa]

In Layman's terms, think of Intel as Force Motors. They would keep updating their one vehicle 'Tempo Traveller' with efficient engines etc. but would never really venture out to do something other than that. When they did, they were late or failed badly. (e.g. Intel Atom)

Nvidia on the other hand is a company with a different mindset. I can't think of a car company to compare them, but a quick look into their recent activities (after 2015) will reveal that they are pretty much serious about AI and Cloud computing and newer technologies.

I was a part of their training programme when our college had signed an MoU with them in Pune. Some of their guys would teach us on Saturdays and their philosophy was that software and hardware need to go hand in hand, so their singular focus on building specific hardware and software for specific things has led them to where they are today (e.g Graphics processor for overlaying images on videos - think of F1 live telecast where race data gets updated and displayed with constant tickers at real time, that needs fast computing, this might be different from a processor and the code needed to process simple Excel row addition running in millions)

I don't know much about Intel these days but I don't think they are getting anywhere with their current or future line-up of processors

[GutsyGibbon]

Originally Posted by ashokrajagopal

Intel did try in the Smartphone market with its Atom series.

They have been a disaster in terms of phone chipsets. Back in the 90s, they were working on a GSM phone. This was when I was with Texas Instruments, and most phones in the world were Nokia or an Ericsson shipped with TI/Omap versions of custom chips. There was so much smack talk of Intel coming in and gobbling up the phone market, but they never even came close to a real launch worthy of big marketing. Internally ( I have heard in the industry and the inter-webs) that they were huge product teams with massive staffing. When Nokia phones had one month battery life, Intel collaborated with tiny companies and managed to create phones with 1 day battery life. They were that off.

Even while competing with Snapdragon (2008 to 2012), the best Intel managed to do was a Lava/Xolo windows phone that was way inferior to the competition. This product was after a mammoth effort of a family/series of chipsets for close to 10 years, that never saw fruition.

[partha379]

Would also like to think this is another classic case of "Innovators dilemma" - their PC business was so big that bidding and competing in any other market including mobile chips was not a priority. Intel had the option of supplying for iPhone which they lost due to various reasons (power, cost etc thanks to architecture cited in other posts), in addition they had this company called Xscale that was doing ARM architecture and they abandoned it in favour of their own low power architecture Atom. They never gave enough focus on the Xscale side and ended up betting on the wrong horse - Atom. Atom single core performance was huge, but read the below 1st person experience of why that wasn't good enough.

I had the opportunity to work on firware/system software for home NAS boxes - option was Atom vs ARM . An ARM based board (ARM processor + memory + wifi chip) was around $10, an Atom chip alone was in the 10s of dollars at that time. A dual core ARM chip with SW speed-up was getting nearly 70-75% of the speed at like 1/10th of the cost of Atom. Now the SW work costs money - but if you can sell thousands of boxes the cost is easy to recoup compared to paying extra for every device you sell. The world moved to this low power-low speed billions of devices tied with SW approach and Intel wasn't ready to play this game.