Intel-Altera, again

formatting link

So I guess Intel is back in the ARM business. Maybe it's just a face-saving way to get their ARM license back.

Or, horror, we'll see an Altera SOC with x86's inside.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin
Loading thread data ...

Stratix86 XI ??

Cheers

Reply to
Martin Riddle

I think we'll see an intel something in it - so it can easily pair up with windows 10 and increase the hold of wintel on the IoT.

Reply to
David Eather

Twenty years ago, Intel took the strategic direction to move away from the Altera type products, they abandoned much of the embedded processor base, because it didn't have the margins of the x86 business and Intel lost interest in it. Now with the PC business flattening out, no Intel presence in cell phones, desperate for future growth, I guess the markets that most of that Altera stuff goes into, that Intel walked away from, now look good.

Reply to
trader4

It sounds like Intel wants the "data center" business which, realistically, is better suited to ARM than x86. Ditto IoT silliness.

One of my engineers knows an Intel person. Iperson complained to her how awful the x86 products are. All the x86s are fluffed-up 8008s, really a 4004 architecture; clumsy, microcoded, register-poor, ugly things. They get performance from extraordinary kluges and superb silicon fab.

It's weird that everything non-x86 that Intel has tried was a failure. Bubble memory, DRAM, iapx32, their own (960?) RISC thing, their lapsed ARM license, Itanic. They are frantic to keep x86 alive, way past its natural death.

Xilinx has better silicon than Altera, but we cut over to Altera because the Xilinx development software is so bad. Intel is famous for sneering at small users, so we fear that we'll have to go back to Xilinx or something.

Trying to get the Thunderbolt chip data sheets from Intel was incredible. We'd need a team of lawyers and proof that we'd buy tens of thousands a year... to see the data sheets! It's no accident that Thunderbolt looks like another failure.

There are rumors already that people are steering away from Altera for new designs. We might. We've done two ZYNQ SOC projects, which were great except for the vile Xilinx software.

--
John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 
 Click to see the full signature
Reply to
John Larkin

Not sure of that, really. Depends on the data center. And I am really not sure how an FPGA onboard helps with a data center.

Starting with the 486 it was all cache all the time for performance.

But I know of three separate directors who regretted not going that way. Just being able to spec DRAM alone( not that you could not for Moto or PPC ) would have probably saved a couple of 'em.

Nothing weird about it - the Wintel bubble was huge.

No accident at all - the only reason 1488/Firewire did well was a defacto "monopoly" ( nothing else had that corner of latency and speed ) and second-source -you could even run 1488 on a PeeCee.

Thunderbolt looks like a Beta offering now ( in the Beta v. VHS sense ).

Apple's slowly pivoting into a fashion company. No kidding; it's really happening. They have gone oh so vertical - didn't get the Henry Ford memo, I guess....

The FPGA software problem will be with us apparently ...forever. Why I'd a' thought by now...

--
Les Cargill
Reply to
Les Cargill

There's a lot of noise about it. Google fpga in data center

An x86 needs all the help it can get.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

Well, that's the end of Altera. Intel will get bored with the whole thing soon enough.

Reply to
krw

I have a raspberry Pi and a 4 year old netbook powered by an Intel Atom. They seem to have similar performance and I would expect that the more modern Atoms have similar power consumption (although the netbook stays powered a pretty long time).

I would bet that for IoT the Atom will work just as well as any of the higher end ARMs. I'm not sure how a combined FPGA/CPU part would be much better for IoT though. I think you can add a *lot* of dedicated hardware for the same silicon as a much less capable FPGA fabric. For IoT the flexibility of FPGAs doesn't buy a lot. Just ask Cypress. They have been pushing "programmable" digital and analog hardware in their PSOC devices for many years and they have yet to set the world on fire. In fact, some of their PSOC devices are hard to distinguish from other MCUs except that they have fewer peripherals.

DRAM? Intel started out making DRAM! They didn't fail at it. Intel dumped DRAM long before others because they saw that it was a commodity business with lots of downside and little upside potential. There are plenty of DRAM companies that either didn't make it or had to be folded into the fewer survivors each year.

Xilinx has always claimed to have better silicon than Altera, but that is not largely true. Most designs in FPGAs only use a fraction of the available logic so that it really matters little just how effective the logic is. What matters is how well the result works which is mostly a software function. The last company I worked for put FPGA design in the software department. Very few designs ever push hard on FPGA hardware.

Lol, the fact that they don't want to share them with you doesn't mean they won't be selling lots of parts. Mostly this is to keep from wasting resources on unprofitable customers and by that I mean they can never make enough profit from you to show up in any of their balance sheets. TI did this to me with some DSP products some years back. I don't think that spelled the end of TI dominance in DSP.

Yeah, so you should clearly avoid the company that helps reduce your development time and improves the quality of the final product. Focus on the unit cost for those qty 1000 a year products.

--

Rick
Reply to
rickman

I think that 99% of this IoT stuff is silly. I don't want my refrigerator to count my calories. But IoT is cited as one reason for the Altera acquisition.

They quit making it.

CPUs will be a cheap commodity soon, too. Intel won't be able to charge kilobucks for a processor when people can license a respectable CPU from ARM and tuck it into the corner of an ASIC for 50 cents.

The fewer people that make Thunderbolt boxes, the fewer people will care to buy CPUs with Thunderbolt ports. IBM sold a lot of computers because all sorts of people could make ISA cards.

It's OK. It turns out that I don't need to make Thunderbolt boxes because my customers don't want it.

Good thing Intel doesn't own Ethernet.

--
John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 
 Click to see the full signature
Reply to
John Larkin

I don't think you know what IoT is.

Because it was not a good profit center. Not because they "failed" at it. They were making DRAM when it was profitable (early days) and got out when it became the same as selling bread and milk with low profit margins.

Define "soon". PCs are clearly not growing much and the market will continue to weaken. But so far there is no challenger to x86 in the PC market. Intel is not in any way in danger of going out of business anytime soon.

Just as with most companies they sell first to the most profitable customers and only to the general market once the beachhead is established. But then you have always overestimated your own importance.

--

Rick
Reply to
rickman

They won't sell any if there's no ecostructure.

If thousands of big and small companies made Thunderbolt peripherials, more people would want to use it. Intel is doing what it can to prevent that.

Looks like USB and Ethernet will be the way computers can get out to the world.

--
John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 
 Click to see the full signature
Reply to
John Larkin

IoT was listed as a possible application for some new CPUs- CPUs with no Ethernet or Wifi or BT etc.0 peripherals and not enough memory to run a decent stack. That was on the investor-oriented press release. They omitted that stuff on the one aimed at engineers. It's not technically a lie- one *could* use such a processor to offload the main one, but..

--
Best regards,  
Spehro Pefhany 
 Click to see the full signature
Reply to
Spehro Pefhany

Given that malware evolves in Lamarckian fashion, it's hard to beat dumb electromechanical stuff for security.

I'll pay extra for a boiler that uses a thermopile to control the pilot, for instance, and there's no way in the world I'm going to put networked surveillance equipment in my house.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Perhaps the idea is to send a message to your cell phone to remind you that you left the refrigerator light on. It makes more sense,

They already are, except x86 and that can't last.

No, but Broadcom thinks they do but they just got bought by Avago, of all companies (guess there are blinkin' lights on Ethernet ports).

Reply to
krw

I have an advanced fridge that turns the light off automatically when I close the door.

But I would cancel my vacation and fly home from Paris if I knew that the fridge light was on.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

That doesn't change the fact that it makes more sense. ;-)

If it can tell you it's on when you're in Paris (why you'd want to go there is another question), you can turn it off from there, too. This Internet thing goes both ways.

Reply to
krw

Actually, I've been to Paris several times, and don't greatly like the place. London is great; Oxford is magical. Truckee is close.

But the IoT thing looks mostly horrible to me... being realtime bombarded with trivial inputs like my pulse rate and the soil moisture in my back yard, and how long my blender has run. Why does a blender need wifi? Twitter just isn't enough stimulation for some people. IoT is another investment bubble, like nanotechnology.

--
John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 
 Click to see the full signature
Reply to
John Larkin

Not everything was a failure. Even on your list above, I wouldn't say Dram was a failure. Intel produced the first Dram chip, created the industry and was a major Dram supplier for more than a decade. When tough competition from Japan was killing everyone in the early 80s, Intel made the strategic decision to withdraw from what had become a money losing business and instead focus it's resources on microprocessors. It was a question of investing huge amounts to remain competitive in a lower margin business or use that capital on microprocessors and similar, where the margins were good. That sure turned out to be the right decision. Most companies would have decided to defend and double down on a business they created and had Intel done that, the results would likely be that Intel would not be the giant it is today.

Eproms/Flash memory has been successful for Intel. I guess now a lot of what they are doing with Flash is with a joint venture with Micron, but they are still in that business.

Microcontrollers, eg the 8051 family were a huge success for a very long time, until Intel decided the margins and long term future just were not there. Same thing with their peripheral chip business.

I agree with Arm, the 960 Risc embedded family, Itanic.... At the point they tried to do the first two, it was already clear that Intel's focus was elsewhere and it was a half-hearted effort. Arm they stumbled into by virtue of being sued by DEC over patents and in the settlement, Intel bought a DEC fab and that product line. Plus, they had already alienated their embedded controller customers, who they had ignored for years. Suddenly showing up with a couple new things, customers just didn't buy it. Also, they had focused their salesforce away from those customers too.

The iTanic sure was a big failure. That was close to their core business and they sure poured a lot of resources into it. It never made a lot of sense to me or apparently most customers. You can add ProShare, which was video conferencing via ISDN or Ethernet back in the early

90s. That was Grove's pet project, they never sold any of it, and it cost Intel hundreds of millions over many years of trying.

I think Intel's biggest mistake was not buying a company or two that was well positioned in some key markets. Cell phones being one example.

Sneer at you? You must be special. Fifteen years ago, they turned their small customers over the reps and disti and they typically don't even have anyone available to sneer at you. lol

Yes, that's what I'm talking about.....

That's interesting. But I see the concern. They wonder if Altera customers will get treated with the interest Intel has shown it's embedded customers that last couple decades.....

Reply to
trader4

This is great:

formatting link

Smart water bottles.

--
John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 
 Click to see the full signature
Reply to
John Larkin

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.