[TheReg] Intel Management Engine: Is this why we have seen slower speed growth?

[frozentundra123456]

Anyone who thinks Intel is purposely slowing CPU performance gains is free to design a faster CPU and rake in all those billions and billions in profit.* Modern CPU design is hard, and doubly so when backwards compatibility is required.


*Beyond the obvious like different clocks and core counts.

Agreed. I think there are several reasons for the slowing in performance increases. The main one is that the easy gains have been made already, and apart from adding more cores, which I think Intel should do on the mainstream, there is no low hanging fruit anymore. Also, apart from a few high end users, there is not really a need for higher performance, and the emphasis, whether we like it or not, has shifted to performance per watt rather than pure performance.

 

[MarkizSchnitzel]

Barriers to entry.

Intel knows this and is making use of it to maximize profit. Their primary interest is just that, maximizing profit. This does not necessarily mean maximizing performance improvement per year. Especially in a monopoly-like environment.

I think unwritten rule in any confrontation is to ALWAYS go with the least possible force needed. Tactical reserve and such.

Exactly what another monopoly,apple is doing (1GB RAM, low mp camera, hd screen, 3g/4g, tiny battery, small screen, nfc..). Only do something when you have to, when it gonna bring most cash.

 

[zir_blazer]

The ones that the whole ME affair hurts the most is the open source Firmware development. Libreboot (A Coreboot fork that doesn't wants to use binary blobs) has this to say about Intel ME.
Curiously, I recall having read somewhere a guy with a Notebook that said that after a Processor upgrade, his machine consistently shutted down at the 30 minutes mark, which is something that ME is supposed to do. If I wasn't lazy I would try to look around for that Thread and try to figure out how exactly that happened.


There are a lot of useful things that you could do with a huge backdoor in your systems, like the Intel Anti Theft feature, which for Notebooks seems like a neccesary evil. However, the point here is that YOU need to be able to control it, and that also means that you need to learn how to setup and use it. This is the part that everyone is missing.
I think that its actually more risky to have the feature left there unused, since what rootkits and malware usually does is take advantage of the default, expected options, and chances are that small variations here or there may screw up a few ones. Secure Boot for example isn't that bad once you understand what it does and the type of rootkits that it can prevent. On Linux you can even sign your own files, and load the key to the Firmware. However, the procedure is a pain in the butt explained on very technical terms, and not a lot user-friendly material to actually use it.

What DOES worries me is that UEFI 2.5 is intending to standarized the Firmware flashing procedure (Usually each Flash ROM has its own flashing procedure, that the flashing application does for you). This means that a single exploit there would potentially allow to brick Motherboards en masse - what mitigated the BIOS overwrite capabilities of the infamous W95.CIH.

 

[Fjodor2001]

There is a small underdog company that has no barriers to the entry and it seems that despite Intel slowing down they are not making much progress in shrinking the huge performance gap they have when compared to Intel. Maybe they are being bribed by Intel to not develop faster processors too.

Barriers to entry include financial barriers such as Capital, Predatory pricing and Investment.

 

[Fjodor2001]

Sounds like something Fjodor would say.

Nice straw man attempt.

 

[mrmt]

Barriers to entry include financial barriers such as Capital, Predatory pricing and Investment.

Incompetence and lack of technical expertise from upper management have far more deleterious effects for established players than these barriers.

 

[III-V]

Nice straw man attempt.

Stop saying inane crap and making a caricature out of yourself, and you'll stop ending up in people's signatures and stop being the butt of everyone's jokes.

 

[gdansk]

Intel would have to be the worst architects in history.

As TFA article says it is beneath the OS. Not competing with the OS. Beneath. A different level. It has its own RAM & ROM, processor, and real-time OS. It exists to manage the main processor. It's really actually quite handy for sysadmins. Having access to those things doesn't mean it uses them all the time.

She's basically trying to say that, if someone (NSA?) wanted to, they could use the ME to compromise the security of modern x86-based systems.

I'm amazed she trusts an FPGA though. Most of the bitstream generators aren't open source. The "trust stick" she proposes couldn't be trusted for long.

 

[Rakehellion]

Because Intel dominates the market and has no competitors.

They could make next year's chip half as fast and what are you idiots going to do, not buy it? Good luck telling your boss you just bought 500 secondhand Dells from ebay.

If they can spend $50million on research this year instead of $500million, that's an amazing business decision.

 

[ShintaiDK]

She runs a business

 

[Rakehellion]

And this, among many other reasons, is why I have said the cardinal human sin is greed, for a very long time now...

Or maybe you're greedy for expecting corporations to give you something for free.

 

[ShintaiDK]

Because Intel dominates the market and has no competitors.

They could make next year's chip half as fast and what are you idiots going to do, not buy it? Good luck telling your boss you just bought 500 secondhand Dells from ebay.

If they can spend $50million on research this year instead of $500million, that's an amazing business decision.

Ignorance is a bliss? You do know if people and companies dont upgrade, Intel goes bankrupt very fast?

The monopoly card is played way too much and by people who obviously dont understand large cash flow companies or business outside a very tiny box. But rather think Intel is selling something like tap water with a more or less static demand.

 

[Maxima1]

Or maybe you're greedy for expecting corporations to give you something for free.

No, it's called increasing stakeholder value instead of shareholder value.

 

[III-V]

Or maybe you're greedy for expecting corporations to give you something for free.

I don't think it's greedy at all, considering how much they've exploited populations.

 

[MongGrel]

Agreed. I think there are several reasons for the slowing in performance increases. The main one is that the easy gains have been made already, and apart from adding more cores, which I think Intel should do on the mainstream, there is no low hanging fruit anymore. Also, apart from a few high end users, there is not really a need for higher performance, and the emphasis, whether we like it or not, has shifted to performance per watt rather than pure performance.

I tend to agree with this.

A lot of the R&D focus seems to have shifted to improving smaller mobile platforms in general.

It appears to me there are still gains beyond made in the serious upper end of workstations, just not probably something the average consumer would use.

 

[mrmt]

She runs a business

My view's on her job are a mixed bag. On one hand her approach to security makes conceptual sense as it is basically "don't trust anything that you didn't/cannot inspect the code and traffic yourself", but OTOH is it viable on this world, or at least is it viable for a small firm like hers to build? Every single new processor out there would have to be scrutinized for backdoors by her firm, and so every NIC card, every USB controller...

 

[Fjodor2001]

Stop saying inane crap and making a caricature out of yourself, and you'll stop ending up in people's signatures and stop being the butt of everyone's jokes.

First a straw man attempt, now this. Reported...

 

[Fjodor2001]

Incompetence and lack of technical expertise from upper management have far more deleterious effects for established players than these barriers.

Regardless of company? All upper management is like that?

You asked a general question as to why other companies did not enter the x86 desktop CPU segment to compete with Intel. And you're saying it's because the management of all other companies lack technical expertise? Has nothing to do with barriers to entry at all? Ok, whatever you say...

 

[mrmt]

Regardless of company? All upper management is like that?

You asked a general question as to why other companies did not enter the x86 desktop CPU segment to compete with Intel. And you're saying it's because the management of all other companies lack technical expertise? Has nothing to do with barriers to entry at all? Ok, whatever you say...

What I said is that there is a company that doesn't face barriers to entry because it is already on the x86 market, all they had to do was to develop good products and guess what, they don't. And despite Intel slowing down they don't seem to be able to catch up Intel on performance.

So before accusing Intel of purposely slowing down on performance development I would look for other features that would make these performance gains inviable, such as the x86 design becoming mature enough to the point that IPC gains will be extremely hard to extract.

 

[Fjodor2001]

What I said is that there is a company that doesn't face barriers to entry because it is already on the x86 market, all they had to do was to develop good products and guess what, they don't. And despite Intel slowing down they don't seem to be able to catch up Intel on performance.

So before accusing Intel of purposely slowing down on performance development I would look for other features that would make these performance gains inviable, such as the x86 design becoming mature enough to the point that IPC gains will be extremely hard to extract.

Just to clarify, what x86 company are you talking about? VIA? AMD?

Also, did you read article I linked to in my previous post? Having an x86 license is not the only barrier to entry that faces the competition. There are e.g. financial barriers as well. Can we expect VIA or even AMD to have the R&D resources of Intel? If not, surely you must agree that it's harder for them to compete with Intel.

The fact that the competition does not have an x86 CPU that beats Intel's does not prove Intel is not dragging its feet and milking the market. What we would need are other companies in Intel's position and with their financial resources competing with Intel. If none of them could improve CPU performance (and perf/watt, etc) at a faster rate than Intel, then yes we could conclude that it wasn't due to barriers to entry. But we don't have that situation with companies in equal position competing.

All companies strive to achieve barriers to entry. Now that Intel has that they would be foolish to not take advantage of it. Nothing strange with that.

But having said that there's also a risk getting too comfortable, only relying on using the benefit of one's market position to maximize profit. The competition has caught up with Intel before when they got too lazy. It can happen again, despite that the competition is in a tough position, so it's harder for them to compete of course.

 

[Maxima1]

What I said is that there is a company that doesn't face barriers to entry because it is already on the x86 market, all they had to do was to develop good products and guess what, they don't. And despite Intel slowing down they don't seem to be able to catch up Intel on performance.

Are you being disingenuous on purpose?

http://www.wsj.com/articles/SB10001424052702303417104579541791542046598

Intel's Sway Drives Up Server-Chip Prices

"The customer base is saying Intel is more reluctant to offer very competitive pricing, because competition is no longer there," said Albert Mu, general manager of Taiwan-based Tyan Computer Corp., a unit of MiTAC International Corp. that makes Intel-based servers but is also backing a rival technology. "This is basically costing the industry a lot."
Intel's position partly reflects the failings of longtime rival Advanced Micro Devices Inc., the only large vendor that uses the same underlying design as Intel. AMD's share of those x86 server chips has dwindled from nearly 25% of the market in 2006 to less than 3%.

 

[III-V]

First a straw man attempt, now this. Reported...

Whatever helps you refrain from taking personal accountability.

 

[mrmt]

The fact that the competition does not have an x86 CPU that beats Intel's does not prove Intel is not dragging its feet and milking the market.

If you think that Intel dragging its feet will still yield better results than AMD and VIA full steam ahead then these two companies have no place on the market.

"The customer base is saying Intel is more reluctant to offer very competitive pricing, because competition is no longer there," said Albert Mu, general manager of Taiwan-based Tyan Computer Corp., a unit of MiTAC International Corp. that makes Intel-based servers but is also backing a rival technology.

What you are saying here is that OEMs like Tyan have less and less leverage because they have to get their products from a single source, but that's not to say that costs are going up to consumers. In fact, TCO is still getting better each generation, and that's the only metric that matters for a server customer.

 

[ShintaiDK]

The TCO cost is also why people can find very cheap older generation Xeons as upgrades for their desktop. Compared to relatively expensive equal desktop parts.

 

[maddie]

It seems that some are only thinking in opposites and not being able to see a range of possibilities.

Its either Intel is purposely slowing down development or IPC gains are almost impossible to extract.
What about benign neglect due to it not being a very high priority? Just enough of a gain between generations to keep sales up but not enough to wow the customer. Almost all say that the average PC can run 90+ % of productivity software and seeing that clients will not upgrade to a higher performing PC, in most cases, if the present one is good enough, we can see Intel making progress at their optimized pace.
Invest in other technologies that have a higher priority.

Do we know from 1st principles that large IPC gains are a thing of the past, or are we inferring this based on Intel's slow improvement?