Linked by Thom Holwerda on Thu 4th May 2017 23:42 UTC
Intel

Last month, Intel's new naming scheme for its Xeon processors leaked. Instead of E3, E5, and E7 branding, the chips would be given metallic names, from Bronze at the bottom-end through Silver and Gold to Platinum at the top. Today, the company made this new branding official as part of a larger shake-up of its Xeon platform.

The next generation of Xeons, due to arrive this summer, will make up what Intel calls the "Xeon Scalable Processor Family." This explains the change in core naming that is accompanying the new branding; the SP suffix is replacing the E, EP, and EX suffixes used in previous-generation Xeons.

Order by: Score:
Improved telemetry?
by judgen on Fri 5th May 2017 06:04 UTC
judgen
Member since:
2006-07-12

"Telemetry is an automated communications process by which measurements and other data are collected at remote or inaccessible points and transmitted to receiving equipment for monitoring."

Why the hell would you want that for example for your workstation?

(intel brags about improved telemetry in the new Xeons)

Edited 2017-05-05 06:05 UTC

Reply Score: 3

RE: Improved telemetry?
by Sauron on Fri 5th May 2017 09:09 UTC in reply to "Improved telemetry?"
Sauron Member since:
2005-08-02

Microsoft have got away with it (for now), so they're all going for it. I expect this is only the start!

Reply Score: 3

RE[2]: Improved telemetry?
by BluenoseJake on Fri 5th May 2017 13:05 UTC in reply to "RE: Improved telemetry?"
BluenoseJake Member since:
2005-08-11

They all do it, singling out MS when they are actually the most transparent is kinda funny. Google has been doing it since day one, and Apple since iOS became a thing.

Reply Score: 1

RE: Improved telemetry?
by Sidux on Fri 5th May 2017 09:18 UTC in reply to "Improved telemetry?"
Sidux Member since:
2015-03-10

It's a thing for some years now. Selling RTU and now integrating them in the Xeon chips is not that far fetched.
https://en.wikipedia.org/wiki/Remote_terminal_unit

They've been trying to include various set of features in a CPU for some time now as well.

Reply Score: 1

RE: Improved telemetry?
by TooShy on Fri 5th May 2017 09:41 UTC in reply to "Improved telemetry?"
TooShy Member since:
2011-03-02

-"Why the hell would you want that for example for your workstation?"

... Maybe because Xeon's line processors are intended for big iron servers not for workstations ?

Reply Score: 1

RE[2]: Improved telemetry?
by laffer1 on Fri 5th May 2017 13:04 UTC in reply to "RE: Improved telemetry?"
laffer1 Member since:
2007-11-09

Intel's line is most certainly for workstations as well as servers. Tell Dell, HP and Lenovo that they should stop selling all workstations. Not to mention, they're also used in smaller servers.

Until ryzen, there wasn't out there in the consumer or small business space for fast parallel systems.

If you actually need parallel speed, a workstation is the way to go. I use one for compiling packages for my OS project.

Reply Score: 3

RE[3]: Improved telemetry?
by Kochise on Fri 5th May 2017 13:42 UTC in reply to "RE[2]: Improved telemetry?"
Kochise Member since:
2006-03-03

Dockered ?

Reply Score: 2

RE[3]: Improved telemetry?
by dark2 on Fri 5th May 2017 14:26 UTC in reply to "RE[2]: Improved telemetry?"
dark2 Member since:
2014-12-30

If you actually need parallel speed, a workstation is the way to go. I use one for compiling packages for my OS project.


Now that dual/quad/6/8 core is common without multi-socket mother boards, there is now reason to pay extra for xeon to use for a personal computer workload. The only advantage it offers is ECC (Error correcting code) memory at the price of a small dip in speed, but that type of memory is more important in servers and has negligible benefits in personal computers.

Reply Score: 1

RE[2]: Improved telemetry?
by unclefester on Sun 7th May 2017 03:00 UTC in reply to "RE: Improved telemetry?"
unclefester Member since:
2007-01-13

... Maybe because Xeon's line processors are intended for big iron servers not for workstations ?


Every Intel workstation on the market uses Xeons, ECC RAM and professional grade video cards. In some situations (eg aviation and medical devices) it is a legal requirement to use approved professional grade hardware.

Reply Score: 2

RE: Improved telemetry?
by ahferroin7 on Fri 5th May 2017 12:14 UTC in reply to "Improved telemetry?"
ahferroin7 Member since:
2015-10-30

Telemetry in the context of a CPU is not the same as in the context of an OS. There are aspects of performance that can't be reliably measured externally to the component itself but have a significant impact on how well stuff runs (the branch-predictor's hit ratio is an example of this). Intel's usage of the word telemetry for this type of thing is probably not the best choice, but this is what they're talking about when they talk about 'improved telemetry', not something sending data to some remote location outside of the system.

Reply Score: 2

RE: Improved telemetry?
by Alfman on Fri 5th May 2017 13:51 UTC in reply to "Improved telemetry?"
Alfman Member since:
2011-01-28

judgen,

"Telemetry is an automated communications process by which measurements and other data are collected at remote or inaccessible points and transmitted to receiving equipment for monitoring."

Why the hell would you want that for example for your workstation?

(intel brags about improved telemetry in the new Xeons)



As with most features these days, the problem isn't necessarily the feature, but rather who controls it! Does the end user/company control it? If so then it can be legitimately used by the end user and company to track their own equipment. The ethical boundary that more and more vendors are crossing isn't that these features are included, but it's that the owners are not in control.

The other thing that's highly concerning is that there's more and more proprietary code running on all our CPUs that we cannot audit. I'm talking about AMT (active management technology) which makes off the shelf CPUs remotely accessible. While this can be very useful in use cases that used to be served by having a DRAC or ILO card (for remote control), it's a real shame that these are black boxes and we really have no idea what's running on them. Even if you install linux and an open EFI, modern CPUs have operating systems that operate with even higher privilege.


In related news, it was revealed this week that the proprietary code in intel's AMT implementation since 2008 has a critical remote exploitation vulnerability.

https://thenextweb.com/insider/2017/05/02/intel-sold-remotely-exploi...

AMT is a management tool that allows an authorized user to remotely manage a machine, giving serial access, and with the right drivers, it can offer a remote-desktop experience.

Typically, AMT requires the user to authenticate with a password – but this vulnerability essentially circumvents that process, giving the keys to the kingdom to anyone with a copy of Metasploit.

If the computer is on a misconfigured network where network port 16992 is accessible to the outside world, it means that anyone sitting anywhere in the world can take advantage of these features. Even if that isn’t the case, someone could attack it from within the network.

Perhaps the most troubling facet of this saga is that the bug – which mercifully, isn’t found in consumer Intel chips – remained undetected for almost nine years. Intel has been selling vulnerable silicon for almost a decade. There must be, quite literally, hundreds of millions of computers at risk.


This is why it's important to run open source code, even for firmware and privileged CPU modes. As a consumer however I continue to be extremely disappointed because while I want to encourage everyone to make an informed choice and promote open technology, the fact of the matter is that the availability of open products is often non-existent. Whether it's intel's AMT/DRAC/Spiderduo, my network printer, my network PDU and UPS, etc I was determined to acquire open products if I could, but very often I come up empty handed. How come it's so difficult to find products with open technology to support with my wallet? I don't have the words to describe just how disappointing this is to me.

Edited 2017-05-05 13:56 UTC

Reply Score: 2

RE[2]: Improved telemetry?
by darknexus on Fri 5th May 2017 17:59 UTC in reply to "RE: Improved telemetry?"
darknexus Member since:
2008-07-15

Yeah, I just found out about this today and have been auditing our entire network. We never used AMT so we were safe, however I did a full audit anyway to be sure. Pain in the butt.
I disagree that open source will solve this problem though. I, as the user, have no way to verify that the published source is identical to the code running on my CPU and, further, don't have the time to audit the code myself. One corrupt individual in the right place, or one corrupt OEM, undermines open source just as easily as it does a "black box."

Reply Score: 2

RE[3]: Improved telemetry?
by Lennie on Fri 5th May 2017 18:58 UTC in reply to "RE[2]: Improved telemetry?"
Lennie Member since:
2007-09-22

Unless you can upload your own binary.

Reply Score: 3

RE[4]: Improved telemetry?
by darknexus on Fri 5th May 2017 19:11 UTC in reply to "RE[3]: Improved telemetry?"
darknexus Member since:
2008-07-15

Unless you can upload your own binary.

Do you think I have time to do that on thousands of machines?

Reply Score: 2

RE[5]: Improved telemetry?
by Lennie on Fri 5th May 2017 19:43 UTC in reply to "RE[4]: Improved telemetry?"
Lennie Member since:
2007-09-22

You would do it when you deploy machines.

Just like people install the BIOS and software image they know works.

Reply Score: 3

RE[4]: Improved telemetry?
by Alfman on Fri 5th May 2017 21:59 UTC in reply to "RE[3]: Improved telemetry?"
Alfman Member since:
2011-01-28

Lennie,

Unless you can upload your own binary.


For clarification, are you suggestion there is actually a way to upload your own binary to AMT or is this just hypothetical?

If there is a way to run your own AMT firmware I'd love to know how!


What sucks for me is that the latest AMT software for my system from HP is dated 2014 on it's website, and I have no idea if they intend to patch these vulnerabilities. I'm upset that I don't have the source code for it.

We often blame consumers for buying proprietary hardware in the first place, but damn it I look for open alternatives. It's not that I didn't know better, most vendors just won't sell it.

Reply Score: 2

RE[5]: Improved telemetry?
by Lennie on Mon 8th May 2017 11:50 UTC in reply to "RE[4]: Improved telemetry?"
Lennie Member since:
2007-09-22

Sorry, hypothetical.

We are using more and more open source / free software and still computing is getting more and more closed.

Edited 2017-05-08 11:51 UTC

Reply Score: 2

RE[6]: Improved telemetry?
by Alfman on Mon 8th May 2017 13:32 UTC in reply to "RE[5]: Improved telemetry?"
Alfman Member since:
2011-01-28

Lennie,

Sorry, hypothetical.

We are using more and more open source / free software and still computing is getting more and more closed.


Indeed, the benefits of open source are lost when machines lock the user out using crypto. The FSF saw this, and developed GPL3 to counter it, but by the time they did, critical projects like linux were firmly planted in GPL2.

The industry doesn't have to evolve this way. I keep fantasizing about how things should be with open & robust technology to encourage 3rd party innovation, but it's quite apparent that profit driven companies have a different agenda.

Reply Score: 2

RE[7]: Improved telemetry?
by Lennie on Mon 8th May 2017 14:24 UTC in reply to "RE[6]: Improved telemetry?"
Lennie Member since:
2007-09-22

But I think ease of use also has something to do with it.

People are willing to give their data to someone else to not have to deal with the maintenance.

Doing it right actually takes extra effort, like:

Chrome can hold your passwords and synchronize them between browsers. We don't know what/how they encrypt.

Firefox does this properly, they use the password to encrypt the data in the browser before sending it the Mozilla servers. Mozilla has no access to the data.

Reply Score: 2

RE[3]: Improved telemetry?
by Alfman on Fri 5th May 2017 19:16 UTC in reply to "RE[2]: Improved telemetry?"
Alfman Member since:
2011-01-28

darknexus,

Yeah, I just found out about this today and have been auditing our entire network. We never used AMT so we were safe, however I did a full audit anyway to be sure. Pain in the butt.


They say consumer devices aren't effected, but I'm not sure what that means because I have consumer devices that nevertheless have AMT. I've made sure the affected ports aren't exposed on the WAN, but the thing is I do use AMT locally.


I disagree that open source will solve this problem though. I, as the user, have no way to verify that the published source is identical to the code running on my CPU


Oh I agree this needs to be verifiable, ideally IT staff could flash it themselves and then have a secure process to periodically verify it hasn't changed.

... and, further, don't have the time to audit the code myself. One corrupt individual in the right place, or one corrupt OEM, undermines open source just as easily as it does a "black box."


Well, the benefit of open source is that white-hat guys (ie security researchers, defcon hackers, etc) can audit it independently. We're not forced to take the (highly biased) manufacturer at their word that the software is secure. When it's open, you get all the same assurances from the manufacturer PLUS those of independent researchers.

It's not good when exploits remain in open source code that nobody's looking at (I'm sure you remember shellshock, poodle, etc). But while open source is not the panacea that will resolve 100% of security problems, it's still a crucial requirement for making security transparent rather than taking companies at their word. Companies go on record all the time bluffing about the security of their systems, screwups have happened so many times even with financial firms, porn sites, retail stores, tech companies, etc that we all need to know that we can't take them at their word.

Reply Score: 3

"Xeon Scalable Processor Family."
by quackalist on Fri 5th May 2017 13:37 UTC
quackalist
Member since:
2007-08-27

Bummer, who knew they weren't before they got metallic branding.

Reply Score: 2

Comment by Flatland_Spider
by Flatland_Spider on Fri 5th May 2017 19:12 UTC
Flatland_Spider
Member since:
2006-09-01

In the future Intel will do away with processor designations and just use price ranges. The ranges will roughly map to the list below:

* "If you have to ask..."
* "I'm sorry, how much?"
* "You're not paying for this, so who cares."
* "Spendy"
* "Pricey"
* "Cheapskate aka Come on. Don't get cheap on me now. You got some dollars there. I know you do, so let's just move up a little."

Edited 2017-05-05 19:12 UTC

Reply Score: 1