Forgot your password?
typodupeerror
Graphics Software Hardware

Smart Monitoring PC Hardware Launched By NVIDIA 82

Posted by Zonk
from the i-see-you've-been-playing-some-portal dept.
MojoKid writes "NVIDIA has just introduced a new open-industry standard for real-time monitoring and control of PC power supplies, chassis, and water cooling systems. Dubbed ESA, which stands for Enthusiast System Architecture, the company hopes the standard will be adopted across the industry. A new wave of ESA compliant hardware that can be monitored and controlled via a standard interface could ensue, like smart health-monitoring power supplies and other components, that would increase system stability and reliability. 'The ESA standard is built around the USB HID (Human Interface Device) specification and has been submitted to the USB-if HID subcommittee for discussion and approval. ESA is essentially a hardware and software interface that takes data collected by analog sensors and converts it to digital information that can accessed via software. Below are a handful of slides taken from an NVIDIA-produced presentation on ESA.'"
This discussion has been archived. No new comments can be posted.

Smart Monitoring PC Hardware Launched By NVIDIA

Comments Filter:
  • Anyone wanna join my club; ATS*?

    * ATS: Ambiguous Three-letter-acronyms Suck

  • Are you sure MojoKid wrote that? If so, he forgot to add the presentation slides to his slashdot submission.
    • Re: (Score:1, Interesting)

      by Anonymous Coward

      Are you sure MojoKid wrote that? If so, he forgot to add the presentation slides to his slashdot submission.
      MojoKid [slashdot.org] does actually appear to be Marco Chiappetta, the author of the hothardware.com article. So he did write that, yes, but did a pretty sloppy job of quoting himself for the synopsis.

  • Hmm.... (Score:4, Funny)

    by EveryNickIsTaken (1054794) on Tuesday November 06, 2007 @11:12AM (#21254525)

    ESA is essentially a hardware and software interface that takes data collected by analog sensors and converts it to digital information that can accessed via software.
    Sounds like a computer to me...
  • by Dareth (47614) on Tuesday November 06, 2007 @11:14AM (#21254555)
    I hope the sensors work better than the ones in my HP printers. While it is may be interesting all the things they can detect, jam in tray 3, under fuser, etc. They really annoy the hell out of the users.

    Well, I would ramble on more about voltage flux warnings etc, but I have to run and change my print toner... I have less than 1000 pages left!
    • Yeah let's hope not.. Otherwise expect messages stating your running out of hard drive space when really you're only using 50%.
    • by vadim_t (324782)

      Well, I would ramble on more about voltage flux warnings etc, but I have to run and change my print toner... I have less than 1000 pages left!

      Well, it depends on how you use the printer. Here I'm sitting near a printer used to print product catalogs. 1000 pages left is a sign of that if there's no extra toner left, ordering some now would be a very good idea.

      Really the right thing to do would be making the threshold configurable. Also depending on the cartridge, the precise amount of toner might be hard to

      • by lukas84 (912874)
        It is configurable on newer HP printers (and Lexmark, and maybe more, but i don't know that).

        And yes, i think 1000 is a pretty good point for a warning. Remember, we're talking about company printers here. Getting a replacement cartridge can be pretty complicated depending on the company, and even then the supplier might not have stock of your specific cartridge type (Probably only with more exotic, older printers).
    • Re: (Score:3, Funny)

      by mikael (484)
      While it is may be interesting all the things they can detect, jam in tray 3,

      I find using ketchup works much better than toner ink - although it does tend to drip a lot.
      • by mcmire (1152897)

        While it is may be interesting all the things they can detect, jam in tray 3,
        I find using ketchup works much better than toner ink - although it does tend to drip a lot.
        It certainly works better than using jam...





        (thanks folks, I'll be here all night)
    • I hope the sensors work better than the ones in my HP printers.

      I hope they work better than their own product [imageshack.us]. The CPU temp depicted is actually 51 degC, according to more accurate Core Temp and Intel TAT monitors. The nForce fan speed is actually ~5000 RPMs, but registers 11000 every few seconds without getting any louder. Occasionally the CPU core and FSB are drastically off.

      Even with nVidia's integrated monitor, most "enthusiasts" (overclockers) end up trusting other independent utilities like SpeedFan and CoreTemp, and disregarding nVidia's monitor. Maybe that's w

  • by Anonymous Coward
    Isn't this what SMBUS / I2C is for?
  • by neokushan (932374) on Tuesday November 06, 2007 @11:19AM (#21254629)
    In NVIDIA's new open-industry standard, you can monitor components inside your computer.

    In soviet Russia, components inside your computer monitor YOU!
    • Re: (Score:2, Insightful)

      by tcolberg (998885)
      And yet on the otherside of the world, the administration's wide definition of wiretapping means that the "components inside your computer monitor YOU" TOO!
    • Sounds so nice to me. But can they make drivers for the graphics card that actually work. Maybe that's not important?
  • by drdanny_orig (585847) * on Tuesday November 06, 2007 @11:21AM (#21254649)
    Wasn't this already accomplished w/ IPMI? At least in more expensive server hardware? What am I missing?
    • Re:IPMI Lite? (Score:5, Informative)

      by Junta (36770) on Tuesday November 06, 2007 @11:54AM (#21255105)
      From my brief look at nVidia's site, it looks like it's more an internal standard than external. IPMI stops at the service processor. The service processor, in turn, used as many arbitrary busses/protocols internally to gather the data and report it back (and while the service processor *might* help you with fan-control versus throttling using OEM commands, the sensor interface strictly speaking only mandates reporting, not controlling). Such a system, for example, would do perfectly fine at reporting on all components explicitly integrated by the manufacturer. However, stick in an arbitrary graphics card, and no sensor value for it's temp/fan will appear as an IPMI-reported device. If a hypothetical standard existed within the case, monitoring may work better without having to buy a single-vendor solution that planned for all that, and you could hypothetically monitor and control arbitrary installed devices. Also, if intelligent about it, more tradeoffs can be made. When you can send instruction to both the GPU and it's cooling device, then for performance you could request that the cooling ramp up whatever it takes, but then go to acoustic and it would throttle the GPU instead of ramping up fans. An IPMI compliant BMC that was also ESA compliant might generate SDRs based on some sort of ESA-discovery.

      nVidia isn't the only game in town talking about internal standards for this, Intel for one definitely has been pushing their own ideas in terms of moving standardization to the interior of the systems.
  • by MonorailCat (1104823) on Tuesday November 06, 2007 @11:25AM (#21254687)
    I currently have to use several programs and a manual fan controller to for half the capability this system offers. To read temps I have to use different programs for motherboard, case, and GPU, and several fans aren't even controllable, in hardware or software. I have little understanding of the temperature distribution in my case. Getting all this information and more into one integrated hardware/software package is a feature I'd gladly pay more for.
    • by kyrio (1091003)
      You've never heard of SpeedFan?
    • I currently have to use several programs and a manual fan controller to for half the capability this system offers. To read temps I have to use different programs for motherboard, case, and GPU, and several fans aren't even controllable, in hardware or software.

      I'd love a nicely integrated solution. I don't have a high-end GPU right now but I went through all kinds of pain just trying to figure out if SMART reports correct data for my drives (more about this [latenightpc.com]). The short answer is that while I think it may have been wrong I still took the time to install some new quieter, high volume fans.

      It seems like it's time for cooling solutions to enter the PnP age. Ever since MS strong-armed the industry into adopting Plug n Play it's been much simpler for computers to

    • Likewise.

      I use SpeedFan, but there is a lot it can't do:
      I can't tell where the motherboard temperature sensors are.
      It doesn't read temps or fan speeds or control fan speeds in the PSU.
      It doesn't read temps or fan speeds or control fan speeds in the GPU (unless I plug the fan into a motherboard fan header.)
      There is no standard hardware for attaching additional temperature probes.

      In my case, some of this is moot - my GPU and PSU don't have fans. I take geek pride in having a quiet single fan system*. There ar
    • Getting all this information and more into one integrated hardware/software package is a feature I'd gladly pay more for.

      Check out Everest Ultimate Edition [lavalys.com]. Here's a screenshot of the sensor page: http://img137.imageshack.us/img137/4020/everestoe0.jpg [imageshack.us]. Everest also shows VERY detailed information on every aspect of your hardware, operating system and software. It's quite an impressive piece of software, the only downside is that it's Windows only.

  • USB port usage... (Score:2, Insightful)

    by Ptur (866963)
    So you'll need a usb port for each item that you want to monitor (PSU, case, cooling,...).
    Apart from usb port shortage, this also means more wires inside the case will be restricting the airflow.
    • by LWATCDR (28044)
      I don't see the need to use USB for this. Why not use I2C or SMbus which based on I2C?
      It already exists and is more than fast enough. Seems like there is already a solution in place that may just need to be improved and not replaced.
      • by Ptur (866963)
        or SPI as a matter of fact....
         
        The only problem is that I2C and SPI are not ideal on longer lengths, and certainly not for inter-component (MBPSU) communication.
        • Re: (Score:3, Insightful)

          by LWATCDR (28044)
          The problem with SPI for this application is that you need a CS line per device. That could be a lot of lines and or wires. Length of run shouldn't be a problem for the sensors in a PC case. The limit is depends on the clock speed but you can easily get a few meters. So with a 6 to 12 foot limit one has to wonder just how big your case is!
          Motherboards already use the SMBus to monitor temps and fans. So why go with USB?
          • by Ptur (866963)
            true... lowering the clock helps. Still, I have seen enough problematic uses due to interference. There's a difference between using I2C or SPI on a PCB or using wires throughout a case. They simply miss signal integrity guarantee.
            • by LWATCDR (28044)
              Well except that SMBus which is I2C is already used on motherboards for temperature sensors and fans so we know that it works.
  • by SuperBanana (662181) on Tuesday November 06, 2007 @11:30AM (#21254753)

    Nothing like primary source material [nvidia.com], folks.

    I really grow tired of Slashdot linking to another site that describes a web page, instead of just linking to the page itself.

    • by trongey (21550)
      If he'd done that he wouldn't have driven all those hits to his own page.
    • Re: (Score:2, Insightful)

      by Anonymous Coward
      ...and when someone does link to the original source document for something like this, people say that the editors are just shilling for the manufacturer. Seriously, they can't win. Say what you will about the quality of a particular article (which, from what I can tell, is poor), but it's very acceptable to link to a new article which reports the contents of a different site, so as to give analysis, compilation of additional primary sources, etc.
    • by keithjr (1091829)
      Does that not mean you also grow tired of Slashdot itself, which is a list of links to posts that describe other pages?

      Sorry, couldn't resist.
    • by Big Jojo (50231)

      ... linking to another site that describes a web page, instead of just linking to the page itself.

      Strange. That's essentially the policy WikiPedia has ... if you put up a good article using only primary sources you'll get dinged because This article or section needs sources or references that appear in reliable, third-party publications. Maybe for disciplines like history that makes sense. But it's a phenomenally stupid policy for technology, where secondary and tertiary sources are both rare and, mo

  • Is it interactive? (Score:3, Insightful)

    by ircmaxell (1117387) on Tuesday November 06, 2007 @11:31AM (#21254781) Homepage
    This would be cool if it allowed real time interactivity with the devices rather than just monitoring. Think about it, being able to monitor and control devices remotely... It can be done now, but it's highly device dependant... With an open protocol, lmsensors would become much easier to install...
  • Calling it "Enthusiast" implies that it isn't for serious use, just hobbyists. I assume they'll want to market this feature to businesses and government buyers eventually. They should've picked a more generic name.
    • by kent_eh (543303)
      They should've picked a more generic name.

      More generic than ESA [wikipedia.org]???
      It looks like one of the most generic TLAs out there.
    • Isn't it really just for the hardcore modders out there who pimp their rigs up with fans and lights and watercooling, etc? They are the only ones who want this sort of control over a PCs internals.
    • In the long run, yes, but they can always roll out a couple generations of technology to try out on the cutting edge stuff then once the kinks are worked out rebrand it as Enterprise System Monitoring or some-such businessy sounding thing.
  • by Danathar (267989) on Tuesday November 06, 2007 @11:37AM (#21254849) Journal
    All I can say is getting LINUX to recognize sensors on various motherboards is a pain in the Ass. Any standardization in this area would be great.
    • What does LINUX stand for?
    • by evilviper (135110)

      All I can say is getting LINUX to recognize sensors on various motherboards is a pain in the Ass.

      First off, this isn't for sensors on the motherboard, but the likes of PSUs and water-cooling systems that don't currently have sensors, or at least not standard.

      Second, you're absolutely right that reading sensors via the Linux means is a nightmare. Setting myself up for a troll mod here, I know, but I have to say the *BSDs have had the sensor problem all sewn up, for quite a while now. 'sysctl -a hw' properl

  • Ummm....SMB? (Score:4, Insightful)

    by Mateorabi (108522) on Tuesday November 06, 2007 @11:43AM (#21254941) Homepage
    Isn't this what the System Management Bus (SMBus) is for? SMB also has the advantage of not requiring hubs to provide multiple ports since its a true multi-slave two wire bus. (Multi master, too.) Why not just provide a breakout connector on the motherboard to chain more devices? It is 100-400kbps but most of the peripherals don't need to report more than a few bits per second of diagnostic info anyway.
    • SMBus is a mess (Score:5, Informative)

      by Wesley Felter (138342) <wesley@felter.org> on Tuesday November 06, 2007 @12:28PM (#21255509) Homepage
      Today's motherboards use I2C/SMBus for sensor access, and it is a total mess. There are hundreds of models of sensor chips that need drivers, and there's no way to know the mapping from sensor numbers to the real world (which fan is "fan 0"?). Then some vendors add microcontrollers that are not on the SMBus and have totally undocumented interfaces (hello AIGuru). I haven't looked at ESA, but hopefully it solves some of these problems.
    • Also PMBus [wikipedia.org], which standardizes much more of the relevant stuff.

      For the record, SMBus maxes out at 100 Kbit/sec. PMBus allows 400 Kbit/sec, as does I2C. Neither of them is really intended to be a multi-master bus, except in the very limited sense that they define a way to send notifications from slave to master. (Though the alternative SMBALERT# mechanism seems much simpler... I think the "notification" thingie was designed to prevent a simple migration path from I2C, and promote fancier Intel southbrid

  • that will be ignored by all the big OEMs and the general populace.

    I don't mean to be a troll, because this is a great sounding idea, but anything I get excited about seems to fizzle long before it even gets a chance to shine.

    I know that I'll support this tech, but, getting other people to see the light will be, quite frankly, impossible.
  • by Bill, Shooter of Bul (629286) on Tuesday November 06, 2007 @12:03PM (#21255215) Journal
    I've just picked up a fault in the AE35 unit. It's going to go 100% failure in 72 hours.
  • Why are they trying to put this in the Human Interface Device part of USB when it isn't for a human interface? I think having this functionality would be great, but I'm curious why it's proposed to be in HID.
  • Why not SNMP? (Score:4, Insightful)

    by smackenzie (912024) on Tuesday November 06, 2007 @12:28PM (#21255499)
    Why don't they fold more video monitors (and enthusiast monitors like water temperature of water coolers) into SNMP or SNMP2? (Or have they already?) Why do we need another standard?

    SNMP is an industry standard, well-supported, flexible with multiple attractive interfaces... pretty sure it can be realtime and has the added bonuses of being networkable by default.
    • by mikkelm (1000451)
      You want to use SNMP for hardware-to-hardware capability querying? Sounds like it'd make more sense to develop some sort of low-overhead SCMP (yeah, you got it right) that SNMP could leverage.
    • by fifirebel (137361)
      Ahem.

      Maybe because SNMP is IP-based. Now you do not want to assign IP addresses for all the components inside your computer, do you?

      And no, rendez-vous or mDNS (or whatever the name-du-jour is for IP autoconfiguration is at Apple today) is not an answer...

      • You don't need an IP for every component, just each computer. You can use your local IP 127.0.0.1 without issue. Really, nVidia should just be working to make this latency-free for local monitoring. By the way, each SNMP instance on each computer can record dozens (or even hundreds) of monitors. Generally with a good SNMP agent / manager, every time you register a device (like a Dell server or an IBM laptop) it adds dozens of things to monitor, some of them extremely low-level.
      • sorry for the potential OT.
        > Now you do not want to assign IP addresses for all the components inside your computer, do you?

        ... actually, I'm thinking you might be on to something here.

        Think IPv6 (where address space wouldn't be a problem) - and monitoring a farm of computers for something (example Fan speed). You could have query:able fans in different PC cases, all by using some sort of internal routing in the pc, some policys what's accessable (firewalling disks, making fans "read only, e.t.c.),
  • Again, this isn't a troll...I'm just a realist.

    I just hope that the standard requires high-quality components, because I can see this backfiring. I have sold ASUS motherboards since I starting selling computers a few years back, and have had a lot of success (read: very very low RMA rate). Even though I choose to include only high-quality components, I have seen a situation where the motherboard -thought- it was running over temperature and began throttling itself (ASUS P4S800-X I think it was)...and this
    • by Pingh (1130313)
      I actually had a very similiar, but on the opposite end of the spectrum. I bought 2 Asus P5K's and set them up with identical processors. The Asus P5K comes with Asus Suite (has all these features like AI Nap, QFan, AI Nos, etc.) On the first board Asus Suite reports the cpu temperature being 15C, and the case being 33C. On the second board Asus Suite reports the cpu temperature being 22C and the case being 33C. These numbers were completely off because even with standard cooling I couldn't bring the cor
      • Wow...that's nuts. I can't imagine that ASUS could squirm its way out of that problem with "Oh, it's the right temp". Below room temp? Clearly an impossibility.

        The P5K boards are supposed to be the upper echelon, no?...that's pretty said that they have such a flaw.

        When you say "case", which sensor are you referring to? Do you mean chipset? Or is there another sensor I'm not aware of?
  • One thing the PC architecture is seriously lacking on is hardware diagnostics. If you have a machine that's crashing with when booted off of CD in your known-stable OS, it's a pain in the rear to track down what piece of hardware is causing the crashes. The almost complete lack of hardware diagnostics (work on a big iron machine once to see how good life can be with real diagnostic tools) means you're almost always reduced to just pulling out/swapping random bits of hardware until it stops crashing. This

If you think the system is working, ask someone who's waiting for a prompt.

Working...