Sand in the Gears of Wintel

For nearly 20 years Wintel, M$ and Intel’s monopolistic PC practice, has demanded ever more RAM every few years in PCs. Now, when attempting to move onto ARM, M$ is making noises that suggest they will actually reduce the footprint in RAM potentially making small cheap computers capable of running M$’s OS. They are saying publicly that “8” will use 200MB less RAM than “7”. Whether this is pre-release PR or fact remains to be seen. One thing is sure, “partners” like Intel and OEMs may not be pleased that people may be able to keep their PCs running like new longer.

We read here often that older PCs are prone to failure. That’s not true except for fans. Old chips are just well broken-in and can tick for a decade or longer. Some early ICs have been running 40 years as good as they were new. If folks can keep airplanes alive for 60 years, getting an IC to last is trivial with no moving parts.

We read here often that PCs need 4gB of RAM at least. That’s not true either, unless you are running that other OS and load a bunch of useless stuff on every boot. Last year, I worked at a place where 8 year old PCs running GNU/Linux were quite useful with only 256MB RAM. Of course they worked better as thin clients than thick but even the terminal server running GNU/Linux only needed about 500MB for the services and 50MB per simultaneous client.

It is a fact that PCs don’t slow down but M$ made them slower with age by constantly increasing the use of RAM (and virtual RAM) and hard drives so that PCs using that other OS ran less efficiently with time. By starting with a lower demand for RAM, that other OS may get its foot in the door of ARMed PCs but it remains to be seen whether ARMed PCs will become less efficient over time with “8”.

I predict that M$ will not be able to cut decades of poor utilization of resources in the next release.

About Robert Pogson

I am a retired teacher in Canada. I taught in the subject areas where I have worked for almost forty years: maths, physics, chemistry and computers. I love hunting, fishing, picking berries and mushrooms, too.
This entry was posted in technology. Bookmark the permalink.

14 Responses to Sand in the Gears of Wintel

  1. I have no problem with load peaks. I have a powerful server. If I have a bigger job to do I run on that server. Here it is idling but in a school running 20 users it will still be idling and my bigger job can take over the role of the idle-loop.

  2. Phenom says:

    No Pogson. You are very, very wrong on this one. A business manager would not mind to have power, sitting idle, as long as when he requires an urgent full year detailed report for a customer, sitting in the conference room next door, the report pops out in a few seconds.

    Theoretically speaking, you cannot meet load peaks, if your system is normally running at 80% load or more. In all situations, servers are expected to run at less than 60%-70% during normal load. Anything above that will cause the system to choke in case of a peak. This is no secret, every system administrator with half a brains will tell you that.

    Now, before Ohio comes to speak his usual nonsense, let me tell you that the load numbers I quote are used by Google as well, when they manage their load balancing.

  3. When I work for an employer who expects me to get the best performance from the investment, idling matters.

  4. oldman says:

    “The more powerful the machine the more it idles. That’s a waste. Waste is foolish.”

    Thats your opinion Pog, and even if I grant that it might be true, it is untimately no business of yours how much I “waste”

    My desktop is a fairly powerful system that may indeed be idle for most of the time, but when I need the paowe as I do when I am composing music

  5. Price, no problem. ARM charges reasonable licensing fees. ARM CPUs cost much less to make/buy so ARMed equipment cost less and FLOSS is cheaper too.

    Performance. Check your idle time. The more powerful the machine the more it idles. That’s a waste. Waste is foolish. If you need that kind of power, put it on a server and use the correct software to manipulate the power and data on the server. Then many users at once can benefit from the investment, a much larger ROI.

  6. Phenom says:

    Btw, just saw the new offerings of ultrabooks from Asus.

    For $999, you get a powerful 11.6″ laptop, which can run for 5+ hours on batteries, and weights about a kilo. For a mere $100 plus you get 7+ hours.
    A device, perfectly suitable for content creation in most of its forms. Now, I am really interested to see how ARMs are going to compete with that – price, performance, and battery life.

  7. oiaohm says:

    Phenom pull head out sand recent patent wars has caused the create of the wild card tizen by hardware makers.

    Yonah solid state caps have fixed the prime cause of board failures. electrolytic capacitor failure. So extending the motherboard failure points. I have a 10 and 15 year old motherboards that were done with solid state caps. They still work perfectly. Even the first IBM XT did not use electrolytic capacitors and it part of the reason why so many of them are still around and are still working without any repairs ever being done to them other than cleaning.

    I guess the caps on your old board that failed were electrolytic.

    Oddly enough a mechanical hard-drive has tolerance more of a write cycle than flash at this stage.

    There are other memory techs competing with flash that do beat hard drive write cycle performance. MRAM FeRAM…. Mostly all magnetic based in some form where flash is a form of capacitance based.

    Capacitance storage devices seams the be the bane of evil for circuits.

    The good quality parts for computers are good for 7 to 8 years by makers quality assurances bar the harddrive and fans.

    AMD has different live ratings on different speed and designs of CPUs. Yes when building if you do full hardware you can assemble highly dependable computers.

    Issue lot of the good parts are expensive.

  8. Fans collect dust and move causing wear. Parts that just sit like stone are much less likely to fail.

    “According to the study, the number one cause of drive failures was simply age. The longer the drive has been in operation, the more likely it will fail. According to the study, drives tended to start showing signs of failure after roughly five to seven years of service, after which there was a significant increase in average failure rates (AFR). The failure rates of drives that failed in their first year of service or shorter was just as high as those after the seven year mark. “

    see Study: Hard Drive MTBF Ratings Highly Exaggerated

    5 to 7 years from a hard drive before it is likely to fail. Many here consider 3 year old PCs old and due for replacement. Where I worked last year we had about 80 hard drives running and only one or two failed and most were 8 years old. What do you think is the MTBF of a chunk of silicon? The oldest PC I had working there was 15 years old and the only thing that made it less useful was a 10mbits/s NIC, not its age. It could still display 1024×768 which was useful.

    Seagate quotes failure rates on hard drives less than 1% per annum. Intel quotes MTBF for SSD drives of 2 million hours.

    You can bet that something moving has a higher failure rate than something sitting still. Fans are highly susceptible to failure. I even had a CPU cooler once that lost a blade and fried the CPU. According to Thermaltake their CPU cooler fans have a MTBF of a few tens of thousands of hours, much less than any CPU at rated temperatures. AMD says, “The rated lifetime of our processors is something between 5 and 10 years.”. You would think they would advertise it if it were good but that’s a hell of a longer time than a fan lasts, especially if there is much dirt in the air.

  9. Yonah says:

    “We read here often that older PCs are prone to failure. That’s not true except for fans.”

    It’s definitely true. In my former life I worked for 4 years doing computer repair as a career, in addition to always being the one to help out a friend or relative diagnose their computer problems. Hard drives crap out, power supplies quit working or get fried, and even electrolytic capacitors dry up or burst open. Dial-up modems are rare now, but I’ve seen more than one motherboard zapped by a lighting strike, sometimes blasting the plastic off IC chips. One of my machines built in 1999 worked for 5 years until one day I returned from the supermarket to a dead motherboard with no visible signs of damage.

    Just how reliable is computer hardware? That’s up to debate. But calming that the only issue with old computers is the fan going bad, that’s just silly.

  10. Mats Hagglund says:

    Still using the same old former Windows Xp pc with 1 GB RAM, harddrive 180 Gb with GeForce 9500 GT. I bought it august 2005, kicked out XP in summer 2008 and started using Linux. 4 partions: Mandriva 2010 KDE, SUSE 11.4 KDE, Ubuntu 10.04 and Linux Mint 9. I even made one separate partition for “storage files”.

    Just think about how effect Linux is. One propably needs just 25-40 GB for one Linux distribution with tens of decent application.

    Without Linux i should have to spend another 1K € for Windows computer. With Linux people really save money – and brains too.

  11. Phenom says:

    I second Contrarian. Windows 8 supports ARM nicely, and offers the best development platform via .NET with all its flavors, subsystems, and tools.

    While ARM might be of some threat to Intel for the mobile devices market, it is absolute none to Microsoft.

    All recent patent wars only made Android look ugly, and 8 a preferable solution to hardware manufacturers.

    Admit it, Pogson. Microsoft arrives to the ARM platform just in time.

  12. lpbbear says:

    Oh Noes! You said something negative about OZ!

    Cue the “Roach Patrol”!

  13. Contrarian says:

    “Now, when attempting to move onto ARM…”

    Whistling past the graveyard again, #pogson? 🙂

    Surely Microsoft is doing far more than “attempting”! ARM device manufacturers are working diligently with Microsoft to be among the first to market with genuine Windows devices that they believe will ignite the ARM PC market as it did the Atom based netbook world.

    “We read here often …”

    What sort of a straw man is that, #pogson? Nothing of the sort has been discussed in any detail here. I don’t recall even the mention of such things although I don’t read every word here.

Leave a Reply