Software architecture changes is major and slow ? server cores ?

Software operating system is currently undergoing major changes from symmetrical processing to hypervisor controlled multitasking processing in real time threads and server cores in virtual memory. This, of course, includes a program to wrote a program, to automatically process simple desire of users. Search engine usage by keyword search is a clear example of user desire fulfilled. But increasingly, people need comparator to get the best answer to a searched result.

Having said that, we can understand why server core built into the virtual memory becomes the extra internal engine to make many decisions for us. Virtual memory can be uses same as ASIC or PGA, risc or cisc.

So, how far is Linux direction in its own virtual technology to handle the user needs of hypervisor operating system with server cores(Microsoft development) in virtual memory in hdd?

We need to understand the user desire to make computers work for each individual?

Footnote:
Recently, WinME had virtual memory increased to 2gb, but internet files of history were only 49mb. The rest is used for server core to run the computer during auto-update of operating system(WinME) changes on the fly(reboot needed to use the changes), while internet operation is run by threads. In multicored computer, it will operate much faster than 32 bit i686 cpu alone.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Many uses of server cores in VM ? Realtime virtual cpus ?

If you only have one cpu core, and you have hdd free space; you could build server cores of virtual cpus to run many realtime tasks in threads, from virtual memory to drams.

First, cpus are registers, ALU plus FPU. 8 bit cpus are around 65,000 transistors. 16 bit i386 are 250,000 transistors. Which means some correlation(bits of memory instead of transistors) to do the virtual cpu in software.

These server cores do not need to be too many data bits across. to handle video bit streams using over sampling, data streams are serial(bit by bit). Analog data compression might require more than 8 bit(24 bit ideal) A/D and D/A conversion.

In the end, it is the virtual cpu function embedded that makes server cores great as hypervisor of the physical cpu. It could be a simple comparator of keystrokes to later select the best info that most users pick(more hits). It could be date sensitive info only. It could be repeat of keywords and associated words(subkeywords). And of course info that was never in your computer before(new info).

Some server cores might do keyboard maps for musical keyboard that you can accompany a movie showing on the screen. Or adding layers of screens for a modern graphics enhanced movie making.

In other words, server cores in virtual memory can give you more multitasking than physical cpus can do with applications in realtime.

Virtual cpu ? Simulated cpu ? Which comes first ?

Virtual cpu is one cpu used on many realtime threads. MIPS recently introduced their cpu architecture with virtual cpu capability(with fixed hard threads).

Simulated cpu is a software file when launched can boot an operating system and run applications. It can be a very small software cpu. Common emulation, interpreters, and virtualizations are not simulated cpu. simulated cpu is now being used in virtual memory in hdd. Now, when you hear a lot of noise from your hdd, it is most likely that your simulated cpu is running your installed operating system in the virtual memory.

Recently, Adobe Flash has a new version. It was downloaded at 6 kBps on my external modem. And installed on the fly without rebooting my computer. My cpu was freed to run other tasks very efficiently.

So, virtual cpu comes first; realtime threads(protected) are installed in virtual memory too. Simulated cpu runs some tasks for internet connections?

Powershell, NAP time, votes in quorum in Win XP etc. ?

Since the development of windows server 2008, Vista and other older operating systems are undergoing a major architectural change. Artificial intelligence is added in 'vote in quorum' for cluster management and scheduling by majority of votes in quorum. This gives the hypervisor the preference of which cluster to use if available(internally or externally in the clouds).

NAP time provides protection for client patching(connection) and antivirus compliance, NAP server can revalidate any changes.

Powershell is command shell, which is a scripting service. It can do a program to write a program in scripting packets in the ,net environment. More powerful than what we do in scripting Posix packets in realtime over 10 years(since 1996) now.

Since these are feature in the development of Windows server 2008 yet to be released. Many Windows users are experiencing auto-update of these features in Vista, XP and WinME. It seems that WinME recovered all the speed and power it used to have, even in service denial situation at the ISP, today, after the latest auto-update.

If you have windows, you probably are wondering whatever happened to you in the past month or so. Microsoft has been experimenting with their operating systems, to get server 2008 ready to roll out for all older and newer windows Os's that windows server 2008 had to service.

Microsoft compression before password at ISP ? How good ?

The availability of Microsoft compression is sorely welcomed. The compression(PPM method) is about 2.1 ratio on downloads. However, it is activated before you submit your password to your ISP account. Which sometimes, when compression has not been launched, can leave your password unrecognized by your ISP account.

So, this is a great help to relieve the ISP of denial of service at busy time.

Enjoy.

ISP's are offering Power Boost feature free of charge to speed up data transmission. Is that Microsoft compression, or part of Vista Ready Boost?

Microsoft virtual memory still needs something; Garbage/Raid5 ?

It is interesting that while you are using internet, you can check sysinfo and see the free dram and free virtual memory spaces available.

And how and why your browser stopped working; when dram free space, goes down to 200kb.

What, no garbage collection to unload dram's 73 mb free space you started with, at the beginning of the internet session?

So, we can now expect another round of Microsoft auto-update to have garbage collection dynamically by a server core in the hdd virtual memory; to free up dram free space, especially during downloading large file from the internet.

Now, we also have to see if garbage collection feature has the vitues of Raid5; if a server core(simulated cpu) is installed. Cache vs backup storage should be a selection of choices? Is download accelerator, Raid5 storage with compression?

WinME with garbage collection and Raid5 webpage storage ?

On this date, my winME has garbage collection, which initiates by a lot of hdd noises. It moves webpages of frequent use to hdd redundant(raid5) permanent storage. Past webpage will be shown first; then you have to refresh to get the latest webpage available from the internet.

At first, it seemed wrong. Why do we have to refresh at all? But if you use Office Live, the webpages are always the same(formats of word processor, spreadsheet and database). So, you work with the format and send the data for upload. This saves data transfer capacity on the cloud. You can work offline, but you don't have the workgroup for collaboration until you use password to upload your data.

I don't use Office Live, so I can not check it out.

The virtual memory file now garbage collected to Raid5 files, which then can service on-demand software(SOA) with a minimum of internet traffic. And the webpages presentation on your own computer will come up faster.

Check it out. If you don't like it; complain here at Tuxmachines; and Bill Gates or Steve Ballmer may hear it directly from you.

Footnote:
Refresh of webpage, should use comparator of list of framebuffers, and only fade in(substitute) the downloaded differentiated framebuffer files, not rewriting the refreshed webpage from scratch(like frame changes in movies?).

Future Linux OS architecture ? Not soon enough ?

It is obvious all OS started with bios. Linux boot into bios with text, installed drivers and applications then startx. Go into window manager then browser.

The new architecture needs to boot into bios, install all utilities, startx and show browser with toolbars and taskbars to operate the computer.

Many components of kernels had to be cut then paste on the icons of the OSbars(not the Beranger deskbars?) in the browser format. Why even have K3b, just have a few icons to do the dvd tasks? The end of the tunnel is near; automation comes to dead stop on one mouse click.

Once we recognize the fact that scripts are the poetic licence to cut and paste, then scripts are condensed to icon for mouse clicks(CnR fame).

It is inevitable; it is only a question of time to train distros to do cut and paste with scripts(automation defined) in kernels(perl), desktops(python) and browsers(php), linked to icons on toolbars and taskbars.

Compatibility is architecture dependent ? Library next ?

Many distros are having fun making their own Linux operating system. Often, their architecture is restricted to certain groups of computers manufactured in a vintage period. Latest computers lack drivers for Linux, but may use diswrapper on Microsoft drivers. Older computers had i386 platform and universal drivers(VGA, PCI, IDE, etc.), but can not use operating system compiled to i686 instructions. For computer made 5-10 years ago, its pot luck depending on platform(cpu class) and architecture of directories and file names. So, Linux open sourced community is fractured and most distros are lacking something or another, or full of bugs.

Until architecture is defined by installers, or Linus and Andrew; most of us can not use Linux without fixing it over time. It seems hopeless with all the standards committees not making any head ways to unify Linux distros for prosperity sake.

Some one had to spear head the unification of Linux dostros to register names of directories and file names to avoid missing bindings. hardware address switching had to be registered to avoid conflicts. Then installers had to offer compiled cpu class of instructions. All applications had to be classified to their cpu compatibility in the installer. Library(dependency issues) had to be classified to the vintage of applications by version numbers regarding backward compatibilties.

USB architecture consideration ? emulating scsi limited ?

When we consider mainframe architecture against x86 terminal type architecture; Linux attempt to boot(suggestion: bios change in shadowed address, then boot on the fly) from scsi emulation, must includes partitioning utilities on fat12, fat 16 and fat32(LBA) on flash cards.

However, USB devices are universal and simplifies computer design, such as printers, data storage devices, displays, media players and communications devices. These can use udev(port addresses of hubs and 128 USB devices) and drivers(data transfer to each USB devices) built in the USB devices itself(firmware).

Until Linux sorted out the varies USB devices and how to handle them; messing around with temporary fixes in file systems are just bugs galore in the future.

Footnote:
128 USB hub connection(serial port data transfer) superseded all other buses(PCI/bridges, etc.) in the number of connections. Thus mainframe computer design is suitable for extreme multitasking with realtime threads on multicored cpus.

Essay on architecture ? componentized drivers ? What next ?

The computer hardware had three classes, embedded systems(controllers, cellphones etc.), X86s(IBM bios architecture) and mainframe computers(cpus, dram address switching and i/o buses).

These architecture are all different in port memory address. So, for Linus and Andrew to offer an operating system kernel for all the hardware; the kernels became bloated. So, some distros offer installers that can componentize the functions. Unfortunately, Linux did not have registered engineering data like IBM x86 handbook. And the constant changes in kernel source codes made compliance difficult, almost impossible for any Linux standards committee to even start to make sense. Bugs galore is made worse by distro diversification attempts to have many new file names for very little code changes and versions. Backward compatibility is lacking. Every 15 minutes new bugs are reported right after you fixed the previous one.

So, if Linus made source code of kernels componentized by functions, then installers can match their hardware architecture. It will not be easy, because you have to understand hardware architecture first based on platform(cpu) and drams, then buses, then storage devices, then peripherals.

Linux standards committee can help, if the operating system is made more compartmentized, and driver files more componentized. Installation(on hdd) can reduce all the codes to those compatible to the hardware, leaving all the excesses on the dvd? Is it thinstation iso, files and kudzu, anaconda and hal/dbus choices?