Microsoft will continue to charge its customers for a single processor license once a new crop of hardware servers with the equivalent of two microprocessors on each chip hit the market next year. Intel’s first dual-core Xeon processor is scheduled to arrive in the first quarter of 2006, a company exec said Monday, meaning that a competing chip from rival AMD will likely arrive several months earlier.
Microsoft math: dual core licensed as one chip
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
I swear this is the third time this has come up here, just seams like massive deja vu.
Anyways, of course it would be considered 1 cpu, those who for some reason think of them as 2 cpu’s are silly. But at least it’s a bit easier to get to that conclusion then it is to think HT is 2 cpus, i’m amazed how many people think it is.
Also even if you did want to think of it as 2 cpus, how would MS ever be able to charge for 2 cpus, since every computer would become dual cpu (as these chips come into all new computers), so there would be no single cpu. MS would stick to what ever a normal computer is for their base offering.
IBM and Oracle lately have announced that they will be charging for two licenses if you’re running their products on dual-core CPUs.
MS will not. Good.
Thank you Microsoft! It is times like this that make me thankful that I chose you as my software provider.
…they will be pin compatible with existing Opterons until they EOL DDR-based Opterons after making the move to DDR2 RAM. AMD plans to stick with standard DDR RAM for the time being until DDR2 has advanced sufficiently.
This is a smart move by Microsoft. Besides that, I believe it is just fair to pay per processor, and not per core.
it makes sense to pay per transistor. this will keep pressure on CPU designers to keep them lean and mean.
good move on microsofts part i cant wait to see the chips in a pc a server is nice but i dont run a server and like to have the bragging rights of a chip my os cant fully utilize
MS made a laudable decission. This solution is the best option for both MS and their customers.
The linux kernel developers will of course charge you double the amount than for a single core chip, exactly zero, that is :-).
It should be payed per operating system being run, not by what your machine is. If you have 10 cpus, it should cost the same as 1 cpu if it is the same operating system (that means same version). You should not be made to pay more cause the program is on a better system. The program gives you no advantage, it is the same program! You already payed more for the system, what right does the software creator have to charge more for something that is not their own work. The fact you have to buy a high cost licence of M$ already is a scam.
“The linux kernel developers will of course charge you double the amount than for a single core chip, exactly zero, that is :-).”
A wonderful argument until you realise you need support in most situations. Of course Linux distributions (that give you reliable, accountable support) don’t charge zero.
The zero dollar Linux kernel is great for hobbyists but not for corporations, most organisations or typical home users.
@matt
Red Hat & SUSE charge the same price for 1 to 8 way boxes. Thus, the price for support is the same for a 1 way and an 8 way box. I;m not sure if they count a dual core as 1 or to 2 CPUs but it doesn’t matter until you get over 4 chips in the box.
You shouldn’t been charged per cpu you should be charged for the operating system. It isnt any of their business what computer your putting it on……
Software has the nice ‘advantage’ of it being released to serve a multitude of purposes and then charging people on their usage of it.
Sometimes this is enforced through serial keys or whatever, but other times just by a license.
So, if any company wants to charge more for multiple CPUs they can, and it has a certain logic to it. They can always say, we put in ‘X’ effort to add mulitple CPU support to our product, so we’re going to charge you more money if you use that feature. Now of course I don’t think companies are going to be consistent and true to this, but that’s the reason.
Microsoft actually charges you per CPU?
Why would any corporation set up an SMP Windows server?
I wonder if Microsoft included these additional costs in with their TCO estimates.
“Because I remember I am a customer and these companies exist because of ME and I don’t bow to everything they want.”
If everybody where like you, their would be no companies. Nobody can live out of nothing. Money rules the world.
You aren’t a customer if you’re not buying. You’re a parasite.
SUSE charge the same price for 1 to 8 way boxes.
Do they? That contrasts with the text of the article, which states that Novell charges on a per-chip basis for SUSE licenses.
“it makes sense to pay per transistor. this will keep pressure on CPU designers to keep them lean and mean.”
The whole RISC less is more idea failed. Why is the Athlon FX-53 at 2.4 GHz faster than a Pentium 4 at 3.6 GHz? The core is more complex (not not talking about all the transistors used in cache, btw more cache is good anyway). The modern x86(64) is the best solution CISC bandwith usage with a internal RISC core running the intructions
“Microsoft actually charges you per CPU?
Why would any corporation set up an SMP Windows server?
I wonder if Microsoft included these additional costs in with their TCO estimates.”
This sort of crap is just uninformed OS bigotry. A very small amount of research would have answered everyone of those questions. The answers are:
Yes. For some time on the OS and database servers at least.
They’d want to for the same reasons you’d set up a SMP Linux or UNIX system (probably most often large databases)
They do include those costs in their TCO estimates and you would have had to read only one to know that (or a good article on one).
You not knowing the answers to these questions is more a sign of your own ignorance than any deficiency in Windows or deviance in TCO calculations.
Just What the hell did somebody think was wrong with my comment?
It would be illogical for MS to charge per processor because even though these cpu’s are or will be dual core, the transistors are all connect together on a single sliver of silicon wafer.
What right do they have to charge you more cause your machine has 2 cpus when it is the same program and code. It is only one instance.
What right do they have to charge you more cause your machine has 2 cpus when it is the same program and code. It is only one instance.
The same “right” they have to charge you at all.
It seems that few people here understand that the real purpose of per-CPU licensing is really just a crude method of charging more to those more likely to be able to pay it.
* In general, SMP machines are purchased by customers with relatively high processing power requirements.
* In general, such companies have higher turnovers, staff counts, client counts, etc
* In general, SMP machine cost more.
Therefore, the reasoning goes, customers who have multiprocessor machines – in general – have more money to spend and more reason to spend it (ie: they have more to lose from something going wrong). Therefore, in an excellect example of the free market in action, they get charged more (and pay it).
The only other remotely feasible way of extracting money in rough proportion to the vict^H^H^H^Hcustomer’s resources is by charging per-user – which is, of course, why software for which per-user licensing makes sense usually charge per user (except in circumstances where, for example, the software is generally tied to some other product, like hardware (eg: Apple, Sun) or support contracts (eg: Sun, Redhat)).