Patent on Clock Calibration

[AJ]

Hi,

I was just checking over a few pages on the wiki and came across Brendan's explanation of measuring the APIC timer rate. Some searching on Google brought up -> This Link <-. I know there are some daft patents about but

Cheers,
Adam

[Solar]

Patents on trivial algorithms and procedures are a FUBAR concept, plain and simple. It should be screamingly obvious to any rational mind. Yet here you are, in danger of being sued for coming up with anything even remotely similar, which is pretty hard not to, now or in twenty years' time.

[Combuster]

Ok, for all the people in the US who were so friendly to try my OS...

GET RID OF IT. ITS ILLEGAL.

This deserves a +1 for anti-americanism

[Brendan]

Hi,

IMHO (and I'm not a lawyer), there's no way this patent would hold up in any court - it shouldn't be hard to find prior art and it's very obvious to "someone trained in the field".

As far as I can tell, in the US the patent is granted by default, and the courts decide whether or not the patent should have been granted later on.

Ok, for all the people in the US who were so friendly to try my OS...

Your code doesn't use a slower clock to calibrate a faster clock. It uses (a simplified version of) the basic ideas present in the NTP protocol (since the early 1980's) to synchronise one clock with another.


Cheers,

Brendan

[AndrewAPrice]

IMHO (and I'm not a lawyer), there's no way this patent would hold up in any court - it shouldn't be hard to find prior art and it's very obvious to "someone trained in the field".

As far as I can tell, in the US the patent is granted by default, and the courts decide whether or not the patent should have been granted later on.

It has been common for large companies (any large (or small) company in general) to take out patents on the most simple things, not to stop competition, but in order to protect the freedom of the idea rather than down the track someone else patents it, rushes their product to the market first, and then tells them they can't use that idea.

Also, it's unlikely a large company would sue a small non-commercial project for violating a patent based on independent work that is not competing with their product. The legal fees plus the negative image would disadvantage the company more than benefit it.

Even if they do decide to take action, the court will decide the outcome on a specific case basis. The court would consider if you're claiming the idea as your own, if you're directly copying the patent or a product rather than independently reach the same solution, and if by allowing you to continue is going to contribute to a loss of profit for the patent holder.

[earlz]

note: the following code infringes on a patent

Code: Select all

int t=timer1_ticks;
int t2=timer2_ticks;
int c=0;
while(t==timer1_ticks){}
c=t2-t;
//c is calibration variable

[Solar]

All that (patent not holding up in court, patents registered by default to "keep the idea open"). But the threat is there, and I want to spent my days programming, not in court. A court trial is a huge financial risk. Personally, I wouldn't be able to go to court, I couldn't even afford talking to a lawyer about it.

So, that the patent would be ruled against the court and that companies might not even use their patent against me doesn't change the fact that this patent policy puts small competitors at a stiff disadvantage. Large companies can afford a legal department, small upstarts cannot.

And as for "small non-commercial ...", there's a track record of e.g. Microsoft sueing companies for patent infringement as soon as they become successful.

Additionally, the raw number of patents already granted makes it next to impossible to check whether you infringe on a patent beforehand - and even if you are successful, it again takes time that small competitors do not have.

[earlz]

Yet another reason why programs should be considered math formulas and therefore not patentable(by US law)

[Combuster]

Ok, for all the people in the US who were so friendly to try my OS...

Your code doesn't use a slower clock to calibrate a faster clock.

Doesn't determining the Local APIC's reference clock against the PIT count as calibrating a faster clock? Or is my English that bad

[Brendan]

Hi,

Ok, for all the people in the US who were so friendly to try my OS...

Your code doesn't use a slower clock to calibrate a faster clock.

Doesn't determining the Local APIC's reference clock against the PIT count as calibrating a faster clock? Or is my English that bad

The method used to synchronise clocks by NTP could be used to synchronise a faster or slower clock (e.g. the local APIC timer) with another faster or slower clock (e.g. the PIT). This is a completely different idea to the patent, and has nothing to do with calibrating a faster clock from a slower clock (even if the code for both entirely different ideas happens to be identical, because patents cover ideas and don't cover implementation).

note: the following code infringes on a patent

No it doesn't. To infringe on the patent, you'd need to do "a series of measurement pairs" (not just one).


Cheers,

Brendan

[AJ]

Hmm...looks like my poor choice of wording in the title and the poor wording of the patent itself have caused some problems!

Systems and methods for efficiently and accurately determining a speed of a faster clock having unknown frequency using a slower clock having a known frequency.

1. A method for using a first clock to calibrate a second clock

[emphasis mine]

So the way the abstract reads, anyone using the PIT to determine the Local APIC Timer is violating the patent. Read the patent claims section and it would seem that the claim does actually relate more to calibration than synchronisation. Either way, I am still amazed that the patent could even exist.

As for prior art - I'm sure that most people (for years before the patent was filed) have used very similar systems for determining things like RPM in physics / engineering lessons and more abstractly, rates of reaction in chemistry. Watchmakers must have been using this for years, along with the people who originally measured vibration rates of quartz. Again, IANAL, but agree that the patent could not possibly stand up in court.

Cheers,
Adam

[Solar]

But you would still have to go to court to challenge it, which underlines the point I made earlier.

To put it more precisely, should anyone sue me for anything I did in e.g. PDCLib, the project would be dead as I couldn't even afford the lawyer. Doesn't matter if it would stand up in court because for one-man-shows like what we do here, going to court is likely not an option.

[AJ]

Absolutely agree. If anyone challenged me about patent infringement, I couldn't even think about putting up a defence. Of course, when I've made my millions...

[AndrewAPrice]

Absolutely agree. If anyone challenged me about patent infringement, I couldn't even think about putting up a defence. Of course, when I've made my millions...

Transfer the project in it's entirety to a fake person claiming to be base in a country not a member of the Patent Cooperation Treaty?

[Solar]

Fraud?

What's worse, under German law I cannot "transfer" a project, as German law has no concept of yielding copyright. You wrote it, it's yours, even if you say it isn't. You can merely give license to other people.