If you 'don't want to be a microsoft', I assume you mean you're willing to be a decent person about this. That can only work if the customer is also a decent person. If you think that, here's what you do.
First off, there are two distinct possibilities. Either they'll be a customer again in the future (in which case they have a vested interest in helping you out) or this was a one time thing. In either case, just ask them honestly what there budget for the project is. Do this/after/ explaining your expenses; that is, how much you want and why you want it, broken up (honestly) into sunk costs, future development, and the stuff for your pocket. Once they quote a budget, if it's reasonable, take 80% of it. If it's not reasonable, say so. That'll make them want to come back.
How would it be a security loss? Any program concerned about security today is already aware that any of its memory may be swapped out at any time... and that swap files can, on many architectures, survive between boots. The only safe way to ensure that stuff you write in memory is not persisted, today, is to clear it by hand. How is this different with MRAM?
How would it be a stability loss? I just don't see it... all this talk about 'but when you reboot, you'll be in the same state.' No, when you reboot, your memory will be in the same state, and your processor will be in the reset state. Does ANY software on ANY platform using DRAM assume that memory is initialized to a known state at startup?
(Probably late enough post that no one will read it, but what the hell.)
The *.99 cent pricing strategy was NOT started as a psychological trick. It was first instituted by Macy's, the big department store, to solve a very specific problem.
Scenario 1:
I go in and buy a $5 whoosit, and bring it to the cashier. She rings me up for $5, I hand her a $5 bill, she smiles, and I walk away.
And then she sticks the $5 bill in her pocket.
Scenario 2:
I go in and buy a $4.99 whoosit, and bring it to the casher. She rings me up for $4.99, I hand her a $5 bill, and she has to open the register to make change. Opening the register registers the sale, and she can no longer slip money into her pocket off the books.
So Macy's decided to make $0.01 less on each sale, in exchange for essentially eliminating (this form of) employee theft.
This problem could have been dealt with in other ways. Taxes now solve it; it's very rare that a price at a store like Macy's will fall on a dollar amount. Alternately, you can just convince the customer to always ask for a receipt... I just got back from China, and one of the things that is big over there (maybe required?) is for receipts to have small lottery scratch-offs on them, where you have some chance of winning a dollar or two. The idea is that this gives the customer motivation to ask for the receipt, which means there's always a record of the sale. Although the beneficiary in the government (the record is used to avoid tax fraud), the idea is the same.
Even though everyone thinks that *.99 pricing is a psychological trick, that was not its original purpose. Whether or not it is effective at this adopted purpose is, as far as I know, an open question.
Please explain how printf will let you debug the following (more than just a binary chop):
1) A race condition. 2) A deadlock. 3) A performance anomaly in I/O bound code. 4) A stack-smasher. 5) Heap corruption. 6) The, uh, printf family of functions.
What memory are you measuring here? You mention shared memory.. does it matter if a program uses 17MB, if 16.9MB is shared with other running programs? Nope -- if you're only running one program, you have enough memory, and if you're running more than one, the amortized cost is low. Are you counting physical memory used, or virtual memory used? If it's virtual memory, who cares if it's 17MB, if only 1MB gets swapped in?
On the one hand, yes, it's grossly implausible that All Computers (who?) should have a patent on this, and I'm sure there are some nice refereed journal articles that count as prior art, along with lots of IBM lab note books, do recall that at one time we didn't have this, and processors did run at bus speed.... up until the 68060 in motorolla land, the 601 in mac land, and the DX2 in intel land, if I remember correctly.
One of the technologies you'll start seeing for high-performance embedded systems (and can find now, in a few places), is core pinouts designed as the mirror image of a standard DRAM memory pinout. With this setup, a CPU can be put on one side of a four (sometimes five) layer circuit board, normally, and a DRAM chip (single chip, so about 1Gb max for most usage; no double channel) can be put directly opposite it, with vias connecting the two. The electrical connection of the signalling wires between the two is extremely good, and allows much higher speed, lower latency memory to be used.
Slashdot Mirror
Google for Tomasulo's algorithm.
Yeah. I beat photoshop in version 5, and I just haven't had the heart to sit through the later sequels. They've hardly changed the plot at all!
Thanks for sharing that with me... but why do you say it's actually coded in Java?
If you 'don't want to be a microsoft', I assume you mean you're willing to be a decent person about this. That can only work if the customer is also a decent person. If you think that, here's what you do.
/after/ explaining your expenses; that is, how much you want and why you want it, broken up (honestly) into sunk costs, future development, and the stuff for your pocket. Once they quote a budget, if it's reasonable, take 80% of it. If it's not reasonable, say so. That'll make them want to come back.
First off, there are two distinct possibilities. Either they'll be a customer again in the future (in which case they have a vested interest in helping you out) or this was a one time thing. In either case, just ask them honestly what there budget for the project is. Do this
How would it be a security loss? Any program concerned about security today is already aware that any of its memory may be swapped out at any time... and that swap files can, on many architectures, survive between boots. The only safe way to ensure that stuff you write in memory is not persisted, today, is to clear it by hand. How is this different with MRAM?
How would it be a stability loss? I just don't see it... all this talk about 'but when you reboot, you'll be in the same state.' No, when you reboot, your memory will be in the same state, and your processor will be in the reset state. Does ANY software on ANY platform using DRAM assume that memory is initialized to a known state at startup?
apple dropped ps2 back in 1998, along with the floppy drive, back in 1998. think about that for a moment.
I keep trying to think about apple dropping ps2... and keep ending up laughing at your confusion.
(Probably late enough post that no one will read it, but what the hell.)
The *.99 cent pricing strategy was NOT started as a psychological trick. It was first instituted by Macy's, the big department store, to solve a very specific problem.
Scenario 1:
I go in and buy a $5 whoosit, and bring it to the cashier. She rings me up for $5, I hand her a $5 bill, she smiles, and I walk away.
And then she sticks the $5 bill in her pocket.
Scenario 2:
I go in and buy a $4.99 whoosit, and bring it to the casher. She rings me up for $4.99, I hand her a $5 bill, and she has to open the register to make change. Opening the register registers the sale, and she can no longer slip money into her pocket off the books.
So Macy's decided to make $0.01 less on each sale, in exchange for essentially eliminating (this form of) employee theft.
This problem could have been dealt with in other ways. Taxes now solve it; it's very rare that a price at a store like Macy's will fall on a dollar amount. Alternately, you can just convince the customer to always ask for a receipt... I just got back from China, and one of the things that is big over there (maybe required?) is for receipts to have small lottery scratch-offs on them, where you have some chance of winning a dollar or two. The idea is that this gives the customer motivation to ask for the receipt, which means there's always a record of the sale. Although the beneficiary in the government (the record is used to avoid tax fraud), the idea is the same.
Even though everyone thinks that *.99 pricing is a psychological trick, that was not its original purpose. Whether or not it is effective at this adopted purpose is, as far as I know, an open question.
I'm sure I'm missing some obvious part of your question, but... nullfs? unionfs?
Except if I do something illegal, it's between the government and myself, not between a third party and myself...
Also, you don't need to round trip one signal at a time, you can have multiple signals in flight. Which makes this whole discussion rather moot.
But journaling filesystems work under the assumption that writes to the hardware become persistant in-order. Caches (can) violate this.
Please explain how printf will let you debug the following (more than just a binary chop):
1) A race condition.
2) A deadlock.
3) A performance anomaly in I/O bound code.
4) A stack-smasher.
5) Heap corruption.
6) The, uh, printf family of functions.
What memory are you measuring here? You mention shared memory.. does it matter if a program uses 17MB, if 16.9MB is shared with other running programs? Nope -- if you're only running one program, you have enough memory, and if you're running more than one, the amortized cost is low. Are you counting physical memory used, or virtual memory used? If it's virtual memory, who cares if it's 17MB, if only 1MB gets swapped in?
Dude, can I get a hit of that?
Didn't quite get the athlon you wanted for christmas?
The pill?
Um? It's a 2048 bit assymmetric key, you don't need to brute force every possiblility. Thanks for playing, though.
On the one hand, yes, it's grossly implausible that All Computers (who?) should have a patent on this, and I'm sure there are some nice refereed journal articles that count as prior art, along with lots of IBM lab note books, do recall that at one time we didn't have this, and processors did run at bus speed.... up until the 68060 in motorolla land, the 601 in mac land, and the DX2 in intel land, if I remember correctly.
Please please please mod parent up as the first one in this article who has given a concise proof that he understands quantum encryption.
Parent may or may not be insightful, but is, sadly, not true. Power usage is linear with clock speed, assuming voltage is constant.
(Didn't read the article.) Are they cache snooping, which would be the obvious thing, or just using a shared cache and ignoring the problem that way?
One of the technologies you'll start seeing for high-performance embedded systems (and can find now, in a few places), is core pinouts designed as the mirror image of a standard DRAM memory pinout. With this setup, a CPU can be put on one side of a four (sometimes five) layer circuit board, normally, and a DRAM chip (single chip, so about 1Gb max for most usage; no double channel) can be put directly opposite it, with vias connecting the two. The electrical connection of the signalling wires between the two is extremely good, and allows much higher speed, lower latency memory to be used.
If everyone left off the last eight words of that comment (or perhaps the last eleven), the world would be a better place.
So, uh, why not multiple widescreen displays?
No, it's a game where those who focus on the value of their content lose to those who focus on the marketing of their content.