Right, but the side receiving the signals from those in the disaster area aren't in a disaster area. BPL noise interferes with the reception of the emergency traffic.
I don't know about the majority of schools, but at my school, which has a quite good CS program, the professors are brilliant, but cannot write good/maintainable code at all.
Ask them about how to approach solving a problem and they can draw the structures and describe the algorithm, but if they can write the code, it's atrocious.
While you are correct that clusters are not the ultimate solution for high performance computing, single-image computers are not a great solution either. They require specific optimizations to be done for the particular system and do not lend for easy system upgrades.
What you are missing is that hardware is the least of your worries when porting. The OS will have different calls to do things. The binary format is different. The same type of function for the OS may do it differently. And so on.
Porting from one OS to another even on the same architecutre is a pain the ass.
I run a gentoo mail server and there have been a few package updates that broke it. Part of it has to with using more obscure packages (pam-mysql, etc), but still, a package shouldn't be marked stable until it's been tested. For some of the server stuff, they just don't test it.
So if two people are having a conversation by shouting across a plaza and I walk to a spot where I can hear one or both sides, I'd be wiretapping?
I know the FCC likes to claim that signals broadcast are not public property, but when you broadcast something, anyone within range is going to pick it up. Why would that be a violation of wiretapping? It's not like you actually tapped into the system at all. It was already giving you the conversation.
You misunderstood the parent. Some earlier versions of distributions assumed that if you wanted to automatically partition a drive that you wanted to erase everything on it rather than just use the free space. Lots of new users accidentally wiped their windows partitions because they didn't know the consequences.
I put linux on my Xserve. It's a server. Linux fit the problem better. No GUI overhead, easier to maintain, etc.
For G5s, the person generally is either using it as a server or for sciTech work, such as long simulations or massive calculations.
The sciTech programs are typically written on linux x86 systems and then someone else gets the job of finding the best system to run them. Putting linux on a G5 ends up being easier than getting the initial developer to port the app. Not to mention it gives a consistant interface for all the systems regardless of architecture. This has the nice side effect of making it easier for the admins to maintain the systems and keep up to date on whats new.
I not sure what's worse, the fact that you took the time to write such a long satirical post, or that you actually were moderated Informative for a while.
> Every other person in developed countries has a pocket-sized radio that will let them talk to anyone in the world.
Of course, amatuer radio is present even in undeveloped countries and in some cases is the only way to communicate with someone other than the locals.
Don't assume that simply because the Internet has changed how communication is done that it is the end-all solution. What happens when an attack brings down the core routers? The Internet isn't quite as capable of routing around failures as everyone hopes.
As for commercial radio, even with walkie-talkies, you can't even come close to the distance, variety of services, independence of infrastructure, or ease of setup that amatuer radio offers.
Quite simply, I have things on my wired home network that I don't want anyone on my AP to access. Using a linux box to handle routing and firewalling between the Internet, wired, and wireless networks does something that software firewalls (like ZoneAlarm) can't do and that would cost over $300 for a hardware firewall to do the same.
If i've already got an old machine laying around from my last upgrade, why waste money on the hardware firewall?
Maybe you should take a minute and actually look at what _you_ are comparing. The scores quoted by Apple for the Intel machines explicitly stated that GCC was used as the compiler to remove the compiler as a variable. The published scores on SPEC's website do not. You are comparing cows to telephone poles.
Now, comparing G5 results with GCC to x86 results with GCC is pretty fair when you are comparing the hardware only, not the software. That is what they claimed to be showing.
Actually, there is a certain period of delay between letters and words. Thus it is quite easy to determine what is intended simply by listening for the pause.
As noted earlier, both systems used GCC 3.3. That is where most of the loss can be attributed.
As for the question about HyperThreading. On a SPEC int or fp run, the test is single threaded. HyperThreading in this case only serves to help with the OS overhead a bit.
In the case of a SPEC rate run, hyperthreading actually hurts performance. This is due to the fact that to fully load down the system, a copy of SPEC is run for each processor the system sees (in the case of a Dual Xeon with HT, 4). Considering SPEC int and fp tests use all the available units of the respective type, HT would be unable to do anything useful as a single thread would be using all the particular units and the second thread on the processor gets stalled.
So, if you disable HT and run only 2 copies of SPEC for a rate run, the overall throughput is greater because the threads complete faster since they aren't stalling.
Having just done some serious testing on a couple of hyperthreading capable machines (dual Xeon 3.06GHz and 3.0GHz P4), I can say a bit about it's effects on programs. If the code is multi-threaded (I didn't read the article to see if his is and this is meant to be really general) it will be distributed over all the "processors" equally. This works great for programs that have 2 very different threads. However, for an app that is very int or very fp intensive in multiple threads, hyperthreading actually hinders overall throughput.
This is due to the fact that hyperthreading is still limited to the number of functional units in the processor. For code that is very intensive on a particular type of unit (int or fp), you basically end up with a stall condition on the virtual processor while all the functional units of that type are used by the first processor.
Hyperthreading is better suited to cases such as a user using a 3d modeling program and a MP3 player. The MP3 player will hopefully end up on one virtual processor and use the int units while the 3d modeling will end up on the other and use the fp units. This would allow both to run in parallel on the same processor.
So, if you are using a very int or very fp intensive, multi-threaded app, turn off hyperthreading. If you are a typical user running many programs that use both int and fp, then turn it on.
It's not so cut and dry when you consider that we aide what the rest of the world considers a terrorist nation.
Re:Thats it, people.
on
NES PC
·
· Score: 2, Funny
Just because the linux people want to be like NetBSD doesn't mean they are wrong.
Re:Hairspray is for girls
on
Potato Bazookas
·
· Score: 4, Funny
And after you done with wussy air, you move on to better things like CO2. We had a friend at a welding supply shop that got up 80lb tanks of CO2. Hook that up to a potato cannon, add a 3/4 turn brass valve and you've got a lot better cannon than air.
The initial test of it shot it out the door of the place i was working, over the parking lot (12 cars), across 5 lanes of traffic, over a Kroger's and associated parking lot, and into the field behind it. We deemed it a success.
Slashdot Mirror
Right, but the side receiving the signals from those in the disaster area aren't in a disaster area. BPL noise interferes with the reception of the emergency traffic.
I don't know about the majority of schools, but at my school, which has a quite good CS program, the professors are brilliant, but cannot write good/maintainable code at all.
Ask them about how to approach solving a problem and they can draw the structures and describe the algorithm, but if they can write the code, it's atrocious.
at least I can still use else if and elif
Thanks for helping get rid of that elsif abomination!
While you are correct that clusters are not the ultimate solution for high performance computing, single-image computers are not a great solution either. They require specific optimizations to be done for the particular system and do not lend for easy system upgrades.
What you are missing is that hardware is the least of your worries when porting. The OS will have different calls to do things. The binary format is different. The same type of function for the OS may do it differently. And so on.
Porting from one OS to another even on the same architecutre is a pain the ass.
I run a gentoo mail server and there have been a few package updates that broke it. Part of it has to with using more obscure packages (pam-mysql, etc), but still, a package shouldn't be marked stable until it's been tested. For some of the server stuff, they just don't test it.
So if two people are having a conversation by shouting across a plaza and I walk to a spot where I can hear one or both sides, I'd be wiretapping?
I know the FCC likes to claim that signals broadcast are not public property, but when you broadcast something, anyone within range is going to pick it up. Why would that be a violation of wiretapping? It's not like you actually tapped into the system at all. It was already giving you the conversation.
You misunderstood the parent. Some earlier versions of distributions assumed that if you wanted to automatically partition a drive that you wanted to erase everything on it rather than just use the free space. Lots of new users accidentally wiped their windows partitions because they didn't know the consequences.
I put linux on my Xserve. It's a server. Linux fit the problem better. No GUI overhead, easier to maintain, etc.
For G5s, the person generally is either using it as a server or for sciTech work, such as long simulations or massive calculations.
The sciTech programs are typically written on linux x86 systems and then someone else gets the job of finding the best system to run them. Putting linux on a G5 ends up being easier than getting the initial developer to port the app. Not to mention it gives a consistant interface for all the systems regardless of architecture. This has the nice side effect of making it easier for the admins to maintain the systems and keep up to date on whats new.
I not sure what's worse, the fact that you took the time to write such a long satirical post, or that you actually were moderated Informative for a while.
> Every other person in developed countries has a pocket-sized radio that will let them talk to anyone in the world.
Of course, amatuer radio is present even in undeveloped countries and in some cases is the only way to communicate with someone other than the locals.
Don't assume that simply because the Internet has changed how communication is done that it is the end-all solution. What happens when an attack brings down the core routers? The Internet isn't quite as capable of routing around failures as everyone hopes.
As for commercial radio, even with walkie-talkies, you can't even come close to the distance, variety of services, independence of infrastructure, or ease of setup that amatuer radio offers.
Obviously you've missed the part about amatuer radio being a sanctioned FCC emergency service. It's there when others fail.
OT, but ECC (elliptic curve cryptography) is as strong as RSA with smaller keys. A 160-bit ECC key is just as strong as a 1024-bit RSA key.
This is possible since the underlying problem in ECC is much more complex and requires more time to attack for a given key length.
Put on OS X, CHUD includes one of the best tools for profiling, Shark.
Quite simply, I have things on my wired home network that I don't want anyone on my AP to access. Using a linux box to handle routing and firewalling between the Internet, wired, and wireless networks does something that software firewalls (like ZoneAlarm) can't do and that would cost over $300 for a hardware firewall to do the same.
If i've already got an old machine laying around from my last upgrade, why waste money on the hardware firewall?
Maybe you should take a minute and actually look at what _you_ are comparing. The scores quoted by Apple for the Intel machines explicitly stated that GCC was used as the compiler to remove the compiler as a variable. The published scores on SPEC's website do not. You are comparing cows to telephone poles.
Now, comparing G5 results with GCC to x86 results with GCC is pretty fair when you are comparing the hardware only, not the software. That is what they claimed to be showing.
Actually, there is a certain period of delay between letters and words. Thus it is quite easy to determine what is intended simply by listening for the pause.
As noted earlier, both systems used GCC 3.3. That is where most of the loss can be attributed.
As for the question about HyperThreading. On a SPEC int or fp run, the test is single threaded. HyperThreading in this case only serves to help with the OS overhead a bit.
In the case of a SPEC rate run, hyperthreading actually hurts performance. This is due to the fact that to fully load down the system, a copy of SPEC is run for each processor the system sees (in the case of a Dual Xeon with HT, 4). Considering SPEC int and fp tests use all the available units of the respective type, HT would be unable to do anything useful as a single thread would be using all the particular units and the second thread on the processor gets stalled.
So, if you disable HT and run only 2 copies of SPEC for a rate run, the overall throughput is greater because the threads complete faster since they aren't stalling.
Having just done some serious testing on a couple of hyperthreading capable machines (dual Xeon 3.06GHz and 3.0GHz P4), I can say a bit about it's effects on programs. If the code is multi-threaded (I didn't read the article to see if his is and this is meant to be really general) it will be distributed over all the "processors" equally. This works great for programs that have 2 very different threads. However, for an app that is very int or very fp intensive in multiple threads, hyperthreading actually hinders overall throughput.
This is due to the fact that hyperthreading is still limited to the number of functional units in the processor. For code that is very intensive on a particular type of unit (int or fp), you basically end up with a stall condition on the virtual processor while all the functional units of that type are used by the first processor.
Hyperthreading is better suited to cases such as a user using a 3d modeling program and a MP3 player. The MP3 player will hopefully end up on one virtual processor and use the int units while the 3d modeling will end up on the other and use the fp units. This would allow both to run in parallel on the same processor.
So, if you are using a very int or very fp intensive, multi-threaded app, turn off hyperthreading. If you are a typical user running many programs that use both int and fp, then turn it on.
Of all the times for me not to have mod points.....
And we support Israel because.......
It's not so cut and dry when you consider that we aide what the rest of the world considers a terrorist nation.
Just because the linux people want to be like NetBSD doesn't mean they are wrong.
And after you done with wussy air, you move on to better things like CO2. We had a friend at a welding supply shop that got up 80lb tanks of CO2. Hook that up to a potato cannon, add a 3/4 turn brass valve and you've got a lot better cannon than air.
The initial test of it shot it out the door of the place i was working, over the parking lot (12 cars), across 5 lanes of traffic, over a Kroger's and associated parking lot, and into the field behind it. We deemed it a success.
Now, propane we were a bit leary to try.
But you could have asked for no peas when you ordered couldn't you?
Which is now known as Tru64. Of course, depending on what manual page you look at it, the name varies from OSF/1 to Digital UNIX to Tru64.