Actually, I think the fact that the metro is a service provided to you is reason enough to justify the searches of bags. Its an agreement you make when you agree to use the subway.
If what you said is true about this being an unreasonable search, then it could also be said that checkpoints for drunks along the highways is unreasonable. By the same means the fact that a visitor visiting a US government intelligence facility should not be searched because that is an unreasonable thing to expect.
The 4th ammendment protects you in your home and in your private life. It does not protect your privacy at someone elses house. If someone wanted to watch every action you took in their residence that is legal, however I think in most states you must first warn them that they're being recorded etc etc. The same goes when you are on corporate grounds, and the same can be implied about a quasi governmental corporation such as the MTA.
The "slippery slope" you discuss is trying to protect rights you do not have. Saying soon we'll have big brother checkpoints is a laugh as we already do. Along our highways are weigh stations and fruit/vegetable inspections, drunk inspections etc etc. Don't think this hasn't been in effect for years.
First, the only difference between the Niagara processor and the rock processor are that rock supports multiple cpus on a system (as far as I know).
Next you claim that XP only supports SMP/CMT/CMP . . . This is true, but that implies it will support any number of threads simultaneously. As far as the OS is concerned the big problem is removing race conditions that will crash the OS. These occur no matter how many threads are running simultaneously (on hardware) as long as its more than one. Theres no reason these OS's won't support this new chip, however it may not natively run faster on them (as it might not have enough execution threads, but thats okay as its the OS and not the application).
As for not being appealing to the general public, I'm not sure. The general public wants numbers, not performance on their applications. If Intel calls this the Intel 32x4Ghzx64bit processor the public will be like "thats got to be better". The architecture is pretty impressive for throughput, but it requires applications to support it.
For games and stuff those will become supported. If you don't think so, look at the xbox 360 and the ps3. If your games isn't multithreaded its performance WILL suck. Multithreading is being forced upon us. We have the option now of 2 fast processors, or 4 processors that run at 75% of the speed of the original two. The answer becomes obvious. these are both the answer to HPC and consumer applications, it just happens that most of the thread contexts will be idle on consumer hardware (or just run {spy,mal}ware).
No mention of intel threading tools were mentioned, although I imagine some are in the works. How helpful these will be is debatable. Much more work is needed to improve TLP. I've been doing research on threading models for database applications (http://www.lehigh.edu/~pcg2 has some information on it). However database applications tend to be parallelized easier than others (although my work is trying to parallelize them further for future architectures as well as compare workloads).
While TLP requires much programmer effort, if an application is well designed it may not be as difficult as thought. You just need to located the primary sections of code and see what can be done, or create pipeline code that supports N-threads running on each stage of the pipeline. Its kind of weird to do, but can work remarkably well, especially when communication overhead is minimal.
not quite actually. The big difference being that the cell architecture has a host processor, and many smaller sub processors. The subprocessors have a backwards memory model (which seems extremely confusing) in that each one has a scratchpad memory. Also each sub processor has a limited instruction set.
What I described consists of many identical processors although the possibility exists for a chip to have one high-ILP core and many high throughput cores to optimize for both single and parallel app). However even with asymetric cores like that they'd all be capable of running any x86 instruction.
Phil
Intel stole the EM64T technology? While I'm not an Intel fanboy, and I own an AMD machine, saying this is a lie. Both AMD and Intel agreed to make x86 processors for IBM (with AMD the backup supplier). It was agreed upon that the architecture was shared by them. Simply making a processor that uses another architecture is not a big deal. AMD uses so many Intel x86 architecture features its not even funny.
They have MMX/SSE/SSE2/SSE3 instructions on their chips. Those are arguably more complex (and important for fp apps) than the 64 bit instructions. How often does a 32 bit integer overflow.... with the exception of memory address instructions (which is a big incentive for x86-64).
As far as putting a memory controller on die... That might be patented by AMD, but if it is it's either covered by a crosslicense agreement between AMD and Intel or is a patent that shouldn't exist. The idea is simple, both companies have made memory controllers and both have made cpus with buses going to the memory controller. It's not that difficult to combine them on a single chip, ASIC's have done similar things for years.
No... I doubt they'll be using the Pentium M core for this redesign. The new push will be for multithreading. The pipeline may shrink a bit, but long pipelines are nice because they allow for very high clock speeds due to low fanouts. When designing high power software going from 4 threads to 16 is often not too difficult. At least if you use the right paradigms. Combined with low-latency communication (L2 cache speeds) this makes for a very powerful combination.
When designing such a machine its important to consider what the software will look like. Is it better to run 16 threads each with a CPI (cycles per instruction) of 1.2 or run 32 threads with a CPI of 1.6? This will actually push us much further back than the P3.
The cores on these processors are far more likely to resemble the original Pentiums. Simple pipelines, in-order execution, minimal instruction level parallelism. When the current P4 superscalar beasts can rarly pull a CPI of 1, whats the point of allowing 4 instructions to execute simultaneously (at least if the core is only executing one thread).
The new push will be to have 8 very simple cores (albeit with advanced SSE4 units with even wider vector instructions such as 256 or 512 bits) and allow each core to run 2 or 4 threads. This won't be hyperthreading as hyperthreading is a form of SMT (although Intel may reuse the name). It will be a form of fine-grained multithreading that allows context switches on L1 or L2 cache misses, as well as other latent operations. Of course their will also be logic to allow all the threads to run equally.
With these processors we'll be able to run 16-32 threads simultaneously (or almost simultaneously). For applications that can be massively threaded this will result in a huge boost in performance. For the single threaded applications that aren't easily parallelizable.. . many of them don't need more power than what a simple 4GHz core can offer them. Those that require more computation than that will likely be reprogrammed to support multi-threading.
This technology will scale tremendously. These new processors will essentially be supercomputers on a chip. I think this because of a presentation I saw by one of the lead P4 architects who was talking about future processors. This will be the future, and the time is now to rethink any applications you currently have and find someone competent in multithreading.
I disagee with your statment that bad handwriting is a sign of lazyness. While bad handwriting is often a sign of lazyness, that is not always the case.
I grew up with parents who did whatever they could to make my handwriting better. It was always pretty horrendous, and even when I tried hard and spent time on every letter it looked sloppier than many peoples bad handwriting. Did this make me lazy?
There was a time when I realized (and eventually convinced them) that there was no point fighting it. My handwriting sucks, I have poor hand eye coordination, etc. Its not a huge deal... If you know how to deal with it. I take mountains of notes by hand and find that I can read most of them, but I don't need to. My notes are normally taken to help me absorb the material as I listen. However if I take the time I can make something thats legible (albeit messy) for others to read.
Such a generalization on bad hand writing is simply not fair to the many of us who can't write well no matter what we do.
Phil
Will this really spell the death for the ISP like speakeasy? Sure the bargain basement DSL providers might be hit, but the telcos know that some people want and are willing to pay for the better support. However they may not have the will to run that. Especially because adding more choices generally confuses teh consumer.
If speakeasy pays the dsl company enough such that the dsl makes MORE profit from renting a dsl line to speakeasy than renting it directly to a customer there's no reason I can think of that they'd disallow it. Speakeasy is a premium service and costs more for a reason. Mostly because it works better than most other options. If people pay the extra money I think the telcos will let speakeasy be, and just make more profit off those customers.
As far as competition goes, this does put DSL providers on a more level playing field with cable modems. If you had the option of spening hundreds of millions of dollars worth of fiber to peoples homes would you do it knowint that the second you do it every other company can now pay you simple usage fees and use your lines? Where is the incentive in that? Where do the companys make back the money they invested? If the profit isn't there it is both immoral and illegal for the CEO to try to implement such a policy. They were elected to their position to make money, not piss it away.
I haven't actually used it enough to compare it, and I have no real experience with Illustrator. I don't have too great of needs when it comes to vector graphics. Mostly making diagrams consisting of circles, arrows, boxes and text. For that purpose you really don't need anything too fancy, but I found OmniGraffle much easier to use and line up objects to make them look good than Dia (although running dia remotely on a dual xeon vs OmniGraffle on my ibook was a big plus for dia).
As for CMYK, I know there are options for it. I haven't played with it much. Check out http://www.omnigroup.com/applications/omnigraffle/ for more details. While its OSX only, I presume you use those enough if youre involved with desktop publishing you have macs available. I'd reccomend downloading their free demo and giving it a try. If nothing else its much cheaper than illustrator depending on your needs.
I've found that for producing vector figures (mostly for research papers) OmniGraffle is pretty amazing. Its not free, but supports far more features than dia (such as helping you auto align and create symmetric figures). Also the interface seems nice and well done.
I must say I thought about getting an iRiver and some of those other music players in the past. They're generally cheaper than ipods, and they support oggs.
Add onto that the fact that I had all my music in.ogg format and it was looking quite nice. However back in October I bought myself an ibook. Now that I wasn't running linux I realized that oggs really are annoying. No one else can listen to them (without you explaining how to do it), itunes didn't support them (and after upgrading to 10.4 there's no longer a quicktime plugin to use). and a myriad of other reasons got to me. I also started to love apple products... they just work.
Anyhow, I then went through the process of ripping my cds once more to mp3s. I must say that it wasted a lot of hard drive space (going for q5 ogg's to 192kbps mp3s). Where an album of oggs took up 40MB the mp3's now took up 70MB. It was significant, but I had enough hard drive space. I've now converted over 100 of my cds to mp3s (and have many many more to go).
Just today actually I broke down and ordered an iPod. Why one of those over an iriver? Mostly because the ipod just works. It works great with itunes, its small, and has a larger capacity (60GB). Not only does iriver not have a hdd mp3 player of that size, but their 20 gb one is thicker than the ipod (although not by much). On top of this I just bought a new car stereo that is ipod compatible. Now I'll be able to use my ipod on trips and control everything from the car stereo. No amount of open standards will allow another device to do the same thing, only market share and demand.
While I loved using oggs, and wish the world switched to them, I've forced myself to stop using them. The rest of the world sometimes dictates actions.
phil
Well I have reasons to believe those 3 might happen.
For reason number one about taxes on the internet, read the article. It said they are considering adding that. And knowing the UN's track record it would likely be a tax on developed nations to help build less devloped (ie often tyranical, undemocratic and socialist/communist) nations build network infrastructure (all the while lining the pockets of those in charge).
For reason number 2, disallowing countries... I don't know how this would work, but lets say china doesn't want taiwan recognized as a nation. They can lobby to bring it down. China may also want to stop american ISPs from blatantly trying to get around firewalls to post information that is pro-freedom. Now they have power to really stop it.
For reason number 3... The UN has no rules protecting freedom of speach. Most countries on this planet have no protection of it. The few we think of that do are vastly outnumbered by the many other countries no one really cares about. Even democratic countries in europe often censor things (not to say the US is perfect but we have yet to go all out on censoring politically motivated speach, most of the cases for free speach that have been disputed here are related to claims of obscenity which is not protected by the constitution).
Comparing the motivation of china, libya, korea, etc to those of the US to enact these things is not a valid argument. I don't need to be paranoid to think that the Chineese don't have american interests at heart. Saying these are tin foil hat claims is absurd. It is valid to worry about how the UN would handle something when they put countries like Libya in charge of the committe for Human Rights. The UN is full of PC crap that would make a disaster.
If you can give me an example where the UN took on a responsibility as large as this, and acted fairly I might understand your argument. However I can't think of any. How many UN projects have involved every nation working to better every nation? Most involve some or most nations working to better other nations, and even many of those are massively corrupt.
Our government may be corrupt, but its managable. It has accountability to our citizens, and our businesses, so there is a limit to the corruption. Plus the constitution still stands behind the government. Sure the activist judges on the courts are working hard to erode those rights, but thats a different matter altogether.
1. What do we charge for use of the TLDs? Someone likely charges access to the tier 1 networks, but thats different. If you think too much is charged for IP addresses and domains, I'd still say they're reasonable.
2. Uh, there is already a.cu domain name. Has been for a while, and I haven't heard of any action about their to try and remove it.
3. There's been no talk of it yet. There are tons of anti-war websites out there (sometimes I feel slashdot is trying to become one). There has been no effort to remove them. Maybe efforts by people to dispute their claims, etc. However those are completely different. Also we do have a 1st ammendment. . . unlike certain international organizations who think china has a better human rights record than the US.
Saying the problems are unchanged is a joke. You gave 3 examples, all of which HAVE NOT HAPPENED in the lifetime of the ineternet so far, and there is no reason to believe they will. And the UN might not change these things, but then again I have no reason to believe it wouldn't. I have no reason to believe the UN will act fairly, no reason to think the UN wouldn't tax it for their own purposes ($$$), no reason to think they wouldn't want to censor it, and no reason to think they wouldn't ban countries they dislike.
Of course if you could give me reasons to trust the UN (that are legitimate) I might consider the idea of changing who owns the internet... . Oh that and specific examples of what ICANN has done so poorly.
The UN is more democratic? The first question is what part of the UN would run it?
I could just imagine the UN Security council trying to run it.... Wait, nothing would ever happen because you need unanimous consent of all nations involved to do anything. Getting some of these to agree is a far fetched notion.
Of course we could always let the general assembly run it. There's a brilliant idea. Give the United States as much say over the internet as every other tiny country in the world. Thats fair and democratic?
They could also do it by population. In that case China has a huge advantage. That would be great.
Under any system we'd allow the wrong people to get our hands on it. Is letting china, libya, cuba, north korea etc telling the rest of the wold what to do with the internet really the definition of DEMOCRATIC PROCESS?
A big part of the problem is the UN has no accountability. When the UN starts using it to push their viewpoint (as the topic said universal net access) what then. What do we do when the internet becomes a vehicle for corruption? Who do we call and say change this? Someone will be getting rich while the internet collapses. Currently ICANN doesn't have the power to tax the net like this, or to create filters etc. In the hands of the UN. . . who knows what power it'll have. The UN has zero accountability. If ICANN tryied this now they'd be stopped in a second.
What I find the most amusing about all of this is how so many Europeans are all about this idea. As if they'd actually have a say over it? It wouldn't be the EU's internet, it would be the world. Under the UN that means security council or general assembly. Tell France or Germany that Uganda has as much say over the nets infrastructure as they do. Or that China has more say (due to bribing other countries) or whatever. The EU would lose out on the deal, but the only possible thing that would make them like it is the fact that it hurts the US more than it hurts them (kind of like Kyoto). Stabbing yourself to hurt the US is not a good idea.
While maybe some more international control could be used for the internet, I would say that there is no reason for the UN to have any say over it. The UN is corrupt and getting worse. There's zero accountability. Whats the best option then? Well tell us legitimate problems with the internet as its being run and maybe we'll examine them on that basis.
Actually its likely that what the band is doing isn't 100% legal either. Most bands don't own their works completely. The label does. Its one of the shitty sides of the industry. Some bands have enough pull that they can offer stuff like mp3s for free, but oftentimes the record label gets nervous about that stuff (as the band isn't likely to make a dime off cd sales, but they will make more money if people here their stuff and goto their concerts). Normally the labels are more civil to their artists, but it all depends.
my ibook regularly will get about 5 hours of life on it. Granted thats with the screen brightness turned down to the lowest level (while still being visible. But I'm also using wireless networking and doing normal stuff like running mail, having 5 windows open in safari, running aim, xchat, etc. Granted I'm not using it to play fancy games or anything on battery life (but who does). My machine also has 768 megs of ram that chew up power... of course that might be offset by the fact that my hard drive is less likely to spin due to lack of disk thrashing.. . . who knows. Also generally listening to it mobily I keep the volume muted.
however there are times when I'll go outside to use it where I have to keep the brightness at the highest setting and I leave sound on etc, and even then I get about 3 and a half hours of usage out of it.
And if Americans are allowed to buy prescription drugs for cheaper from Canada should the lawmakers who changed the law be charged for the deaths of people 10 years down the road for medicines that were never researched because the drug companies could no longer make the profits that allowed them to spend billions of dollars per year in research?
In that case I guess they're damned if you do damned if you dont.
The problem is that no matter how effiecient you design your ALU you can not account for the fact that a significant portion of many applications runtimes are waiting on the memory hierarchy. Main memory costs are expensive (in terms of cycles). Current trends are pushing clock speeds up, and so the procesorry/memory gap will continue to widen.
This is true because P ~= A*C*V^{2}*F where A=die area, C=capacitance, V=voltage and F=frequency. Traditionally area was the one constant in this equation (making larger dies doesn't make economic sense). However now P is the real constant. F just has to be scaled back in its growth rate, and within 5-10 years (when V is likely to stop decreasing) it will be scaled back further.
However we will see increasing processor frequencies. With that will be a larger processor/memory gap. And while some applications are cache aware, most programmers are not going to program with cache in mind, and even when they do L1 through Ln cache levels have a latency associated with them. Allowing 4 threads to quickly shuffle around the core on an L1 cache miss just makes sense. Regardless of how awesome your ALU is this will improve performance. While 4 threads per processor may not yield more performance than 3 or 2 (depending on what the threads are doing), the performance of the system as a whole is no worse than it would otherwise be.
I for one welcome these new multi-threaded monsters.
This is true. On a 500MHz machine OOP makes a huge difference. However when we move to a 4GHz machine that requires 400 cycles to access main memory, 25 cycles to access L2 cache and 4 cycles to access L1 cache, the difference between OOP and in-order starts to fall away.
Even the best code on the best processors of today aren't getting a huge speedup from OOP. Also just because the processor is in order doesn't mean a memory/fp/int instruction can't all be run in parallel depending on how its designed (however they must be retired in order).
The primary factor however is the memory hierarchy. If most applications are waiting on main memory or cache half of the time, even the most efficient processing can only speedup the processor by 50% (Amdahl's law).
Phil
in this case there isn't really a problem. Considering that if you're compiling more than 1 file the compilation is parallelizable. If you just change one file than only that one file has to be compiled. If the code in that file is so complex that a single simple core running at 4GHz takes too long to compile the file than you need to reexamine how you're writing your code. If it takes long enough that its annoying something is wrong.
However that being said, I'm not really sure how gcc divides up the building phase. If thats not parallelizable than there is a good chance you may have to wait, I have no idea.
Actually from what I've heard, the entire industry is moving in this direction. The whole idea of out of order processors (OOP) has become outdated. OOP was great. Enabled massive single threaded performance, however the costs (in terms of area and heat dissipation) is enormous.
I just came back from the DaMoN workshop where the keynote was delivered by one of the lead P4 developers. He explained the future of microprocessors and said that the 10-15% extra performance that OOP enables just isn't worth it. The Pentium 4 has 3 issue units, but the way things are rarely issues more than 1 instruction per cycle.
We can squeeze more performance out of them, but not much. The easiest method is to go dual core. However if an application must be multithreaded to enable the best performance, what would you rather have . . . 2 highly advanced cores, or 8-10 simple cores that can issue half as many instructions per cycle as the dual core design. Than consider the fact that each core enables 4 threads to run (switch on cache miss/access). It doesn't take a rocket scientist to see that overall performance is improved with this.
The other option is the hybrid core. A single really fast x86 core combined with multiple simpler x86 cores. That way single threaded apps can run fast (until they're converted) and you can get overall throughput from the system without blowing away your power budget on OOP optimizations.
Granted most of this is in the future (within the next 5 years), but IBM's going that way (ala Cell), its within Intels roadmap, Sun is pushing that route etc. I assume AMD has plans to create a supercomputer on a chip . . . unless they wish to be obsoleted.
At first I didn't think the Intel switch was a good idea, now I'm kind of neutral. One thing I still find odd though - why Intel of all people? Why not AMD?
While I buy AMD processors and you buy AMD processors, that doesn't make them the best under every circumstance. For apple they need guaranteed production, guaranteed availability of chipsets, guaranteed compatibility, and almost equally important want access to the Intel's compiler for xcode.
Intel's compiler from the start gives a sizable speedup compared to gcc. This often means that Intel processors run faster than AMD processors. Also for the transition Intel is a more well known and respected company than AMD. Once the transition is complete there's nothing to stop Apple from switching to AMD processors on some of their line, but they don't have to. With x86 (and commodity hardware) the future is much more open. Intel makes quite good processors. The current pentium 4 generation is a bit of a flop and Intel has sort of admitted it (heat issues with the Prescotts, etc). However the future seems brighter, and is likely a path Apple wishes to pursue.
This idea actually has a lot of potential. Processors no longer run their native instruction sets. Some processors like the Itanium are closer to it, but the x86's core is a risc core. As such it would likely be easier for Intel to change the instruction decode logic to cover it.
I'm not sure how the decode logic works on the Pentium M, but on the P4/Xeon's it uses a tracecache, which essentially eliminates decode logic further down the pipeline (and even on most instruction loads). If this exists in the P-M this could be relatively "simple" to implement by 2006/2007. Its a lot of code to fix, and the bytesex of the processors will have to be a major concern.
The other quesstion is who owns transmeta's IP? Something similar could be used for the initial transition and later change to a more pure PPC solution. This is purely speculation, but knowing how todays processors are designed, and how much of a pain shifting architectures is, wouldn't it make more sense to change the processor to support the ISA than to change the billions of lines of code to support the ISA? The era of reconfigurable computing is rapidly approaching, and while this isn't really reconfigurable computing a dual mode PM might not be far off (just specify upon boot, or even at program loadtime whether the machine is emulating the x86 or the PPC ISA).
It really depends what you want. I have 2 19inch CRTs on my desk at home. I love the things, they're sharp, look good, run at high resolutions etc. However they have their drawbacks. As a college student who tends to move a couple times a year, the idea of lugging those things up 2 flights of stairs is painful. Also moving a CRT is a good way to break them. One of mine is starting to blur and the color doesn't look as good as it used to. As its a high quality CRT (Samsung 900IFT) the only plausible reason is that I've moved it multiple times which results in bouncing them around more than they should.
Also the cost of CRTs to most people is greater. If you are a business considering either 17inch CRTs (or even possibly 19in ones) or 15-17in LCDs, the LCDs are a highly attractive option.
Sure the initial cost in an LCD is greater, but the average person can move an LCD themselves (moving CRTs can be painful). On top of that they use less power, and save desk space. Maybe your home desk space isn't lacking, but in many offices it is.
As far as the future goes... how much innovation has there been in CRTs in the past 5 years? For the most part its stagnated. Maybe slightly higher resolutions, flatter tubes have become standard etc. With LCDs we've seen brighter lcds, fewer dead pixels, higher resolutions etc. They can't quite match CRTs (yet), but at the rate they're improving they'll be there soon.
I'm even tempted to make the switch. Its funny because I don't game much, but I don't mind occasional lan parties. While I'd like my 19in CRT at a lan party, the idea of lugging that thing around is far from appealing. Instead I'll borrow a 15in LCD from lab and save my back.
I'm all in favor of IPv6 acceptance, and I'd like to start seeing it deployed in parallel with IPv4, and waiting for the advances to happen. But I'd like to understand why the ISPs haven't done it yet on purely capitalist grounds. Maybe it's because it's just too little to justify the expense.
On purely capitalist grounds the ISP's don't want to do this. Especially the Tier 1 guys who own the majority of IP addresses. Its the laws of supply and demand. If they keep the supply of IP addresses constant (which it is now) and demand for them keeps increasing (regardless of NAT routers, every cable modem gets an ip, as does every dsl modem, etc otherwise your local ISP would get too many complaints that XYZ doesn't work) they can charge more per IP address. The law of supply and demand dictates that buying extra static IP addresses cost as much as they do. Well, that and the fact that few people want it so it costs the ISPs extra to enable it for one person specifically.
What do you think will happen as more ips are demanded. These major backbone ISPs charge more for addresses. Of course they can only go so far before new Tier 1 IPv6 ISPs come into existance, but that will take a long time. If you have a monopoly on a limited resource thats in high demand (thats constantly increasing) would you give it up?
Slashdot Mirror
Actually, I think the fact that the metro is a service provided to you is reason enough to justify the searches of bags. Its an agreement you make when you agree to use the subway.
If what you said is true about this being an unreasonable search, then it could also be said that checkpoints for drunks along the highways is unreasonable. By the same means the fact that a visitor visiting a US government intelligence facility should not be searched because that is an unreasonable thing to expect.
The 4th ammendment protects you in your home and in your private life. It does not protect your privacy at someone elses house. If someone wanted to watch every action you took in their residence that is legal, however I think in most states you must first warn them that they're being recorded etc etc. The same goes when you are on corporate grounds, and the same can be implied about a quasi governmental corporation such as the MTA.
The "slippery slope" you discuss is trying to protect rights you do not have. Saying soon we'll have big brother checkpoints is a laugh as we already do. Along our highways are weigh stations and fruit/vegetable inspections, drunk inspections etc etc. Don't think this hasn't been in effect for years.
Phil
First, the only difference between the Niagara processor and the rock processor are that rock supports multiple cpus on a system (as far as I know).
Next you claim that XP only supports SMP/CMT/CMP . . . This is true, but that implies it will support any number of threads simultaneously. As far as the OS is concerned the big problem is removing race conditions that will crash the OS. These occur no matter how many threads are running simultaneously (on hardware) as long as its more than one. Theres no reason these OS's won't support this new chip, however it may not natively run faster on them (as it might not have enough execution threads, but thats okay as its the OS and not the application).
As for not being appealing to the general public, I'm not sure. The general public wants numbers, not performance on their applications. If Intel calls this the Intel 32x4Ghzx64bit processor the public will be like "thats got to be better". The architecture is pretty impressive for throughput, but it requires applications to support it.
For games and stuff those will become supported. If you don't think so, look at the xbox 360 and the ps3. If your games isn't multithreaded its performance WILL suck. Multithreading is being forced upon us. We have the option now of 2 fast processors, or 4 processors that run at 75% of the speed of the original two. The answer becomes obvious. these are both the answer to HPC and consumer applications, it just happens that most of the thread contexts will be idle on consumer hardware (or just run {spy,mal}ware).
Phil
No mention of intel threading tools were mentioned, although I imagine some are in the works. How helpful these will be is debatable. Much more work is needed to improve TLP. I've been doing research on threading models for database applications (http://www.lehigh.edu/~pcg2 has some information on it). However database applications tend to be parallelized easier than others (although my work is trying to parallelize them further for future architectures as well as compare workloads).
While TLP requires much programmer effort, if an application is well designed it may not be as difficult as thought. You just need to located the primary sections of code and see what can be done, or create pipeline code that supports N-threads running on each stage of the pipeline. Its kind of weird to do, but can work remarkably well, especially when communication overhead is minimal.
Phil
not quite actually. The big difference being that the cell architecture has a host processor, and many smaller sub processors. The subprocessors have a backwards memory model (which seems extremely confusing) in that each one has a scratchpad memory. Also each sub processor has a limited instruction set. What I described consists of many identical processors although the possibility exists for a chip to have one high-ILP core and many high throughput cores to optimize for both single and parallel app). However even with asymetric cores like that they'd all be capable of running any x86 instruction. Phil
Intel stole the EM64T technology? While I'm not an Intel fanboy, and I own an AMD machine, saying this is a lie. Both AMD and Intel agreed to make x86 processors for IBM (with AMD the backup supplier). It was agreed upon that the architecture was shared by them. Simply making a processor that uses another architecture is not a big deal. AMD uses so many Intel x86 architecture features its not even funny.
They have MMX/SSE/SSE2/SSE3 instructions on their chips. Those are arguably more complex (and important for fp apps) than the 64 bit instructions. How often does a 32 bit integer overflow.... with the exception of memory address instructions (which is a big incentive for x86-64).
As far as putting a memory controller on die... That might be patented by AMD, but if it is it's either covered by a crosslicense agreement between AMD and Intel or is a patent that shouldn't exist. The idea is simple, both companies have made memory controllers and both have made cpus with buses going to the memory controller. It's not that difficult to combine them on a single chip, ASIC's have done similar things for years.
Phil
No... I doubt they'll be using the Pentium M core for this redesign. The new push will be for multithreading. The pipeline may shrink a bit, but long pipelines are nice because they allow for very high clock speeds due to low fanouts. When designing high power software going from 4 threads to 16 is often not too difficult. At least if you use the right paradigms. Combined with low-latency communication (L2 cache speeds) this makes for a very powerful combination.
.. . many of them don't need more power than what a simple 4GHz core can offer them. Those that require more computation than that will likely be reprogrammed to support multi-threading.
When designing such a machine its important to consider what the software will look like. Is it better to run 16 threads each with a CPI (cycles per instruction) of 1.2 or run 32 threads with a CPI of 1.6? This will actually push us much further back than the P3.
The cores on these processors are far more likely to resemble the original Pentiums. Simple pipelines, in-order execution, minimal instruction level parallelism. When the current P4 superscalar beasts can rarly pull a CPI of 1, whats the point of allowing 4 instructions to execute simultaneously (at least if the core is only executing one thread).
The new push will be to have 8 very simple cores (albeit with advanced SSE4 units with even wider vector instructions such as 256 or 512 bits) and allow each core to run 2 or 4 threads. This won't be hyperthreading as hyperthreading is a form of SMT (although Intel may reuse the name). It will be a form of fine-grained multithreading that allows context switches on L1 or L2 cache misses, as well as other latent operations. Of course their will also be logic to allow all the threads to run equally.
With these processors we'll be able to run 16-32 threads simultaneously (or almost simultaneously). For applications that can be massively threaded this will result in a huge boost in performance. For the single threaded applications that aren't easily parallelizable
This technology will scale tremendously. These new processors will essentially be supercomputers on a chip. I think this because of a presentation I saw by one of the lead P4 architects who was talking about future processors. This will be the future, and the time is now to rethink any applications you currently have and find someone competent in multithreading.
Phil
I disagee with your statment that bad handwriting is a sign of lazyness. While bad handwriting is often a sign of lazyness, that is not always the case. I grew up with parents who did whatever they could to make my handwriting better. It was always pretty horrendous, and even when I tried hard and spent time on every letter it looked sloppier than many peoples bad handwriting. Did this make me lazy? There was a time when I realized (and eventually convinced them) that there was no point fighting it. My handwriting sucks, I have poor hand eye coordination, etc. Its not a huge deal... If you know how to deal with it. I take mountains of notes by hand and find that I can read most of them, but I don't need to. My notes are normally taken to help me absorb the material as I listen. However if I take the time I can make something thats legible (albeit messy) for others to read. Such a generalization on bad hand writing is simply not fair to the many of us who can't write well no matter what we do. Phil
Will this really spell the death for the ISP like speakeasy? Sure the bargain basement DSL providers might be hit, but the telcos know that some people want and are willing to pay for the better support. However they may not have the will to run that. Especially because adding more choices generally confuses teh consumer.
If speakeasy pays the dsl company enough such that the dsl makes MORE profit from renting a dsl line to speakeasy than renting it directly to a customer there's no reason I can think of that they'd disallow it. Speakeasy is a premium service and costs more for a reason. Mostly because it works better than most other options. If people pay the extra money I think the telcos will let speakeasy be, and just make more profit off those customers.
As far as competition goes, this does put DSL providers on a more level playing field with cable modems. If you had the option of spening hundreds of millions of dollars worth of fiber to peoples homes would you do it knowint that the second you do it every other company can now pay you simple usage fees and use your lines? Where is the incentive in that? Where do the companys make back the money they invested? If the profit isn't there it is both immoral and illegal for the CEO to try to implement such a policy. They were elected to their position to make money, not piss it away.
Phil
I haven't actually used it enough to compare it, and I have no real experience with Illustrator. I don't have too great of needs when it comes to vector graphics. Mostly making diagrams consisting of circles, arrows, boxes and text. For that purpose you really don't need anything too fancy, but I found OmniGraffle much easier to use and line up objects to make them look good than Dia (although running dia remotely on a dual xeon vs OmniGraffle on my ibook was a big plus for dia).
/ for more details. While its OSX only, I presume you use those enough if youre involved with desktop publishing you have macs available. I'd reccomend downloading their free demo and giving it a try. If nothing else its much cheaper than illustrator depending on your needs.
As for CMYK, I know there are options for it. I haven't played with it much. Check out http://www.omnigroup.com/applications/omnigraffle
Phil
I've found that for producing vector figures (mostly for research papers) OmniGraffle is pretty amazing. Its not free, but supports far more features than dia (such as helping you auto align and create symmetric figures). Also the interface seems nice and well done.
Phil
I must say I thought about getting an iRiver and some of those other music players in the past. They're generally cheaper than ipods, and they support oggs. Add onto that the fact that I had all my music in .ogg format and it was looking quite nice. However back in October I bought myself an ibook. Now that I wasn't running linux I realized that oggs really are annoying. No one else can listen to them (without you explaining how to do it), itunes didn't support them (and after upgrading to 10.4 there's no longer a quicktime plugin to use). and a myriad of other reasons got to me. I also started to love apple products... they just work.
Anyhow, I then went through the process of ripping my cds once more to mp3s. I must say that it wasted a lot of hard drive space (going for q5 ogg's to 192kbps mp3s). Where an album of oggs took up 40MB the mp3's now took up 70MB. It was significant, but I had enough hard drive space. I've now converted over 100 of my cds to mp3s (and have many many more to go).
Just today actually I broke down and ordered an iPod. Why one of those over an iriver? Mostly because the ipod just works. It works great with itunes, its small, and has a larger capacity (60GB). Not only does iriver not have a hdd mp3 player of that size, but their 20 gb one is thicker than the ipod (although not by much). On top of this I just bought a new car stereo that is ipod compatible. Now I'll be able to use my ipod on trips and control everything from the car stereo. No amount of open standards will allow another device to do the same thing, only market share and demand.
While I loved using oggs, and wish the world switched to them, I've forced myself to stop using them. The rest of the world sometimes dictates actions.
phil
Well I have reasons to believe those 3 might happen.
For reason number one about taxes on the internet, read the article. It said they are considering adding that. And knowing the UN's track record it would likely be a tax on developed nations to help build less devloped (ie often tyranical, undemocratic and socialist/communist) nations build network infrastructure (all the while lining the pockets of those in charge).
For reason number 2, disallowing countries... I don't know how this would work, but lets say china doesn't want taiwan recognized as a nation. They can lobby to bring it down. China may also want to stop american ISPs from blatantly trying to get around firewalls to post information that is pro-freedom. Now they have power to really stop it.
For reason number 3... The UN has no rules protecting freedom of speach. Most countries on this planet have no protection of it. The few we think of that do are vastly outnumbered by the many other countries no one really cares about. Even democratic countries in europe often censor things (not to say the US is perfect but we have yet to go all out on censoring politically motivated speach, most of the cases for free speach that have been disputed here are related to claims of obscenity which is not protected by the constitution).
Comparing the motivation of china, libya, korea, etc to those of the US to enact these things is not a valid argument. I don't need to be paranoid to think that the Chineese don't have american interests at heart. Saying these are tin foil hat claims is absurd. It is valid to worry about how the UN would handle something when they put countries like Libya in charge of the committe for Human Rights. The UN is full of PC crap that would make a disaster.
If you can give me an example where the UN took on a responsibility as large as this, and acted fairly I might understand your argument. However I can't think of any. How many UN projects have involved every nation working to better every nation? Most involve some or most nations working to better other nations, and even many of those are massively corrupt.
Our government may be corrupt, but its managable. It has accountability to our citizens, and our businesses, so there is a limit to the corruption. Plus the constitution still stands behind the government. Sure the activist judges on the courts are working hard to erode those rights, but thats a different matter altogether.
Phil
1. What do we charge for use of the TLDs? Someone likely charges access to the tier 1 networks, but thats different. If you think too much is charged for IP addresses and domains, I'd still say they're reasonable.
.cu domain name. Has been for a while, and I haven't heard of any action about their to try and remove it.
.. . Oh that and specific examples of what ICANN has done so poorly.
2. Uh, there is already a
3. There's been no talk of it yet. There are tons of anti-war websites out there (sometimes I feel slashdot is trying to become one). There has been no effort to remove them. Maybe efforts by people to dispute their claims, etc. However those are completely different. Also we do have a 1st ammendment. . . unlike certain international organizations who think china has a better human rights record than the US.
Saying the problems are unchanged is a joke. You gave 3 examples, all of which HAVE NOT HAPPENED in the lifetime of the ineternet so far, and there is no reason to believe they will. And the UN might not change these things, but then again I have no reason to believe it wouldn't. I have no reason to believe the UN will act fairly, no reason to think the UN wouldn't tax it for their own purposes ($$$), no reason to think they wouldn't want to censor it, and no reason to think they wouldn't ban countries they dislike.
Of course if you could give me reasons to trust the UN (that are legitimate) I might consider the idea of changing who owns the internet.
Phil
The UN is more democratic? The first question is what part of the UN would run it?
I could just imagine the UN Security council trying to run it.... Wait, nothing would ever happen because you need unanimous consent of all nations involved to do anything. Getting some of these to agree is a far fetched notion.
Of course we could always let the general assembly run it. There's a brilliant idea. Give the United States as much say over the internet as every other tiny country in the world. Thats fair and democratic?
They could also do it by population. In that case China has a huge advantage. That would be great.
Under any system we'd allow the wrong people to get our hands on it. Is letting china, libya, cuba, north korea etc telling the rest of the wold what to do with the internet really the definition of DEMOCRATIC PROCESS?
A big part of the problem is the UN has no accountability. When the UN starts using it to push their viewpoint (as the topic said universal net access) what then. What do we do when the internet becomes a vehicle for corruption? Who do we call and say change this? Someone will be getting rich while the internet collapses. Currently ICANN doesn't have the power to tax the net like this, or to create filters etc. In the hands of the UN. . . who knows what power it'll have. The UN has zero accountability. If ICANN tryied this now they'd be stopped in a second.
What I find the most amusing about all of this is how so many Europeans are all about this idea. As if they'd actually have a say over it? It wouldn't be the EU's internet, it would be the world. Under the UN that means security council or general assembly. Tell France or Germany that Uganda has as much say over the nets infrastructure as they do. Or that China has more say (due to bribing other countries) or whatever. The EU would lose out on the deal, but the only possible thing that would make them like it is the fact that it hurts the US more than it hurts them (kind of like Kyoto). Stabbing yourself to hurt the US is not a good idea.
While maybe some more international control could be used for the internet, I would say that there is no reason for the UN to have any say over it. The UN is corrupt and getting worse. There's zero accountability. Whats the best option then? Well tell us legitimate problems with the internet as its being run and maybe we'll examine them on that basis.
Phil
Actually its likely that what the band is doing isn't 100% legal either. Most bands don't own their works completely. The label does. Its one of the shitty sides of the industry. Some bands have enough pull that they can offer stuff like mp3s for free, but oftentimes the record label gets nervous about that stuff (as the band isn't likely to make a dime off cd sales, but they will make more money if people here their stuff and goto their concerts). Normally the labels are more civil to their artists, but it all depends.
Phil
my ibook regularly will get about 5 hours of life on it. Granted thats with the screen brightness turned down to the lowest level (while still being visible. But I'm also using wireless networking and doing normal stuff like running mail, having 5 windows open in safari, running aim, xchat, etc. Granted I'm not using it to play fancy games or anything on battery life (but who does). My machine also has 768 megs of ram that chew up power... of course that might be offset by the fact that my hard drive is less likely to spin due to lack of disk thrashing.. . . who knows. Also generally listening to it mobily I keep the volume muted.
however there are times when I'll go outside to use it where I have to keep the brightness at the highest setting and I leave sound on etc, and even then I get about 3 and a half hours of usage out of it.
Phil
And if Americans are allowed to buy prescription drugs for cheaper from Canada should the lawmakers who changed the law be charged for the deaths of people 10 years down the road for medicines that were never researched because the drug companies could no longer make the profits that allowed them to spend billions of dollars per year in research?
In that case I guess they're damned if you do damned if you dont.
Phil
>
The problem is that no matter how effiecient you design your ALU you can not account for the fact that a significant portion of many applications runtimes are waiting on the memory hierarchy. Main memory costs are expensive (in terms of cycles). Current trends are pushing clock speeds up, and so the procesorry/memory gap will continue to widen.
This is true because P ~= A*C*V^{2}*F where A=die area, C=capacitance, V=voltage and F=frequency. Traditionally area was the one constant in this equation (making larger dies doesn't make economic sense). However now P is the real constant. F just has to be scaled back in its growth rate, and within 5-10 years (when V is likely to stop decreasing) it will be scaled back further.
However we will see increasing processor frequencies. With that will be a larger processor/memory gap. And while some applications are cache aware, most programmers are not going to program with cache in mind, and even when they do L1 through Ln cache levels have a latency associated with them. Allowing 4 threads to quickly shuffle around the core on an L1 cache miss just makes sense. Regardless of how awesome your ALU is this will improve performance. While 4 threads per processor may not yield more performance than 3 or 2 (depending on what the threads are doing), the performance of the system as a whole is no worse than it would otherwise be.
I for one welcome these new multi-threaded monsters.
Phil
This is true. On a 500MHz machine OOP makes a huge difference. However when we move to a 4GHz machine that requires 400 cycles to access main memory, 25 cycles to access L2 cache and 4 cycles to access L1 cache, the difference between OOP and in-order starts to fall away. Even the best code on the best processors of today aren't getting a huge speedup from OOP. Also just because the processor is in order doesn't mean a memory/fp/int instruction can't all be run in parallel depending on how its designed (however they must be retired in order). The primary factor however is the memory hierarchy. If most applications are waiting on main memory or cache half of the time, even the most efficient processing can only speedup the processor by 50% (Amdahl's law). Phil
in this case there isn't really a problem. Considering that if you're compiling more than 1 file the compilation is parallelizable. If you just change one file than only that one file has to be compiled. If the code in that file is so complex that a single simple core running at 4GHz takes too long to compile the file than you need to reexamine how you're writing your code. If it takes long enough that its annoying something is wrong.
However that being said, I'm not really sure how gcc divides up the building phase. If thats not parallelizable than there is a good chance you may have to wait, I have no idea.
Phil
Actually from what I've heard, the entire industry is moving in this direction. The whole idea of out of order processors (OOP) has become outdated. OOP was great. Enabled massive single threaded performance, however the costs (in terms of area and heat dissipation) is enormous.
I just came back from the DaMoN workshop where the keynote was delivered by one of the lead P4 developers. He explained the future of microprocessors and said that the 10-15% extra performance that OOP enables just isn't worth it. The Pentium 4 has 3 issue units, but the way things are rarely issues more than 1 instruction per cycle.
We can squeeze more performance out of them, but not much. The easiest method is to go dual core. However if an application must be multithreaded to enable the best performance, what would you rather have . . . 2 highly advanced cores, or 8-10 simple cores that can issue half as many instructions per cycle as the dual core design. Than consider the fact that each core enables 4 threads to run (switch on cache miss/access). It doesn't take a rocket scientist to see that overall performance is improved with this.
The other option is the hybrid core. A single really fast x86 core combined with multiple simpler x86 cores. That way single threaded apps can run fast (until they're converted) and you can get overall throughput from the system without blowing away your power budget on OOP optimizations.
Granted most of this is in the future (within the next 5 years), but IBM's going that way (ala Cell), its within Intels roadmap, Sun is pushing that route etc. I assume AMD has plans to create a supercomputer on a chip . . . unless they wish to be obsoleted.
Phil
While I buy AMD processors and you buy AMD processors, that doesn't make them the best under every circumstance. For apple they need guaranteed production, guaranteed availability of chipsets, guaranteed compatibility, and almost equally important want access to the Intel's compiler for xcode.
Intel's compiler from the start gives a sizable speedup compared to gcc. This often means that Intel processors run faster than AMD processors. Also for the transition Intel is a more well known and respected company than AMD. Once the transition is complete there's nothing to stop Apple from switching to AMD processors on some of their line, but they don't have to. With x86 (and commodity hardware) the future is much more open. Intel makes quite good processors. The current pentium 4 generation is a bit of a flop and Intel has sort of admitted it (heat issues with the Prescotts, etc). However the future seems brighter, and is likely a path Apple wishes to pursue.
Phil
This idea actually has a lot of potential. Processors no longer run their native instruction sets. Some processors like the Itanium are closer to it, but the x86's core is a risc core. As such it would likely be easier for Intel to change the instruction decode logic to cover it.
I'm not sure how the decode logic works on the Pentium M, but on the P4/Xeon's it uses a tracecache, which essentially eliminates decode logic further down the pipeline (and even on most instruction loads). If this exists in the P-M this could be relatively "simple" to implement by 2006/2007. Its a lot of code to fix, and the bytesex of the processors will have to be a major concern.
The other quesstion is who owns transmeta's IP? Something similar could be used for the initial transition and later change to a more pure PPC solution. This is purely speculation, but knowing how todays processors are designed, and how much of a pain shifting architectures is, wouldn't it make more sense to change the processor to support the ISA than to change the billions of lines of code to support the ISA? The era of reconfigurable computing is rapidly approaching, and while this isn't really reconfigurable computing a dual mode PM might not be far off (just specify upon boot, or even at program loadtime whether the machine is emulating the x86 or the PPC ISA).
Phil
It really depends what you want. I have 2 19inch CRTs on my desk at home. I love the things, they're sharp, look good, run at high resolutions etc. However they have their drawbacks. As a college student who tends to move a couple times a year, the idea of lugging those things up 2 flights of stairs is painful. Also moving a CRT is a good way to break them. One of mine is starting to blur and the color doesn't look as good as it used to. As its a high quality CRT (Samsung 900IFT) the only plausible reason is that I've moved it multiple times which results in bouncing them around more than they should.
Also the cost of CRTs to most people is greater. If you are a business considering either 17inch CRTs (or even possibly 19in ones) or 15-17in LCDs, the LCDs are a highly attractive option.
Sure the initial cost in an LCD is greater, but the average person can move an LCD themselves (moving CRTs can be painful). On top of that they use less power, and save desk space. Maybe your home desk space isn't lacking, but in many offices it is.
As far as the future goes... how much innovation has there been in CRTs in the past 5 years? For the most part its stagnated. Maybe slightly higher resolutions, flatter tubes have become standard etc. With LCDs we've seen brighter lcds, fewer dead pixels, higher resolutions etc. They can't quite match CRTs (yet), but at the rate they're improving they'll be there soon.
I'm even tempted to make the switch. Its funny because I don't game much, but I don't mind occasional lan parties. While I'd like my 19in CRT at a lan party, the idea of lugging that thing around is far from appealing. Instead I'll borrow a 15in LCD from lab and save my back.
Phil
On purely capitalist grounds the ISP's don't want to do this. Especially the Tier 1 guys who own the majority of IP addresses. Its the laws of supply and demand. If they keep the supply of IP addresses constant (which it is now) and demand for them keeps increasing (regardless of NAT routers, every cable modem gets an ip, as does every dsl modem, etc otherwise your local ISP would get too many complaints that XYZ doesn't work) they can charge more per IP address. The law of supply and demand dictates that buying extra static IP addresses cost as much as they do. Well, that and the fact that few people want it so it costs the ISPs extra to enable it for one person specifically.
What do you think will happen as more ips are demanded. These major backbone ISPs charge more for addresses. Of course they can only go so far before new Tier 1 IPv6 ISPs come into existance, but that will take a long time. If you have a monopoly on a limited resource thats in high demand (thats constantly increasing) would you give it up?
Phil