Sorry, I should have said that AMD sold a line of 32 bit CPU's for socket 754, not that socket 754 was exclusively 32bit. There are 64bit CPUS for socket 754.
How quickly we forget what terrible choices the Intel fanboys had before Core 2 Duo shipped. CPU's that ran too hot, consumed too much power and had worse, much worse in some cases, performance than its AMD counterparts. AMD clearly had the upper hand for performance for the first time, and took advantage of it to make some much needed cash.
AMD put out 3 different socket sets to maximize their profits- socket 754 for low end, non-64 bit computing, and single channel memory, socket 939 for mainstream users, and socket 940 for server and extreme users. All marchitechture, but all forgiven because the AMD users could buy dual cores that weren't just space heaters. Yeah, the price for the good stuff wasn't any cheaper, but the benefits were so obvious that only the Intel/Dell fanboys "stayed the course" or at least, held off from buying.
Then Intel releases a near perfect CPU, great performance, good heat, medium power, just no upgrade to memory acces. Intel fanboys rejoice and finally upgrade. Middle of the roaders feel like they have a choice. AMD is suddenly left in the position it had occupied for all those years, second place. Yeah, it has a lot of options, and is still competitive for server stuff, but it's no longer a lock for the desktop user.
Amd reverts to what worked for them previously- move all desktops to the same socket and give that socket a lot of upgrade life. Since DDR2 is finally available in quanity, and at speeds that actually don't produce a slower OS than using DDR 400, AMD decides to make the change to DDR2. Save for the recent attempt to make money, AMD users have been able to buy one socket for the majority of AMD cpus available at that time, and that provides them some marginal sales, for those users who want a chance at a later CPU upgrade.
SO, socket AM2 is released at a time where it doesn't make much sense to upgrade for AMD fanboys. Intel fanboys are buying all the core 2 duo's their pocketbooks can handle, and middle-of-the-roaders are picking and choosing, just like always, versus performnce and price. AM2 is not cheaper than Intel solutions; the real deals for AMD are the clearance of older socket 754/939 stuff. Any real wonder that AM2 sales, at the moment, have been lackluster? As I see it, AMD took the long view, and released AM2 for the upcoming K8L and newer stuff. They'll take whatever sales they can get, but they aren't overly worried about sales right now. I mean, Dell is finally selling AMD's and I'd bet that AMD is waiting on that cash cow to come in.
I wouldn't worry about maxing out gigabit ethernet with only 4 ATA disks in there, unless they're RAIDed for speed or you happen to be pulling a different file off each disk simultaneously. Hell, gigabit ethernet is only an optional feature on the boxes anyway, probably because they're better suited to low-access storage/backup or light fileserving than constantly spanking them for transfer - plenty of usage scenarios that fit that nicely.
The 750GB sata disks are rated for 70MB/s sustained transfer rate. Each. Even line speed for a single gige is only 128 MB/s. I'd like to think that you be using the storage in such a fashion that you could access all of the disks at the time, whether via raid or some concurrent access patterns. The connectivity (lack thereof) is a big limiting factor when considering this box.
Yeah, I caught that the gige was optional. Sheesh.
Here's the Motherboard Info:
Motherboard/Processor:
* 1GHz VIA C3 CPU
* VIA CLE266 Northbridge
* VIA VT8237 Southbridge
* DDR266 RAM - Up to 1GB
* 2 USB 2.0 ports
* 1 Serial port
* 1 Parallel port
* 1 VGA port
* PS2 mouse & keyboard ports
Anybody have performance numbers for these units? A 1GHZ CPU can be hard-pressed to run an OS, serve disk and support a gige connection at full throughput. I'd be weary of looking at these for a data center without knowing how fast they can serve out the disk over a single gige connection. In fact, I see a distinct lack of information about this unit functions as a "storage node". Are you buying a 1U, 3.0TB node on which you need to install an OS and fileserver? Doesn't look like it would have the horsepower to run an iSCSI driver in additional to software raid drivers and still produce any real transfer speeds.
While a rack of these sounds nice in cost/wattage terms, it appears that you would have just purchased a cluster of storage nodes. A cluster of storage nodes with no way to present the available 120TB's as any kind of coherent storage space. You might be able to run Lustre, PVFS or GFS on them, if that's even possible, but that's a level of complexity the price and performance don't warrant.
If you figure in the cost of a Storage Engineer and lack of performance, this looks less appealing at the full rack level. Doesn't mean some PHB's won't buy into the the whole "Cheap Cluster Disk!" theme though. I pity the sysadmins who get a 120TB of raw disk and 40 more nodes to admin.
It's pretty simple- the constitution reserves all rights to the people or States, save those which are carefully delegated to the federal government. Amendment X specifically states this: The powers not delegated to the United States by the Constitution, nor prohibited by it to the States, are reserved to the States respectively, or to the people.
The bill of rights is not an exhaustive list- it's just a of rights which the founders thought were worth mentioning specifically. You have a right to privacy by default- at least, a right that the federal government may not abridge.
We tweak the kernel for long-haul IP traffic, certain types of error messages (thousands of warnings per thousands nodes equals gigabytes of logs per day, otherwise), and other things. We also update and build custom drivers for our gige and fiber channel cards. All of which isn't possible with the Windows versions.
We also have support contracts from our software vendors (IBM, Redhat and SuSE) in case we run into issues where some other professional opinions and expertise make a difference. We tend to get quick answers and good solutions from these vendors- I have no idea what dealing with MS tech support would be like.
Mod the parent up- this is the basis for the legality of the site.
Russian copyright law is, shudder not the same as US copyright law! What horrors! Why, if it's not legal here, courtesy of our fine Congress, it must not be legal anywhere!
What a crock of logical fallacy. That's the biggest point overlocked in this debate- Russia has no reason to conform with US laws, and if they choose not to, they aren't doing anything illegal. It's bad enough that the RIAA and MPAA can push through crap laws like the Sony Bono Copyright extension, how much worse is it that they have so many people brainwashed into believed that those laws hold everywhere.
Per Russian law, which is subject to change, what they are doing is perfectly legal. Don't like it? Tough.
Sure, for an 8 port switch, where all the computers have a direct connection. Consider the issues involved for a router with a 128 machines all trying to cross-communicate. Or larger collections of computers that might need to use multiple sets of switches to span the entire system.
On a Force10 switch, with 2 nodes on the same blade: tg-c844:~ # ping tg-c845
PING tg-c845.ncsa.teragrid.org (141.142.57.161) from 141.142.57.160 : 56(84) bytes of data.
64 bytes from tg-c845.ncsa.teragrid.org (141.142.57.161): icmp_seq=1 ttl=64 time=0.148 ms
64 bytes from tg-c845.ncsa.teragrid.org (141.142.57.161): icmp_seq=2 ttl=64 time=0.146 ms
64 bytes from tg-c845.ncsa.teragrid.org (141.142.57.161): icmp_seq=3 ttl=64 time=0.145 ms
64 bytes from tg-c845.ncsa.teragrid.org (141.142.57.161): icmp_seq=4 ttl=64 time=0.144 ms
The same nodes using a myrinet connection: tg-c844:~ # ping tg-c845-myri0
PING tg-c845-myri0.ncsa.teragrid.org (172.22.57.161) from 172.22.57.160 : 56(84) bytes of data.
64 bytes from tg-c845-myri0.ncsa.teragrid.org (172.22.57.161): icmp_seq=1 ttl=64 time=0.051 ms
64 bytes from tg-c845-myri0.ncsa.teragrid.org (172.22.57.161): icmp_seq=2 ttl=64 time=0.044 ms
64 bytes from tg-c845-myri0.ncsa.teragrid.org (172.22.57.161): icmp_seq=3 ttl=64 time=0.044 ms
64 bytes from tg-c845-myri0.ncsa.teragrid.org (172.22.57.161): icmp_seq=4 ttl=64 time=0.043 ms
The latency gets below 10 usec with the use of special drivers, this is just using the 2.4 Linux tcp stack. What's even scarier about the Myrinet is that I can have all 900+ machines talking at the same time with no drop in latency- we have that network spec'd for full bisection bandwidth. Try that on 900 nodes on a gige network, let alone a 100baseT.
As was mentioned here earlier, ethernet is nice for networks that change. Once you have a significant number of machines attached, and the number of switches and routers gets past 1, ethernet loses it's equivalence in latency.
Define low cost? Myrinet with less than 10 microsecond latency is normally considered to be the least expensive option.
You can check their price lists, but an 8 port solution (with 8 HBA's) will set you back over $8k, not including the fiber.
Sorry, we're talking about what you can get today. Next year, I'll be able to buy and it'll be better than what's available now.
The point of the article is that it's possible to get good value for your money now. I've run dual core for 3 months at home and dual proc for 3 years at work. I own a small computer computer business. I have hard time telling anyone to buy a single core if any dual core is available for under $150. The second core just makes such a big difference for regular usage for OS level tasks.
Would I prefer AMD to have an offering in that price range? Sure, but they don't, and at the moment, it's easy to see where your money is best spent right now. The fact the D805 overclocks insanely for some chips isn't even relevant for most home users, and all business users- because they won't be overclocking.
Is Intel going to regret selling these for under $150? Probably, but if you can't beat them on performance (ignoring the occasional superchip), you sell in volume with good values. That's how AMD survived during the K5 and K6 years.
Waiting for the next best thing is kinda silly in the computer world. Well, waiting more than a week, anyways.:)
Re:unresolved technical concerns (FORD on biodiese
on
Tiny Biodiesel Reactors
·
· Score: 2, Informative
Biodiesel B20 (20% biodiesel/80% petroldiesel) already has 45 million road miles of testing with no side effects.
ASTM already has standards for a 20% blend.
Go to Biodiesel.org's Fact sheets and have look for yourself. If you were to use 100% biodiesel, some of your quoted concerns would need to be addressed. Not that big a deal- just need to replace pure rubber for fuel lines, check and replace fuel filters for diesels that have already been in service, and preheat/keep warm any diesel driven vehicles if it gets really cold outside.
What's really spiffy is the possibility that small kits of these could be used right on the farm to make more self-sufficient farming possible for remote areas of the world. A tractor might run for 20 years, but bringing in diesel is a yearly event.
Scaling capability for most parallel programs has a lot to do with how "large" of a simulation you run. If you run a 100,000 atom simulation on more than 100 nodes (let's assume perfect spatial decomposition), you've got less than 1,000 atoms of force evaluation work per node, it becomes quite challenging to scale well once the amount of work per processor gets very small. If you simulate 1,000,000 atoms, then you have more work per processor, and the GigE system will fare much better than it would with 100,000 atoms. Unless you've got a very low latency interconnect, you're going to find it extremely hard to maintain anything near linear speedup once you've only got a few hundred atoms per node, regardless of the MD package you're using.
And given that the abstract of their paper states that a 1 million atom simulation was accomplished, they probably had excellent speedup.
Most likely, the simulation was run on NCSA's Tungsten cluster, which has a low latency Myrinet interconnect, just to provide the capability to run NAMD at scale on thousands of cpus.
Just to be clear, Linpack is the code that is used as the basis for top500 runs. If you want to
get a decent rating, your cluster must be able to run Linpack well. BLAS helps optimize the score
the cluster can achieve.
We use BLAS for most of our Linpack runs, as it is the fastest set of generic libraries for the purpose.
And you can ignore the complexity of coding for a P4, when you consider how much fun it is to
code for Itanium, and to do it correctly.
Maybe George Didn't Use Enough Test Files?
on
OpenOffice Bloated?
·
· Score: 2, Insightful
I'm sure that George did a full range of tests, of files from size ~10K up to the 100MB monster. I'm not sure why he didn't publish the results of the other files, but surely MS Office showed the same disproportional speed increase and lower memory use for all types of spreadsheets and word documents...?
I read the blog entry, and this looks like an attempt to report the worst case (corner case) benchmark, for whatever reasons George might find useful. A fuller set of tests and results might make his case that MSOffice is better a little more convincing.
Hard to be convincing about the usefulness of MSOffice over OpenOffice, if you're going to ignore the fact that MSOffice isn't available for some portion of the users, and choose to report a single data point as if it were conclusive.
That's why I bought a Canon PIMAX 4000. Has 5 seperate ink tanks, and I can get refills for less than $3.00 a tank. All they do is hold ink, and lots of it.
Given that an IP4000 can now be had new for ~$100, why buy HP or Lexmark again?
I got tired of paying HP 1/2 the price of the printer every time I refilled my DJ895.
EULA's apply to how the purchaser of the software may use the software AND whether or not the purchaser may redistribute the software. You may only use the software in the fashion that the "owner" decries. You can't redistribute it legally at all.
GPL applies to how the "purchaser" may redistribute the software only. GPL makes no claims about use. You may use GPL software in any fashion you wish. You may redistribute it under the terms of the GPL.
Distribution of Copyrighted material is so thoroughly legislated that GPL is in no danger of "repeal", regardless of any court decisions based on EULA and usage restrictions. Being able to redistribute copyrighted material is entirely at the discretion of the owner, as RIAA and the MPAA so often remind us.
Point 2 from the letter, please forgive if I failed to reproduce the formatting correctly.
"SCO Has a Superior Kernel - SCO OpenServer 6 includes the UNIX System
V Release 5 (SVR5) kernel, the result of more than 25 years of
high-end development work that has created a proven track record of
stability and reliability. With our latest release, OpenServer
provides support for up to 32 processors, 64 GB of memory, terabyte
file sizes, and full support for multi-threaded applications. Linux
is still young from an operating system perspective. I would
challenge any kernel out there to match us head-to-head."
Hmmm, sitting on our Altix at the moment, I see:
512 processors, 2TB RAM, and several filesystems over 10TB, all usable running under a single copy of Linux. The kernel is 2.4.21 from RH, patched with SGI Propack
We have 10 seperate jobs running, anywhere from 8 to 128 cpus in each. I think we compare pretty well.
Hey, I was in the USMCR, so I can sympathize with you and the amount of danger you're when your troops are out in the street. Anything that can give the military troops an edge should be considered.
There are two caveats.
Looking at the posts, I'm getting the sense that most of folks are more worried about civilian crowds here in the US, and what something like this would do to a group of peaceful protesters that the police or National Guard have been ordered to disperse. As a commander, you surely remember the incident at Kent State, where lethal force was used against a crowd of students. Giving local authorities this kind of "non-lethal" weapon does not give any citizen in the US a warm fuzzy about our right to assemble peaceably.
The second point concerns the US appearance before the world. This weapon, in the hands of uncontrolled troops, could conceivable save the lives of soldiers at the cost of indiscriminately killing the civilians. Given that our military (Well, the Army, anyways) has already admitted losing control of their troops at Abu Gareb, it's not that much of a stretch to see any new "area of effect" weapon being misused at some point. Heck, it's hard enough to get the job done in the best of circumstances, so please forgive us if we show some wariness when a new technology is announced that will provide a "shortcut" for troops.
As I said, any new weapon that gives US troops more options in the field should be considered. Please remember however, that there are plenty of video clips floating about that show some US troops to be less than responsible- not every grunt is in the Corps and can be expected to have Marine discipline, but every US troop on the ground still represents the entire country, and some of them are not demonstrating the professionalism we hope to see, even under averse conditions.
Don't think that the average slashdot user wants to see anyone killed, but posters do have higher awareness of the potential side effects of new technology. In some respects, whatever is said here on/. won't matter at all. If the new weapon is made available for you guys, the military will use it or not, as the situation dictates. You'll modify how it gets used according to your experiences with it in combat, and if it makes a difference, it will keep being used. Nothing anyone here says will make a difference in that.
The bottom line is this: As long as the Marine Corps, and the rest of the military remember that they represent all of the U.S. in their actions, we will cut you guys more than enough slack to get the job done. Semper Fi.
Um, we actually have 937 total nodes, each of which is at least a dual cpu node. We have 8 login nodes that are quad cpu. (slightly older 1.3 Itanic II's)
The 887 nodes break down into 256 dual 1.3 and 631 dual 1.5's. The other nodes function as support, storage or mgmt.
If you do the math (931 * 2 + 8 * 4) we get 1894 Itaniums in a cluster.
More impressive is the fact that we had gige and myrinet for everynode in the cluster waaay back in 2003.:)
Plus the 1024 Itaniums in the Altix.
I'm not going to talk about why we bought Itanic, my boss is looking at me.:) Suffice it to say that we had good reasons to look at Itanic, did our research on all available options and put the cluster up for open bid.
I have a nice cluster with ~1800 Itanium II's. It's fast, the CPU's stable, and it runs on Linux. I have a lot of hands-on experience with it.
A couple of points that seem to have been missed when looking at why the itanium less widespread:
each CPU is quite large, having a square surface area for the unit about 2" x 5" and it's about 2" high
That area includes a voltage regulater and the passive cooling fans
It doesn't include any of the necessary active cooling
If you add these physical factors to the points already made about heat, power and EFI bios, it's obvious to say that Itanium won't run in your mini-ATX destop or laptop. This isn't a slam on the design, as it was never designed to run in those form factors, but it's hard to see how any cpu today is going to have a wide use if it isn't available for dual use for destop and servers. Once you eliminate the desktop market, (and I'm going to lump the workstation market in with the servers) the number of places you can sell these processors drops considerably.
Once you start adding in the lack of Windows support for itanium, the strides that the 86_64 architechture has made in capability, and the low numbers of current adopters, it's not looking like Itanium will ever gain widespread acceptance.
Perhaps the fact that the author is a Microsoft MVP and that Microsoft is bringing out a competitor to bittorrent could be seen as evidence that the purpose of the article is to spread Fear, Uncertainty and Doubt about bittorrent, rather than to simply inform the public about a new threat to security.
Given that the videos are being distributed in a self-extracting wrapper, it's certainly possible that this method of deploying their adware will not be as obvious as it might otherwise be. At least, not that obvious for the first time you accept one of these crappy licenses.
Interesting timing on the part of Slashdot to post both of these articles at the same time, no? I mean, it's not like Slashdot would actually advertising revenue from Microsoft...
Well, this was "The Register" (No, I'm not going to put a link to that shining example of yellow journalism.). What did you expect from them? A serious reporting of the just the facts, with the usual levening of background information and a cautious dash of perspective?
It's always a good idea to consider the source, when something offends your common sense and drives up your blood pressure.
Unfortuantely, that won't work anymore- as Newegg has a policy of pulling any negative reviews from users for specific items. You can still vote a bad score, but they won't keep a negative review on line.
They have a leg to stand on, (don't want a clueless user spewing hate mail) but its still a bit lacking in ethics. Otherwise, your methods would work well.
Slashdot Mirror
http://en.wikipedia.org/wiki/Socket_754 has a breakdown of socket 754 limitations.
I don't believe that either socket 939 or 940 had any 32 bit CPUs available.
AMD put out 3 different socket sets to maximize their profits- socket 754 for low end, non-64 bit computing, and single channel memory, socket 939 for mainstream users, and socket 940 for server and extreme users. All marchitechture, but all forgiven because the AMD users could buy dual cores that weren't just space heaters. Yeah, the price for the good stuff wasn't any cheaper, but the benefits were so obvious that only the Intel/Dell fanboys "stayed the course" or at least, held off from buying.
Then Intel releases a near perfect CPU, great performance, good heat, medium power, just no upgrade to memory acces. Intel fanboys rejoice and finally upgrade. Middle of the roaders feel like they have a choice. AMD is suddenly left in the position it had occupied for all those years, second place. Yeah, it has a lot of options, and is still competitive for server stuff, but it's no longer a lock for the desktop user.
Amd reverts to what worked for them previously- move all desktops to the same socket and give that socket a lot of upgrade life. Since DDR2 is finally available in quanity, and at speeds that actually don't produce a slower OS than using DDR 400, AMD decides to make the change to DDR2. Save for the recent attempt to make money, AMD users have been able to buy one socket for the majority of AMD cpus available at that time, and that provides them some marginal sales, for those users who want a chance at a later CPU upgrade.
SO, socket AM2 is released at a time where it doesn't make much sense to upgrade for AMD fanboys. Intel fanboys are buying all the core 2 duo's their pocketbooks can handle, and middle-of-the-roaders are picking and choosing, just like always, versus performnce and price. AM2 is not cheaper than Intel solutions; the real deals for AMD are the clearance of older socket 754/939 stuff. Any real wonder that AM2 sales, at the moment, have been lackluster? As I see it, AMD took the long view, and released AM2 for the upcoming K8L and newer stuff. They'll take whatever sales they can get, but they aren't overly worried about sales right now. I mean, Dell is finally selling AMD's and I'd bet that AMD is waiting on that cash cow to come in.
Yeah, I caught that the gige was optional. Sheesh.
Here's the Motherboard Info:
Motherboard/Processor:
* 1GHz VIA C3 CPU
* VIA CLE266 Northbridge
* VIA VT8237 Southbridge
* DDR266 RAM - Up to 1GB
* 2 USB 2.0 ports
* 1 Serial port
* 1 Parallel port
* 1 VGA port
* PS2 mouse & keyboard ports
Anybody have performance numbers for these units? A 1GHZ CPU can be hard-pressed to run an OS, serve disk and support a gige connection at full throughput. I'd be weary of looking at these for a data center without knowing how fast they can serve out the disk over a single gige connection. In fact, I see a distinct lack of information about this unit functions as a "storage node". Are you buying a 1U, 3.0TB node on which you need to install an OS and fileserver? Doesn't look like it would have the horsepower to run an iSCSI driver in additional to software raid drivers and still produce any real transfer speeds.
While a rack of these sounds nice in cost/wattage terms, it appears that you would have just purchased a cluster of storage nodes. A cluster of storage nodes with no way to present the available 120TB's as any kind of coherent storage space. You might be able to run Lustre, PVFS or GFS on them, if that's even possible, but that's a level of complexity the price and performance don't warrant.
If you figure in the cost of a Storage Engineer and lack of performance, this looks less appealing at the full rack level. Doesn't mean some PHB's won't buy into the the whole "Cheap Cluster Disk!" theme though. I pity the sysadmins who get a 120TB of raw disk and 40 more nodes to admin.
It's pretty simple- the constitution reserves all rights to the people or States, save those which are carefully delegated to the federal government. Amendment X specifically states this:
The powers not delegated to the United States by the Constitution, nor prohibited by it to the States, are reserved to the States respectively, or to the people.
The bill of rights is not an exhaustive list- it's just a of rights which the founders thought were worth mentioning specifically. You have a right to privacy by default- at least, a right that the federal government may not abridge.
We tweak the kernel for long-haul IP traffic, certain types of error messages (thousands of warnings per thousands nodes equals gigabytes of logs per day, otherwise), and other things. We also update and build custom drivers for our gige and fiber channel cards. All of which isn't possible with the Windows versions.
We also have support contracts from our software vendors (IBM, Redhat and SuSE) in case we run into issues where some other professional opinions and expertise make a difference. We tend to get quick answers and good solutions from these vendors- I have no idea what dealing with MS tech support would be like.
Russian copyright law is, shudder not the same as US copyright law! What horrors! Why, if it's not legal here, courtesy of our fine Congress, it must not be legal anywhere!
What a crock of logical fallacy. That's the biggest point overlocked in this debate- Russia has no reason to conform with US laws, and if they choose not to, they aren't doing anything illegal. It's bad enough that the RIAA and MPAA can push through crap laws like the Sony Bono Copyright extension, how much worse is it that they have so many people brainwashed into believed that those laws hold everywhere.
Per Russian law, which is subject to change, what they are doing is perfectly legal. Don't like it? Tough.
On a Force10 switch, with 2 nodes on the same blade:
tg-c844:~ # ping tg-c845
PING tg-c845.ncsa.teragrid.org (141.142.57.161) from 141.142.57.160 : 56(84) bytes of data.
64 bytes from tg-c845.ncsa.teragrid.org (141.142.57.161): icmp_seq=1 ttl=64 time=0.148 ms
64 bytes from tg-c845.ncsa.teragrid.org (141.142.57.161): icmp_seq=2 ttl=64 time=0.146 ms
64 bytes from tg-c845.ncsa.teragrid.org (141.142.57.161): icmp_seq=3 ttl=64 time=0.145 ms
64 bytes from tg-c845.ncsa.teragrid.org (141.142.57.161): icmp_seq=4 ttl=64 time=0.144 ms
The same nodes using a myrinet connection:
tg-c844:~ # ping tg-c845-myri0
PING tg-c845-myri0.ncsa.teragrid.org (172.22.57.161) from 172.22.57.160 : 56(84) bytes of data.
64 bytes from tg-c845-myri0.ncsa.teragrid.org (172.22.57.161): icmp_seq=1 ttl=64 time=0.051 ms
64 bytes from tg-c845-myri0.ncsa.teragrid.org (172.22.57.161): icmp_seq=2 ttl=64 time=0.044 ms
64 bytes from tg-c845-myri0.ncsa.teragrid.org (172.22.57.161): icmp_seq=3 ttl=64 time=0.044 ms
64 bytes from tg-c845-myri0.ncsa.teragrid.org (172.22.57.161): icmp_seq=4 ttl=64 time=0.043 ms
The latency gets below 10 usec with the use of special drivers, this is just using the 2.4 Linux tcp stack. What's even scarier about the Myrinet is that I can have all 900+ machines talking at the same time with no drop in latency- we have that network spec'd for full bisection bandwidth. Try that on 900 nodes on a gige network, let alone a 100baseT.
As was mentioned here earlier, ethernet is nice for networks that change. Once you have a significant number of machines attached, and the number of switches and routers gets past 1, ethernet loses it's equivalence in latency.
For some people, that's cheap. If not, sorry.
The point of the article is that it's possible to get good value for your money now. I've run dual core for 3 months at home and dual proc for 3 years at work. I own a small computer computer business. I have hard time telling anyone to buy a single core if any dual core is available for under $150. The second core just makes such a big difference for regular usage for OS level tasks.
Would I prefer AMD to have an offering in that price range? Sure, but they don't, and at the moment, it's easy to see where your money is best spent right now. The fact the D805 overclocks insanely for some chips isn't even relevant for most home users, and all business users- because they won't be overclocking.
Is Intel going to regret selling these for under $150? Probably, but if you can't beat them on performance (ignoring the occasional superchip), you sell in volume with good values. That's how AMD survived during the K5 and K6 years.
Waiting for the next best thing is kinda silly in the computer world. Well, waiting more than a week, anyways. :)
ASTM already has standards for a 20% blend.
Go to Biodiesel.org's Fact sheets and have look for yourself. If you were to use 100% biodiesel, some of your quoted concerns would need to be addressed. Not that big a deal- just need to replace pure rubber for fuel lines, check and replace fuel filters for diesels that have already been in service, and preheat/keep warm any diesel driven vehicles if it gets really cold outside.
What's really spiffy is the possibility that small kits of these could be used right on the farm to make more self-sufficient farming possible for remote areas of the world. A tractor might run for 20 years, but bringing in diesel is a yearly event.
And given that the abstract of their paper states that a 1 million atom simulation was accomplished, they probably had excellent speedup.
Most likely, the simulation was run on NCSA's Tungsten cluster, which has a low latency Myrinet interconnect, just to provide the capability to run NAMD at scale on thousands of cpus.
We use BLAS for most of our Linpack runs, as it is the fastest set of generic libraries for the purpose. And you can ignore the complexity of coding for a P4, when you consider how much fun it is to code for Itanium, and to do it correctly.
I read the blog entry, and this looks like an attempt to report the worst case (corner case) benchmark, for whatever reasons George might find useful. A fuller set of tests and results might make his case that MSOffice is better a little more convincing.
Hard to be convincing about the usefulness of MSOffice over OpenOffice, if you're going to ignore the fact that MSOffice isn't available for some portion of the users, and choose to report a single data point as if it were conclusive.
Given that an IP4000 can now be had new for ~$100, why buy HP or Lexmark again?
I got tired of paying HP 1/2 the price of the printer every time I refilled my DJ895.
EULA's apply to how the purchaser of the software may use the software AND whether or not the purchaser may redistribute the software. You may only use the software in the fashion that the "owner" decries. You can't redistribute it legally at all.
GPL applies to how the "purchaser" may redistribute the software only. GPL makes no claims about use. You may use GPL software in any fashion you wish. You may redistribute it under the terms of the GPL.
Distribution of Copyrighted material is so thoroughly legislated that GPL is in no danger of "repeal", regardless of any court decisions based on EULA and usage restrictions. Being able to redistribute copyrighted material is entirely at the discretion of the owner, as RIAA and the MPAA so often remind us.
Why do so many people get these two confused?
"SCO Has a Superior Kernel - SCO OpenServer 6 includes the UNIX System
V Release 5 (SVR5) kernel, the result of more than 25 years of
high-end development work that has created a proven track record of
stability and reliability. With our latest release, OpenServer
provides support for up to 32 processors, 64 GB of memory, terabyte
file sizes, and full support for multi-threaded applications. Linux
is still young from an operating system perspective. I would
challenge any kernel out there to match us head-to-head."
Hmmm, sitting on our Altix at the moment, I see:
512 processors, 2TB RAM, and several filesystems over 10TB, all usable running under a single copy of Linux. The kernel is 2.4.21 from RH, patched with SGI Propack
We have 10 seperate jobs running, anywhere from 8 to 128 cpus in each. I think we compare pretty well.
There are two caveats.
Looking at the posts, I'm getting the sense that most of folks are more worried about civilian crowds here in the US, and what something like this would do to a group of peaceful protesters that the police or National Guard have been ordered to disperse. As a commander, you surely remember the incident at Kent State, where lethal force was used against a crowd of students. Giving local authorities this kind of "non-lethal" weapon does not give any citizen in the US a warm fuzzy about our right to assemble peaceably.
The second point concerns the US appearance before the world. This weapon, in the hands of uncontrolled troops, could conceivable save the lives of soldiers at the cost of indiscriminately killing the civilians. Given that our military (Well, the Army, anyways) has already admitted losing control of their troops at Abu Gareb, it's not that much of a stretch to see any new "area of effect" weapon being misused at some point. Heck, it's hard enough to get the job done in the best of circumstances, so please forgive us if we show some wariness when a new technology is announced that will provide a "shortcut" for troops.
As I said, any new weapon that gives US troops more options in the field should be considered. Please remember however, that there are plenty of video clips floating about that show some US troops to be less than responsible- not every grunt is in the Corps and can be expected to have Marine discipline, but every US troop on the ground still represents the entire country, and some of them are not demonstrating the professionalism we hope to see, even under averse conditions.
Don't think that the average slashdot user wants to see anyone killed, but posters do have higher awareness of the potential side effects of new technology. In some respects, whatever is said here on /. won't matter at all. If the new weapon is made available for you guys, the military will use it or not, as the situation dictates. You'll modify how it gets used according to your experiences with it in combat, and if it makes a difference, it will keep being used. Nothing anyone here says will make a difference in that.
The bottom line is this: As long as the Marine Corps, and the rest of the military remember that they represent all of the U.S. in their actions, we will cut you guys more than enough slack to get the job done. Semper Fi.
The 887 nodes break down into 256 dual 1.3 and 631 dual 1.5's. The other nodes function as support, storage or mgmt.
If you do the math (931 * 2 + 8 * 4) we get 1894 Itaniums in a cluster.
More impressive is the fact that we had gige and myrinet for everynode in the cluster waaay back in 2003. :)
Plus the 1024 Itaniums in the Altix.
I'm not going to talk about why we bought Itanic, my boss is looking at me. :) Suffice it to say that we had good reasons to look at Itanic, did our research on all available options and put the cluster up for open bid.
A couple of points that seem to have been missed when looking at why the itanium less widespread:
- each CPU is quite large, having a square surface area for the unit about 2" x 5" and it's about 2" high
- That area includes a voltage regulater and the passive cooling fans
- It doesn't include any of the necessary active cooling
If you add these physical factors to the points already made about heat, power and EFI bios, it's obvious to say that Itanium won't run in your mini-ATX destop or laptop. This isn't a slam on the design, as it was never designed to run in those form factors, but it's hard to see how any cpu today is going to have a wide use if it isn't available for dual use for destop and servers. Once you eliminate the desktop market, (and I'm going to lump the workstation market in with the servers) the number of places you can sell these processors drops considerably.Once you start adding in the lack of Windows support for itanium, the strides that the 86_64 architechture has made in capability, and the low numbers of current adopters, it's not looking like Itanium will ever gain widespread acceptance.
And just where and how do you think he obtained that family fortune? Face it folks- we're living in a new age of robber barons.
35% of all internet traffic is second only to tcp/ip
Given that the videos are being distributed in a self-extracting wrapper, it's certainly possible that this method of deploying their adware will not be as obvious as it might otherwise be. At least, not that obvious for the first time you accept one of these crappy licenses.
Interesting timing on the part of Slashdot to post both of these articles at the same time, no? I mean, it's not like Slashdot would actually advertising revenue from Microsoft...
It's always a good idea to consider the source, when something offends your common sense and drives up your blood pressure.
Unfortuantely, that won't work anymore- as Newegg has a policy of pulling any negative reviews from users for specific items. You can still vote a bad score, but they won't keep a negative review on line. They have a leg to stand on, (don't want a clueless user spewing hate mail) but its still a bit lacking in ethics. Otherwise, your methods would work well.