So... 1) Laptops should not be allowed to attach internally unless they only touch the internet by VPN tunnelling into the the network and going out to the internet on proxy servers. 2) SQL servers should NOT be able to talk to just any system on the internal network either - only front end or app servers that transact with them and only though internal firewalls...
Any decent course covering 3 tier architectures will tell you that beyond your external firewalls there should be internal filewalls behind which the db layer is. Web layer (and perhaps the app layer) may be exposed, but database servers should ONLY comunicate across the interior firewalls and only with those front end servers and then only on a few well-defined ports. That so many MsSQL servers were exposed shows sloppy (and this always equates to cheap) design discipline. Archiects and contractors aren't allowed to build buildings this way - why do we allow systems slackers to do so? Too bad we live in a time when the fed gov't is so captive to business that no regulatory initiative, not matter how needed, is going to fly...
Yeah, I believe the L3 cache on P970 is off die. ITP2 has big transistor count because (in the case of Madison) it has SIX MEGABYTES of ON-DIE L3 cache.
Intel doesn't put it there because it likes wasting wafer real-estate - on die cache acceses though a faster and wider data path...
Boy, Intel can't win. First people hammer (excuse the expression) it for making a 64-bit processor with massive parallelism (the Explicitly Parallel Instruction Computing, or EPIC architecture) which introduces a whole different way of using the compiler/silicon relationship (which generates best perf times our there on practically all SPEC tests) - AND THEY GET YELLED AT FOR NOT BEING A DIRECT EXTENSION OF x86 ARCHITECTURE AND CHANGING THINGS TOO MUCH.
Now this fellow thinks because there's some compatibilty with old 32-bit, that it's NOT RADICALLY DIFFERENT ENOUGH. WHAT DO YOU WANT? QUANTUM COMPUTING? (Intel's saving that for 2010...;-) )
The heat tiles on the shuttle have to be virtually 100% perfect for safe re-entry and as 110 or so safer re-entries have proven, NASA has done a good job of keeping the tile sets fixed. But...
I fear that something happened to them this time, either through:
1) Sloppy re-tiling - loose adhesive, incorrect placement, etc. Some QA check procedure breaks down (al la the Hubble lens)
2) Launch debris (eg big chunks of ice) falling off the fuel tank (this happens all the time) and hitting a vunerable spot (like the leading edge of a wing where the tiles curve and are probably most vunerable. I wonder if there any way for the crew in orbit to do a visual inspection of the high (bottom) face of the shuttle to check for damage (-not sure they would be able to DO anything about it, perhaps that's why there's no procedure to look...if it were bad you wouldn't want to know)
3) (Hate to say this) Sabotage... This could occur several possible ways, either intentially poor install at the re-tiling works (good question: Is this the first time Columbia has flown since it's last retiling?) Or perhaps it was damaged remotely (rifle bullet) while on the pad or being transported? New Republic has an execellent article recently about the fad of inexplicably legal huge.50 cal rifles that can take down an airliner at two miles. Probably wouldn't need something this lethal to target the shuttle from a few miles away and simply break or crack a few tiles in a way that visual inspection wouldn't pick up. Normally I'd consider the sabotage scenario paranoid, but there was an Israeli astronaut aboard...
Not sure from your comments why you would never buy an Intel cpu for personal use again.
If they made one that performed well at a price you found acceptable, why wouldn't you?
Just wondering...
I have yet to see a firm definition of how the PR rating is defined currently (or ever). They've had two auditors at least doing it, Arthur Andersen first then PriceWaterhouseCoopers (I guess, no matter how much they like the integrity of the original auditor, when they go out of business you need to find the next best thing). But for the life of me, I can't find a clear equation listed anywhere indicating what the input benchmarks are and how the final PR number is calculated. (Please reply with links if you can find it). If I were a cynical fellow, I'd say that the PR rating is not (if it ever was) a real calculated measure of SOMETHING and not just a way for AMD marketing to say where they'd LIKE to position a cpu in the market against the other guys....
First the perf myth is dispelled, watch for the management myth to go down.
Sun has been living with an outdated out-of-band management scheme based on serial ports which requires technially stupid solutions like hooking system RS232 to an external terminal server to dumb-down than dumb-up the interface for remote LAN-based access. There are all sorts of much more powerful out-of-band management solutions in the Intel world (intelligent, independently powered cpu modules with network connection for much more robust in and out-of-band mgmt AND powerful default baseboard mgmt). Sun manages to perpetuate the management myth due to ignorance and wishful thinking - and needs to, since they've obviouly lost the price perf and even raw perf crown a while ago...
Let me clarify a couple of popular misconceptions. Facts:
1) Intel compilers improve code performance (over GNU compliers) on both Intel (PIII and P4) and AMD (Athlon) processors due to supporting SSE and SSE2 instructions and other extensions. Although this perf gain will be greater on Intel cpus.
2) gcc maintainers have been unwilling to put Intel or AMD specific optimizations in the code -there's no secret instructions, just unwillingness to use the published stuff (check out the 100s of docs, forums and other stuff at developer.intel.com (where to get your non-commecial compiler downloads))
Since they just announced the Cometa consorium whose goal is put 802.11b hotspots all over the US, I suppose they will be subject to Dep't of Homeland Insecurity actions....
Funny, I thought the nice part about the DIGITAL ver was that I WOULD be able to access it - unlike the original Doomsday book, which, despite your claim, would take considerable effort (travel, etc) to make 'accessable' to ME.
Who cares - Intel is up to a 533MHz FSB
on
AMD Delays Hammer
·
· Score: 1
Let me get this straight:
The only performance metrics Intel issues are non-subjective GHz numbers. You'll never see public benchmarks from Intel against AMD (their lawyers won't let them do it) - and benchmarks against Pentium III will always be current SSE2-code tests that will indeed, legitimately show strong speed improvement of Pentium 4 over P3.
AMD on the other hand, uses a synthetic, subjective "PR" rating system, the authenticity of which was audited by Arthur Anderson (see http://www.theinquirer.net/?article=1207 ) which may or may not reflect an intention to compare against Intel or other competitors or other AMD products.
Which do you think is more misleading?
"The news publishers who say "yes" say that turning off graphics in your web browser should be illegal too." Presumably then, using a text-only browser like Lynx would be automatically illegal.
Serial redirect (in BIOS a la Intel 440GX, PC Weasel and Sun (in a slightly more powerful way)) IS NOT WHAT HE ASKING FOR. KVM is a totally transpearent remote way to access any system operation including GUI graphics (which the Sun-o-phile seems not to have grasped). Serial redriection via terminal servers IS A SUBSET of what you can do with KVM.
Intel hasn't made chips on.25m (250nm) in ages and the only.18m(180nm) stuff is legacy Pentium III and Celeron dies. BTW, get with the trend and start talking namometers....
DRM requirements for computers is like forcing the automakers to build your car so it must stripsearch you for weapons anytime you stop at a bank - so you won't rob it.
It's awkward, unnesscessary and wrongheaded.
Push this message...
Wrong, Itanium IS binary compatible with IA-32code
on
AMD's 64-Bit Chip
·
· Score: 1
I don't know where the author got this piece of misinformation. Perhaps it's a FUD distortion from AMD relating to the fact that Itanium was not designed to run 32-bit apps, so it's not terribly efficient (read slower) in doing so -whereas Opteron, which is basically a 32-bit processor with address extensions is more efficient at running legacy code. Presumably, Hammer will not perform as well on real 64-bit apps as Itanium since it lacks massive parallelism and other architectural differences that Intel felt they needed to put into IA64.
There are lots replies here about how various OSes and mobos can detect fan outages and do graceful (or other) shutdowns... THAT'S IRRELIVANT. Our poster, you'll note, had the power supply fan shutdown. Since nothing except very high end server power supplies are instrumented (or initiate shutdowns themselves on fan failure)... YOU (AND YOUR MOBO) HAVE NO IDEA WHEN YOUR POWER SUPPLY FAN DIES
(...Unless you can tell the difference in the sound. Everytime I see a really old PC which is suspiciously quiet, I poke a hand in back and - sure enough - the supply fan is dead.)
I don't worry much about old Intel Pentium or even K6 CPU machines, but I fear some of the hotter Athlon units may become fire hazards in their old age. (See Tom's Hardware's flaming chip videos)
Why was this mod'd down? Profiling like this IS the starting point for a simulation. An astute observation. I'm not modding today, but would someone please mod this up?
Thanks
The notion that Microsoft could learn from how IBM has handled Open Source ignores the fundamental difference between Ms and IBM. IBM has cleverly decided that hardware AND software margins are nice to have, but they are primarily a vehicle for services revenue. While Ms has a non-trivial consulting organization, (minuscule in comparison to IBM Global Services, though) it is chartered as a cost-recovery group (they try to bill enough to pay for themselves) but they are not a profit-and-loss center. MCS is there to plug in expertise where needed to advance strategic goals which all boil down to selling more and more lucrative software. Even if Microsoft owned ALL of computer systems consulting business (Windows AND UNIX/Linux) worldwide I don't believe it would not begin to approach the revenues it now receives from software. From a business point of view, "doing an IBM" and moving software to OpenSource hoping to make money on services would be insane for Ms.
The best hope is to get Ms to consider co-operating with key OpenSource projects like Ximan Mono so that the future MS world of.Net does not get walled off from OS inputs.
Obviously, none of the would-be hardware contractors that MS talked to would have built it and sold it themselves - Ms has the brand and would sell it. The negotiation would have revolved aruond how little Dell (or anybody else) would have charged Ms for the box, thus determining how much of a loss MS would eat on each box.
Intel has consistantly FOUGHT digital content mgmt from CEO on down (See recent Business 2.0 article asking Is this Man (Andy Grove) is Pirate )
AMD has said nothing on this and argued on Microsoft's side in antitrust court - and MS is very intent on installing in content mgmt.
Wake up.
Slashdot Mirror
So...
1) Laptops should not be allowed to attach internally unless they only touch the internet by VPN tunnelling into the the network and going out to the internet on proxy servers.
2) SQL servers should NOT be able to talk to just any system on the internal network either - only front end or app servers that transact with them and only though internal firewalls...
Any decent course covering 3 tier architectures will tell you that beyond your external firewalls there should be internal filewalls behind which the db layer is. Web layer (and perhaps the app layer) may be exposed, but database servers should ONLY comunicate across the interior firewalls and only with those front end servers and then only on a few well-defined ports.
That so many MsSQL servers were exposed shows sloppy (and this always equates to cheap) design discipline.
Archiects and contractors aren't allowed to build buildings this way - why do we allow systems slackers to do so?
Too bad we live in a time when the fed gov't is so captive to business that no regulatory initiative, not matter how needed, is going to fly...
Yeah, I believe the L3 cache on P970 is off die. ITP2 has big transistor count because (in the case of Madison) it has SIX MEGABYTES of ON-DIE L3 cache.
Intel doesn't put it there because it likes wasting wafer real-estate - on die cache acceses though a faster and wider data path...
Boy, Intel can't win. ;-) )
First people hammer (excuse the expression) it for making a 64-bit processor with massive parallelism (the Explicitly Parallel Instruction Computing, or EPIC architecture) which introduces a whole different way of using the compiler/silicon relationship (which generates best perf times our there on practically all SPEC tests) - AND THEY GET YELLED AT FOR NOT BEING A DIRECT EXTENSION OF x86 ARCHITECTURE AND CHANGING THINGS TOO MUCH.
Now this fellow thinks because there's some compatibilty with old 32-bit, that it's NOT RADICALLY DIFFERENT ENOUGH.
WHAT DO YOU WANT? QUANTUM COMPUTING? (Intel's saving that for 2010...
Why was the 1 in 75 comment modded down? If this is true that's good information...
The heat tiles on the shuttle have to be virtually 100% perfect for safe re-entry and as 110 or so safer re-entries have proven, NASA has done a good job of keeping the tile sets fixed. But... .50 cal rifles that can take down an airliner at two miles. Probably wouldn't need something this lethal to target the shuttle from a few miles away and simply break or crack a few tiles in a way that visual inspection wouldn't pick up. Normally I'd consider the sabotage scenario paranoid, but there was an Israeli astronaut aboard...
I fear that something happened to them this time, either through:
1) Sloppy re-tiling - loose adhesive, incorrect placement, etc. Some QA check procedure breaks down (al la the Hubble lens)
2) Launch debris (eg big chunks of ice) falling off the fuel tank (this happens all the time) and hitting a vunerable spot (like the leading edge of a wing where the tiles curve and are probably most vunerable. I wonder if there any way for the crew in orbit to do a visual inspection of the high (bottom) face of the shuttle to check for damage (-not sure they would be able to DO anything about it, perhaps that's why there's no procedure to look...if it were bad you wouldn't want to know)
3) (Hate to say this) Sabotage... This could occur several possible ways, either intentially poor install at the re-tiling works (good question: Is this the first time Columbia has flown since it's last retiling?) Or perhaps it was damaged remotely (rifle bullet) while on the pad or being transported? New Republic has an execellent article recently about the fad of inexplicably legal huge
Not sure from your comments why you would never buy an Intel cpu for personal use again.
If they made one that performed well at a price you found acceptable, why wouldn't you?
Just wondering...
I have yet to see a firm definition of how the PR rating is defined currently (or ever). They've had two auditors at least doing it, Arthur Andersen first then PriceWaterhouseCoopers (I guess, no matter how much they like the integrity of the original auditor, when they go out of business you need to find the next best thing).
But for the life of me, I can't find a clear equation listed anywhere indicating what the input benchmarks are and how the final PR number is calculated. (Please reply with links if you can find it).
If I were a cynical fellow, I'd say that the PR rating is not (if it ever was) a real calculated measure of SOMETHING and not just a way for AMD marketing to say where they'd LIKE to position a cpu in the market against the other guys....
Title of this item is off topic...
First the perf myth is dispelled, watch for the management myth to go down.
Sun has been living with an outdated out-of-band management scheme based on serial ports which requires technially stupid solutions like hooking system RS232 to an external terminal server to dumb-down than dumb-up the interface for remote LAN-based access.
There are all sorts of much more powerful out-of-band management solutions in the Intel world (intelligent, independently powered cpu modules with network connection for much more robust in and out-of-band mgmt AND powerful default baseboard mgmt). Sun manages to perpetuate the management myth due to ignorance and wishful thinking - and needs to, since they've obviouly lost the price perf and even raw perf crown a while ago...
Let me clarify a couple of popular misconceptions. Facts:
1) Intel compilers improve code performance (over GNU compliers) on both Intel (PIII and P4) and AMD (Athlon) processors due to supporting SSE and SSE2 instructions and other extensions. Although this perf gain will be greater on Intel cpus.
2) gcc maintainers have been unwilling to put Intel or AMD specific optimizations in the code -there's no secret instructions, just unwillingness to use the published stuff (check out the 100s of docs, forums and other stuff at developer.intel.com (where to get your non-commecial compiler downloads))
Since they just announced the Cometa consorium whose goal is put 802.11b hotspots all over the US, I suppose they will be subject to Dep't of Homeland Insecurity actions....
Funny, I thought the nice part about the DIGITAL ver was that I WOULD be able to access it - unlike the original Doomsday book, which, despite your claim, would take considerable effort (travel, etc) to make 'accessable' to ME.
Just getting to 333?
Let me get this straight:
The only performance metrics Intel issues are non-subjective GHz numbers. You'll never see public benchmarks from Intel against AMD (their lawyers won't let them do it) - and benchmarks against Pentium III will always be current SSE2-code tests that will indeed, legitimately show strong speed improvement of Pentium 4 over P3.
AMD on the other hand, uses a synthetic, subjective "PR" rating system, the authenticity of which was audited by Arthur Anderson (see http://www.theinquirer.net/?article=1207 ) which may or may not reflect an intention to compare against Intel or other competitors or other AMD products.
Which do you think is more misleading?
"The news publishers who say "yes" say that turning off graphics in your web browser should be illegal too."
Presumably then, using a text-only browser like Lynx would be automatically illegal.
Serial redirect (in BIOS a la Intel 440GX, PC Weasel and Sun (in a slightly more powerful way)) IS NOT WHAT HE ASKING FOR. KVM is a totally transpearent remote way to access any system operation including GUI graphics (which the Sun-o-phile seems not to have grasped).
Serial redriection via terminal servers IS A SUBSET of what you can do with KVM.
Intel hasn't made chips on .25m (250nm) in ages and the only .18m(180nm) stuff is legacy Pentium III and Celeron dies.
BTW, get with the trend and start talking namometers....
DRM requirements for computers is like forcing the automakers to build your car so it must stripsearch you for weapons anytime you stop at a bank - so you won't rob it.
It's awkward, unnesscessary and wrongheaded.
Push this message...
I don't know where the author got this piece of misinformation. Perhaps it's a FUD distortion from AMD relating to the fact that Itanium was not designed to run 32-bit apps, so it's not terribly efficient (read slower) in doing so -whereas Opteron, which is basically a 32-bit processor with address extensions is more efficient at running legacy code. Presumably, Hammer will not perform as well on real 64-bit apps as Itanium since it lacks massive parallelism and other architectural differences that Intel felt they needed to put into IA64.
There are lots replies here about how various OSes and mobos can detect fan outages and do graceful (or other) shutdowns...
THAT'S IRRELIVANT.
Our poster, you'll note, had the power supply fan shutdown. Since nothing except very high end server power supplies are instrumented (or initiate shutdowns themselves on fan failure)... YOU (AND YOUR MOBO) HAVE NO IDEA WHEN YOUR POWER SUPPLY FAN DIES
(...Unless you can tell the difference in the sound. Everytime I see a really old PC which is suspiciously quiet, I poke a hand in back and - sure enough - the supply fan is dead.)
I don't worry much about old Intel Pentium or even K6 CPU machines, but I fear some of the hotter Athlon units may become fire hazards in their old age. (See Tom's Hardware's flaming chip videos)
Why was this mod'd down? Profiling like this IS the starting point for a simulation. An astute observation. I'm not modding today, but would someone please mod this up? Thanks
The notion that Microsoft could learn from how IBM has handled Open Source ignores the fundamental difference between Ms and IBM. IBM has cleverly decided that hardware AND software margins are nice to have, but they are primarily a vehicle for services revenue. While Ms has a non-trivial consulting organization, (minuscule in comparison to IBM Global Services, though) it is chartered as a cost-recovery group (they try to bill enough to pay for themselves) but they are not a profit-and-loss center. MCS is there to plug in expertise where needed to advance strategic goals which all boil down to selling more and more lucrative software. Even if Microsoft owned ALL of computer systems consulting business (Windows AND UNIX/Linux) worldwide I don't believe it would not begin to approach the revenues it now receives from software. From a business point of view, "doing an IBM" and moving software to OpenSource hoping to make money on services would be insane for Ms. .Net does not get walled off from OS inputs.
The best hope is to get Ms to consider co-operating with key OpenSource projects like Ximan Mono so that the future MS world of
Obviously, none of the would-be hardware contractors that MS talked to would have built it and sold it themselves - Ms has the brand and would sell it. The negotiation would have revolved aruond how little Dell (or anybody else) would have charged Ms for the box, thus determining how much of a loss MS would eat on each box.
Intel has consistantly FOUGHT digital content mgmt from CEO on down (See recent Business 2.0 article asking Is this Man (Andy Grove) is Pirate )
AMD has said nothing on this and argued on Microsoft's side in antitrust court - and MS is very intent on installing in content mgmt. Wake up.