I agree. There are people on Slashdot who are not in favor of F/OSS, and have to make ridiculous claims to make a point.
RMS has been and still is a very important person for the F/OSS movement.
"Why would they start working on a new chip that moves it off again?"
Please! They are not moving it off again!
Please understand! This is an upgrade to their exististing Xeon MP server design and these chips can fit into _existing_ servers. There will be new Xeon MP chips based on Nehalem ready late next year.
Xeon MP chips takes much more time to validate that desktop chips.
"Still it has to cost something to expect a 4MB cache. It's not free."
Ofcourse, not free, but adding cache is a fairly easy and cost effective thing to do. It does cost die space, but does not cause yield problems. Intel have to compensate with more cache to hide memory latency due to lack of integrated memory controller, but they can afford to do so.
Intels Conroe/Woodcrests with 4MB cache are 142 mm2 in 65nm, AMD's dual core chips are I think 199mm2. AMD though have the memory controller on the die and is on 90nm. They will most likely add more cache when they go to 65nm.
Logic parts of a die are much more sensitive to problems, especially critical speed path problems. Caches have been done using redundancy for many years now, and scrapping a die because of faults in the cache parts are very rare today.
The Cell CPU is much more prone to yield problems than a SRAM cell for example.
1. Intel has been working on this for some years, and will begin to release CPUs with integrated memory controllers and routers in 2008.
3. No, not very expensive since they are on 65nm now.
"[*] Don't look at the retail cost for the true margins they make [if any] on the cores. Selling at a loss or near loss is not a new tactic."
Not sure if that is something you really think or if you are trolling. Intel is still making lots of money and obviously has much better margins than AMD.
Microsoft doesn't call it anything specifically, you just open the drive/volume/partition the way you open a file by specifying for example "\\.\PHYSICALDRIVE0".
Others (customers/users of Windows) sometimes call it raw device or raw partition but I have not seen that Microsoft use the term raw device.
HP has not killed VMS.
HP is the leading seller of x86 Linux boxes.
AMD64 is not (at least yet) near high-end. IBM seem very calm about adopting Opteron (only offering a HPC 2-way server), while HP have real Opteron servers for sale. Actually HP is currently the only tier-1 vendor who has an in-house design for Opteron. IBM's and Sun's Opteron boxes are re-branded Newisys boxes.
Windows 2000 was released as a release candidate for Alpha and it was a real 64-bit Windows with 64-bit pointers just like the current Itanium versions of Windows.
Slashdot Mirror
You are correct. More common hardware is lots cheaper and can be used for more tasks.
I agree. There are people on Slashdot who are not in favor of F/OSS, and have to make ridiculous claims to make a point. RMS has been and still is a very important person for the F/OSS movement.
No. He can use GNU/Linux. In fact it isn't very often he himself talk abut HURD.
I agree. Many people don't understand what he says.
I wonder which of these is true:
Where oh where did you get that from? I friend wiped the disk and installed GNU/Linux. Is that unconscionable behavior?
"Why would they start working on a new chip that moves it off again?" Please! They are not moving it off again! Please understand! This is an upgrade to their exististing Xeon MP server design and these chips can fit into _existing_ servers. There will be new Xeon MP chips based on Nehalem ready late next year. Xeon MP chips takes much more time to validate that desktop chips.
What is the point of commenting on an article you haven't read?
AMD is going quad-core in 65nm, and some parts will have large L3 caches.
"Still it has to cost something to expect a 4MB cache. It's not free."
Ofcourse, not free, but adding cache is a fairly easy and cost effective thing to do. It does cost die space, but does not cause yield problems. Intel have to compensate with more cache to hide memory latency due to lack of integrated memory controller, but they can afford to do so.
Intels Conroe/Woodcrests with 4MB cache are 142 mm2 in 65nm, AMD's dual core chips are I think 199mm2. AMD though have the memory controller on the die and is on 90nm. They will most likely add more cache when they go to 65nm.
Logic parts of a die are much more sensitive to problems, especially critical speed path problems. Caches have been done using redundancy for many years now, and scrapping a die because of faults in the cache parts are very rare today.
p df/manufacturing.pdf
The Cell CPU is much more prone to yield problems than a SRAM cell for example.
This PDF is a bit old but good, and the use of redundancy has increased since then.
http://developer.intel.com/technology/itj/q41997/
"More transitors lowers your yield though, cost++"
Cache transistors don't lowers yield. Logic transistors do.
1. Intel has been working on this for some years,
and will begin to release CPUs with integrated memory controllers and routers in 2008.
3. No, not very expensive since they are on 65nm now.
"[*] Don't look at the retail cost for the true margins they make [if any] on the cores. Selling at a loss or near loss is not a new tactic."
Not sure if that is something you really think or if you are trolling.
Intel is still making lots of money and obviously has much better margins than AMD.
Microsoft doesn't call it anything specifically, you just open the drive/volume/partition the way you open a file by specifying for example "\\.\PHYSICALDRIVE0". Others (customers/users of Windows) sometimes call it raw device or raw partition but I have not seen that Microsoft use the term raw device.
"NO raw device access "
That is wrong.
Windows have raw device, it just isn't called that.
Yes, ofcourse. They want to push POWER as much as possible.
HP has not killed VMS. HP is the leading seller of x86 Linux boxes. AMD64 is not (at least yet) near high-end. IBM seem very calm about adopting Opteron (only offering a HPC 2-way server), while HP have real Opteron servers for sale. Actually HP is currently the only tier-1 vendor who has an in-house design for Opteron. IBM's and Sun's Opteron boxes are re-branded Newisys boxes.
You are wrong on both.
Windows 2000 was released as a release candidate for Alpha and it was a real 64-bit Windows with 64-bit pointers just like the current Itanium versions of Windows.