You may want to reconsider your belief that MIPS processors are slower than Intel or AMD processors. Check out the SPEC CPU FP numbers here.
Yes, the MIPS processor only runs at 400 Mhz, but the PIII's don't approach the all-around FP speeds of the R12K until the Intel reaches about 1 Ghz. Its a good illustrator of the Mhz lie. On integer benchmarks, the MIPS isn't quite as far ahead, but the 400 Mhz R12K is similar to a 733 PIII.
Oh yeah, one other thing. The multiple OSes running on one "system" is what sgi calls partitioning. This is where a customer might have a 4x32p system running during the daytime and reboot to a 128p system for night time batch. However, when in 4x32p mode, it looks like 4 different systems (like each of the nodes in a beowulf) -- bigsys1, bigsys2, bigsys3, and bigsys4. When in 128p mode, its just a single bigsys.
Login to your beowulf node 1 and try to access memory (r/w) on node 2. Can't do it without message passing or shared memory libraries. Clusters require special programming to go node to node.
Login to a 3000, you don't even know what node your on, in fact the system doesn't give you any impression that its any different from a small up or mp. The thing that tips you off is the load of 200, 400, or 500+. Depending on whats going on on the system, your process may be migrated from one node to another without you noticing. On the 3000, any process on any processor can access every page on every node -- all through regular memory references.
Aah, yes. The $64000 question. The answer to this is NUMA and hypercube structured interconnect. Check out the specs. Its not an SMP. It is shared memory like an SMP. Looks and acts like an SMP at all processor counts.
Actually, one of the benefits of the system's modular building blocks is that one rack can hold up to 32 processors. This is double the density of the o2k. The refrigerator comment is correct though. 4p in 4U is bound to kick off some heat.
RHAT's recent market capitalization is roughly 5B on a stock price of ~$70 with $12M in revenue. They have about 140 employees and their book value is negative.
Compare this with SGI. Their market capitalization is roughly half RHAT's (which illustrates the silliness of RHAT's stock value). SGI had reveunes of 2.75B last year and has about 9000 employees (after layoffs). SGI is a much bigger company, they just layed-off about 8X RHAT's workforce. They also have half a billion in cash.
In short, RHAT's stock value may be high, but high more on hopes more than anything. Compared with assests and sales, their stock is worth little in real cash. Considering how quick the open source movement is, how fast could RedHat be replaced by a newer better distro? When will their support services start paying off?
A better buyout target is Mandrake. They make a better RedHat than RedHat.
Is the comparison to a Cray fair when you consider the inter-node communication/bandwidth needs?
In clustering and parallel computation, bandwidth counts. My guess is that a different application that requires much more communication between nodes, the T3E would step on the Netfinitys. 100 megabit ethernet does it for low-communication jobs, but what about those that require much more intensive inter-node communication.
Slashdot Mirror
Yes, the MIPS processor only runs at 400 Mhz, but the PIII's don't approach the all-around FP speeds of the R12K until the Intel reaches about 1 Ghz. Its a good illustrator of the Mhz lie. On integer benchmarks, the MIPS isn't quite as far ahead, but the 400 Mhz R12K is similar to a 733 PIII.
Oh yeah, one other thing. The multiple OSes running on one "system" is what sgi calls partitioning. This is where a customer might have a 4x32p system running during the daytime and reboot to a 128p system for night time batch. However, when in 4x32p mode, it looks like 4 different systems (like each of the nodes in a beowulf) -- bigsys1, bigsys2, bigsys3, and bigsys4. When in 128p mode, its just a single bigsys.
Login to a 3000, you don't even know what node your on, in fact the system doesn't give you any impression that its any different from a small up or mp. The thing that tips you off is the load of 200, 400, or 500+. Depending on whats going on on the system, your process may be migrated from one node to another without you noticing. On the 3000, any process on any processor can access every page on every node -- all through regular memory references.
Aah, yes. The $64000 question. The answer to this is NUMA and hypercube structured interconnect. Check out the specs. Its not an SMP. It is shared memory like an SMP. Looks and acts like an SMP at all processor counts.
Actually, one of the benefits of the system's modular building blocks is that one rack can hold up to 32 processors. This is double the density of the o2k. The refrigerator comment is correct though. 4p in 4U is bound to kick off some heat.
RHAT's recent market capitalization is roughly 5B on a stock price of ~$70 with $12M in revenue. They have about 140 employees and their book value is negative.
Compare this with SGI. Their market capitalization is roughly half RHAT's (which illustrates the silliness of RHAT's stock value). SGI had reveunes of 2.75B last year and has about 9000 employees (after layoffs). SGI is a much bigger company, they just layed-off about 8X RHAT's workforce. They also have half a billion in cash.
In short, RHAT's stock value may be high, but high more on hopes more than anything. Compared with assests and sales, their stock is worth little in real cash. Considering how quick the open source movement is, how fast could RedHat be replaced by a newer better distro? When will their support services start paying off?
A better buyout target is Mandrake. They make a better RedHat than RedHat.
Compare:
SGI's Profile
RHAT's Profile
Is the comparison to a Cray fair when you consider
the inter-node communication/bandwidth needs?
In clustering and parallel computation, bandwidth
counts. My guess is that a different application
that requires much more communication between
nodes, the T3E would step on the Netfinitys. 100
megabit ethernet does it for low-communication
jobs, but what about those that require much more
intensive inter-node communication.