You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
opteron vs xeon
- Thread starter chiguy
- Start date
Quote from nitro:
If you have one CPU and it dies, you are down. If you have two CPUs and one dies, the operating system and applications continue to run as if the machine were a single CPU machine.
Servers don't necessarily have two CPUs for speed, but for redundancy in case of failure.
That is not quite accurate, Nitro. Intel supports a feature on some server boards that they tout as processor redundancy. With this feature the second CPU is in standby, and at the time of failure the last valid state from the primary processor is copied to the standby. This is a different mode of operation than the typical MP scenario, which has both processors running simultaneously. Intel's redundancy feature can work as advertised in some cases. Handling a CPU failure gracefully with MP is a very delicate piece of software engineering, and it is something your desktop OS is not likely to do. Any failure that results in corruption is not likely to be handled gracefully in either scenario. I can say from experience that corruption is in the majority of such scenarios.
-segv
Quote from JayS:
Will a dual core use 4 gigs of ram in either the AMD or Intel chips. I'm looking at a custom build and wondering if it will make use of all of it. Its only going to cost like $130 to go from 2 gig to 4.
The are 2^32 possible memory addresses in a 32 bit system, so the maximum size of the memory is 4GB. However, there are operating system specific limitations regarding memory. Windows operating systems have a default VM of 2GB, as do many UNIX-like systems. In some versions of Windows, the VM can be extended an additional 1GB. In the case of UNIX-like systems (FreeBSD, Linux), one can alter the memory map to make more of the reserved space available to userland applications. Having 4GB in the system is probably overkill for you, unless you are a developer.
-segv
Clarification on 4G memory with Windows XP Pro:
Kernel will use 2G for itself, paging, etc
Applications can use the other 2G (total among them)
You can 'tweek' BOOT.INI with the /3GB switch, which will adjust the Kernel to 1G and Applications to 3G, but in my experience this degrades overall performance.
Kernel will use 2G for itself, paging, etc
Applications can use the other 2G (total among them)
You can 'tweek' BOOT.INI with the /3GB switch, which will adjust the Kernel to 1G and Applications to 3G, but in my experience this degrades overall performance.
Quote from segv:
That is not quite accurate, Nitro. Intel supports a feature on some server boards that they tout as processor redundancy. With this feature the second CPU is in standby, and at the time of failure the last valid state from the primary processor is copied to the standby. This is a different mode of operation than the typical MP scenario, which has both processors running simultaneously. Intel's redundancy feature can work as advertised in some cases. Handling a CPU failure gracefully with MP is a very delicate piece of software engineering, and it is something your desktop OS is not likely to do. Any failure that results in corruption is not likely to be handled gracefully in either scenario. I can say from experience that corruption is in the majority of such scenarios.
-segv
I remember way way back when we were using clusters of dual Athlon MP machines co-located and every once and a while my TOP would show one CPU instead of two. One burned out and she kept on going. And with regards to CPU performance: Sometimes its not the hardware, but the quality of the software you build and/or use.
Harping back to the ISP connection as a possible choke point for data retreival: I don't recall now cometothinkofit of ever being able to digest downloads at a fraction of the speed served up. However, in the early days of DSL my dialup connection was much more reliable. It may have been slow but it was current, while the DSL connection was fast but lagging some 2 to 3 hours in its refresh rate. I was constantly calling customer support reminding them that their service was inadequate. Today I am happy to say that my DSL service refresh rate is instaneous - I know because I program live web content.
So, the hardware out there when configured properly can handle any download: The Service providers both Cable and DSL have gotten their acts together (at least in this area), which leaves only the content provider - as the usual suspect for lagging delivery of information!
So, the hardware out there when configured properly can handle any download: The Service providers both Cable and DSL have gotten their acts together (at least in this area), which leaves only the content provider - as the usual suspect for lagging delivery of information!