Can someone please explain to me the difference between single-core, dual processor technology and dual-core, single processor technology?
Dual-Core vs. Dual Processor
- Thread starter pythagoras
- Start date
A processor is a single chip put into a motherboard. A core is what is inside the processor. So if you have a Dual Processor machine you have 2 physical chips inside of it. If you have a dual core machine you still only have one chip inside.
It is possible to have a Dual Processor machine where each processor is a dual core chip. This would be a 4 core system, but still only a dual processor machine.
Basically the number of physical sockets you have placed a chip into = the number of processors. Each processor can have more than one core inside of it, hence dual core and whatnot.
It is possible to have a Dual Processor machine where each processor is a dual core chip. This would be a 4 core system, but still only a dual processor machine.
Basically the number of physical sockets you have placed a chip into = the number of processors. Each processor can have more than one core inside of it, hence dual core and whatnot.
Thanks for the fast reply. But what about the differences in the technology (in laymen's terms)? The reason I'm asking is because it seems to me that both "ideas" are essentially the same: one processor/core helps out the other processor/core, etc.
memory bandwidth (in the AMD implementation, at least) is also different - as physically separate chips have dedicated RAM...which is not necessarily an advantage.
in intel's case, there is essentially no difference (until conroe, anyways). they put two cpu dies on the same physical chip and both chips must go to the northbridge, first, before they can talk to the other processor, which is exactly what happens if there are two physically separates cpu dies on different chips. on the other hand, amd actually manufactures the dual core chips as a single die and has a special communication layer connection the cpu dies. they can talk directly to each other and grab information without needing to go through the northbridge... on a dual processor amd system, they must go through the north bridge first, so they can talk to each other. the physically separated cpus are actually slower than the dual core versions.
the capability for dual core (or any infinite number of cores for that matter within manufacturing capabilities) *should* be faster, but that all depends on the designers implementation. the more cores on a single die, the more complex the communication becomes. overhead increases. a lot more taking place. even in intel's less efficient implementation, you are still better off with dual core vs. physical dual cpu... you can get the same performance for cheaper! or put dual core in both slots for quad core... this is just the next step in cpu speed ramp ups... they've hit a wall with straight clock speeds and now they must work around it. intel's new core duo 2 and merom server chips (which can be up to quad core) are designed as a single package, so they can communicate directly to each other. processors are always evolving.
the capability for dual core (or any infinite number of cores for that matter within manufacturing capabilities) *should* be faster, but that all depends on the designers implementation. the more cores on a single die, the more complex the communication becomes. overhead increases. a lot more taking place. even in intel's less efficient implementation, you are still better off with dual core vs. physical dual cpu... you can get the same performance for cheaper! or put dual core in both slots for quad core... this is just the next step in cpu speed ramp ups... they've hit a wall with straight clock speeds and now they must work around it. intel's new core duo 2 and merom server chips (which can be up to quad core) are designed as a single package, so they can communicate directly to each other. processors are always evolving.
just fixing a minor typo, but good post overall.
As far as quad cores go, the workstation in my sig (High Def) that I use at work uses 2 AMD dual core processors. Windows sees all 4 cores, and typically spreads what I am doing accross all 4 to reduce the workload on a single core. So it runs all 4 at 25% instead of 1 at 100% unless I tell it to. Because these are AMD chips, all 4 cores can talk to each other at optimal speeds due to their hypertransport technology. But I suspect the internal communication is even faster between the 2 cores in each processor than between the cores on another processor. Since cores 0/1 are in processor 1 and Cores 2/3 are in processor 2. In the end the difference is negligible.
merom is the laptop version isn't it? whoops. i appreciate the catch krotchy! yeah, i figured that unless you are running your cpus at 100% ALL the time, you will not likely notice much difference between a dual core and dual cpu setup (all else being equal).
Contrary to popular belief, AMD dual cores cannot communicate to each other on die. They have to go through the northbridge and back just like Intel processors. Yonah and Conroe are the only processors known to be able to communicate directly with each other without leaving the processor itself.
http://www.xbitlabs.com/articles/cpu/display/dualcore-dtr-analysis.html
http://www.xbitlabs.com/articles/cpu/display/dualcore-dtr-analysis.html
you should check page 3 of the link you gave me. it clearly states the processors can communicate with each other,
"The two cores not only reside on the same silicon wafer but are also connected to the crossbar switch via a system request interface. All requests for data in system memory pass through the switch, so we can expect that the cores communicate between each other without the mediation of the system or memory bus."
the crossbar is the on die cpu communication link.
some more pretty pictures here:
http://www.pcper.com/article.php?aid=141&type=expert
"The two cores not only reside on the same silicon wafer but are also connected to the crossbar switch via a system request interface. All requests for data in system memory pass through the switch, so we can expect that the cores communicate between each other without the mediation of the system or memory bus."
the crossbar is the on die cpu communication link.
some more pretty pictures here:
http://www.pcper.com/article.php?aid=141&type=expert
They state that they expect it to work that way. That is how we all thought it would work as well. Their tests show otherwise.
"So, the results of reading unmodified data make me think that there?s no fast reading directly from another core?s cache. "
"So, I have to state that I can?t find any indication of direct data transfers from one execution core to another in the Athlon 64 X2 processor. According to my tests, the most recent copy of data is always read from system RAM. This must be a limitation of the MOESI protocol implementation. The following seems to happen when data are accessed: on receiving a read request probe read that the second core puts on the system bus, the first core performs a write-back of the modified cache line into memory. After this write or at the same time with it, the requested line is transferred to the second core. If the data in the first core?s cache haven?t been modified, they are read from system RAM. Why is there no direct transfer between the cores via the crossbar switch? Ask AMD?s engineers about that!
"ooh, i see what you are saying. looks like amd somehow did not implement that correctly? or the crossbar is used for something else i guess. then it comes down to memory controller speed and fsb bandwidth. i also saw a review where an amd dual core opteron was a few percentage points faster than the equivalently clocked (and revision) single core dual physical processor system. i need to go find that. i bet the onboard memory controller, to which both cpus have the same access, helps boost performance ever so slightly. but, i doubt anyone could notice a difference between a dual core vs. dual processor with every thing else being equal.
thanks for the interesting article, dguy. i should have read it more carefully.
thanks for the interesting article, dguy. i should have read it more carefully.
Well to give you somewhat of a picture about the Crossbar:
L2 cache has latency in the 10s of ns.
Crossbar operates at die frequency. So even at around 2500Mhz, the latency is still theoritically around 200ns.
L2 cache has latency in the 10s of ns.
Crossbar operates at die frequency. So even at around 2500Mhz, the latency is still theoritically around 200ns.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 22K
-
-
Facebook
Twitter