qcmadness

引用:

原帖由 Henry 於 2011-8-17 02:59 發表

Increase in bandwidth needs the help of architecture in order to have improvements.
860 and 950 basically is the same chip.

If as you said, bandwidth is not so important, I think CUDA cards in datac ...

if you know tesla has a much lower bandwidth than their desktop counterparts, do you still insist that bandwidth should be as important?

you cannot cut some memory channels in gpus easily because it is tied with ROPs and other parts such as L2 cache that are considered architectural...
you are comparing things that are vastly different

because 860 and 950 are architectually similar, it just proves that adding another memory channel YIELDS LITTLE.

qcmadness

引用:

原帖由 Henry 於 2011-8-17 02:59 發表

Increase in bandwidth needs the help of architecture in order to have improvements.
860 and 950 basically is the same chip.

If as you said, bandwidth is not so important, I think CUDA cards in datac ...

and the large bandwidth is only useful when 2-way and 4-way processors are used

you still don't know why intel and amd wants more memory channels for servers

for desktop, even single channel memory will not scarify much unless IGP / iGPU is used

qcmadness

you still need some backing benchmarks to prove that adding more memory channels does help

qcmadness

http://www.anandtech.com/show/45 ... ing-the-best-ddr3/8

引用:

I think we confirmed what we pretty much knew all along: Sandy Bridge's improved memory controller has all but eliminated the need for extreme memory bandwidth, at least for this architecture. It's only when you get down to DDR3-1333 that you see a minor performance penalty. The sweet spot appears to be at DDR3-1600, where you will see a minor performance increase over DDR3-1333 with only a slight increase in cost. The performance increase gained by going up to DDR3-1866 or DDR3-2133 isn't nearly as pronounced.

SB nowhere requires another 20GB/s memory bandwidth when ~20GB/s is enough.

Henry

引用:

原帖由 qcmadness 於 2011-8-17 03:52 發表
http://www.anandtech.com/show/45 ... ing-the-best-ddr3/8



SB nowhere requires another 20GB/s memory bandwidth when ~20GB/s is enough.

I have already read the review long time ago.
As I said, depends on arch, OPs performance, cores, also the tasks.
If CPU cannot feed the RAM bandwidth, CPU is the bottleneck.
If CPU need to wait for RAM, RAM is the bottleneck.

Now SB-E has 6-8 cores, some may have even 10-12 cores (in DP/MP), can they be satisfied with same amount of bandwidth (2 channel) as those has 4 cores?
I don't think so, though 4 channel may be an overkill in single CPU system.

Can those GF110 or GF104 cores be satisfied with only e.g. 1/5 of their enormous bandwidth (20% RAM freq, same bits) under Tesla platforms, I am really eager to know.
As those platforms claimed much more FLOPs can be juiced out than normal CPUs.

qcmadness

引用:

原帖由 Henry 於 2011-8-17 04:48 發表

I have already read the review long time ago.
As I said, depends on arch, OPs performance, cores, also the tasks.
If CPU cannot feed the RAM bandwidth, CPU is the bottleneck.
If CPU need to wait for  ...

1. you still have not shown any tasks which scales well with memory bandwidth
2. SB-E for desktop only 4 and 6 oores. 8 cores is your IMAGINARY product
3. http://www.anandtech.com/bench/P ... 5&i=221.222.223 is this clear enough?

Henry

引用:

原帖由 qcmadness 於 2011-8-17 05:01 發表


1. you still have not shown any tasks which scales well with memory bandwidth
2. SB-E for desktop only 4 and 6 oores. 8 cores is your IMAGINARY product
3. http://www.anandtech.com/bench/Product/313? ...

1. I am sorry I cannot show this moment, but does not scale well =/= does not scale. (Something similar should be the first test in your link 3)
2. 8 core is still available in LGA2011 form, and I cannot say nobody will use it in single CPU.
3. I would like to know why the result in 1st has difference if it is not bandwidth dependent, though I am convinced with the results.

qcmadness

引用:

原帖由 Henry 於 2011-8-17 05:39 發表

1. I am sorry I cannot show this moment, but does not scale well =/= does not scale. (Something similar should be the first test in your link 3)
2. 8 core is still available in LGA2011 form, and I ca ...

1. of course it scales, but is the scaling meaningful is another question
2. i don't think so, server socket for 4 qpi connections is different
3. decompression / compression is particularly bandwidth sensitive

qcmadness

Edit: both server and desktop versions of SB-E have LGA 2011
but compatibility is unknown and only 2 QPI lanes are available...

bebird

this post becomes very technical

Henry

引用:

原帖由 qcmadness 於 2011-8-17 11:58 發表


1. of course it scales, but is the scaling meaningful is another question
2. i don't think so, server socket for 4 qpi connections is different
3. decompression / compression is particularly bandwid ...

Video compression is also compression (under different algorithm), that's why I say it should be bandwidth sensitive.
MP side should also use LGA2011 if not got wrong, as to take over LGA1567.

[ 本帖最後由 Henry 於 2011-8-17 16:15 編輯 ]