Not populating all of the memory channels for a CPU reminds me of the saying "You can't put 10lbs of 'stuff' into a 5lb box"

You can, but if you're lucky the handles come off when you try to lift it. If you're unlucky, the bottom fall out after you've lifted it.

At work we've got weight limits for various containers. The general opinion is that "If it fits, how can it be over weight?" Of course some loads are much, much, much heavier than others, even though they're heaps smaller. Density is a concept some people have problems coming to terms with.

---

Grant
Darwin NT

Density is a concept some people have problems coming to terms with.

I don't know, I run into some QUITE dense people ;)

Tom

---

A proud member of the OFA (Old Farts Association).

An update: I now have 6 dimms (3 dimms per socket) and now runtimes are ranging from just under 1 hr to 2.5 hrs depending on the type of work being processed using the AVX app

i.e I added 2 more dimms this week(last week for us Australians due to the time difference)
Running 29 cpu tasks + 1 GPU
new PCM data:

|---------------------------------------||---------------------------------------|
|--             Socket  0             --||--             Socket  1             --|
|---------------------------------------||---------------------------------------|
|--     Memory Channel Monitoring     --||--     Memory Channel Monitoring     --|
|---------------------------------------||---------------------------------------|
|-- Mem Ch  0: Reads (MB/s):    -1.00 --||-- Mem Ch  0: Reads (MB/s):  2642.42 --|
|--            Writes(MB/s):    -1.00 --||--            Writes(MB/s):  1504.65 --|
|-- Mem Ch  1: Reads (MB/s):  3632.90 --||-- Mem Ch  1: Reads (MB/s):  2644.07 --|
|--            Writes(MB/s):  1998.68 --||--            Writes(MB/s):  1505.33 --|
|-- Mem Ch  2: Reads (MB/s):  3628.68 --||-- Mem Ch  2: Reads (MB/s):    -1.00 --|
|--            Writes(MB/s):  2003.15 --||--            Writes(MB/s):    -1.00 --|
|-- Mem Ch  3: Reads (MB/s):  3518.19 --||-- Mem Ch  3: Reads (MB/s):  2647.91 --|
|--            Writes(MB/s):  1942.43 --||--            Writes(MB/s):  1510.73 --|
|-- NODE 0 Mem Read (MB/s) : 10779.77 --||-- NODE 1 Mem Read (MB/s) :  7934.40 --|
|-- NODE 0 Mem Write(MB/s) :  5944.26 --||-- NODE 1 Mem Write(MB/s) :  4520.70 --|
|-- NODE 0 P. Write (T/s):     901985 --||-- NODE 1 P. Write (T/s):     521928 --|
|-- NODE 0 Memory (MB/s):    16724.03 --||-- NODE 1 Memory (MB/s):    12455.10 --|
|---------------------------------------||---------------------------------------|
|---------------------------------------||---------------------------------------|
|--                 System Read Throughput(MB/s):      18714.17                --|
|--                System Write Throughput(MB/s):      10464.96                --|
|--               System Memory Throughput(MB/s):      29179.13                --|
|---------------------------------------||---------------------------------------|

A big improvement in memory bandwidth, particularly writes.

---

Grant
Darwin NT

A big improvement in memory bandwidth, particularly writes.

Indeed it is, I wonder what the difference is when I add an additional dimm per socket

A big improvement in memory bandwidth, particularly writes.

Indeed it is, I wonder what the difference is when I add an additional dimm per socket

What app generates the memory chart output?
I can run it on mine for comparison.
Intel states the max memory bandwidth for these CPUs is 51.2 GB/s.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

A big improvement in memory bandwidth, particularly writes.

Indeed it is, I wonder what the difference is when I add an additional dimm per socket

What app generates the memory chart output?
I can run it on mine for comparison.
Intel states the max memory bandwidth for these CPUs is 51.2 GB/s.

You kinda need to compile it yourself: https://github.com/opcm/pcm

A big improvement in memory bandwidth, particularly writes.

Indeed it is, I wonder what the difference is when I add an additional dimm per socket

What app generates the memory chart output?
I can run it on mine for comparison.
Intel states the max memory bandwidth for these CPUs is 51.2 GB/s.

You kinda need to compile it yourself: https://github.com/opcm/pcm

Looks like their last build before moving it to github has vanished.
There are compiled exes but none of the other required files. :/
Figures I don't have an environment setup to build things since I torn apart my VM server with it.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

A big improvement in memory bandwidth, particularly writes.

Indeed it is, I wonder what the difference is when I add an additional dimm per socket

What app generates the memory chart output?
I can run it on mine for comparison.
Intel states the max memory bandwidth for these CPUs is 51.2 GB/s.

You kinda need to compile it yourself: https://github.com/opcm/pcm

Looks like their last build before moving it to github has vanished.
There are compiled exes but none of the other required files. :/
Figures I don't have an environment setup to build things since I torn apart my VM server with it.

Thats a shame, I don't know how legal it is to upload my compile binaries and the dll

An update: I now have 6 dimms (3 dimms per socket) and now runtimes are ranging from just under 1 hr to 2.5 hrs depending on the type of work being processed using the AVX app

i.e I added 2 more dimms this week(last week for us Australians due to the time difference)
Running 29 cpu tasks + 1 GPU

Those are very respectable cpu times. It looks as if a "AMD Radeon R9 200 Series" should run gpu tasks in the 15 minute range for AVX.

Tom

---

A proud member of the OFA (Old Farts Association).

It looks as if a "AMD Radeon R9 200 Series" should run gpu tasks in the 15 minute range for AVX.

?
AMD Radeon R9 200 Series is a video card.
AVX is a CPU instruction set.

---

Grant
Darwin NT