login
Search
"This reduces native kernel max memory support from around 127 TB to around 120 TB. We also limit the Xen hypervisor to ~7 TB of physical memory - is that wise in the long run? Sure, current CPUs support 40 physical bits [1 TB] for now so it's all theoretical at this moment. My guess is that CPU makers will first extend the physical lines all the way up to 46-47 bits before they are willing to touch the logical model and extend the virtual space beyond 48 bits (47 bits of that available to kernel-space in practice - i.e. 128 TB). So eventually, in a few years, we'll feel some sort of crunch when the # of physical lines approaches the # of logical bits - just like when 32-bit felt a crunch when physical lines went to 31 and beyond."
In the future there will
In the future there will only be a need for around five computers, and shit-loads of cell phones.
And just how far away is that?
How far away are 128TB systems? Suppose that 1TB RAM systems are the peak today and that we double the peak every 18 months. That's 7 doublings, or about 10.5 years. Hmmm.... That seems like it's actually not all that far away.
Surely there must be an intervening "diminishing returns."
--
Program Intellivision and play Space Patrol!
Check the top 500
The top PC is roadrunner:
"The Roadrunner system has 98 terabytes of memory, and is housed in 278 refrigerator-sized, IBM BladeCenter® racks occupying 5,200 square feet."
It runs linux, so we might be pretty close already.
Cluster != computer
No, it isn't close. Roadrunner isn't SSI.
--
:wq
1TB RAM systems are not the
1TB RAM systems are not the peak. There are shipping systems with 10TB
of RAM in a single image.
orly?
Link?
Is that all directly addressable RAM?
Wow! I was sticking my finger in the wind guessing 1TB, figuring that on a logarithmic scale, a factor of 2 doesn't throw me off too much. Factor of 10 though is rather noticeable.
So that 10TB all directly addressable by each CPU? (I assume that's what you mean by "single image.") If so, then color me impressed!
--
Program Intellivision and play Space Patrol!
Yes it's all addressable
And since someone asked for a Link. The biggest box is the Altix 4700, which according
to http://www.sgi.com/products/servers/altix/4000/ supports upto 128TB single image
(note that is IA64, not x86). That is the theoretical maximum, but we know of running systems with >10 TB memory. Linux needed some special tunings to scale that high.
There are also a couple of other systems which support memory in the multi TB range.
The largest running x86 system I know of has 512GB. I had to fix a bug a couple
of years ago in the x86-64 port to go that high. It fortunately was a one-liner.
Also note that physical address bits are not only needed for memory. The large NUMA systems tend to have holey physical address maps too, so they need more physical space. And then there is memory mapped IO which also needs physical address space, sometimes a lot of it for special devices. e.g. consider some interconnect link that allows you
to do RDMA to large areas of memory of a huge cluster. Ideally you would want
to have address space for all of that.
If you go back to consumer space for example if you have a graphics card with 1GB
of memory you better have physical space for 1GB to address it directly.
I want my...
I want my 7 TB back you asshole!
Instead of shrink it, make it bigger.
Make so you can have petabyte or exabyte.
Where do you trolls come
Where do you trolls come from? Give me some background on what idiot makes such claims when they don't understand what's going on. FYI, the 7TB is reserved for the Xen hypervisor on Xen based platforms. You, obviously being too dim to 1) use something like Xen and 2) ever afford 120TB of RAM, will never run into this limitation before you escape your parents basement.
Dude, there is such a thing
Dude, there is such a thing as humour in this world.