>From a simulation standpoint, no, there isn't a limit. When you choose a
device ini file, it sets the density for each DRAM device and the storage
per rank. So if you wanted a very large memory system with a low density
DRAM device, the simulator would just give you many ranks per channel.
However, many ranks per channel are not really feasible in real life (2-8
ranks per channel is common depending on the DIMM and speed grade).

I wish we could figure out further what is wrong with a single channel
configuration -- I hope it's not a bug.

-Paul





On Fri, May 10, 2013 at 4:58 PM, Seyed Majid Zahedi <
[email protected]> wrote:

>  Paul,
>
> I increased the number of channels to two, and was able to simulate for 8G
> of RAM.
> Is there any limitation on the capacity that each channel can handle?
>
> Best,
> Seyed
>
>
>
> On 05/07/2013 11:54 PM, Paul Rosenfeld wrote:
>
> That's rather odd. The marss.dramsim branch is a bit out of date with
> respect to marss HEAD -- perhaps that's the problem? The DRAMSim2 code is
> so deep in the simulation code of marss that it shouldn't have a direct
> impact on something like launching QEMU.
>
>  I have been meaning to push out a bunch of changes but I'm currently
> busy working on other things.
>
>
> On Mon, May 6, 2013 at 5:23 PM, seyed <[email protected]> wrote:
>
>> Hi,
>>
>> I'm trying to simulate a machine with more than 4G memory size.
>> running 'free -m' on my own machine gives me:
>>              total       used       free     shared    buffers     cached
>> Mem:         64419      18475      45943          0        366       7571,
>> which means that I have enough memory on my own machine to launch qemu
>> with
>> more than 4G memory.
>> I tried 20G memory size with marss (without DRAMSim2), and it works.
>> However, running the same config with marss.dramsim doesn't work.
>> Actually, it gets stuck after redirecting to /dev/pts/*.
>> I also tried 8G and 12G, but non of them worked for me.
>> Does anyone have the same experience?
>>
>> Best,
>> Seyed
>>
>>
>> _______________________________________________
>> http://www.marss86.org
>> Marss86-Devel mailing list
>> [email protected]
>> https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel
>>
>
>
>
_______________________________________________
http://www.marss86.org
Marss86-Devel mailing list
[email protected]
https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel

Reply via email to