[gem5-users] Avoiding L1 cache/ forcing L1 misses in my protocol

Hi everyone, I am using gem5 that simulates an architecture that takes some information from a core and passes into the Network Interface. I use Mesh topology and Two-Level MESI Protocol. I use ruby and garnet for my memory and interconnect modelling. My need is to bypass L1 cache for data

[gem5-users] Avoiding L1 cache/ forcing L1 misses in my protocol

Hi everyone, I am using gem5 that simulates an architecture that takes some information from a core and passes into the Network Interface. I use Mesh topology and Two-Level MESI Protocol. I use ruby and garnet for my memory and interconnect modelling. My need is to bypass L1 cache for data

Re: [gem5-users] Garnet 2.0: Torus is being Deadlocked for 256 nodes with injection rate = 0.14

The lectures on NoC deadlocks on my website might help understand the problem: http://tusharkrishna.ece.gatech.edu/wp-content/uploads/sites/175/2016/10/L05-Deadlocks-I.pdf http://tusharkrishna.ece.gatech.edu/wp-content/uploads/sites/175/2016/10/L06-Deadlocks-II.pdf

Re: [gem5-users] coherence state machine - 1 event per clock

Thanks Jason! I see that this option called ports and it does work. Is there any other impact of this option on the implementation or only on the transition_per_cycle ? About the example, on my case it should be split to at least 2 different clocks, but I understand the cases that 2 transition

[gem5-users] Problem about PerfectSwitch

I want to ask about PerfectSwitch. PerfectSwitch is a router or arbiter? http://www.m5sim.org/Simple And Why PerfectSwitch need to have routing table? I know PerfectSwitch has Round Robin scheduling to choose a virtual network perform and the switch will see the message destination. But Why it

Re: [gem5-users] Garnet 2.0: Torus is being Deadlocked for 256 nodes with injection rate = 0.14

Plus... The default weight based routing selects the free VC. If so, then why I need to do the VC partitioning as you mentioned in the Torus network. *VC Selection (VS)*: The winner of SA selects a free VC (if HEAD/HEAD_TAIL flit) from its output port. I think this is a very important issue for

Re: [gem5-users] Garnet 2.0: Torus is being Deadlocked for 256 nodes with injection rate = 0.14

Thanks a lot for reply. This is little bit terrible news for me. However, as far I know garnet1.0 don't have the deadlock issue with Torus. Please let me know how can I implement a VC partitioning scheme. Is it possible? I can configure the routing algorithm with particular channel selection,

Re: [gem5-users] Garnet 2.0: Torus is being Deadlocked for 256 nodes with injection rate = 0.14

Hi Faisal, The Torus topology deadlocks as it has rings in each dimension unless one implements a VC partitioning scheme or bubble flow control. That's why I removed torus from the default topologies provided by garnet2.0. If you implement torus, you will have to implement deadlock freedom.

[gem5-users] Garnet 2.0: Torus is being Deadlocked for 256 nodes with injection rate = 0.14

Dear All, I like to simulate the synthetic traffic analysis for Torus for 256 nodes with uniform traffic. However, the network is showing latency degradation after 0.14 injection rate (flit latency = 33.044985 for 0.14 and flit latency = 38.244770 for 0.13 ), which could be the possible case of

[gem5-users] coherence state machine - 1 event per clock

Hello, I'm a new gem5 user and simulating the coherent protocol of MESI with two level of caches. I see on the ProtocolTrace log file that several transition of the same address can occur at the same cycle. See below: 7263 0L2Cache UnblockMT_SB>SS [0x400,