why xyz is has better performance than westfirst routing?

[cgraider]

Hello Noxim family , I came across strange problem simulating and evaluating routing algorithms. XY routing receives more packets and more throughput while westfirst has much less , while base on papers westfirst must have better results. what am i missing?

[mpalesi]

Hi,

What kind of traffic are you using? Please note that deterministic XY routing outperforms the other routing algorithms under uniform random traffic.

Regards,

Mau

On Fri, Dec 3, 2021 at 5:58 PM cgraider @.***> wrote:

Hello Noxim family , I came across strange problem simulating and evaluating routing algorithms. XY routing receives more packets and more throughput while westfirst has much less , while base on papers westfirst must have better results. what am i missing?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/davidepatti/noxim/issues/132, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABOP7WTDCF6MUP66TFO62STUPDZMFANCNFSM5JKEO2FQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

[cgraider]

Dear Mau, Thank you so much for pointing that out, yes its uniform random, since i have your attention here can i ask you few questions? 1a. I need a bit guide toward using SPLASH-2 or LDPC into Noxim. based on previous issues here, dr patti has pointed out that somehow traffic behavior must be mimicked by traffic table, is there any sample code to take a look at ? 1b. mainly my problem with real traffic is on what basis i should modify Packet injection rate 2a. To calculate new router area overhead i know i must use hardware synthesizers, is there any code for current architecture (E.g. Verilog) which implement NoP? 2b.Must papers use Synopsys design compiler, which i cannot afford, is there any free alternative which is trusted by journals and easy to work with?

Dr Palesi it is such an honor to talking to you. I have read few of your papers. Best regards. Mohammad

[mpalesi]

Hi Mohammad,

Find my answers inline below

On Sun, Dec 5, 2021 at 12:55 AM cgraider @.***> wrote:

1a. I need a bit guide toward using SPLASH-2 or LDPC into Noxim. based on previous issues here, dr patti has pointed out that somehow traffic behavior must be mimicked by traffic table, is there any sample code to take a look at ?

Unfortunately we cannot provide a traffic table for SPLASH-2 or LDPC benchmarks. You should use one of the many system-level simulators (es., GEM5) to trace the communication messages among cores and then use the traffic table option in noxim to mimic that traffic. Traffic table is simply a text file that reports in each line the following three information: source node id, destination node id, average packet injection rate

1b. mainly my problem with real traffic is on what basis i should modify

Packet injection rate

You should compute the packet injection rate to be included in the traffic table by following, for instance, the technique described in Catania V., Salvatore M., Palesi M., Patti D. (2018) Packetization of Shared-Memory Traces for Message Passing Oriented NoC Simulation. In High Performance Computing. ISC High Performance 2018. Lecture Notes in Computer Science, vol 10876. Springer, Cham. https://doi.org/10.1007/978-3-319-92040-5_16

2a. To calculate new router area overhead i know i must use hardware

synthesizers, is there any code for current architecture (E.g. Verilog) which implement NoP? 2b.Must papers use Synopsys design compiler, which i cannot afford, is there any free alternative which is trusted by journals and easy to work with?

Unfortunately we cannot locate our vhdl implementation of NoP. However, the NoP scheme it's very simple and should not be difficult implement it from scratch starting form the description of NoP algorithm reported in the paper. Regarding the synthesis tools, unfortunately, to the best of my knowledge there are not free alternatives to commercial tools.

Regards,

mau

[cgraider]

Thank you Dr mpalesi. I got all the answers. I mark this thread as Closed.