I wrote a simple driver for Tachyon, for the benchmarking framework. It appears that the performance difference between us and V8 is even less on my home machine (Core 2 Quad):
Mean values (s) Benchmark Tachyon Google V8 fib 7.86 5.63 loop_sum 5.01 4.53
- Maxime
Afficher les réponses par date
Really cool! (I'm not being sarcastic... really!)
I'd like to see a "gcc -O2" column... but that can wait.
Just to come back to an earlier comment by Bruno, it would be interesting to be able to give a name to the columns (for example: "Tachyon no opt", "Tachyon full opt", "gcc -O1", "gcc -O2", etc).
Marc
On 2010-12-08, at 7:24 PM, Maxime Chevalier-Boisvert wrote:
I wrote a simple driver for Tachyon, for the benchmarking framework. It appears that the performance difference between us and V8 is even less on my home machine (Core 2 Quad):
Mean values (s) Benchmark Tachyon Google V8 fib 7.86 5.63 loop_sum 5.01 4.53
- Maxime
Tachyon-list mailing list Tachyon-list@iro.umontreal.ca https://webmail.iro.umontreal.ca/mailman/listinfo/tachyon-list
On 2010-12-08, at 7:24 PM, Maxime Chevalier-Boisvert wrote:
I wrote a simple driver for Tachyon, for the benchmarking framework. It appears that the performance difference between us and V8 is even less on my home machine (Core 2 Quad):
Mean values (s) Benchmark Tachyon Google V8 fib 7.86 5.63 loop_sum 5.01 4.53
- Maxime
Tachyon-list mailing list Tachyon-list@iro.umontreal.ca https://webmail.iro.umontreal.ca/mailman/listinfo/tachyon-list
Another comment... For being able to quickly compare systems, it is often better to have the "relative measure". Call one of the columns (the first?) the "reference", and report the others relative to it. If you think the absolute measure is important, then add the relative measure next to it in parentheses. Another style I like is the one in the following page: http://www.iro.umontreal.ca/~gambit/bench.html . The fastest time is highlighted in green (number of milliseconds) and the other values are relative to the fastest time.
Marc
On 2010-12-08, at 7:24 PM, Maxime Chevalier-Boisvert wrote:
I wrote a simple driver for Tachyon, for the benchmarking framework. It appears that the performance difference between us and V8 is even less on my home machine (Core 2 Quad):
Mean values (s) Benchmark Tachyon Google V8 fib 7.86 5.63 loop_sum 5.01 4.53
What are the commands to generate this table? I ran "make perf" but no .html file is generated. Also, the .json file that is generated only contains time for Tachyon.
Marc
I haven't added a command to the makefile. You can call ./bench/d8.sh -genReport <json_file_name>.
If there are no times for V8 (empty array?) it must mean that the V8 driver isn't working on your system for some reason. You can print the standard output of the child process by uncommenting line 72 in bench/platforms.js
- Maxime
On 10-12-08 08:40 PM, Marc Feeley wrote:
On 2010-12-08, at 7:24 PM, Maxime Chevalier-Boisvert wrote:
I wrote a simple driver for Tachyon, for the benchmarking framework. It appears that the performance difference between us and V8 is even less on my home machine (Core 2 Quad):
Mean values (s) Benchmark Tachyon Google V8 fib 7.86 5.63 loop_sum 5.01 4.53
What are the commands to generate this table? I ran "make perf" but no .html file is generated. Also, the .json file that is generated only contains time for Tachyon.
Marc
Tachyon-list mailing list Tachyon-list@iro.umontreal.ca https://webmail.iro.umontreal.ca/mailman/listinfo/tachyon-list
-
Marc Feeley -
Maxime Chevalier-Boisvert