search

Quick700 / Quick700

QuickChick / FuzzChick project for CIS-700
Apache License 2.0
2 stars 1 forks source link

Slides For Friday #14

Closed Chobbes closed 5 years ago

Chobbes commented 5 years ago

Are people okay with beamer, or would you prefer to use something else like PowerPoint?

I don't know how to create a powerpoint, but if somebody can send me an invite link to one I'm happy to start adding slides, and more than happy to start beamer slides.

@liyishuai @moqingyan, thoughts?

moqingyan commented 5 years ago

The beamer thing looks pretty cool, though I am more familiar with google slides, I don’t mind take it a try.

On Nov 27, 2018, at 10:59 AM, Calvin Beck notifications@github.com wrote:

beamer

Chobbes commented 5 years ago

Oh, I forgot that google slides was an option. We could do that too.

Chobbes commented 5 years ago

Alright. Nobody screamed. I'm adding a beamer presentation to the git because logging into a specific google account is a pain.

https://github.com/Quick700/Quick700/tree/master/presentation

Feel free to change up the look or whatever. I just like orange.

Also, remember, if you add a code example you need \begin{frame}[fragile], at least with lstlisting, otherwise it does weird things. If you are confused ask for help!

Chobbes commented 5 years ago

I've added a bunch of slides. Feel free to add more, but we should coordinate somewhat on how much time it's going to take.

I will not be available after today.

Chobbes commented 5 years ago

@liyishuai @moqingyan how's your progress?

liyishuai commented 5 years ago

I've added performance results to the slides. My next step is to mention results on DeepWeb.

Chobbes commented 5 years ago

@liyishuai okay. I think you should probably mention some concrete numbers if possible.

liyishuai commented 5 years ago

Leo told me that the timing of experiments were not accurate. I’ll try to describe the quantitative results as much as possible.

Chobbes commented 5 years ago

If the timings for those experiments were not accurate then should we even report on them? That doesn't sound very good, especially if we didn't even run those experiments.

If the numbers are inaccurate to the point where we shouldn't report those numbers, I don't think we should report a conclusion derived from those numbers.

liyishuai commented 5 years ago

I can rerun the experiments on my side, and report if the results agree on order of magnitude.

Chobbes commented 5 years ago

I believe one issue was that Leo was running with instrumentation on both QuickChick and FuzzChick, and you would actually want to compare with QuickChick with no instrumentation because the instrumentation is a 2x slowdown?

Was there any other issue in the methodology?

liyishuai commented 5 years ago

Another source of error is that the experiments were run in parallel, which magnifies the bottleneck effect of disk access.

Chobbes commented 5 years ago

@moqingyan do you know how long it takes you to get through the slides you added? It takes me about 10 minutes to get through mine.

moqingyan commented 5 years ago

@Chobbes I tried to time it once and it takes 12 minutes. I think I can practice it down to 10 minutes and probably get rid of some slides as well.

moqingyan commented 5 years ago

I deleted some of the slides and the time I tried to present is around 8 minutes.

Chobbes commented 5 years ago

We're probably right on target then, that's great!