dash runs fast for a while, but becomes slow later

[amyzx]

Hi, I learnt your dash-ex hash tables and have two questions.

1. Why the number of stash bucket is 2, which is different from the description in your paper.

2. Run time seems strange.

I did not change the parameters in your run.sh, where the warm-up workload is 10M, and benckmark workload is 190M.

In initial several experiments, dash runs pretty slow with the same shell script. But later, it seems that dash-ex is stuck after loading warm-up 5M data.

I feel confused because I have already run dash-ex about 6 times, where dash-ex works pretty well. I try to rebuild the project, but it is still stuck in loading warm-up phase.

The following is my DCPMM's states.

Could you give me some suggestions about how to figure out the reason? Looking forward to your reply, thanks!

[baotonglu]

Hi,

For (1), our paper also stated that the number of stash buckets is 2. Where is the difference? For (2), possibly your PM pool has been full after running 6 times. So you should first delete the pool and create the new pool for Dash.

Baotong

[amyzx]

Hi, @baotonglu Thanks for your quick reply!

For 1, sorry, I checked your paper again. It was stated in experiment part that you used 2 stash buckets. So you are right. I thought the stash bucket should be 4 because of the description in Section 4.3.

For 2, I am a beginner who just starts learning DCPMM, could you share some links where I can find the command to clean PM pool? In dash, is it possible to add function to clean pool after every experiment? I find you have a function called Close_pool in allocator.h.

I print the pmem_ex.data information by using command pmempool info -s pmem_ex.data. But it seems that it has a large available space. But my program still is stuck during loading phase, this time it can only load 40000 warm-up records.

In addtion, could you explain the meaning of depth_count in struct Directory?

[baotonglu]

Deleting the pool before running the experiment is ok, like this. Depth_count is used for directory halving, which is not fully supported in the current version of Dash. So don't care about it now.

Baotong

[amyzx]

Hi, I have already delete the file using rm command. But it still cannot run normally. I print the idx in Load function, and find that it stops when tries to insert _worklod[31985].

Are there any other possible solutions?

[baotonglu]

"I feel confused because I have already run dash-ex about 6 times, where dash-ex works pretty well." => So it worked well previously?

[amyzx]

Yes, it works well when I run the experiment during the initial times. I run it successfully about the first 6 times, just modifying the pm directory in shell script run.sh and test_pmem.cpp.

[baotonglu]

I guess you should give up your modifications, clone our repo again, build and test it. Without enough information, I cannot help you figure out the reason.

[amyzx]

Ok, I will have a try. Thanks!