# Within-chain parallelization with brms

**URL:** <https://discourse.mc-stan.org/t/within-chain-parallelization-with-brms/18693>\
**Category:** brms\
**Created:** [October 15, 2020, 2:04am UTC](https://discourse.mc-stan.org/t/within-chain-parallelization-with-brms/18693 "2020-10-15T02:04:27Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Gang](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.mc-stan.org/gang/32/13572_2.png) [@Gang](https://discourse.mc-stan.org/u/Gang)\
**Post date:** [October 15, 2020, 2:04am UTC](https://discourse.mc-stan.org/t/within-chain-parallelization-with-brms/18693/1 "2020-10-15T02:04:28Z")

</div>

I see the following suggestion in the vignette about within-chain parallelization using brms:

> For a given Stan model one should usually choose the number of chains and the number of threads per chain to be equal to the number of (physical) cores one wishes to use.

Suppose I have 24 CPUs on a computer. I would choose 6 threads per chain for a model using 4 chains to take full advantage of potential speed gain. I expected to see that all the 24 CPUs would be active most of the time. However, it does not seem to be the case: when I checked the CPU usage, I rarely saw more than 4 CPUs were used.

If I want to simultaneously run 5 separate models, can I still choose 6 threads per chain for each of the 5 models? Is this considered hyper-threading?

---

<div class="post-metadata">

**Author:** ![wds15](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.mc-stan.org/wds15/32/908_2.png) [@wds15](https://discourse.mc-stan.org/u/wds15)\
**Post date:** [October 15, 2020, 7:59am UTC](https://discourse.mc-stan.org/t/within-chain-parallelization-with-brms/18693/2 "2020-10-15T07:59:38Z")

</div>

> [@Gang](#):
>
> However, it does not seem to be the case: when I checked the CPU usage, I rarely saw more than 4 CPUs were used.

How large is your data set? The minimal grainsize enforced is 100 which limits parallelism in some cases with few data rows.

> [@Gang](#):
>
> If I want to simultaneously run 5 separate models, can I still choose 6 threads per chain for each of the 5 models? Is this considered hyper-threading?

You should not fire off more Stan threads at the same time than you have physical CPU cores… but there can always be exceptions to this “rule of thumb”.

---

<div class="post-metadata">

**Author:** ![Gang](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.mc-stan.org/gang/32/13572_2.png) [@Gang](https://discourse.mc-stan.org/u/Gang)\
**Post date:** [October 15, 2020, 12:45pm UTC](https://discourse.mc-stan.org/t/within-chain-parallelization-with-brms/18693/3 "2020-10-15T12:45:48Z")

</div>

> [@wds15](#):
>
> How large is your data set?

There are 32005 rows and 4 columns in the dataset in the long format of the data.frame.

> [@](#):
>
> The minimal grainsize enforced is 100 which limits parallelism in some cases with few data rows.

Could you elaborate how is grainsize defined?

> [@](#):
>
> You should not fire off more Stan threads at the same time than you have physical CPU cores

In my test of a model with 4 chains and 8 threads per chain, I didn’t see more than 4 CPUs involved during the few times I checked the CPU usage. The runtime was 4 times shorter compared to the original job with no within-chain parallelization, and this is why I thought I might be able to run some simultaneous models through hyper-threading.

---

<div class="post-metadata">

**Author:** ![wds15](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.mc-stan.org/wds15/32/908_2.png) [@wds15](https://discourse.mc-stan.org/u/wds15)\
**Post date:** [October 15, 2020, 1:59pm UTC](https://discourse.mc-stan.org/t/within-chain-parallelization-with-brms/18693/4 "2020-10-15T13:59:40Z")

</div>

The default grainsize is max(100, data rows / (2 \* # of threads requested)… I do not think you need to bother with this.

Getting a 4x speedup is very good with 8 threads per chains.

Just try out things. It‘s super complicated to tell people what to do here… do you want greatest throughput or shortest walltime for model runs will determine what you are going to do.