Predictor | Crown Width/Root Flare Predictor

Here is a brief example

1. Create your smoothCon object, using x

When you call nooffset you are simply subtracting the offset from the linear predictor.

If you want them in the same plot, you can pull the data from your fit with trt_fit1[["plots"]][[1]]$data$fit and plot them yourself. I looked at the plot style from the mgcViz github. You can add a second axis or scale as necessary.

Yes, data misalignment could explain your 2x slowdown for small arrays that fit in L1d. You'd hope that with every other load/store being a cache-line split, it might only slow down by a factor of 1.5x, not 2, if a split load or store cost 2 accesses to L1d instead of 1.

But it has extra effects like replays of uops dependent on the load result that apparently account for the rest of the problem, either making out-of-order exec less able to overlap work and hide latency, or directly running into bottlenecks like "split registers".

ld_blocks.no_sr counts number of times cache-line split loads are temporarily blocked because all resources for handling the split accesses are in use.

When a load execution unit detects that the load splits across a cache line, it has to save the first part somewhere (apparently in a "split register") and then access the 2nd cache line. On Intel SnB-family CPUs like yours, this 2nd access doesn't require the RS to dispatch the load uop to the port again; the load execution unit just does it a few cycles later. (But presumably can't accept another load in the same cycle as that 2nd access.)

• https://chat.stackoverflow.com/transcript/message/48426108#48426108 - uops waiting for the result of a cache-split load will get replayed.
• Are load ops deallocated from the RS when they dispatch, complete or some other time? But the load itself can leave the RS earlier.
• How can I accurately benchmark unaligned access speed on x86_64? general stuff on split load penalties.

The extra latency of split loads, and also the potential replays of uops waiting for those loads results, is another factor, but those are also fairly direct consequences of misaligned loads. Lots of counts for ld_blocks.no_sr tells you that the CPU actually ran out of split registers and could otherwise be doing more work, but had to stall because of the unaligned load itself, not just other effects.

You could also look for the front-end stalling due to the ROB or RS being full, if you want to investigate the details, but not being able to execute split loads will make that happen more. So probably all the back-end stalling is a consequence of the unaligned loads (and maybe stores if commit from store buffer to L1d is also a bottleneck.)

On a 100KB I reproduce the issue: 1075ns vs 1412ns. On 1 MB I don't think I see it.

Data alignment doesn't normally make that much difference for large arrays (except with 512-bit vectors). With a cache line (2x YMM vectors) arriving less frequently, the back-end has time to work through the extra overhead of unaligned loads / stores and still keep up. HW prefetch does a good enough job that it can still max out the per-core L3 bandwidth. Seeing a smaller effect for a size that fits in L2 but not L1d (like 100kiB) is expected.

Of course, most kinds of execution bottlenecks would show similar effects, even something as simple as un-optimized code that does some extra store/reloads for each vector of array data. So this alone doesn't prove that it was misalignment causing the slowdowns for small sizes that do fit in L1d, like your 10 KiB. But that's clearly the most sensible conclusion.

Code alignment or other front-end bottlenecks seem not to be the problem; most of your uops are coming from the DSB, according to idq.dsb_uops. (A significant number aren't, but not a big percentage difference between slow vs. fast.)

How can I mitigate the impact of the Intel jcc erratum on gcc? can be important on Skylake-derived microarchitectures like yours; it's even possible that's why your idq.dsb_uops isn't closer to your uops_issued.any.

Code readability, short-circuiting and it is not guaranteed that Ord will always outperform a || operand. Computer systems are more complicated than expected, even though they are man-made.

There was a case where a for loop with a much more complicated condition ran faster on an IBM. The CPU didn't cool and thus instructions were executed faster, that was a possible reason. What I am trying to say, focus on other areas to improve code than fighting small-cases which will differ depending on the CPU and the boolean evaluation (compiler optimizations).

You are mixing up the order of y_true and y_pred in brier_score. Here is a working example:

From the caret documentation, we see that variable importance in linear models corresponds to the absolute value of the t-statistic for each covariate. So, we can manually compute it, as I do in the code below.

lm() automatically converts categorical variables as dummies. So, to get the importance of each covariate, we have to sum over dummies. I did not find a way to automate this, so if you want to apply my solution to a different set of variables, you need to be careful in choosing the items of t.stats to be summed.

Finally, we can use results for plotting. I just used the baseline function for a bar plot, but you can customize it as you want (maybe also using the ggplot2 package for better visualization).

Ps when you provide a reproducible example, remember to load all the needed packages.

Pps summing over dummies may be sensitive to the base level of the dummy we are using (i.e., the level we omit from the regression). I do not know if that could be an issue.

By writing predictor_iris <- "Species", you are basically saving a string object in predictor_iris. Thus, when you run lda_ex, I guess you incur in some error concerning the formula object in train(), since you are trying to predict a string using vectors of covariates.

Indeed, I tried the following toy example:

We can do the following (I'm using dplyr 1.0.7 but anything >= 1.0.0 should work):

Create a function that scales