Elevation errors after update to 9.20 from 8.00

I noticed the same thing, lower total elevation gain during my mountain bike rides. Today I went on a ride and wore my 5X on my left wrist and my old 910XT on my right wrist. Two very interesting things happened.The first is my 5X died after 1:10. The battery was at 85% when I started the ride so it is VERY strange that it died after only 70 minutes. The other difference was the elevation. The 5X reported 73 meters of gain while the 910XT showed 356 meters. HUGE difference. If you look at the attached graphs you can see the detail from the 910. I was mountain biking so there should be constant changes in elevation like the 910 shows. If you look at the graph from the 5X it is almost smooth with no detail of the elevation changes. It seems like there is something very wrong with the 5X.

Just as an FYI, this is also affecting the D2 Charlie watches, which are based on the Fenix 5X platform. And if you think it's distressing to have altitude errors when riding or hiking then imagine how I feel flying with this on my wrist! And since we've seen reports of this on the 935 and now the Fenix 5 Plus series I'm inclined to believe this is a software issue, and that it's taken so long tells me that they either don't care or it's something bigger than just a small bug somewhere. Either way, it sure would be nice to have an official acknowledgement.

Seems Garmin is working on a fix now as of the 935 forum. See here.

Seems Garmin is working on a fix now as of the 935 forum. See here.

That's great. I hope it is a global change. I wish they would officially acknowledge and provide an eta. My ride today was clocked at 700k and missing 1300k. It's unusable at this point.

I will add another example. I rode with both my 5X and my old FR 910XT on at the same time. Everything was almost identical with the exception of the elevation gain and graph. It showed 64m gain with the 5X and 179m gain with the 910XT. See the attached picture for the comparison. Actually the first 16 minutes are fine (very similar) but after that the 5X graph gets very smooth and loses all of the elevation detail. I have sent the examples to Garmin support. Hopefully they can fix it soon.

Yesterday I hiked with my kids 19,7 km with elevation ascent 973 m measured. I used Strava for elevation correction and it comes to 1212 m. So it looked exactly like again 25% less measured by watch compared to reality, so clearly the bug we are discussing here..
However, when I put the same route into route planner on Locus, the application calculated 962 m, what is much closer to what the watched measured than to what Strava calculated. Unfortunately we did not have any other device to compare the results, but this measurement looks just fine. I use 5x with latest beta 9.72.

Yesterday I hiked with my kids 19,7 km with elevation ascent 973 m measured. I used Strava for elevation correction and it comes to 1212 m. So it looked exactly like again 25% less measured by watch compared to reality, so clearly the bug we are discussing here..
However, when I put the same route into route planner on Locus, the application calculated 962 m, what is much closer to what the watched measured than to what Strava calculated. Unfortunately we did not have any other device to compare the results, but this measurement looks just fine. I use 5x with latest beta 9.72.

Sorry but " measured by watch compared to reality" is completely wrong! As has been noted in a number of discussions, not to mention on the sites on many of the apps that provide the "corrected elevation" feature, this function is designed to correct primarily for watches that do not have a barometric elevation capability and use only GPS. This is why the default for barometric elevation watches is generally no elevation correction. To add to this, every site does elevation correction differently (as you note in your post the difference), plus it also depends on the data base and location (thus the quality of the data or resolution). Hilly areas will "correct" even more poorly as the horizontal location inaccuracy might put you down over a steep hill so any geo-data lookup will be off. Most sites also use a smoothing algorithm as well.
I'm not saying that some are not having issues with the elevation (I don't seem to be) but, in general, a barometric altimeter will be the most accurate and responsive. There is little point in comparing or assuming correct any "elevation corrected" data.

Michael yes, I wanted to say in other words that it looks I may not have ths issue with elevation.

I'll add my 2c. It's definitely a software issue, not even a question. I have data from multiple devices 3HR, 5X, and EDGE using different fw releases.

Anyone not having the issue are doing activities that don't have fast elevation changes. Again, my gravel rides that have a straight are a little off and require correction but is -10-20%. Fast ele changes I am -60% or more.

I had a technical, fast 10m ride yesterday. My watch posted 243ft. Last year running the identical (fixed course) I posted 902ft. Strava corrected posted 809ft. So you can see that the corrected and last year are very similar and I would accept the results. 243, not a chance. We did over 250 in the opening mile of the ride.

Regarding accuracy, I get things may not be precise but in previous versions my watch was functional and I would classify as very good. It may be slightly off but overall it did not require any corrections from Strava, etc.