Apr 18 2019 06:12 PM - edited Apr 19 2019 02:48 PM
Microsoft Edge Canary latest ver. Version 75.0.133.0 (Official build) canary (64-bit).
so I was trying to upload a 40 GB video file to YouTube but due to a network error it got halted. as you know YouTube lets you resume your video upload if you choose the same file to reupload within the next few days. but before it resumes, YouTube needs to check the file first, so it needs to read the whole file in order to know where it needs to resume the upload from. so at first my 40GB file was on my hard disk and in task manager it showed that Edge canary's disk usage is 50 MB/s. then i decided to copy the file to my SSD drive, (C:\) but again i saw that Edge canary is only using 50 MB/s disk usage even though it's accessing the file on an M.2 super fast SSD with read/write speeds of more than 1000 MB/s each.
Apr 19 2019 08:45 AM
@HotCakeX Do you see different results with the old Edge (Spartan) or with Chrome?
Chrome's sandbox architecture may impact disk read speeds for local files, although I'd be surprised to learn that the impact is more than a few percent.
Are you relying exclusively on the "Disk Usage" column in Task Manager for your analysis?
Apr 19 2019 12:11 PM
Apr 19 2019 01:00 PM
Apr 19 2019 01:03 PM
Apr 19 2019 02:00 PM
Apr 19 2019 02:15 PM
I poked at this a bit more, and made a little test page to benchmark the file read capabilities:
https://bayden.com/test/diskbenchmark.html
This simple page allows you to pick a local file, then the browser reads it into an arrayBuffer. This eliminates the network from the equation, and while it doesn't eliminate all other variables (memory allocation costs, etc), it does provide a less noisy picture. The page itself calculates the number of bytes read and milliseconds elapsed to provide a more precise measure of disk throughput.
When I use this page on a big file on Windows 10 RS5, I note that almost all of the disk activity is charged to "Antimalware Service Executable" and NOT the browser process; the browser process shows little disk activity at all.
While the results fluctuate (and this is unsurprising given how disk caches work, etc), I do see significant differences in effective throughput based on browser versions.
I haven't yet tried this on multiple machines.
Edge 18 RS5 (Spartan) -----------------
Read 2,139,348,308 bytes in 4651 ms
459,975,985 bytes/s
Firefox 68 -----------------
Read 2,139,348,308 bytes in 4394 ms
486,879,451 bytes/s
Chrome 73 Stable -----------------
Read 2,139,348,308 bytes in 2004 ms
1,067,539,075 bytes/s
Chrome 75 Canary -----------------
Read 2,139,348,308 bytes in 32868 ms
65,089,092 bytes/s
Edge 75.131 Dev -----------------
Read 2,139,348,308 bytes in 10102 ms
211,774,728 bytes/s
Edge 75.134 Canary -----------------
Trial #1
Read 2,139,348,308 bytes in 24106 ms
88,747,544 bytes/s
Trial #2
Read 2,139,348,308 bytes in 2426 ms
881,841,841.715 bytes/s
Trial #3
Read 2,139,348,308 bytes in 11456 ms
186,744,789.455 bytes/s
Apr 19 2019 02:22 PM
Apr 19 2019 02:42 PM
Apr 19 2019 02:51 PM - edited Apr 21 2019 07:14 PM
I edited my original post to emphasize on YouTube only.