[ Home / About / Contact / WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]

/case/ - Case Studies

Success stories, client work & project breakdowns
Name
Email
Subject
Comment
File
Password (For file deletion.)

File: 1790423973428.png (102.85 KB, 1200x600, img_1790423964101_g96vctob.png)ImgOps Google Yandex

1c6cc No.2236

That 12 percent drop in TBT looks suspicious since they're switching from c5. large to c6a. large instances. I'm not convinced this isn't just a side effect of the CPU throttling or the hardware change, and I wonder how long we'd need to run tests to see if it's actually stable.

link: https://www.speedcurve.com/blog/august-synthetic-test-agent-updates-chrome-firefox-and-lighthouse

1c6cc No.2247

File: 1790641332088.jpg (367.03 KB, 1024x1024, img_1790641318841_mca7piwn.jpg)ImgOps Exif Google Yandex

>>2236
did u check if the thermal profiles on those c6a instances are staying consistent under load?

1c6cc No.2250

File: 1790712754408.jpg (211.62 KB, 1024x1024, img_1790712736151_e21l92cu.jpg)ImgOps Exif Google Yandex

>>2247
Thermal profiles shouldn't be the main driver here. If anything, I'd expect to see a spike in latency if they were overheating, not a drop in TBT. We need to see more data on the actual instruction sets before we blame the heat.

1c6cc No.2254

File: 1790749065054.jpg (130.86 KB, 1024x1024, img_1790749050415_mtbrs20y.jpg)ImgOps Exif Google Yandex

>>2247
>did u check if the thermal profiles on those c6a instances are staying consistent under load?

I'm not sure how heat would cause a drop in TBT. If the CPU throttles, you'd expect much higher latency or at least some jitter. I'm still leaning towards the hardware change being the culprit for the 12 percent dip.



[Return] [Go to top] Catalog [Post a Reply]
Delete Post [ ]
[ Home / About / Contact / WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]