pikdum's blog

5600X vs 5800X3D on Linux

Recently bought a 5800X3D to replace my 5600X, and ran some benchmarks to see if it was worth it.

Update: It wasn't worth the 8% performance loss with most regular tasks. Returned it for a full refund.

Benchmarks

Did three quick ones, two work related and one gaming:

  1. starting PostgreSQL and applying a lot of migrations
  2. building a Next.js application
  3. running Cyberpunk 2077 on all the presets

Running Pop!_OS 22.04, on 2022-12-02.

PostgreSQL

5600X

./loop.sh  19.87s user 3.77s system 95% cpu 24.709 total
./loop.sh  19.75s user 3.53s system 95% cpu 24.336 total
./loop.sh  20.02s user 3.63s system 95% cpu 24.808 total

5800X3D

./loop.sh  22.02s user 3.83s system 95% cpu 27.086 total
./loop.sh  21.90s user 3.70s system 95% cpu 26.883 total
./loop.sh  21.87s user 3.77s system 95% cpu 26.879 total

Comparison

Total time, smaller is better.

. 5600X #1 ▏ 24.709 ██████████████████████▊
  5600X #2 ▏ 24.336 ██████████████████████▍
  5600X #3 ▏ 24.808 ██████████████████████▉
5800X3D #1 ▏ 27.086 █████████████████████████
5800X3D #2 ▏ 26.883 ████████████████████████▊
5800X3D #3 ▏ 26.879 ████████████████████████▊

On average, the 5800X3D was 91.35% the performance of the 5600X, or 8.65% slower.

Next.js

5600X

npm run build  210.73s user 7.92s system 307% cpu 1:11.06 total
npm run build  216.67s user 7.51s system 305% cpu 1:13.29 total
npm run build  214.16s user 7.20s system 311% cpu 1:10.98 total

5800X3D

npm run build  247.76s user 8.57s system 323% cpu 1:19.18 total
npm run build  242.66s user 8.39s system 333% cpu 1:15.27 total
npm run build  248.84s user 8.10s system 323% cpu 1:19.38 total

Comparison

Total time, smaller is better.

. 5600X #1 ▏ 1:11.06 ██████████████████████▍
  5600X #2 ▏ 1:13.29 ███████████████████████
  5600X #3 ▏ 1:10.98 ██████████████████████▎
5800X3D #1 ▏ 1:19.18 ████████████████████████▉
5800X3D #2 ▏ 1:15.27 ███████████████████████▋
5800X3D #3 ▏ 1:19.38 █████████████████████████

On average, the 5800X3D was 92.09% the performance of the 5600X, or 7.91% slower.

Cyberpunk 2077

These were all done at the various graphics presets, with a resolution of 2560x1440 and a 1070 Ti.

I should've just turned off FSR to help control things better. =/

5600X

Preset Name Average FPS Min FPS Max FPS Time Number of Frames FSR 2.1
Low 41.19 32.92 50.94 64.22 2645 Auto
Medium 40.88 31.56 51.93 64.24 2626 Auto
High 40.78 31.08 50.99 64.22 2619 Auto
Ultra 21.29 17.91 27.99 64.17 1366 Quality
Steam Deck 31.32 24.55 38.67 64.22 2011 Auto

5800X3D

Preset Name Average FPS Min FPS Max FPS Time Number of Frames FSR 2.1
Low 41.93 32.56 52.19 64.22 2693 Auto
Medium 41.75 32.41 53.80 64.23 2682 Auto
High 41.58 31.43 52.38 64.21 2670 Auto
Ultra 31.48 25.18 39.31 64.20 2021 Quality
Steam Deck 41.66 32.24 52.65 64.20 2675 Auto

Comparison

Frames per second, higher is better.

.        5600X Low ▏ 41.19 ████████████████████████▌
       5800X3D Low ▏ 41.93 █████████████████████████
      5600X Medium ▏ 40.88 ████████████████████████▎
    5800X3D Medium ▏ 41.75 ████████████████████████▉
        5600X High ▏ 40.78 ████████████████████████▎
      5800X3D High ▏ 41.58 ████████████████████████▊
       5600X Ultra ▏ 21.29 ████████████▋
     5800X3D Ultra ▏ 31.48 ██████████████████▊
  5600X Steam Deck ▏ 31.32 ██████████████████▋
5800X3D Steam Deck ▏ 41.66 ████████████████████████▊

For a preset, the 5800X3D was:

I've crossed out the Ultra and Steam Deck preset benchmarks, since I'm pretty sure I messed up the benchmarks there. Noticed that it wasn't consistently logging the same FSR option I had selected in the output, too. If I ever do this again, I'll try and do some more benchmarks and control things better.

Was it worth it?

No.