14900KS vs 9800X3D tests in 5 CPU-limited games. RT/PT enabled at 4k w/ DLSS Ultra Performance.
Surprisingly, 14900KS was faster. Oblivion and Stalker2 were much better on intel. Cyberpunk is close despite the 9800X3D winning the benchmark by a big margin.
Then the 1% ๐
Comparison video for the Oblivion fix on 5800X3D and RX9070 showing 50% improvement overall.
Most of the system settings can be seen in the CapFrameX overlay.
https://t.co/AEuDKnOsOT
@GamingOptimiz Same comparison from the 5800X3D RX9070 system. Had to get OBS working on it, and included the system settings this time.
https://t.co/AEuDKnOsOT
@CapFrameX 14900KS(DDR5 8400) improvement is similar to 14700KF(DDR4 4200) at around 20%
But it's only 15% compared to the noHT+8E cores config that I use on it daily.
@CapFrameX Took a single run on KF with older result thrown in. 2nd runs on 9850X3D were slightly slower.
14700KF - 20% faster
9850X3D - over 30% faster ๐
The fewer cores or AMD arch. gets more improvement.
14900KS (noHT 8E cores ) is the remaining one to test, probably on tuesday.
@GamingOptimiz I also used the gamepass version but not RAM/VRAM constrained with 5800X3D on 32GB 3800MT/s and 9070.
https://t.co/4OnpCWxS6p
But I use a CPU-bound scene deliberately chosen for CPU testing with raytracing.
@SKHYJINX Just a quick test and it's amazing being 50% faster in the scene I use for benchmarking.
1st is with the fix, 2nd is a rerun I took after removing it and bottom is yesterday's run.
Hope that everything still works correctly. ๐
@Timebringer@SKHYJINX There should be but I simply set the capture hotkey and duration after its install, and that's all you really need.
Launch a game and it should show up with the overlay, press the capture hotkey and benchmark for the duration set. That's it ๐
@beyond_fps@CapFrameX I agree that the scene I use is not going to be representative of normal gameplay because it was deliberately chosen for CPU bottleneck.
Video for the custom 1min run I do.
https://t.co/gFcyEbtzmQ
@Sebasti66855537 โบ๏ธ got it from @SKHYJINX tweet.
And a correction regarding the lows, I keep 5% instead of the 0.1% since they are more random.
This one is with 0.1% and another run with the fix.
@SKHYJINX Just a quick test and it's amazing being 50% faster in the scene I use for benchmarking.
1st is with the fix, 2nd is a rerun I took after removing it and bottom is yesterday's run.
Hope that everything still works correctly. ๐
@SKHYJINX Interesting. I have been using it for CPU bottleneck comparisons with RT/PT enabled.
RPL leads X3D and also AMD GPUs have a distinct advantage in it.
@mpr_reviews My custom run video. It has worse performance in another town square up north. Drops to 49fps in overlay at around 2min
https://t.co/8SIou88BTM
@mpr_reviews I compared the PCGH save and a Novigrad save on 9850X3D and 14900KS.
PCGH save was faster was on AMD, but my run was very similar with intel slightly ahead on the lows.
https://t.co/0SP6p2UVM7
@CapFrameX@PCGH_Redaktion Just did quick tests of the pcgh save and my own custom save in Novigrad on X3D/14KS and interesting results.
pcgh save is much faster on X3D but mine is slightly slower.
btw greatly appreciate the new look with system info.