[SOLVED] Benchmarking some CPUS : Is going for lower resolution important ?

[BosnieYT]

Hey everyone, I wanted to benchmark for fun some of my CPU's and I noticed that some of them are really close to each other in 1440p. I heard that I should benchmark them at 720p/1080p to see how well the CPUs handle games. Is this true ?
Thanks for the help

 

[geofelt]

Once you get into 1440P and higher, the graphics card becomes the limiting factor, telling you little about cpu capability.
For a simple bench test try CPU-Z, look for the bench tab.

 

[SamirD]

If the purpose of benchmarking is to see the differences between them, searching the cpu on passmark and cpuworld will be a lot easier.

If the purpose of benchmarking is to see if they are working correctly, then you can use various different tools to do so and compare them with other results of that same testing tool. However, I would use multiple tools as a single tool would not be a definitive measurement.

 

[lga1156_ftw]

I have no idea what they did on passsmark for CPU benchmark testing. All of my old first gen i7 cpu scores (x58, lynnfield xeon / i7) dropped to half after some update they made. Using ANY other benchmark like 3d mark , userbenchmark give still same results. I would not trust passmark anymore for CPU benchmarking.

There is problems with userbenchmark too with theyre "testing". They seem to lean on bashing amd instead of intel lol.

I would only use cinebench r15 or r20 and maybe 3d mark, maybe passmark and userbenchmark are biased and bought.

 

[lga1156_ftw]

Actually found the reason, they started using new CPU instructions : https://www.passmark.com/forum/pc-hardware-and-benchmarks/46748-cpu-benchmarks-huge-changes

"- Using new CPU instructions (e.g. AVX512) only available in modern CPUs. "

This twists the results of 1 third of cpus used currently, i have no idea what is the point. Getting 3000 score on passmark and 225 fps in fortnite and 275-300 in csgo. This absolutely makes no sense at all why would they do this.

 

[Karadjgne]

Resolution has nothing to do with the cpu. Cpu deals with the game code. The gpu deals with resolution and detail levels.

If a cpu can pre-render 100 frames, that's what you get whether it's 480p or 5k. The difference being whether the gpu is capable of populating the amount of pixels per frame and shoving it on screen.

A 2080ti is used for the gpu because it eliminates potential gpu limits at lower resolutions. By the time you get to 1440p and up, the workload shifts from the cpu to the gpu making maximum fps all but non existant because of gpu limitations.

1080p is used mostly as a basis because the vast majority of people still use 1080p monitors, so results are quantifiable, 1440p is also used because of its rise in popularity with realism gamers and the use of flagship gpus.

Just as almost all tests are done with single player games, the game code itself being the limit, with multi-player games, the limits are varied according to the particular drops, server limits, AI etc. Tests have to be done in a controlled environment, multi-player is anything but controlled.

 

[SamirD]

Actually found the reason, they started using new CPU instructions : https://www.passmark.com/forum/pc-hardware-and-benchmarks/46748-cpu-benchmarks-huge-changes

"- Using new CPU instructions (e.g. AVX512) only available in modern CPUs. "

This twists the results of 1 third of cpus used currently, i have no idea what is the point. Getting 3000 score on passmark and 225 fps in fortnite and 275-300 in csgo. This absolutely makes no sense at all why would they do this.

Interesting. I saw the change too and initially it was terrible, but then they put back the original data have a new line now for 'cpu comparison' or something like that.

No benchmark is absolute. They are only good for relative comparisons of whatever they are measuring.

 

[Karadjgne]

They have to update periodically, same as testbenches get revamped periodically, or the old reliable Pentium 4 that was originally used would be somewhat lackluster. It's called progress. Many cpus such as the 9900k can use all 3 AVX, not just AVX and AVX2, which was primarily a HEDT thing and games are starting to use more AVX tech because of its particle adaption to physX photo-realistic explosions. BF5 would look kinda lame if a Panzer came busting through a wall and it shattered into only half a dozen large chunks instead of a gazillion pieces.

Games are evolving, platforms are evolving, the testing software has to evolve too or become obsolete.