Zainstaluj Steam
zaloguj się
|
język
简体中文 (chiński uproszczony)
繁體中文 (chiński tradycyjny)
日本語 (japoński)
한국어 (koreański)
ไทย (tajski)
български (bułgarski)
Čeština (czeski)
Dansk (duński)
Deutsch (niemiecki)
English (angielski)
Español – España (hiszpański)
Español – Latinoamérica (hiszpański latynoamerykański)
Ελληνικά (grecki)
Français (francuski)
Italiano (włoski)
Bahasa Indonesia (indonezyjski)
Magyar (węgierski)
Nederlands (niderlandzki)
Norsk (norweski)
Português (portugalski – Portugalia)
Português – Brasil (portugalski brazylijski)
Română (rumuński)
Русский (rosyjski)
Suomi (fiński)
Svenska (szwedzki)
Türkçe (turecki)
Tiếng Việt (wietnamski)
Українська (ukraiński)
Zgłoś problem z tłumaczeniem
Zasady dyskusji i wytyczne
Zgłoś ten post
However if buying today, buying something older then Ryzen 2xxx series or Intel 8th/9th gen just wouldn't make sense
not so common knowledge as you might think if you read all the threads daily... xD
used 3rd-6th gen i5/i7 can be found for cheap and do good starter/budget gaming build if you add a decent gpu
but if you plan on spending over $600 on a new build, no reason not to go 8th gen intel i5, ryzen 2xxx or better
The only reason why Intel CPUs get more performance is becase they clock higher.
And, especially after the seccurity patches, Intel is really not a good one to buy in the budget market. (i.e i5-9400F.) Because its on par with the 2600 now, and the 2600 can take the lead with overclocking.
And, before anyone comments about >This is the IPC >procceds to link UBM.
You're just wrong.
Because UBM takes it all from averages, because people enter CPU that have been overclocked, or throttled to ♥♥♥♥, it's done on averages, not definitive.
It's a good baseline, but if you want to look at what the actual IPC then go check out Hardware Unboxeds video on the subject.
As for cores and threads, its not >how many do I use?
Its >how many do I need. / how many threads does my game scale with.
Most games will only NEED 6 threads at most to function properly. (No stutters or fps problems)
And most games will scale to about 4-6 threads. (Getting noticeable fps increases with each core added.)
More threads are a benefit, because windows will spread the load over loads of threads, causing less stutters, fps drops, and frametimes. And prevents anything takes some of the cpu away from the gaame (if you had just 4 cores for example) then the game wouldnt be effected, because there os plenty of extra threads.
So you dont NEED more than 6c. But its nice to have more.
Plus, in the future, games could utilise more threads.
https://www.youtube.com/watch?v=doGp5LI6igg
So calling it single core performance is wrong.
You need a combination of good ipc and high clockspeed for good single core performance.
btw just as clarification singlecore performance is a combination of ipc and clockspeed if im correct?
https://www.youtube.com/watch?v=m6RWEnUsLSY
i think there is nothing wrong with comparing ryzen to intel or gtx/rtx vs rx/vega if it is properly done and not misleading... aslong as intel can clock higher while maintaining the same ipc it is a better chip or not? not yet factoring $/performance ofc...
Just thinking about (a fictional) Intel CPU that costs 30% more, clocks 20% higher and in the end only offers 15% better performance in games. But people claim Intel got "better IPC" because of that.
that would be really misleading...
another interesting video:
https://www.youtube.com/watch?v=QXdgYlIXhP8
These aren't IPC tests. IPC is the rate that the cpu executes an instruction set i.e. instructions per second. Calling a test that includes i/o and interrupts an IPC test is wrong because an IPC test is usually cpu single-thread only. Multi-threaded can be tested but unless all threads are maxed out, it isn't meaningful. The unit of measurement is instructions per cycle, nothing else.
Depends on the game really, I was just playing the Division 2 free weekend on my old FX8350 GTX1070 ultra presets 1080p very smooth 60+ fps, for kicks I dropped the quality down to low 1080p and was getting 126ish fps I didn't take notice of the ram use.
FC 5, FCND and ROTR my ram easily creeps up around the 10GB+ mark when playing for more than 5min.
https://www.youtube.com/watch?v=0MS6ydxg7FQ
Yup cores/threads have gone up ^.
Ps, I'm not debating that the FX is behind.
To be exact:
IPC = Instruction Per Cycle
Cycle per Second = Clockrate
So you not have to look at a certaint ime for those tests but at your clockrate aswell. And to figure out the real IPC is hard as there not only things like turbo boost, multi thread tasks but also stuff like Speedstep...
Right but it comes all down to theories you can think about for years without anything meaningfull. In the end you have to do benchmarks to figure otu performance as nothing on the paper will tell you. High IPC is meaningless aswell as high clockrate if in the end the programm is so slow that it cant use either of that to its full potencial. The very reason why many old games perform so poor on new hardware has the programm loops are timed for much slower clocked and lower IPC hardware.