How to test CPU?
My fleas, I write what I do not know how to test percent. how?http://www.playground.ru/blogs/other/istoriya_o_tom_kak_amd_ryzen_stal_igrovym_protsessorom-246507/
it is advisable to test with one memory and one frequency
program it's just a peck-peck
games too. it is better to have these kind of slay with a huge stock of videoplace and there to screw everything is ultra maximum resolution and extremely ultra minimal without Sunga.
so like peck-peck noble MSAA in some engines is a deferred method and can be performed with a large time delay and the next stage after creation of all scenes. also it's great testing, if so how roughly speaking the processor may okazatsya work on a 4 stroke more and kick need to get rid of agrovolokno amount of data through the defective linnie Express Express controllers in the case of the bourgeois system kokokoko else 1151 socket
to test the ultra minimal sense osobo not so in this case mainly being tested or the total bandwidth to either the cache or memory.
of course, in all cases, nor should the bakery to work anything outside the Internet
MunchkiN 616
As it turns out it. I'm gonna try your Prots give defecate in WatchDogs2. strictly direct resolution of 800 by 600 or a little higher 1280 by 720 has no wheels?
Set the minimum resolution (800x600 etc.), disable the settings that load of GP (smoothing, etc.) - that is, to create the conditions most load on the graphics card.
The point is that the CPU was loaded to the max and was drawing as many frames a graphics card don't have to limit it. Ideally, should be close to 100% CPU usage and incomplete utilization of GP - that is, the conditions of bottlneck.
If the CPU is loaded enough, and the card works on full is more of a benchmarking GP.
Ganond
Here's how to test and see max fps
And here
And at max settings you'll test the video card, she's dying.... now, if you had 1080 or 1080TI no one would not say a word !
warp 37
This rarely happens to 100% at low ,God forbid 80% ! For 1080TI I'm afraid not even enough 6800K because they are dispersed so that the card is not in idle ))))
HAAKS
if 800x600 is that you can. but as a rule custom resolution does not appear full screen and MSPU does not work there
and so the theory is simple. the performance of Coca is the game almost always rests on the GPU. therefore, the goal of testing is to understand the performance of ultrametal in isolation as from the video card, considering that she several times overlaps the limit performance of the rest of the system. it is endlessly productive and of course productive CPU part.
when testing the card the opposite way, but there cannot be made infinitely productive processor. therefore, there should be extremely cool CPU overclocking for objectively pure test. for the happiness of a graphics card is not so strong so close to the limits of current top-end systems.
MunchkiN 616
thank you for the answer . graphics on max or is it at least something I'm confused) warp advises to Set the minimum resolution (800x600 etc.), disable the settings that load of GP (smoothing, etc.).I've always thought that the way the test percent. and you covetously everything on max except to put razresheniya or I not correctly understood you.
Enough to run any game, to cut down anti-aliasing and reduce the resolution to 720p, to find a segment where the video card will squander less than 99%. And profit.
For example World of Conflict good bench contains testit processor and loads all of the flows.
I also would have been interesting to see for example on the speed of caches and RAM in AIDA. His acceleration, temperature, overclocking RAM. But you do not know how, sorry.
HAAKS
technically, it is not always possible to know exactly what to chart and how many loads the CPU so it is wise to put as much as possible and test at minimum resolution. minimum resolution in order to just unload the brine and Shader units. this does not mean, however, that the performance will not be utakata in graphics in terms of shaders that work with some samples not tied to the resolution. for example, the simplest is the shadows shads the map. you can for example think that for example materials with transparency hamper the GPU. for example if the grass is still there shining and the shadow. removing travon you can save a lot of resources to the GPU but there is gravely for the CPU and a separate transparency, for example has some load on the CPU but not comparable with a small card.
a smoothing smoothing strife. usually turn it off because the CPU in fact it can influence within one percent and the GPU by 50%. and there are all sorts of supersampling which strictly speaking do not require a delay in the formation of the frame. for example ssaa. a good Vysoka of SSAA in fact turn 720p a vacuum in close to 2K for the video card. therefore we can say that some settings can be squeamish.
Vinni-Pukh
Know how and I like the blog set? at least 50 people fleas I looked and there is no incentive.
Ganond
Ganond wrote:
and I like the blog set?
no. What to put ? The blog is so-so.
Ganond
I think you podeshevel with the motherboard, just noticed that the power subsystem is that its like a cheap nurse, hence the exorbitant voltage to the CPU when overclocking.
Man I bought asus b350 and risen 5 1600 and complains that wound up on 3.7 at 1.37 volts, then the voltage is being chased but generally prohibitive need.
You have 10 phases to take something of type asus prime x370.
Ganond
The best acceleration. The better the power supply the lower the voltage of the processor and generally more stable and better chasing. 1.4 volt he's already worth the money and the power consumption is very high. Even on tests Riesen 7 shows how, after 3.8 it starts food to eat if not in themselves.
Just compare the ASUS PRIME X370 10 power phases, and you have 6, some even 5. ASRock X370 TAICHI at all 17. ASUS CROSSHAIR VI HERO 12 phases.
Just under acceleration on Intel Z270 with 5 phases is not very suitable and they are cheaper all the info collected when the mother chose. With Riesen similar situation.
You realize there he and 1.37 volts but then he even 1.45 is not very stable at the frequency of 3.8 already, he is guilty of the type rayzen 5 this rejection of 7 so the bad copies of a bad chase, but it seems like his mother just under acceleration is not particularly suitable..
If something here in the picture an example of where to look if you're not interested before that.
Spoiler
Ganond
Ganond wrote:
Well duck and go to hell. Intol same crap.
Where?))))
I'm the first person who compares your hardware with the test obzorschika, and another test cut. There are clowns like you!