hello everybody, we're here for a new video. this time i came up with a very important consideration about overclocking and using this type of liquid cooling cooler, i discovered a very important feature that could be affecting several computer overclocking systems. most likely and especially intel processors may be suffering from the lack of a small adjustment, not a problem, but adjustment that needs to be done in the bios setup of the motherboard in order
to get the correct result. i think this video is one of the most important ones about overclocking. include, most likely, it will tip over considerably several reviews made. there is an important adjustment that is being ignored and there is to be with these modern processors and i will show later. i was previously going to do a video comparing technical specifications with this cooler master masterliquid pro 120,
but doing a comparative analysis with amd fx and intel processors that do not overclock and get a performance with that same cooler corsair h45 being better, being that my performance was poor in overclocking. i found out later that it is not a problem with these liquid cooling coolers and it is not necessary to make a big change in them, i have seen many modifications where they put more liquid. i would even comment on this because i saw an interesting recommendation, but after figuring
out a setup problem in the bios setup that happens on the gigabyte boards i have noticed many reviews and specialized websites that show overclocking settings did not comment on a very important point in the configuration of bis setup and that it affects considerably when overclocking. so much so that systems without overclocking, processors locked, do not suffer as much with regard to the problem sufferers unlocked processors to overclock. this is because certain
settings will not be tampered with. so we see the people having a very good result with standard clocks, while overclocking the system already gave a problem. here, the ambient temperature is around 24 â° c and the air humidity is 80%. i access many specialized websites and see what people talk about when it comes to setting and temperature settings and what to watch for overclocking. we usually see people talking about vcore and other voltages that are critical to maintaining system stability. okay, this is true.
but there is an exaggeration in this part that is affecting at the time when overclocking efficiently. so let me mention one interesting thing: by the tests i'm doing now, it's possible to achieve good clocks with lower vcore and i'm doing it now with the computer i use. i've done tests, left images to show the tests made to make the video shorter, took screens of the test to show the result. so here you go, summarizing everything that people have been asking about vcore for certain clocks. so, 4.4ghz at 1.15 vcore, ok.
this is quite true, leaving the processor clock down is more interesting vcore, especially when using a high performance liquid cooling. on this xtremesystems.org site, the forum, which is specialized and where they have done many tests, they have high performance liquid cooling, or those air coolers that are very strong. we have a summary that a user did talking about the tests done with vcore in relation to the processor clock. there are other parameters on this site with an overclocking guide for the gigabyte z97 system. they tell you how to do the procedures
and i did not find the essential feature to be able to do an overclocking configuration with stability in the system using a liquid cooling of 120mm or 140mm. that question remains about the capacity of this type of cooler. initially i thought these coolers would not pay. it was a bad idea because i see a lot of people saying that, that's a problem. but all this is because of a detail that none of these sites speak. the sites usually talk about voltages. here, they give tips on uncore, base clock and etc. they are not wrong,
but there is a problem they did not realize and this has been affecting the overclocking of many machines. most likely all intel's 3rd to 6th generation processors that staff have been trying to overclock have not reached as much stability as possible, even using strong cooling. so i checked the tips and found nothing talking from a fundamental point. to understand let's go into an interesting fact. i have a 4-generation processor, 4790k. i was having problems with the old air cooler, hyper t4 which is a weak cooler because of the low pressure exerted on the processor base. it is necessary
to have good pressure at the base between the cooler and the processor. then we have a good performance. so i needed to change my cooler because of the heat in the area where i live. the temperature has risen sharply, reaching 35c. i ended up doing the liquid cooler replacement. i saw a lot of people saying that these 120mm liquid cooled coolers were not good and that it was important to make modifications. everybody said they had a problem. when i went to install mine, i started to have some problems. most of the problems were due to my inexperience regarding the installation of this type of cooler and other problems in relation to the spreading of the thermal paste.
however, there is a fundamental detail here and this will virtually zero many reviews, will invalidate them completely, due to overclocking done incorrectly. look closely at this figure that shows an intel 4th generation processor, those of 3rd are much the same, and the upper generation are similar. realize that these processors have several parts, these parts are increasingly sophisticated and the bios system can, as the manufacturer of the processor allows, have their own configurations. look at the part where the memory controller system is. the part where the mouse is flashing. this part, when overclocking, but not paying attention to some parameters of the bios, ends up doing without intention, not exactly an overclock, is that this system works
with different clocks as the memory is faster or slower. however, it gets locked in the configured clock, if it adjusts a frequency x, it gets locked at that same frequency x. so we have to be careful with that part too, especially when the bios, probably in the case of gigabyte there are bigger problems. these medium boards, as in my case. i have observed that the staff made those overclocking tips on top-of-the-line boards much more focused on overclocking and had inferior performance and did not find it possible to get 4.5ghz with room temperature stability of 25c with stability and liquid cooling of 120mm. it is possible, incredible as it may seem. i will show what affects the system when overclocking.
now here, we remember the cooler i'm using, a corsair h45. i'm using the standard fan (top motor) which is kind of noisy, but the fault, incredible as it may sound in performance is not that fan. it is possible to be practically without noise even in the heaviest benchmark. i will show images of benchmarks that have been made for a few hours and that are heavy and i am no longer having the problems of before. i'll even remove the previous videos because of this discovery. so this is the used cooler. remembering that i increased two more fans because my computer is closed with very bad airflow. i have a 120mm fan and 1200rpm at most and i leave it like this,
because it is not noisy. this fan helps prevent air stagnation. we start by seeing the bios setup. my board is a gigabyte ga-z97-ud3h-bk, so my base clock is at 100.19mhz because the board does a standard overclocking. see vcore at 1,176v. the temperature at 35 â° c. previously, there was instability and could not initialize the system, so benchmarking only 1,176v and 4.5ghz on all cores, completely impossible. see how the core temperature is. here, just the point where i want to get.
i even punched them to notice. we have the performance upgrade. by default, the gigabyte card leaves in auto. many times i left in auto or even increased, because when we do some overclocking, in some configurations, it will throw up performance upgrade, it is a percentage. the options range from 20 to 100%. but there's a detail: every time we change performance upgrade, it changes other settings. in one of them, and that is very important, is the memory frequency where it is exactly where it configures in the memory controller's clock. see the memory i use, a kingstom kvr of 1600mhz, the maximum of this memory is 1800mhz.
i do not have this bios profile for 1800mhz. from 1600 already jumps to 1833mhz or higher. when i set up performance upgrade at 40%, it leaves the memory clock at 1600mhz, according to the base clock (see that it has a slight overclock in the base clock). hat way, the memory controller will work on an expectation of accessing a memory of 1600mhz. perfect, that's the way it should work. in this situation, the memory controller will not overheat. if the memory clock was higher it probably would not be a problem, i say probably because i can not test it. i do not have faster memories to find out. but in relation to the processor, every time the memory is slower in its clock and the memory manager
is in a higher expectation of clock, it will overheat. in the next screen, we have the performance upgrade at 60%, it changed two things: cpu frequency and memory frequency. that is, the memory controller. we now have 1868mhz on the memory controller while kingston's memory is 1600mhz. the controller will be working with memory at this high 1868mhz clock and will overheat. this will make the system unstable. it will unsettle both the parent company and the cores.
so we have core crashes, which is actually because the controller is going crazy trying to access data, thinking that it is accessing a memory of 1868mhz and is actually accessing a 1600mhz. so, it stays the entire time in its cycles trying to synchronize with memory and can not. this overheats the memory controller and overheats the processor, the cores will ask for data from the controller that is at 1866mhz. see how perfomance upgrade is at 100%.
the controller goes to 2136mhz. if you leave it that way, even though the system boots up, running normally, it happens to be expecting the controller to be working with higher clock memories, but the memory is slow, it's only 16x in its multiplier. see that the multiplier went up, if i leave in auto similar fact happens.
0 comments:
Post a Comment