mik91392009 , you can put 2080 there, this is at least a two-fold increase, and this is without a CPU, which can still be put up to 9,9900k.

The catch is that there are no sensors on VRAM and VRM, and if you installed a smaller thermal pad than it was, then over time, VRAM or VRM will simply burn out and you will know it after the fact :) Therefore, temperature measurements after changing the thermal pads, if they fall, can more likely talk about the opposite is that not enough heat is removed from these elements.
Of course, it’s another matter if you are sure that the thermal pads were of exactly the same size.

Post has been editedtukreb - Yesterday, 01:20

Well, you have to test at your own risk :) it is confirmed that only 7970 and 8970 will be started from AMD

bureee
it’s better to take 7970 then, and better 8970 there is 4 GB of memory.

Yes :) apparently Total War somehow uniquely loaded the CPU in a random, unpredictable way.



Looks like a term paper :)

xutruy_Xixil,
It is very difficult to say, just play or work while it works stably :)

As an example, not long ago I decided to do an undervolting, put -150 all the tests went off with a bang, all the games, in general I didn’t know the problems for several days, and then I decided to install a new game Total war.
The first 4 hours of the game passed without problems, and then suddenly caught the BSOD, I was even surprised, I thought, maybe the curves are drava, then again, after a couple of hours of BSOD. In general, BSOD popped up in this game without any obvious signs, and then I remembered that I did undervolting, well, I raised the voltage to -100 and BSOD completely disappeared.

Therefore, I advise you to find what looks stable and use it while it works, if it breaks, it means that it is not configured correctly. :)

Posted on 08/02/2019, 03:53:


I would try to start memtest (new) and play with the tests, maybe there is a quick test without CPU cache.
P.Shttps: //www.memtest86 ....... support /ver_history.htm
DDR4 support appeared since version 6;)

Post has been editedtukreb - 02.08.19, 03:55

xutruy_Xixil,
Well, you used the 2014 program for the test :) (memtest86 4.3.7) I doubt that it works correctly with your CPU :)

memtest86 + also has not been updated since 2014-2015 and it certainly does not know how to forcibly turn off the CPU cache, memtest86 + generally writes a lot of unreliable information, for example, it says that it works in ECC, but actually not :)

By the way, a paid one of the latest versions of memtest86 can be found on torrents (at least in 2018);)

P.S. And yes, only through UEFI in paid memtest86 do all modern chips work and disabling the cache.
Those who have only legacy, I can’t test :), because by default the old version 4.3.7 will be loaded.

Post has been editedtukreb - 02.08.19, 03:39

xutruy_Xixil,
Well, I answered you above :) In general, for the selection of timing, the most ideal solution is the game benchmarks or games that are very demanding from memory, for one and the stability of all the hardware you will find out;)

xutruy_Xixil,
I advise you to test memory directly in Windows, where the data is actually in memory, for example, create a RAM disk and upload movies, for example. Although the catch is that you can’t check all the memory, but you don’t need to do this :) And if you check all the chips, then only with the CPU cache disabled.

Posted on 08/02/2019, 03:12:

xutruy_Xixil,
Of course, this is great speed! The weakest link in the CPU is slow memory and disks.

xutruy_Xixil,
same as 1 GB of text packed in 1 MB. If the data is the same or similar, why keep a copy? :)

No, I'm not worried about it, it's not AMD :)
G Skill F4-2400C16-16GRS

Posted on 08/02/2019, 03:01:

xutruy_Xixil,
I tell you as it is, I ate fish on these tests, I had to find memory tests that would identify the problem by type of work. As it turned out, only with the CPU cache disabled, did the problematic bars fail the memory test. Here it is. It’s the same with disks, with a standard test it shows sky-high numbers, and when you break through the cache it shows the actual state of the disk (with the exception of checking the surface, here fortunately programs can work directly in most cases).

Post has been editedtukreb - 02.08.19, 03:03

xutruy_Xixil,
this program thinks so :) In reality, everything goes through the CPU cache (and there remains if there is enough space), because of this, those who really test RAM do not use memtest86 +, but use the test via RAR (or something I did not understand the likeness of this, because I prefer memtest with disabling the cache) or memtest86 where you can disable the cache to really work with memory, not the CPU cache.

Here is one of the reasons, the curves of SSD tests in crystaldisk, they test at 1GB, when in the SSD (nvme basically) the cache is already more than 2 GB, in the end they test the cache, but they don’t get to the SSD.

Post has been editedtukreb - 02.08.19, 02:56

This test does not support new technologies :) too old, then it is better to test the utility through Windows.

Here the catch of these tests is that they do not test the RAM, but the processor cache, and only the latest versions of memtest86 and the Windows utility allow you to disable the CPU cache, but the tests will go there veryooooooo :)

The biggest mistake of those who test memory, they forget that all modern CPUs work through the cache, and then into memory, and therefore they test the CPU cache :)

Post has been editedtukreb - 02.08.19, 02:46

I hope thishttps://www.memtest86.com/and not a free analogue? :)
The free analog of MemTest86 + does not support the latest technologies and is not able to load memory to failure.

It was possible to test 100% defective memory that fell out during RAR unpacking.
So with Memtest86 +, she passed all the tests within a week, but in the paid program, she fell out within 2 days.

This is not a brand, but Chinese consumer goods, like Acer / HP.

More or less normal beech only Clevo, Alienware and most models at MSI and Dell.

And for me, it’s a mystery why people think Asus / Acer / HP are good beeches and keep buying them: D Trust in advertising less.

Post has been editedtukreb - 02.08.19, 01:37

It seems that if you short to ground, PWM will be informed about the maximum speed - 9999 RPM;)

dismit , Most likely it will be normal, it looks like one of those problems where the laptop is crooked with usb)

dismit put from a flash drive? : D
If yes, then look for DVD, this will be the easiest option.

Yura344,
for the test you need to set 16 GB in order to break through the cache, then you will find out the real disk speed, now he could just work out the cache in the end and get such numbers.

crow1994,
the sample is connected through the north bridge aka CPU, the south bridge has long been used only for disks and peripherals.

Post has been editedtukreb - 30.07.19, 02:23