Hardlimit test bank

[cobito]

In the coming days I will prepare new rankings (which will not appear on the front page but on a specific rankings page). I would appreciate it if you could tell me which statistics you would find interesting. Initially I thought that a ranking of the best micros for games would be good.

Regarding CPUs for games I am not very sure about the characteristics. My idea is as follows (correct me if I am wrong):

1. I think that today, having more than 2 or 3 cores does not make any difference but if video broadcasting or recording services are used, having 4 cores is recommended.
2. In games, it is important to have power in arithmetic operations. I thought of giving the following weight: 75% for integer and floating point tests and 25% for the test of generic tasks (is the memory test omitted?).

In this case, results with 4 execution threads would start to be well received (they always are, but now they would be more) even from micros that have more than 4 cores since the frequency is not the same in many cases.

What do you think?

[cobito]

@krampak about the results of the 8700k that you sent a while ago, do you have the micro underclocked?

[krampak]

No, it was stock @cobito

[rul3s]

Hello @cobito!!!
First of all, congratulations on this initiative, I just found out now, as soon as I can I'll pass it on to all my PCs.
For now, I wanted to pass it on to the Ryzen 3 1200 and I see that it detects 1 CPU, 1 CORE, 4 Threads, check that out because it seems like there's a bug with core detection on the Ryzen.

Best regards!

[cobito]

Thanks for the report, @rul3s. Initially, the core/thread detection was designed only to select the number of threads to run and it's not something that's been polished too much. It also fails on some LGA775 micros. Until about a month from now, I won't be able to put in the hours that the test bench needs (apart from that bug you mentioned because there are quite a few in the central), but I'll take note.

I'm eagerly waiting for all your validations. This thread is very long so I'll say it again: validations in all modes are welcome.

[Yorus]

I might be missing something, but I can't log in to validate.

[cobito]

@yorus It should be working again now

[Yorus]

Nothing, I can't do it. It's supposed to be the same user and same password as the forum, right? I've tried in Chromium and IE11.

[Fassou]

@yorus Try putting the first letter in uppercase, I remember that at the beginning, depending on how you logged in, the username appeared, so something does influence it

[Yorus]

I have put it in both ways. I can't test it again today, but I'll try it from work tomorrow.

[cobito]

@fassou @yorus Apologies for the dizziness, there have been some changes in the local network where the server is located and I made the mistake of making the validation system dependent on the local IP: that's why it was failing. It's already been fixed.

[Yorus]

Thanks @cobito!, tomorrow I will test several devices.

[Yorus]

I have already tried it and it is perfect, although when validating a result it opened the YouTube page with a video of Rick Astley!

[cobito]

@yorus do you have oc on the micro or the memories?

[krampak]

Have you changed anything in the validation system @cobito? I see that on the October raffle page, everything is now invalidated.

[cobito]

@krampak What has changed since then is the structure of the database. It invalidates everything because it believes that test 4 is still the one for generic operations when, since version 1.2, test 4 is a memory test (tests 3 and 4 were swapped).

If you are interested in having some kind of statistics or verification of the validations in the style of the raffle page, let me know and I will implement it. From here to a week, I will be back with everything.

[Yorus]

@cobito said in Hardlimit test bench:

@yorus do you have oc on the micro or the memories?

Are you referring to the 4590s from the firm?. It's all in stock, micro at 3Ghz (3,7 with turbo) and memories at 1333.

[cobito]

@yorus OK, I'll take note. It seems like a bug in the validation system. Until a solution is applied, if it happens again you will have to repeat the test because the validation will be registered as "defective" and it will not be possible to send it.

Anyway, if it is what I think it is, the probability of the failure repeating is very, very low.

[Yorus]

I hadn't looked at the data in detail, you mean where it says Reported max frequency: 7000 MHz?

[cobito]

@yorus no, esa información es solo orientativa. Era si habías hecho oc a algún componente.