How can I tell good CPU sample?

[Octopuss]

I'm going to assemble my upgrade with 5800X shortly and was wondering how can I tell whether my CPU is shit or higher up on the food chain.
What should I use for testing and what parameters am I looking for?

 

[Deleted member 193596]

look how far you can go with the curve optimizer.
my 5800x had some games crashing at -15 and bluescreens/reboots at -20

 

[Bill_Bright]

I'm going to assemble my upgrade with 5800X shortly and was wondering how can I tell whether my CPU is shit or higher up on the food chain.

I'm confused here. It sounds like you have already decided on the 5800X - or perhaps already purchased it. Have you?

The 5800X is an excellent processor. Are there better? Sure. Do you want to spend the money?

 

[Octopuss]

Yes I have, lol. I have all the components at home, ready to be installed, once I am finished with a notebook for my mother in law the witch.

 

[tabascosauz]

If you want to concretely measure silicon quality you probably still want to set all-core and see how high you can get (or how low you can get the volts). If you don't run it long-term, you'll still be fine. Unfortunately because everyone's hyper focused on PBO2, you won't have much to compare to aside from launch reviewers who did all-core testing, but you'll have a general idea. If your chip can't do 4.5 @ 1.35V or can do 4.8 @ 1.35V, etc.

Curve optimizer isn't something I'd say is accurately reflective of silicon quality, too many variances between boards in boosting behaviour, reporting telemetry and stock Vcore, as well as continued AGESA shenanigans.

Obviously, all-core only tells you the quality of the poorest quality core on die, but every chip's distribution of core quality is different so setting per-core curve means all comparisons go out the window anyways

 

[Bill_Bright]

Yes I have, lol. I have all the components at home, ready to be installed, once I am finished with a notebook for my mother in law the witch.

Two words of advice.

1. Never, as in never ever post derogatory comments about an in-law - especially a mother-in-law. It will, eventually, get back to them and then bite you in the a$$.

2. The day after you buy any high-tech item, it becomes obsolete with something better already out there. That applies to TVs, smartphones, audio reproduction equipment, computers and more. Therefore, I recommend you do NOT run any tests because they serve no purpose but to remind you something better is out there.

Build your computer and enjoy your games. That's my advice.

 

[Octopuss]

The idea behind wanting to do a test of some sort is the various reports of people having completely shitty CPUs that run hot (even for 5800X) and not even boosting too well or something. It's not like I would be returning mine and trying to play with luck anyway, but just to know where I stand.

 

[Bill_Bright]

Here's the problem as I see it. Regardless how much Prime95, or whatever benchmark, performance or stress test you choose, claims or markets their products to represent the real world, none do! Only the real world can do that and for that matter, only how YOU use your system in the real world can do that.

All you have to do is play your favorite game and use Core Temp to monitor your temps. If it does not over-heat and/or throttle back due to heat, enjoy your game.

All great scores on those tests really give you is bragging rights, or disappointment. This is why true comparative reviews should always be done in "double-blind" tests. When not done in this manner, the human brain tends to play evil, wicked, nasty head game on you.

 

[neatfeatguy]

It doesn't sound like you're looking to squeeze every little bit of every Mhz out of an overclock attempt, so use the system how you'd normally use it and monitor the temps.

If you're using the system for heavy gaming, play games.
If you're video editing, start editing stuff.
If you're transcoding things, get to doing it.
If you're using it to stream porno for hours on end, stream your porno.

In all honesty, all I did with my 5900x was run CineBench R20 and compare the results to other 5900x that have been reviewed to make sure mine was working at the same performance level. I game with mine and transcode videos (so much faster than my old i5-4670k) to my plex server. Temps right now tend to max out around 80C for my CPU - I have to pick up a couple more fans so the temps should drop a little for me once everything is all said and done.

 

[tabascosauz]

The idea behind wanting to do a test of some sort is the various reports of people having completely shitty CPUs that run hot (even for 5800X) and not even boosting too well or something. It's not like I would be returning mine and trying to play with luck anyway, but just to know where I stand.

The 5800X is hot from a variety of reasons. If you're on air, you'll have a hard time keeping it at the temps you're used to from 2000/3000/4000 chips. There's no avoiding that fact.

There are a few solutions you can try to mitigate the heat, including curve optimizer, ECO mode, and dialing in your own lower PPT/EDC settings. Just depends on the ST/MT perf you're willing to sacrifice for your use case.

 

[sneekypeet]

I think the guy who made the AMD RAM timing thingy software also made a test that gives a "result" as to the quality of the chips. Its accuracy is another matter.

This guy.... https://www.techpowerup.com/forums/threads/ryzen-dram-calculator-by-1usmus.246327/