LONG-TERM Instability with Ryzen 1600x on a x470 chipset?!

[PatXioPC]

My system preset:
Ryzen 1600x @4.0ghz 1.45V SB Voltage 1.22V 
memory DDR4-2666 OC SoC 1.03V
OS: Windows 10 pro (an older version)
Cooler: Masterliquid lite 120 push-pull fan config

Hey, so I was OCing my 1600x with a b350 board, and since it couldn't maintain a stable frequency of 3.8ghz at 1.402V and memories up to dual channel 2666mhz, I recently upgraded to an ASUS x470-F mobo, so can as well upgrade my chip in future. But after I got it, I OC'd it easily to 4.05ghz at 1.45V (note that anything below that voltage wasn't stable inside the OS) and the post was successful but it caused instability with my windows drivers, crashed with a BSOD and got it corrupt. Now when I reinstall my OS, I'm on an older version of it and neither way it was stable. 

I've learned recently that I've degraded my cpu because of running it with higher voltages with my previous mobo. You may wonder why, well, because my previous b350 board had a very crappy VRM and power delivery solution, it wouldn't post unless you make the voltages higher. It generally wasn't designed for efficient operation under lower voltages. Now since I've degraded my cpu, it will be stable at higher voltages like 1.45 for now and in future maybe will have to go more. 

Now just want to know if selling this CPU is a good choice, getting the 2700x so my system gets totally futureproof. And are the 2nd gen chips can hit higer core ratio or oven ref core under the same voltages besides the silicon lottery fact?

[nick name]

I certainly hope you don't plan on selling that CPU unless you are going to be very honest with the buyer. 

 

And your last sentence needs to be proof read again please. I cannot understand what you're asking. 

[TheGlenlivet]

Why?

Do you need 4.0+ performance?  What are you doing with it?  The voltage you are running is very high for long term use.  

[PatXioPC]

9 minutes ago, nick name said:

I certainly hope you don't plan on selling that CPU unless you are going to be very honest with the buyer. 

 

And your last sentence needs to be proof read again please. I cannot understand what you're asking. 

For the last words, just wanted to know if the 2nd gen chips are any better in terms of hitting higher clocks under the same voltage that a 1st gen cpu wouldn't go.

[tzxazrael]

basically, it sounds like you fried your CPU by overdoing it. if you reset everything to stock settings, and the chip is no longer stable, then it's already dying and soon to be dead outright. selling it in that condition without informing the buyer beforehand would basically be scamming them. put it up on your shelf, to remind you of the dangers of overclocking. 

 

if you get a 2700X, just don't overclock it. it has more than enough power at stock to handle almost any task, particularly gaming.

 

as for if the 2nd gen chips can go higher than 1st gen... i'd think it's all still just a matter of "silicon lottery" still.

[PatXioPC]

2 minutes ago, TheGlenlivet said:

Why?

Do you need 4.0+ performance?  What are you doing with it?  The voltage you are running is very high for long term use.  

4ghz does make a difference in my workflow.  Like just when I'm encoding a 4k to h.264 4k preset or 1080p, I see up to 50 secs of difference when I have my memory at 2666mhz and cpu at 4.05ghz and since I have to do this several times a day, it is a thing to consider having. Into the program interface of a CAD, simulation, 3d render slightly makes difference but in terms of 2d images, it does feel like it.

[TheGlenlivet]

1 minute ago, PatXioPC said:

4ghz does make a difference in my workflow.  Like just when I'm encoding a 4k to h.264 4k preset or 1080p, I see up to 50 secs of difference when I have my memory at 2666mhz and cpu at 4.05ghz and since I have to do this several times a day, it is a thing to consider having. Into the program interface of a CAD, simulation, 3d render slightly makes difference but in terms of 2d images, it does feel like it.

I have seen reviews of the 2700X performing better at stock settings than overclocked.  That might actually work for you...

[nick name]

The 2700X can be made to perform above 4.0GHz with the right motherboard using XFR2 and Precision Boost Overdrive. I'm not certain to what degree you can make it stable at speeds above 4.0GHz with your workflow, however. It'll get pretty hot under heavy loads. If you are simply running it stock it tries to stay around 3.9GHz during heavy workloads, but will downclock after building up some heat. 

[nick name]

Here is an example of what to expect using Cinebench. 

 

 

[PatXioPC]

10 minutes ago, TheGlenlivet said:

I have seen reviews of the 2700X performing better at stock settings than overclocked.  That might actually work for you...

Well, what I've learned from my experience is that the raw single core performance is as important as better MC scores in a lot of applications. And applications that use 2-4 threads will benefit by having a better core ratio rather than boosting clocks for just one core on stock settings. But definitely, I want to keep an eye on what voltage I'm setting it with. I definitely don't want to get misled like I was before. 

[TheGlenlivet]

1 minute ago, PatXioPC said:

Well, what I've learned from my experience is that the raw single core performance is as important as better MC scores in a lot of applications. And applications that use 2-4 threads will benefit by having a better core ratio rather than boosting clocks for just one core on stock settings. But definitely, I want to keep an eye on what voltage I'm setting it with. I definitely don't want to get misled like I was before. 

You are correct.  The one thing to consider though, especially with Ryzen+, is the Precision boost technology it has that boosts cores as needed for different applications.  The second Generation of this technology seems to be effective enough that benchmarks I saw with all core OCs were only about as good as stock settings.  

[nick name]

What several people are mentioning about stock is true to a certain degree. You will want to leave the ability for the CPU to control its own speed by not setting a static core clock, however, you can manipulate XFR2 and PBO options to allow for it to boost itself higher and longer. You will need to mess with it a little bit to find what is stable in your working environment and hardware.

 

Another Cinebench example:

 

With this run I used the Performance Enhancer Level 4 (This is a pre-set setting specific to the Asus Crosshair VII) and combined it with a voltage offset of around -.017v (I can't remember precisely). I am using the offset to try to bring down temperatures. 

 

In that Cinebench screenshot the score in 40th place is completely stock CPU speeds with overclocked RAM. The score in 30th position is the run with the above described settings. You can see the score in 30th is listed at the stock 3.7GHz. Keep in mind that the score would be higher without HWiNFO64 and Chrome open. You'll also see the max temp in HWiNFO is lower than the previous screenshot I posted because of the undervoltage I applied (or at least I think so). 

 

 

 

 

[PatXioPC]

14 hours ago, nick name said:

What several people are mentioning about stock is true to a certain degree. You will want to leave the ability for the CPU to control its own speed by not setting a static core clock, however, you can manipulate XFR2 and PBO options to allow for it to boost itself higher and longer. You will need to mess with it a little bit to find what is stable in your working environment and hardware.

 

Another Cinebench example:

 

With this run I used the Performance Enhancer Level 4 (This is a pre-set setting specific to the Asus Crosshair VII) and combined it with a voltage offset of around -.017v (I can't remember precisely). I am using the offset to try to bring down temperatures. 

 

In that Cinebench screenshot the score in 40th place is completely stock CPU speeds with overclocked RAM. The score in 30th position is the run with the above described settings. You can see the score in 30th is listed at the stock 3.7GHz. Keep in mind that the score would be higher without HWiNFO64 and Chrome open. You'll also see the max temp in HWiNFO is lower than the previous screenshot I posted because of the undervoltage I applied (or at least I think so). 

 

 

 

 

Thanks for your help and your practical examples. 

But just want to know one more thing, under all the cores full utilization, what the temps reach max at stock and what the temps may hit with 40x multiplier with 1.36V? And will 120mm AIO be enough to cool it down?

[nick name]

8 hours ago, PatXioPC said:

Thanks for your help and your practical examples. 

But just want to know one more thing, under all the cores full utilization, what the temps reach max at stock and what the temps may hit with 40x multiplier with 1.36V? And will 120mm AIO be enough to cool it down?

Ahhh I thought you would ask about that. If you scroll up there is a screenshot I took during the run where the CPU was running at 42.5 and the core temps were 79.5*. I am running a 360 AIO and I believe the minimum you should use is a 240. I say that, but I have never seen anyone use a 120 on a 2700X so it may work for short periods, but I would recommend going 240 at the minimum.  

 

Lemme check a 40x at 1.36v for you. I will be back with that answer.

[nick name]

This is a screenshot after about 8 back to back runs. It maxed out around 75.4* temp.