• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

*** Official Ryzen Owners Thread ***

you need to stress all areas of the cpu, not just the fpu section. imo fpu stressing is good for testing worst case thermal, not much else. certainly not for hours on end.
 
When using Prime, Blend is good for general high load on CPU and RAM. But if you want to really stress Ryzen CPU with Prime, It's suggested to run 28.10 or latest BETA with custom FFT config (Min & Max = 128, with run FFTs in place selected.

That and Realbench, HCI Memtest and gaming is what I tend to use.

Talk to me in noob friendly terms. :D

I am comfy with Prime 95 used it for years. I want to stability test my Ryzen. Whats the 'proper way'?

I should be using blend instead then? I've always just gone with small fft as I thought that was the 'proper way' to do it.
 
Talk to me in noob friendly terms. :D

I am comfy with Prime. I want to stability test my Ryzen. Whats the 'proper way'?

I should be using blend instead then?

"Proper way"? Is dependant on the PCs use IMO.

I just tackle it from several directions.
Mass CPU usage and power (Custom Prime and/or AVX IBT)
RAM (HCI Memtest one instance for each thread equating to around 90% of total RAM)
General/combination (Prime Blend, Realbench)
General use (encoding, ripping, transcoding and gaming).

For me, stability happens over time in a few phases.

Other tools for testing include GSAT, Y-Crunch and OCCT.
 
"Proper way"? Is dependant on the PCs use IMO.

I just tackle it from several directions.
Mass CPU usage and power (Custom Prime and/or AVX IBT)
RAM (HCI Memtest one instance for each thread equating to around 90% of total RAM)
General/combination (Prime Blend)
General use (encoding, ripping, transcoding and gaming).

For me, stability happens over time in a few phases.

Other tools for testing include GSAT, Y-Crunch and OCCT.

So if I use Prime Blend then to test CPU stability I still run it for 8hrs?

Do I have to set custom settings? Why isn't out the box settings OK?
 
So if I use Prime Blend then to test CPU stability I still run it for 8hrs?

Do I have to set custom settings? Why isn't out the box settings OK?

Blend doesn't really stress the CPU that much, for Ryzen more so as it's just not geared up for it. It's a generic "get your PC working through a lot and see if it fails" sort of test, it's still valid as it still puts a high load on CPU and RAM.
The custom setting i mentioned uses a lot less RAM but hammers the CPU a load more.

There's no one tool for the job, i suggest you get a toolbox set up with some different hammers :p
 
Blend doesn't really stress the CPU that much, for Ryzen more so as it's just not geared up for it. It's a generic "get your PC working through a lot and see if it fails" sort of test, it's still valid as it still puts a high load on CPU and RAM.
The custom setting i mentioned uses a lot less RAM but hammers the CPU a load more.

There's no one tool for the job, i suggest you get a toolbox set up with some different hammers :p


OK, so I will go with 'custom FFT config (Min & Max = 128, with run FFTs in place'.

Is it still necessary to run for 8hrs?
 
OK, so I will go with 'custom FFT config (Min & Max = 128, with run FFTs in place'.

Is it still necessary to run for 8hrs?

No, 1-2 hours would suffice using that given the power usage. Run time differs from one stress test to another IMO. Though a lot is personal preference and your own conscience. For example, if you ran Blend for 8 hours, what's to say it wouldn't fail at 8 hours 2 minutes. You draw your own line at the end of the day. We can only guide.
 
Agreed. There's nothing more painful than leaving it stress testing for hours, thinking its done then firing up a game and insta crashing. Varied tests, much better
 
No, 1-2 hours would suffice using that given the power usage. Run time differs from one stress test to another IMO. Though a lot is personal preference and your own conscience. For example, if you ran Blend for 8 hours, what's to say it wouldn't fail at 8 hours 2 minutes. You draw your own line at the end of the day. We can only guide.

Cool. Well that makes it way more manageable then.
 
This hardly scientific but I'm happy with the lower noise levels and temps with moving to a decent air cooler. I also got a new notes app on my iPad so making use of that ;)

Quiet but cool is CPU fan at around 1000 rpm, the front intakes at around 1000 rpm and the rear fan exhausting at around 800 rpm. Air can escape from the top as I removed my 2 fans from there to test and I prefer the machine without them.

xtjjaqZ.jpg

Thanks for the comprehensive reply @ElliorR, appreciated - above and beyond.

You must be pleased, given the quiet and cool results - and similar ambient temp - plus the removal of 2 additional fans.

Air can escape from the top as I removed my 2 fans from there to test and I prefer the machine without them

Have you run any tests with your top/sound proofing cover placed back on the case (if it has one?) - may not have a drastic affect on temps and may render your machine virtually silent even under load?

I also got a new notes app on my iPad so making use of that ;)
My wife follows me around the house with that - making notes for the builders - that screen is going to give flashbacks in the future...
 
Ok I am lining up for a second attempt at a Ryzen build.

What does everyone think of the following:

- Asus ROG Strix B350-F
- R5 1600
- 16GB Teamgroup 8 Pack edition 3200Mhz
- Asektek AM4 mouting bracket (so I can use my existing Antec Kuhler 620)

I already have a Samsung 960 EvoM2 NVME SSD to add in so that is covered.

The only thing I am not sure about is whether I should spend the extra on a 1600X to get the guaranteed additional clock speed out of the box + possibly 1-200mhz on the top end.

The system will be used for Gaming, video editing and streaming (I may end up using my existing laptop as a dedicated streaming device / capture PC which will take the load off of the new system).
 
Ok I am lining up for a second attempt at a Ryzen build.

What does everyone think of the following:

- Asus ROG Strix B350-F
- R5 1600
- 16GB Teamgroup 8 Pack edition 3200Mhz
- Asektek AM4 mouting bracket (so I can use my existing Antec Kuhler 620)

I already have a Samsung 960 EvoM2 NVME SSD to add in so that is covered.

The only thing I am not sure about is whether I should spend the extra on a 1600X to get the guaranteed additional clock speed out of the box + possibly 1-200mhz on the top end.

The system will be used for Gaming, video editing and streaming (I may end up using my existing laptop as a dedicated streaming device / capture PC which will take the load off of the new system).

I'd say for about £20 more you could get this bundle here with the 1600, X370 Taichi and same 3200 ram... (remove the water cooler)

https://www.overclockers.co.uk/8pac...bundle-amd-1600-3.90ghz-bu-00j-8p.html#t=a3b1
 
Thanks I didn't see that bundle

I am kind of confused how it manages to be the same price in basket as what I had specced considering the Tachi is ~£100 more

:confused:

I have it in my basket now.

Does anyone know whether these bundles (where shown as in stock) qualify for the next day postage?
 
Oh yes I get the free postage, I meant the "order by X, delivery available tomorrow" option. I was looking to order tomorrow morning for delivery Sunday. The individual parts state that is possible but I don't know about the bundle. It shows as in stock when in my basket (with the relevant options). I assume they are literally seperate SKU's within the warehouse / sale system thus in stock means the same as the individual items. (For example they have a shelf where they keep the R5 1600 CPU's which are binned for 3.9Ghz and simply pick one).

Just seen that I could get the 1700 @ 3.9Ghz bundle for another £70.

Hmmm, options.
 
The more I think about it, with the higher binning of the 1600x, if you are getting a different cooler anyway, that might be better investment. I have only started thinking this the past few days though.

Especially as you find the 1600x on special offer quite often.
 
Last edited:
No, 1-2 hours would suffice using that given the power usage. Run time differs from one stress test to another IMO. Though a lot is personal preference and your own conscience. For example, if you ran Blend for 8 hours, what's to say it wouldn't fail at 8 hours 2 minutes. You draw your own line at the end of the day. We can only guide.

Just out if interest how do you know to test Ryzen using Prime 95 using 128 min max, in place ffts etc? Can you use the same for other CPU's?

Just Priming now. I set 1.45v and 4Ghz using your Prime 95 settings above and PC immediately fell over. :p

Guess I'll dial in 3.9Ghz then.
 
Last edited:
Back
Top Bottom