• 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 ***

Smffy said:
I'm sure some people haven't actually used their Ryzen rig since buying it months ago for all the 12 hour testing :)
Click to expand...

Just to shock everyone mine is still running at stock since building it a month ago.... :O I was on holiday last week and got back at the weekend so still finishing off the build. (Got a 1700, mobo, RAM and 1080)

Although once everything is installed and ready to use i'll probably push it a bit. :cool:
 
SupraWez said:
I looked at rapid mode but reading through the doc it only works on one drive, I have a 960 nvme for OS and two 850 for games and other stuff, to be honest I havent noticed any frame glitches, how would you describe it?

Also, are people using he default Windows 10 driver for the SSD or actually loading a Samsung driver, I am on what ever came with the Windows 10 install.
Click to expand...

I have just installed windows on a 960 Evo NVME 256GB. Using magician it reports 3181MB/s, 1457MB/s and 200/250k IOPS, this is with standard windows 10 drivers AFAIK on a C6H board with stock 1700. Not yet installed motherboard specific drivers.
 
nkata said:
I have just installed windows on a 960 Evo NVME 256GB. Using magician it reports 3181MB/s, 1457MB/s and 200/250k IOPS, this is with standard windows 10 drivers AFAIK on a C6H board with stock 1700. Not yet installed motherboard specific drivers.
Click to expand...

When I finished my Windows 10 install I used CrystalDiskMark to check all SSD drives and they were reporting advertised speeds without any Samsung drivers so left them as is :)
 
For those of you with an MSI X370 Pro Gaming Carbon, they just released BIOS version 1.7 based on the updated Agesa 1006 code.

Just updated mine and everything works ok.
 
Rossi~ said:
Indeed, one instance per thread. So R7 chips should be running 16 threads using at least 850MB of RAM each (for 16GB).

As for % coverage, It depends where you want to draw the line. If at least 1000% what happens if it errors at 1100%? is that not stable? If at least 2000%, what happens if it fails at 2100% etc etc, where do you draw the line? Sure the longer the better, but at least 500% will give you perfect stability for every day use that I'm sure of. IIRC from some forums that HCI recommend at least 500% coverage and 1000% is considered "Golden".
So unless you're using your PC for time consuming number crunching/rendering/compiling/folding etc (which you really wouldn't be pushing the RAM for anyway), I would just stick to 200-250% for quick testing to see if you're on your way with 500% to be the goal for hitting stability and/or 1000% to REALLY REALLY be sure.
Click to expand...

That's just not true. I can easily hit 500% and have done 1000% thinking I'm stable. Then I play division for a few hours with some guys from here and will be getting repeated crashes. Down clock my RAM, no more crashes.
Run HCI overnight on the unstable setting and find errors between 1000 and 2000%.
 
prjwebb said:
That's just not true. I can easily hit 500% and have done 1000% thinking I'm stable. Then I play division for a few hours with some guys from here and will be getting repeated crashes. Down clock my RAM, no more crashes.
Run HCI overnight on the unstable setting and find errors between 1000 and 2000%.
Click to expand...

Ryzen being Ryzen, could be CPU/DF/Power stability issue, not RAM.
 
gavinh87 said:
lolwut?
Click to expand...

The higher the CPU clock the harder it is to achieve higher RAM clocks without raising voltages further. i.e. if you find a nice high overclock on the CPU and get it stable and only then start to overclock the RAM, you'll notice that you'll find that the CPU will require higher voltage than previously required, either by the base voltage or increase in LLC.
This is one of the reasons (other than termination resistance, signalling issues and compatibility etc) why people have spent so long trying to get high RAM overclocks.

Is is likely due to the fact that the MEMCLK is linked to the DFICLK (Data Fabric interconnect), which means the higher the MEMCLK you use, the higher the DFICLK, which affects CPU stability.

Overclocking via the REFCLK/BCLK can help a lot here, but has it's own issues where PCI-E is concerned.
 
Rossi~ said:
The higher the CPU clock the harder it is to achieve higher RAM clocks without raising voltages further. i.e. if you find a nice high overclock on the CPU and get it stable and only then start to overclock the RAM, you'll notice that you'll find that the CPU will require higher voltage than previously required, either by the base voltage or increase in LLC.
This is one of the reasons (other than termination resistance, signalling issues and compatibility etc) why people have spent so long trying to get high RAM overclocks.

Is is likely due to the fact that the MEMCLK is linked to the DFICLK (Data Fabric interconnect), which means the higher the MEMCLK you use, the higher the DFICLK, which affects CPU stability.

Overclocking via the REFCLK/BCLK can help a lot here, but has it's own issues where PCI-E is concerned.
Click to expand...

I am aware of how it works it was that statement you made. Ryzen being ryzen lol.
 
gavinh87 said:
I am aware of how it works it was that statement you made. Ryzen being ryzen lol.
Click to expand...

I pre-ordered Ryzen and boy has it been a rollercoaster since release :p
People saying "There's a 4000MHz option in the BIOS now, why can't i use it" and AMD stating up 4000MHz in their community update a little while a go, just shows that people really do expect better RAM clocks than having to settle for 2966MHz. But sadly, I don't see that it's possible to go anywhere near 4GHz with this revision of Ryzen. Well, someone has got single channel loose timings 4GHz with 1x4GB... lol. But there shouldn't be so many people struggling to get 3200MHz down at this stage.
There's still a lot of work to get the compatibility down though as I do expect the boards to support 2dpc dual rank 3200MHz without too much faffing, but sadly it's still and issue for many. I gave up with 32GB RAM, and dropped to 16GB, I may go back for another go now the BIOS has matured. I hope they get it sorted.

That's just RAM as well. There's people out there that have weird stuttering issues, a specific stress app errors but no others do, Asus ROG CH6 users (myself included) have cold boot issues randomly where the memory fails training on POST, sometimes. It's most annoying. It's okay though because ASUS released a setting where you can set the memory to train again X times after failing..

I've got my settings settled for this BIOS, I'll live with the cold boot issue for now as it's my only existing issue. I've been using my PC normally for 2 weeks now and it's good to just use it again lol. This is what I went through on the previous BIOS and I'll go through it all on the next release too. Getting better in small increments, don't expect them to be game changing.

Edit:
Positive note bough, as far as performance is concerned, it's bloody great, especially for the cost. Though I would like the cost difference since release back to me please AMD, for being a BETA tester ;)
I'm annoyed at how Ryzen was rushed and delivered which has resulted in quite a lot of pain, but I don't regret my decision tbh, it's been equally as fun as it has frustrating :)
 
Last edited:
Rossi~ said:
I pre-ordered Ryzen and boy has it been a rollercoaster since release :p
People saying "There's a 4000MHz option in the BIOS now, why can't i use it" and AMD stating up 4000MHz in their community update a little while a go, just shows that people really do expect better RAM clocks than having to settle for 2966MHz. But sadly, I don't see that it's possible to go anywhere near 4GHz with this revision of Ryzen. Well, someone has got single channel loose timings 4GHz with 1x4GB... lol. But there should be so many people struggling to get 3200MHz down at this stage.
There's still a lot of work to get the compatibility down though as I do expect the boards to support 2dpc dual rank 3200MHz without too much faffing, but sadly it's still and issue for many. I gave up with 32GB RAM, and dropped to 16GB, I may go back for another go now the BIOS has matured. I hope they get it sorted.

That's just RAM as well. There's people out there that have weird stuttering issues, a specific stress app errors but no other do, Asus ROG CH6 users (myself included) have cold boot issues randomly where the memory fails training on POST, sometimes. It's most annoying. It's okay though because ASUS released a setting where you can set the memory to train again X times after failing..

I've got my settings settled for this BIOS, I'll live with the cold boot issue for now, but i've been using my PC normally again for 2 weeks now and it's good to just use it again lol. I'll go through it all on the next release though.
Click to expand...

It has been a *********** of a release.
3200 should be achievable for all, but it is not plug and play like the intel systems. This is the price we pay for getting something so early on and paying considerably less than the competition.
 
Yeah, have actually been using my pc more and more over the past week.

I will never buy into a platform this early again. I really do wish I had gone x99 and 68XX route.

Sure, would have paid more for less, but the time saved with hassle would have been invaluable.

Now that I have it (mostly) running well, it's pretty decent. Octo core, 64GB, NVME, combined with a huge 43 inch screen makes for a multitasking powerhouse - had I known about threadripper earlier, I may have waited... but certainly not swapping now.

And I say (mostly) because, what did pass 3 hours of occt before, now fails at that magical point of just shy of 2 hours again... have dropped the frequency a notch, upped the voltage a notch, and am not even bothering to test it. I am soooooooo sick of testing.

Edit: oh, and those annoying usb dropouts - but I know others on oc.net are having that too.
 
eyetrip said:
Yeah, have actually been using my pc more and more over the past week.

I will never buy into a platform this early again. I really do wish I had gone x99 and 68XX route.

Sure, would have paid more for less, but the time saved with hassle would have been invaluable.

Now that I have it (mostly) running well, it's pretty decent. Octo core, 64GB, NVME, combined with a huge 43 inch screen makes for a multitasking powerhouse - had I known about threadripper earlier, I may have waited... but certainly not swapping now.

And I say (mostly) because, what did pass 3 hours of occt before, now fails at that magical point of just shy of 2 hours again... have dropped the frequency a notch, upped the voltage a notch, and am not even bothering to test it. I am soooooooo sick of testing.

Edit: oh, and those annoying usb dropouts - but I know others on oc.net are having that too.
Click to expand...

You've done the right thing, find max and knock it down a notch.
I had a usb dropout one a wireless usb adapter but that was constantly connecting and reconnecting like every 5 seconds, that is now in the bin.
 
8 Pack said:


3600 CL17 not sure why CPU Z is showing CL18 but for sure bios is cl17-17-17
Click to expand...
Decent results.
Any luck with a CPU overclock in place?
*Edit, actually looking at it, those high timings are having an adverse effect on the efficiency score.
Mine at 3466CL14:
34gvzi0.png
 
Last edited:
prjwebb said:
Probably a case of Ryzen not doing odd numbers and rounding up to an even number ^^
Click to expand...

There does seem to be something going on with the Taichi BIOS with tCL when using 15 and 17. 15 always comes through as 16 and 17 always as 18. Very odd.
 
You must log in or register to reply here.
Back
Top Bottom