The AMD Ryzen All In One Tread /Overclocking/Memory Speeds & Timings/Tweaking/Cooling

I must be weird I like my MSI x470,running intel approved 3200 DDR4 memory on the AMD chip with tight timings and can overclock past 3200Mhz with tight timings.Board has been good enough for me.It has only been a few days on this setup and I need to give it a few more days to make sure it is reliable but this MSI x470 was cheap and works fine.

 

Honestly that is why I did wait on Ryzen to get it. I got my setup in Sep 2017. The ASRock X370 Killer SLI A/C was the damn best board I could find under $140 for the features and looks I wanted. If it can support Ryzen 2, you bet your ass Ill be still this board. It has been solid as a rock, especially once I figured out the dual rank setup for 3000 mhz ram.

@gerardfraser Thats because its a X470 board and MSI really has been getting alot of flak for good reason. Their bios are garbage on X370 boards. They are worse then the B350 boards!

 

AMD Will Trade Your Intel Core i7-8086 Win For A 16-Core 1950X Threadripper

Our competitor recently kicked off a sweepstakes to celebrate the first 40 years of the x86 processor by giving away 8,086 limited edition processors.

We appreciate the advancements they’ve helped drive with the x86 architecture over the last four decades. But, we’re ready to take it from here. That’s why we’re giving 40 performance-hungry enthusiasts in the U.S. an opportunity to celebrate the next 40 years of high-performance computing by trading in their commemorative processor prize for our CPU that enables you to work, play and create with heavy metal.

The first 40 U.S.-based winners of the Intel® 8th Gen Core i7-8086k Limited Edition Sweepstakes (which opened on June 7, 2018 and closed on June 8, 2018) (“Competitor Sweepstakes”) to complete certain steps will be offered the opportunity to exchange their new, in-box prize processor for a flagship 16-core AMD Ryzen™ Threadripper™ 1950X processor built for the gamers and creators who need a processor that can do it all.
Those steps will be communicated on this webpage on June 25 at 1:00:00 PM EDT. Check back on June 25 at 1:00:00 PM EDT for complete details on how to participate.

-> https://wccftech.com/amd-will-trade-your-intel-core-i7-8086-win-for-a-16-core-1950x-threadripper/?tqid=iKaiPCMhAkwBfV4F7PackrJcSskFGuqJ5hiLp80$

 

Just a little something weird to share.

Decided to see if I could push the all core oc a bit further. No luck. Along the way noticed that all of a sudden settings that used to be stable, no longer were. Have been wrestling with the issue all week. Even setting the oc to 3.9 and running 3200 was such the chore to get stable. About Wednesday evening, finally got 3466 stable again at both CAS14&15. But the performance was awful. Started to think I pushed the cpu too far and it now was starting to show wear 'n tare. Was so fed up last night with the whole issue I just said the heck with it and set the settings that used to be stable and blew up some stuff in Destiny 2 Iron Banner matches. Amazingly the rig held together without a crash or hiccup. This morning decided to go for gold and set my CAS14 timings and ran a Corona Renderer and Aida64. It showed that performance was almost on par with my previous results but still shy. Here's the weird part, The Stilt always suggested to set Trdrd_Scl and Twrwr_Scl to 2 if the sticks could take it. Decided to play a little and set Trdrd_Scl to 3 and Twrwr_Scl to 2. Aida64 reads were 200ish lower but the rest was the same, including the latency. The thread count in Corona Render went up 5k. So the time was still 1:56 but instead of 4,179,xxx it now was 4,184,xxx. The submission of the results lists the incorrect cpu speed. The listing shows 3.76GHz and not the 3.9 it actually is.

When get home, already tested the timings for stability, going to set Twrwr_Scl to 3 as well. Tcke needed to be raised to 6 for it all to work. Latency initially tested out to 63.9 according to Aida64. So going to be interesting to see how Corona Renderer works out along with CB15. It's all about the benchmark scores since the fps is not going to be anything noticeable. Best part is, not adjusting the voltage or resistance settings one bit. So the "extra" performance is all from the timings.

 

Do not give too much trust to 4th digit value in any benchmark. With Corona, I have around 4,310,xxx +-6,000. But if I give both of its processes above normal priority, It goes even to 4,371,xxx. Yes, it can freeze OS for like 30s at time. But that's the sole problem why score is not stable.
It is so good at utilizing CPU, that anything in OS that can randomly eat some CPU cycles will affect score unless bench has above normal priority.
(If it left one CPU core or at least 1/2 of core to OS, then it would have more stable results.)

 

Shortly after post above. My MSI X470 Gaming Pro Carbon AC kicked the bucket.
Thing lights up LED for CPU, then it goes dark and RAM LED does not lit.
If I do series of reset button clicks, it gets to RAM. But LED for test is lit permanently and immediately there are 3 long beeps from buzzer... =Faulty RAM.
Tested all sticks one by one, slot by slot. Same results. Tried all available BIOS reset methods. Still same.

So I did test All 4 sticks in my APU system and they were just fine. Next possible thing was IMC in CPU. So I put it into APU system too. And again everything starts up just fine.
(No dGPU needed as it goes straight to windows as I have disabled password there and I have static IP handed by my DHCP based on WiFi MAC. Therefore after few seconds, I just took cellphone and remoted in.)

Only other possible fault is PSU, but making bets between MSI board and EVGA PSU... I'll test PSU in old intel system, just to be sure. But I think MSI's build quality is going down even more than their crappy BIOSes.

And the fun part here is, that system was as finely tuned as it could have been. RAM exactly 1.35V, CPU 1.30125V, SoC 1.0250V. I wonder what would be result if I left CPU on auto. As that did put SoC to 1.175V and CPU to 1.512V while boosting.
Or even better, If I used MSIs one click OC which puts CPU to 4.2GHz on all cores and Core voltage to 1.475V.

Edit: And EVGA PSU is now in system I wrote this entire post on. Good bye MSI board. (I did not like you anyway.)

 

My system reboot problem seems to be fixed now I think. It did not seem to be processor related, but rather something going on with the windows monitor power-off feature. That is the only change I made. I ran driver verifier for hours and hours, no reboots, the system would not reboot when being used, but only when the computer was on (not in sleep/hibernate) and idle. Once the monitor goes the sleep, the system would then give bugcheck reboots every hour or so. If I turn monitor shut off to "never", the computer will stay up and running without any reboots.

Not sure if this was windows and Nvidia or what, but updating from driver 397.93 to the latest seems to have fixed the problem.

 

Since ASRock X470 Taichi is already available here, I think I'll, get that. But while board is on RMA and I stated that I would like to get refunded, It is on MSI's repair center to decide if they send me board or money. So, I kind of expect another board as a troll.
If they refund board, then I'll check that ASUS too.

 

ZEN Mini + LiquiD Beast

 

AMD Ryzen Threadripper First Gen HEDT Processor Prices In Constant Free Fall:
1950X With 16 Cores Under €650, 1920X With 12 Cores Under €550

Prices for the AMD Ryzen Threadripper 1st generation processors were already so disruptive for the HEDT market but looks like you can found them at much cheaper rates nowadays.
The Ryzen Threadripper 1950X can now be bought for less than €649.99 (EU) and $749.99 (US) respectively.

PS. That's why we Love AMD/ATI

 

He guys thanks for the help,

My pc lay in pieces (so couldn't replay) and the new cooler arrived so tested everything again and more properly.
Came to the following results (which are even stanger XD):
I overclocked the core to 3.8mhz on 1.3v and tested with aida64 stress test. first 3 hours of only fpu (heaviest) and than 3 hours everything (cpu,ram,fpu,cache). the system past no problem. Than i decided to look at the write,copy etc speeds in aida64 which came with the following results: (read, write, copy).

Cpu stock, ram 2800-16-16-16-36: 43.000, 43.250 and 41.500
Cpu 3.8ghz ram 2800 " " : 38.500, 38.500 and 38.250

Than tried 3.6 ghz same results and 3.4 Ghz (stock but set manualy) and also the same results...
tried to up the soc voltage from 1.05 to 1.15 no difference. tried 2800-18-18-18-40 and still the same results... anyone any idea what might be the issue?

 

It is the voltages being set. The SOC voltage does not have nearly the impact the cpu voltage does. I found that at a given speed, the cpu voltage setting appears to follow a bell curve in a way; once you pass a certain point, the performance begins to drop according to Aida64's cache benchmark test. Not only does too much voltage affect Ryzen's stability, it also hampers performance.

 

I did some more testing and it seems copy and write speeds decrease while overclocking no matter what voltage i use. This is true with an overclock on the ram and with just the stock xmp profile is this some usual behaviour on ryzen or could i try something else? I used 1.25v to test with 3.4ghz (same as stock) and 3.6 ghz with 1.25-1.3 volts.

 

When setting the ram timings, what timings are being left on Auto? Easier to list what settings are not left on Auto?

 

when the ram is overclocked to 2800 all timings where set only the resistance settings weren't set. When using 2800-18-18-18-40 only these timings where set. I tried also 2133 with auto settings and the results were the same the copy/write/read seems to degraded somewhat linearly with the overclocked cpu speed.

Would that be enough info?

 

That is not typical Ryzen behavior that I have seen or read. Most likely something to do with the bios. Was this behavior present when using other bios versions? I have nothing, grasping at straws. Hopefully someone more knowledgeable will have a clue.

 

AMD Hosting Threadripper 2 Event At The End Of July In Maranello



AMD (via Videocardz) has an event happening in late-July in the beautiful landscape of Maranello.
This also happens to be the hometown of La Ferrari and considering AMD announced it would be sponsoring the Scuderia Ferrari F1 in April earlier this year, the implications are of an epic press-only event (with fun stuff like a factory tour).

The company announced it would be sponsoring the Scuderia F1 team a couple of months back and the biggest take away from this is that they finally have the capital for this kind of marketing again – which is absolutely great news for tech enthusiasts. AMD is back in full form.

The entire event will apparently be about Threadripper 2000 series and I am sure some tangential relationship to Ferrari’s racing car and philosophy will be established during this time.
The exact date is unknown and all we heard is “end of July” which we can take to mean the last week of the month.
All of the press that received the invites are under NDA and TR2 was listed as the official reason for the event (that doesn’t mean there can’t be any hidden surprises from the company).

This is what the company had to say about its collab with Ferrari:

As part of the collaboration AMD will also be making available their technology portfolio to Scuderia Ferrari.

“The AMD passion for high-performance computing products perfectly matches with Scuderia Ferrari’s dedication to high-performance racing leadership,” said John Taylor, chief marketing officer at AMD. “Scuderia Ferrari is a world-class innovator and brand. We are excited to partner with them in linking our high-performance brands to accelerate the strong momentum around our products.”

Now we haven’t actually heard anything of the sort, and it is pure speculation on my part, but it occurs to me that this would be a great time to introduce automotive related products should AMD have any.
I can see no causal link between Threadripper 2 and Ferrari, apart from the obvious and generic luxury brand-high performance-raw horsepower brand connections.

Threadripper 2 maxes out at 32 cores, and there isn’t a V32 engine in existence that Ferrari uses – so there won’t be a direct core-to-valve analogy to be found anywhere (unless they try to equate V8s to CCXs) for petrol heads and tech enthusiasts.
We can just wait and see what the company has in store for us once the event actually happens.



PS. Yup i have my own World record in GTR2:
Car - Ferrari Maranello 650HP
Track - Barcelona GP 2003
Time - 1.27.287
(ahh the Good old Times )

 

Sorry it took me a little time but i checked the bios again and it seems like i missed the bios released in june i had the bios of may installed (probably did not check the naming good enough). I've installed the latest bios and the problems seem to be solved still strange behaviour though. This weekend I'll most likely try to stresstest everything thanks for the help though. the first indication by some aida testing seems 3.8ghz @1.30-1.325V seems to be the target and the ram seems to be running fine at 2800-16-16-16-36.

thanks!

 

Hi all, Just returned to AMD after 13 years (first PC with Athlon 64 3000+). Got 2700X+ Asus Prime Pro X470 2*4 G.skill V 2400mhz.
I have one problem, Getting only 1640~1720 in CB R15,
So I thought it's the stock cooler, so I bought Antec Kuhler k240, but the score is the same, Max all core turbo is 3815~3911 MHz with max Vcore @ 1.52V.
I think I lost the Silicone Lottery :-(.