AMD RYZEN Builders Thread

[IRobot23]

I have no real numbers to tell you. PC3200 is WAY better than 2133 or 2400, but cas 16 vs 14 ? not sure the numbers on that. Buy the best you can afford (within reason) This stuff looks "decent"

Benchmarks do like CAS, games do not.
Better to have 3466MHz CL16 than 3200MHzCL14.

SR will be slower at same freq, but you can reach higher freq with SR than DR.

Trial AIDA 3,6GHz

 

[IRobot23]

How much worse is RAM like this for Ryzen exactly? https://www.newegg.com/Product/Prod...-_-DesktopMemory-_-20232394-S3A4D&ignorebbr=1

Ram prices are all over the place, even for the same speeds! I'm guessing it's over timings/latency.

If you are on budget, get 8GB.
Some people are overpaying DDR4.
I just built few system with R5 1600/3446MHz(samsung E die) 2x4GB/ and little better board... which is much better than getting R5 1400/1500X with same or worse board and 3200MHz DDR4 CL14. For gaming of course.

 

[IndyColtsFan]

I think I've given up on trying to get a 100% stable overclock on my Ryzen 1700X. Though I need to do more testing, it seems "gaming stable" at 3.8-3.9 GHz, but I need to nail that down a little more. Prime95 or any computationally heavy activity (like D.C.) will cause it to crash eventually (crashes range from black screening to powering off). I'm pretty sure it would be 100% stable at 3.7 GHz.

I think whenever I decide to upgrade my main box, I might seriously pay someone to overclock it - I just don't have the time or patience for this anymore, lol.

 

[richierich1212]

I think I've given up on trying to get a 100% stable overclock on my Ryzen 1700X. Though I need to do more testing, it seems "gaming stable" at 3.8-3.9 GHz, but I need to nail that down a little more. Prime95 or any computationally heavy activity (like D.C.) will cause it to crash eventually (crashes range from black screening to powering off). I'm pretty sure it would be 100% stable at 3.7 GHz.

I think whenever I decide to upgrade my main box, I might seriously pay someone to overclock it - I just don't have the time or patience for this anymore, lol.

Have you tried different memory settings/voltage?

 

[IndyColtsFan]

Have you tried different memory settings/voltage?

I've left memory settings alone for the most part, along with the bus speed (or whatever the correct terminology is). I did just read something interesting over at overclockers which might be my issue. I'm pretty sure that when my system is shutting down, it is due to the motherboard temp protection option. Someone over at Ocers disabled it and his problems disappeared since the Taichi seems to not report temps properly. Thoughts?

 

[VirtualLarry]

I think I've given up on trying to get a 100% stable overclock on my Ryzen 1700X. Though I need to do more testing, it seems "gaming stable" at 3.8-3.9 GHz, but I need to nail that down a little more. Prime95 or any computationally heavy activity (like D.C.) will cause it to crash eventually (crashes range from black screening to powering off). I'm pretty sure it would be 100% stable at 3.7 GHz.

I think whenever I decide to upgrade my main box, I might seriously pay someone to overclock it - I just don't have the time or patience for this anymore, lol.

Is that on air, or water? I think pushing for 3.8Ghz or more on stock air is pretty-much a no-go in terms of stability, not the least because of temps.

I had to go CLC water, to get 3.8Ghz stable, although PrimeGrid in BOINC really makes the CPU take a pounding, temp-wise, even under water.

 

[IndyColtsFan]

Is that on air, or water? I think pushing for 3.8Ghz or more on stock air is pretty-much a no-go in terms of stability, not the least because of temps.

I had to go CLC water, to get 3.8Ghz stable, although PrimeGrid in BOINC really makes the CPU take a pounding, temp-wise, even under water.

Air and in a compact case, which makes it very challenging. I've got dual 120 mm Coolermaster Jetflos as intakes and 92 mm push-pull fans in a dual tower Noctua with an 80 mm exhaust a few inches behind it, which does a good job of pushing out hot air. The temps under heavy load were in the mid to upper 60s when I was testing today with CPU voltages of 1.35-1.3625 and LLC at levels 2 and 3. I could've sworn it was more stable before, so maybe I need to use LLC at level 1 and move voltage settings back to OC mode rather than Stable mode.

The black screens are probably the result of needing more voltage in certain cases but the shutdowns indicate that the motherboard overtemperature protection is kicking in. The latest BIOS changed which sensor is used for this value and even though it is an improvement, I think that it still isn't accounting for the 20 degree offset on the X chips (does that sound right?) and therefore may be shutting down when, in fact, the CPU is in no danger. I do have a spare 40 mm fan and a spare slim 120 mm fan which I could conceivably mount to move more air over the VRMs as well, which might also help with the MB temp sensors.

 

[Reinvented]

I think I've given up on trying to get a 100% stable overclock on my Ryzen 1700X. Though I need to do more testing, it seems "gaming stable" at 3.8-3.9 GHz, but I need to nail that down a little more. Prime95 or any computationally heavy activity (like D.C.) will cause it to crash eventually (crashes range from black screening to powering off). I'm pretty sure it would be 100% stable at 3.7 GHz.

I think whenever I decide to upgrade my main box, I might seriously pay someone to overclock it - I just don't have the time or patience for this anymore, lol.

List out all your specs, so we can give you a baseline to try.

I seem to be having problems with my X370 Taichi. I just recently got a new PSU, and got some CableMod cables. Computer has problems booting with my overclock, but stock has no problems. It will turn off immediately once I press the power button and do this a few times and then boot stock. If my side panel was crushing the cables, it would have a problem booting as well right? Does the X370 settings have a boot up vcore or something? Not sure what's going on.

Edit: I think I solved it. It works just fine if I have AM4 Advanced Boot Training left to Auto. Seems to be okay now.

 

[.vodka]

So, I've been doing some fine tuning.

First I took that 3.9GHz fixed OC and turned it into a pstate OC. Changed P0 FID to 9B (3875MHz, easier to stabilize), left everything else as it is, disabled XFR and set a +0.1425v offset (base 1.1875v for the 1700) to reach ~1.31v under load taking vdroop into account, using LLC level 2. Voltage without load at that pstate is ~1.33-1.35v which is fine for a 24/7 OC. I'm not comfortable with more than that considering the long term, and 100.2*38.75 is around 3883MHz, and P95 stable. Also enabled relaxed EDC throttling, with enough cooling it's not a problem and increases performance.

CPU drops down to 1.5GHZ 1v at idle, perfect.

Up next, I did some memory fine tuning. BIOS defaults are ridiculously loose and there is a lot of performance left on the table. This thread is amazing. Considering my memory kit is a 3200C14 factory OC'd to 3466C16, I found The Stilt's recommended timings for B-die kits and applied them. 1.1v vSOC, 1.375v memory:

Ryzen loves fast memory and tight timings, performance at 3882MHz + agressive 3200C14 is better than 3900MHz with 3466C16 untuned defaults, as seen in CPUz, AIDA and cinebench.That's 1T without geardown mode, which is true 1T mode. BGSA is also enabled because I'm using 1DPC and both DIMMS are single rank. Latency is down 6-7ns, quite the improvement.

This is mostly stable, ran 16 instances of HCI memtest with 900MB each, made it to 450% coverage to have one error, the other 15 instances ran to 750% without problems then I closed them all. I need to keep stress testing.

I can't get 3200C14 to cold boot (PSU unplugged -> turn on) either setting primary timings to 14-14-14-30 1T leaving the rest auto or using the agressive timings. It will boot without problems if I don't unplug the PSU... there are some C6H BIOSes with cold boot fix, I guess that will make its way to the next official release. More AGESA revisions will continue to make things better... it's alright. I can always load the agressive profile after a cold boot and go on with my stuff.



The Stilt says P95 v28.10 with 128k FFTs in place is the proper P95 to use for Ryzen stress testing (CPU based at least) and it shows:

29.2 with official Ryzen support doesn't hit 150w. It seems the heatsink I'm using is just fine for this worst case, temperatures seem to be more than in check. VRM temperatures are fine, too.

 

[krumme]

So, I've been doing some fine tuning.

First I took that 3.9GHz fixed OC and turned it into a pstate OC. Changed P0 FID to 9B (3875MHz, easier to stabilize), left everything else as it is, disabled XFR and set a +0.1425v offset (base 1.1875v for the 1700) to reach ~1.31v under load taking vdroop into account, using LLC level 2. Voltage without load at that pstate is ~1.33-1.35v which is fine for a 24/7 OC. I'm not comfortable with more than that considering the long term, and 100.2*38.75 is around 3883MHz, and P95 stable. Also enabled relaxed EDC throttling, with enough cooling it's not a problem and increases performance.

CPU drops down to 1.5GHZ 1v at idle, perfect.

Up next, I did some memory fine tuning. BIOS defaults are ridiculously loose and there is a lot of performance left on the table. This thread is amazing. Considering my memory kit is a 3200C14 factory OC'd to 3466C16, I found The Stilt's recommended timings for B-die kits and applied them. 1.1v vSOC, 1.375v memory:

Ryzen loves fast memory and tight timings, performance at 3882MHz + agressive 3200C14 is better than 3900MHz with 3466C16 untuned defaults, as seen in CPUz, AIDA and cinebench.That's 1T without geardown mode, which is true 1T mode. BGSA is also enabled because I'm using 1DPC and both DIMMS are single rank. Latency is down 6-7ns, quite the improvement.

This is mostly stable, ran 16 instances of HCI memtest with 900MB each, made it to 450% coverage to have one error, the other 15 instances ran to 750% without problems then I closed them all. I need to keep stress testing.

I can't get 3200C14 to cold boot (PSU unplugged -> turn on) either setting primary timings to 14-14-14-30 1T leaving the rest auto or using the agressive timings. It will boot without problems if I don't unplug the PSU... there are some C6H BIOSes with cold boot fix, I guess that will make its way to the next official release. More AGESA revisions will continue to make things better... it's alright. I can always load the agressive profile after a cold boot and go on with my stuff.



The Stilt says P95 v28.10 with 128k FFTs in place is the proper P95 to use for Ryzen stress testing (CPU based at least) and it shows:

29.2 with official Ryzen support doesn't hit 150w. It seems the heatsink I'm using is just fine for this worst case, temperatures seem to be more than in check. VRM temperatures are fine, too.

1.31v for 3.9 seems very low voltage for 1700 to me. Doesnt r7 use something like 1.45 for xfr/boost clocks?
I would say anything below 1.38 is perfectly safe. I remember amd saying 1.4 or 1.45 was safe forblong term. 350 boaeds have trouble over 1.4 so they are limited here.
Give it 1.375
Soc 1.15

 

[IndyColtsFan]

List out all your specs, so we can give you a baseline to try.

My specs are:
CPU: Ryzen 1700X
BOARD: Asrock X370 Taichi, v3.0 BIOS
HSF: Noctua NH-D9L with dual Noctua 92 mm fans in push pull config, using Noctua's thermal paste.
CASE: Riotoro CR1080 (compact, dual-chamber mini-tower)
RAM: 16 GB Corsair DDR4-3000 Vengeance LPX RAM
VIDEO: EVGA 1060 6GB (one of the small, single fan models)
DRIVES: 480 GB MyDigitalSSD BPX NVMe boot drive, 500 GB EVO 840 SSD, 2 TB 2.5" Firecuda
INTAKE: Two Coolermaster 120 mm Jetflos in the motherboard chamber, one 80 mm Coolermaster in the drive/PSU chamber
EXHAUST: One rear-mounted 80 mm Coolermaster and one top-mounted 80 mm Coolermaster (MB chamber)
PSU: Some modular OCZ (StealthXtreme or ModXtreme) from my parts closet, which is either a rebranded Fortron or PC Power and Cooling.

I seem to be having problems with my X370 Taichi. I just recently got a new PSU, and got some CableMod cables. Computer has problems booting with my overclock, but stock has no problems. It will turn off immediately once I press the power button and do this a few times and then boot stock. If my side panel was crushing the cables, it would have a problem booting as well right? Does the X370 settings have a boot up vcore or something? Not sure what's going on.

Edit: I think I solved it. It works just fine if I have AM4 Advanced Boot Training left to Auto. Seems to be okay now.

Are your memory speeds correct?

A couple of observations from my build:
1. From their A-Tuning app, it says EZ OC is not supported. Why is that? In the past I've let MB auto OCs find a starting point and I'd tweak from there.
2. Seeing occasional event log errors about certain processor power features not being supported and needing a firmware upgrade from the manufacturer. Any ideas?

On an unrelated note, I am about to throw this iPad across the room. I am so sick of its keyboard doing whatever the hell it wants.

 

[IEC]

1.31v for 3.9 seems very low voltage for 1700 to me. Doesnt r7 use something like 1.45 for xfr/boost clocks?
I would say anything below 1.38 is perfectly safe. I remember amd saying 1.4 or 1.45 was safe forblong term. 350 boaeds have trouble over 1.4 so they are limited here.
Give it 1.375
Soc 1.15

I run 3.8GHz @ 1.25V, and can run 3.9GHz at around 1.32V (without significant LLC). So with LLC 1.31V sounds possible - he just got a good 1700.

 

[Reinvented]

My specs are:
CPU: Ryzen 1700X
BOARD: Asrock X370 Taichi, v3.0 BIOS
HSF: Noctua NH-D9L with dual Noctua 92 mm fans in push pull config, using Noctua's thermal paste.
CASE: Riotoro CR1080 (compact, dual-chamber mini-tower)
RAM: 16 GB Corsair DDR4-3000 Vengeance LPX RAM
VIDEO: EVGA 1060 6GB (one of the small, single fan models)
DRIVES: 480 GB MyDigitalSSD BPX NVMe boot drive, 500 GB EVO 840 SSD, 2 TB 2.5" Firecuda
INTAKE: Two Coolermaster 120 mm Jetflos in the motherboard chamber, one 80 mm Coolermaster in the drive/PSU chamber
EXHAUST: One rear-mounted 80 mm Coolermaster and one top-mounted 80 mm Coolermaster (MB chamber)
PSU: Some modular OCZ (StealthXtreme or ModXtreme) from my parts closet, which is either a rebranded Fortron or PC Power and Cooling.



Are your memory speeds correct?

A couple of observations from my build:
1. From their A-Tuning app, it says EZ OC is not supported. Why is that? In the past I've let MB auto OCs find a starting point and I'd tweak from there.
2. Seeing occasional event log errors about certain processor power features not being supported and needing a firmware upgrade from the manufacturer. Any ideas?

On an unrelated note, I am about to throw this iPad across the room. I am so sick of its keyboard doing whatever the hell it wants.

My stuff is solved. It boots up every single time now using A-XMP profiles. Just need to make sure AM4 Boot Training is on for any overclock.

As for your stuff, the only thing holding you back is cooling in this case. Said it before, better cooling allows for more overclock headroom. Larry had roughly the same problem. But, it can be easily remedied by getting an AIO/cooler that has a higher heat capacity (240mm radiator).

 

[IndyColtsFan]

List out all your specs, so we can give you a baseline to try.

My specs are:
CPU: Ryzen 1700X
RAM: 16 GB Corsair DDR4-3000 Vengeance LPX RAM
HSF: Noctua
VIDEO: EVGA 1060 6GB (one of the small, single fan models)
DRIVES: 480 GB MyDigitalSSD BPX NVMe boot drive, 500 GB EVO 840 SSD, 2 TB 2.5" Firecuda


I seem to be having problems with my X370 Taichi. I just recently got a new PSU, and got some CableMod cables. Computer has problems booting with my overclock, but stock has no problems. It will turn off immediately once I press the power button and do this a few times and then boot stock. If my side panel was crushing the cables, it would have a problem booting as well right? Does the X370 settings have a boot up vcore or something? Not sure what's going on.

Edit: I think I solved it. It works just fine if I have AM4 Advanced Boot Training left to Auto. Seems to be okay now.[/QUOTE]

My stuff is solved. It boots up every single time now using A-XMP profiles. Just need to make sure AM4 Boot Training is on for any overclock.

As for your stuff, the only thing holding you back is cooling in this case. Said it before, better cooling allows for more overclock headroom. Larry had roughly the same problem. But, it can be easily remedied by getting an AIO/cooler that has a higher heat capacity (240mm radiator).

The temps aren't that high though.

 

[Reinvented]

What is your target OC? Mine was roughly 3.8 - 3.9 at the lowest voltage I could get it stable at. Which for me on my 1700 ended up being 1.3 or so.

I suggest you try the following:

Manual
3800 or 3900
1.325

Stable
Fixed
1.325
LLC Level 2
Auto
LLC Level 2

Disable global c-states
P-states disable anything after 3
Disable C6 and CnQ

 

[IndyColtsFan]

What is your target OC? Mine was roughly 3.8 - 3.9 at the lowest voltage I could get it stable at. Which for me on my 1700 ended up being 1.3 or so.

I suggest you try the following:

Manual
3800 or 3900
1.325

Stable
Fixed
1.325
LLC Level 2
Auto
LLC Level 2

Disable global c-states
P-states disable anything after 3
Disable C6 and CnQ

Thanks! I'll give these settings a shot.

 

[DrMrLordX]

http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=threads/amd-ryzen-builders-thread.2499342/page-178#post-39018578

Compare the following with the above. I said I'd do some AIDA runs this weekend, so . . .

https://www.dropbox.com/s/q83vlvlf421f88e/AIDA3200.png?dl=0

https://www.dropbox.com/s/tsf8u4t86n9yob5/AIDA3333.png?dl=0

https://www.dropbox.com/s/0aqzbko8u2jnk5o/AIDA3466.png?dl=0

edit: decided to run CBR15 on this setup. Lowered my tRC to 44 and then got the following:

https://www.dropbox.com/s/o78j5kah2acbd78/1800x4000CPU3466RAMCineBenchR15.png?dl=0

That beats my score of 1795 @ 4075 MHz that I posted in the R15 thread. Goes to show that fast RAM can do on these machines. Note that I don't think the tRC change was the only factor involved. I've gotten 1800 even on this machine @ 4 GHz before, I just never bothered to screenshot it.

 

[StefanR5R]

Is anybody aware of an X370 µATX board which is able to bifurcate the PCIe lanes from the CPU to PCIe 3.0 x8 + x8?

The only X370 µATX board which I have found so far is the Biostar Racing X370GT3. The PCIe slots of that one are organized similar to B350 µATX boards: One x16 slot from the CPU, the rest are PCIe 2.0 slots from the chipset (x1, x1, x4). The remaining PCIe 3.0 x4 lanes from the CPU are routed to an M.2 slot.

 

[krumme]

Couldnt get my Corsair lpx 3000C15 to run at stated speed with newest agesa, but applying and correcting standard DOCP settings read in bios to readings from Typhoon on the actual Hynix chips got it stable it seems.
RC, FAW and RRDL was set to agressive on the auto settings, and more agressive than what the chips was binned for? What is going on here?

Intel black box XMP / smoke and mirrors or what else?

Part Number for the Hynix chips
H5AN8G8NAFR-TFC
Frequency CAS RCD RP RAS RC FAW RRDS RRDL
1499 MHz 15 17 17 35 52 33 6 8

 

[VirtualLarry]

New HWMonitor out at cpuid.com , now my ASRock AB350M Pro4 mobo doesn't display 2.x Vcore anymore. Now it seems sane, and VSoC is called out separately.

 

[VirtualLarry]

So, seeing how Samsung B-die DDR4 is sooooo expensive, are there any cheaper DDR4 kits, with high bandwidth, that would allow a high CCX / IF clock, while still not having horrible latency?

I tried overclocking this 16GB (2x8GB) kit of Team Dark DDR4-2400. I tried slamming it up to XMP enabled, DDR4-2800, 1.350V, 18-18-18-38-72. It would POST every time, but wouldn't boot Windows 10 64-bit. Gave RAM-related BSODs.

So, I clocked it down to 2667, still at 1.350V, 16-18-18-38-72.

Would like a nice 3000 or 3200 kit. If I have to get B-die, I will, but ... are there cheaper alternatives?

 

[Sonikku]

Not that I have the money to these days even if I wanted to, but there's no way I'd build a Ryzen system right now with RAM prices being what they are. ANY ddr 4 ram seems fudged, but 3200 speed 14 latency ddr 4 is so expensive that a 16gb kit costs about as much as a Ryzen processor itself! And three times the cost of that popular Asrock pro 350 motherboard.

 

[Malogeek]

Would like a nice 3000 or 3200 kit. If I have to get B-die, I will, but ... are there cheaper alternatives?

Any alternatives you really have to go from the motherboard's QVL.

 

[Malogeek]

If you are on budget, get 8GB.

I don't think anyone here should be recommending 8Gb of RAM unless they're doing fairly simple productivity, in which case I question why they'd be building a custom PC anyway.

 

[scannall]

So, seeing how Samsung B-die DDR4 is sooooo expensive, are there any cheaper DDR4 kits, with high bandwidth, that would allow a high CCX / IF clock, while still not having horrible latency?

I tried overclocking this 16GB (2x8GB) kit of Team Dark DDR4-2400. I tried slamming it up to XMP enabled, DDR4-2800, 1.350V, 18-18-18-38-72. It would POST every time, but wouldn't boot Windows 10 64-bit. Gave RAM-related BSODs.

So, I clocked it down to 2667, still at 1.350V, 16-18-18-38-72.

Would like a nice 3000 or 3200 kit. If I have to get B-die, I will, but ... are there cheaper alternatives?

Here is the QVL list for your motherboard. AB350M Pro4 QVL Memory list