AMD RYZEN Builders Thread

[IEC]

Yep, definitely post the P/N. If it really hits the 2933 strap out of the box with XMP then it would help a lot of people on a budget. Assuming all the ICs used in that part are the same (which is not always a safe assumption, especially for the less expensive SKUs)

Better yet, use Thaiphoon Burner and post the screenshot. That way we have the P/N *and* the ICs used.

 

[Mockingbird]

Can you reply with a link if this $90 memory that runs at that speed ? and what motherboard ?

Gigabyte GA-AX370 Gaming K7

Corsair Vengeance LPX 16GB (2x8GB) DDR3 C15 (CMK16GX4M2B3000C15) [with SK Hynix]

 

[IEC]

As for my F4-3200C14Q-32GVK kit, each stick is reported as F4-3200C14-8GVK individually, with XMP-3200 settings of 14-14-14-34-48-1T (but MSI boards XMP default to 14-14-14-34-75-1T).

They use K4A8G085WB aka "Samsung B-die" ICs.

GVK = Ripjaws V, Black
GVR = Ripjaws V, Red

So you don't necessarily need to pay a premium for the TridentZ version if you are targeting 3200 CL14.

 

[Justinbaileyman]

Do you mean Project Quantum? It had no relation to AM4 or ryzen whatsoever, running a 4790K.

Yes that is the one I was talking about. Pretty cool if you ask me. To bad you cant buy them. I am wanting to buy a case that doesn't look like a computer case but is something interesting to look at.My wife says the computer case sitting next to our living room t.v. is an eyesore.I bought one of those HTPC cases thinking that would help but she didnt like that either.

 

[imported_jjj]

Gigabyte GA-AX370 Gaming K7

Corsair Vengeance LPX 16GB (2x8GB) DDR3 C15 (CMK16GX4M2B3000C15) [with SK Hynix]

That mobo lists quite a lot of kits at 2933-3200 http://download.gigabyte.eu/FileList/Memory/mb_memory_ga-ax370-gaming-k7.pdf
They even list a bunch of DS kits at 3200.

 

[Markfw]

Gigabyte GA-AX370 Gaming K7 Corsair Vengeance LPX 16GB (2x8GB) DDR3 C15 (CMK16GX4M2B3000C15) [with SK Hynix]

Newegg has that on sale for $120, where did you find it for $90 ?

 

[Mockingbird]

Newegg has that on sale for $120, where did you find it for $90 ?

Amazon

I think prices went up.

 

[Markfw]

Amazon

I think prices went up.

$120 at Amazon also.

So you can get PC3000 cas 15-17-17-35 for $120
or
PC3200 cas 14,14,14,34 for $180

Both at todays prices. So, yes its 50% more, but the ram is faster and lower timings.
You get what you pay for.

 

[NesuD]

As for my F4-3200C14Q-32GVK kit, each stick is reported as F4-3200C14-8GVK individually, with XMP-3200 settings of 14-14-14-34-48-1T (but MSI boards XMP default to 14-14-14-34-75-1T).

They use K4A8G085WB aka "Samsung B-die" ICs.

GVK = Ripjaws V, Black
GVR = Ripjaws V, Red

So you don't necessarily need to pay a premium for the TridentZ version if you are targeting 3200 CL14.

Yes I saw those as well but the aesthetics have value to me. Also there was some talk of the heatspreaders on the ripjaws V being kind of flimsy and bending easily. I haven't built a blinged up box with a window in a long time. Last time I did we were using Cold Cathode tubes for internal lighting, wrapping the cable bundles in UV cable wrap, and putting on color shifting paint jobs and air brushing the exteriors ourselves. It was fun to take them outside on a sunny day and wax em like it was the family car. Mark has been using the Trident Z with great success so I was sold on it. Add in The heat spreader design and colors being a good aesthetic match with my motherboard and being able to control the lighting effects with Aura and it was the perfect part for me.

 

[ux12]

Gigabyte (I have an ax370 k7 and an 1800X with a Phanteks TC14PE cooler) seems to use a "CPU" temperature calculated in mysterious ways for fan control.
It fluctuates a ton based on different workloads and often reads below ambient. As a result, to keep my fans spinning at what I think are logical speeds (ramped down at idle, ramp up on load), my fan curve in the bios for the cpu is set to something stupid like 20% at 30C and 80% at 40C. Basically the min/max is very narrow as the temperature input used to control it is very unrealistic.

I first noticed this behavior while running IBT (linpack). Basically, I had my fan curves set to cover the more natural range of temperatures (low speed at 30, medium at 50, and max at 65 or so for my cooler), but my fans would never ramp up and the cpu would throttle to 3.5Ghz down from 3.6Ghz when Tctl reaches 72C or so.
All this time, the "CPU" temperature never shooting past 44Cish. I know about the offset built into tctl for the X cpus, but the throttling has me a bit puzzled.

Anyway, after redoing my fan profiles, at least the fan ramps up, but I still throttle to 3.5 in IBT.

I'm not really sure where I'm going with my post here but I guess I need a better cooler? Or maybe Phanteks has poor mounting pressure... I took the cooler off and paste was spread well but the contact area looked very tiny. Maybe too much of a convex surface on these coolers?
I know it cools decently, but the poor paste contact mark and the throttling combined with the mystery temperature values really has me concerned.

 

[krumme]

Gigabyte GA-AX370 Gaming K7

Corsair Vengeance LPX 16GB (2x8GB) DDR3 C15 (CMK16GX4M2B3000C15) [with SK Hynix]

What is the timings? Trc? And is it with new agesa?
I have same ram i think and can try in a few days on asus b350m with new bios.

 

[EXCellR8]

even with the "sale" on the FlareX i still cannot shell out that much for 16GB... especially with that tacky heat spreader. what happened to subtle head spreaders? sheesh.

 

[NesuD]

even with the "sale" on the FlareX i still cannot shell out that much for 16GB... especially with that tacky heat spreader. what happened to subtle head spreaders? sheesh.

Yea that heatspreader makes me throw up just a little every time I look at it. Fugly is the only word I have to describe it and that feels inadequate. If i have a case with a window and it is all lit up everything has to look good or whats the point.

 

[Crono]

even with the "sale" on the FlareX i still cannot shell out that much for 16GB... especially with that tacky heat spreader. what happened to subtle head spreaders? sheesh.

You'd love the Team Night Hawk RAM I'm using, then

 

[IEC]

If you look closely, the "V" part of it almost looks like the symbol for Vega...

 

[Crumpet]

If you look closely, the "V" part of it almost looks like the symbol for Vega...

Don't, I have mighty needs.. My R9 290 tri X is literally on its last legs. I had to remove the small OC from it the other day because it started artifacting on bootup, her days are numbered.

For anyone looking at AIO coolers, I can definitely recommend the NZXT Kraken X62.

Taking my overclock and running it under a full system stress test and keeping the cpu around 72c... on silent mode.

 

[JimmiG]

Gigabyte (I have a sn ax370 k7 and an 1800X with a Phanteks TC14PE cooler) seems to use a "CPU" temperature calculated in mysterious ways for fan control.
It fluctuates a ton based on different workloads and often reads below ambient. As a result, to keep my fans spinning at what I think are logical speeds (ramped down at idle, ramp up on load), my fan curve in the bios for the cpu is set to something stupid like 20% at 30C and 80% at 40C. Basically the min/max is very narrow as the temperature input used to control it is very unrealistic.

I first noticed this behavior while running IBT (linpack). Basically, I had my fan curves set to cover the more natural range of temperatures (low speed at 30, medium at 50, and max at 65 or so for my cooler), but my fans would never ramp up and the cpu would throttle to 3.5Ghz down from 3.6Ghz when Tctl reaches 72C or so.
All this time, the "CPU" temperature never shooting past 44Cish. I know about the offset built into tctl for the X cpus, but the throttling has me a bit puzzled.

Anyway, after redoing my fan profiles, at least the fan ramps up, but I still throttle to 3.5 in IBT.

I'm not really sure where I'm going with my post here but I guess I need a better cooler? Or maybe Phanteks has poor mounting pressure... I took the cooler off and paste was spread well but the contact area looked very tiny. Maybe too much of a convex surface on these coolers?
I know it cools decently, but the poor paste contact mark and the throttling combined with the mystery temperature values really has me concerned.

Your board probably has SenseMI skew enabled, which is why you sometimes see temps below ambient. As noted earlier in this thread, the offset is not actually a constant +20C, it's variable, which makes reading temps even more confusing and inaccurate. My fan curve is also totally messed up - Low: 40C, Medium: 45C, High: 50C. There's about a 10C temperature range if you want your system to be quiet at idle and ramp up the fans at full load.

I have the same Phanteks cooler, and it's fine. I get about 64C Tdie running Prime95 Small FFT and the CPU boosts to 3.7 GHz. It throttles to 3.5 GHz running OCCT Linpack, which is normal and probably due to power/current limitations rather than temps, as Tdie was only 54C. If your CPU boosts to 3.7 GHz running Prime95 Small FFT, your temps are fine, whatever the readings show.

 

[Mockingbird]

What is the timings? Trc? And is it with new agesa?
I have same ram i think and can try in a few days on asus b350m with new bios.

15-17-17-35

BIOS: F2

 

[Valantar]

You'd love the Team Night Hawk RAM I'm using, then

My god, those are fugly. For your sake I hope you don't have a windowed case

 

[Mockingbird]

CORSAIR Vengeance LED 16GB (2 x 8GB) DDR4 3000 (PC4 24000) CMU16GX4M2C3000C15B for $106.24 with promo code VENGEANCE15

Available in 3 different LED colors (Red, Blue, and White)

https://www.newegg.com/Product/Prod...ion-_-cables-_-na-_-na&Item=20-236-133&cm_sp=

 

[IEC]

Just an FYI, I've noticed that the UEFIs using the latest AGESA 1004a seem to be setting the Trc timing on the RAM higher. Between 2.3 and 2.4 on my MSI board it went from 73 (already high) to 75 (even higher). While overall AIDA64 latency did go down (71ns vs 75ns) because of the latency improvement from the AGESA code, it wasn't the full 6ns expected because of timings (and maybe subtimings) being relaxed. I can't tell if the timings being relaxed is AGESA-related or simply mobo makers trying to make more kits of RAM run stably on their boards. Just something to be aware of.

 

[krumme]

Just an FYI, I've noticed that the UEFIs using the latest AGESA 1004a seem to be setting the Trc timing on the RAM higher. Between 2.3 and 2.4 on my MSI board it went from 73 (already high) to 75 (even higher). While overall AIDA64 latency did go down (71ns vs 75ns) because of the latency improvement from the AGESA code, it wasn't the full 6ns expected because of timings (and maybe subtimings) being relaxed. I can't tell if the timings being relaxed is AGESA-related or simply mobo makers trying to make more kits of RAM run stably on their boards. Just something to be aware of.

Thanx. I expected that given that earlier results with higher freq often had a higher trc. I am pretty sure that if i could change trc on my 3000 hynix corsair lpx ram i could go from 2667 to 2933 but i am looking forward to try it in a few days if the new agesa and bios allows/change it.

 

[CuriousMike]

I think I've misunderstood AMD's "Tips for building a better AMD Ryzen system" memory suggestion.

I had read their suggestion as my memory is Samsung b-die, but it looks like it's not. And re-reading their post, it's not directly saying that my memory is b-die.

So, this might help to explain why I've been incapable of hitting 3200mhz on the memory.

1. I thought I had b-die per https://community.amd.com/community...4/tips-for-building-a-better-amd-ryzen-system

2. My memory

3. What Thaiphoon Burner says ( this program read the memory the first time I tried; every other time it is coming back empty.)

 

[richierich1212]

I think I've misunderstood AMD's "Tips for building a better AMD Ryzen system" memory suggestion.

I had read their suggestion as my memory is Samsung b-die, but it looks like it's not. And re-reading their post, it's not directly saying that my memory is b-die.

So, this might help to explain why I've been incapable of hitting 3200mhz on the memory.

1. I thought I had b-die per https://community.amd.com/community...4/tips-for-building-a-better-amd-ryzen-system

Usually a dead give away for Samsung B-die are uniform timings such as 14-14-14/15-15-15/16-16-16.
Faster timings compared with other memory sticks at the same rated speeds usually means Samsung memory too.

 

[ScottAD]

I think I've misunderstood AMD's "Tips for building a better AMD Ryzen system" memory suggestion.

I had read their suggestion as my memory is Samsung b-die, but it looks like it's not. And re-reading their post, it's not directly saying that my memory is b-die.

So, this might help to explain why I've been incapable of hitting 3200mhz on the memory.

1. I thought I had b-die per https://community.amd.com/community...4/tips-for-building-a-better-amd-ryzen-system

2. My memory

3. What Thaiphoon Burner says ( this program read the memory the first time I tried; every other time it is coming back empty.)

There are two versions of Corsair LPX RAM, ver 5.39 is Hynix. Ver 4.31 is Samsung.

The 3600 kit is ver 4.31 the 3200 YMMV