AMD Ryzen 3000 Builders Thread

[Markfw]

Well to be fair there are huge variations between boards and AGESA versions. Previously on 1003AB my 3700X barely went over 60C under load (Prime95 LargeFFT). After flashing it to 1003, the same chip now heats up to 75C. Performance is also more consistent. It's as if there is a power budget difference between BIOS versions. Under both BIOS'es PBO or AutoOC did little to nothing. It's as if there is an invisible hand governing power/heat limits of the chip. And in all likelihood, I believe there is such a hand.

Well, its one thing to complain about one motherboard mfg's bios, and quite another to blame the chip of being non-compliant.

 

[tamz_msc]

Seriously, thats within random variance... You are just trolling now

No, it means that the multiplier is at 45.75x, not 46.00x. You cannot call it random variance as its just one sample.

 

[DisEnchantment]

3 and 4) If you don;t need PCIE4, then get an x470 board, one with good VRMs, that processor draws a lot, and a better board would make it more stable.
5) You want quiet ? get a 240,280 or 360 AIO.

I decided to go for X570, they might be the first to get BIOS updates for stability. I regretted buying a B350 last time because it was the last to get BIOS updates. My 1700X was quite unstable for many months after I bought on March 2017.
Which is the a good X570 at the moment that does not cost an arm and a leg?

Will my Corsair Hydro Series™ H110i 240 mm be still OK after 2 years of usage?

Stability and quietness are more important to me than pushing the few extra MHz.

 

[B-Riz]

No, it means that the multiplier is at 45.75x, not 46.00x. You cannot call it random variance as its just one sample.

I have explained this before with my CB20 vs prime95 tests, max boost depends on the load / weight of what the CPU is doing.

AND, BIOS's are still an issue.

Come back and complain after our beta test and the 3950X is released, that is the real release of Zen2...

https://www.reddit.com/r/Amd/comments/cfli2n

 

[B-Riz]

PSA for those of you thinking about lapping your Ryzen 3000.

In one word: ugh

This chip (3900x) has the most infuriatingly difficult-to-remove nickel plating I've ever experienced. My 7700k was a cakewalk compared to this thing. Two 220-grit sheets of carbide wet/dry paper and I still didn't quite get rid of all the nickel plating. Got metal shavings and crap all over the pins too. What a fiasco. Fortunately, I was able to remove the brownish-grey schmutz from the pins with a child's toothbrush (Equate brand) and some 91% isopropyl alcohol. So it probably still works. I hope. I'll have to test-boot it later or . . . something.

Not sure if I'm going to bother lapping it any further since the only goal was to get rid of the nickel . . . not to make it smooth. Conductonaut will resurface the IHS anyway. Or at least it should. CLU always did on copper.

edit: gah, chip seems to be dead. No post. Time to order another one! Yay! Got one from a scalper. This is getting expensive.

Use lighter fluid next time for cleaning, aka "charcoal starter" at Wal Mart, etc.

https://en.wikipedia.org/wiki/Naphtha

 

[DrMrLordX]

You are mistaking me for someone else. I have not lapped the IHS.

Yeah don't do that.

Use lighter fluid next time for cleaning, aka "charcoal starter" at Wal Mart, etc.

https://en.wikipedia.org/wiki/Naphtha

Doubt it would have helped. Once my new 3900x is here, if it works, I may delid the dead victim to see how much crap I got under the IHS. That's likely what killed it. The pins were clean.

 

[B-Riz]

Yeah don't do that.



Doubt it would have helped. Once my new 3900x is here, if it works, I may delid the dead victim to see how much crap I got under the IHS. That's likely what killed it. The pins were clean.

Well, lighter fluid evaporates pretty quickly, so, if you lap and need to clean well, a wash down with it would not entail the hand wringing of using something that takes a while to dry out.

Or, you lap with some light weight oil on the sand paper instead of water, then clean off with the lighter fluid.

Is lapping these chips going to make over a 1*C difference in cooling?

 

[DrMrLordX]

Is lapping these chips going to make over a 1*C difference in cooling?

Depends on what you are doing. Lapping IHS and HSF in a conventional setup with "normal" paste is good for maybe 1-2C, depending on which paste you use. In the "old days" when people were still using Shin Etsu products, stuff like G751 did not respond well to lapped surfaces, while x23 variants (notably x23-7783D) did. Sometimes you could get very narrow bondline thickness without air bubbles or gaps.

If you use liquid metal (CLU, Conductonaut), bare copper surfaces are essentially resurfaced, so the lapping doesn't much matter. But Conductonaut in particular likes to pump out when applied to nickel (it has no affinity for nickel). My goal was to expose copper on the IHS so that I could use Conductonaut between IHS and copper waterblock to avoid pump-out. Using LM products between copper surfaces tends to "soft weld" the surfaces together.

 

[coffeemonster]

4575 MHz != 4600 MHz, so it's not boosting as advertised.

I mostly agree with your stance on the advertised boost speed being an issue right now, but this is being too pedantic. I've seen these half or 3/4 frequencies many times on my first gen and it's not unreasonable to round those up for marketing.

 

[yepp]

So peeps lucky enough to grab a 3900x, is it true one chiplet boost higher than the other thus better binned? Or is it due to workload?

 

[Kenmitch]

Which is the a good X570 at the moment that does not cost an arm and a leg?

I think it's a crap shoot currently.

I have the MSI MPG Gaming Plus running the AMD ComboPI1.0.0.3a update that was released at launch. I'm having none of the issues that some users are experiencing, but I'm sporting only sporting the 3600. There hasn't been any uEFI updates yet, but I think that's probably a good thing in the end.

 

[Kenmitch]

So peeps lucky enough to grab a 3900x, is it true one chiplet boost higher than the other thus better binned? Or is it due to workload?

Yes it's possible that one chiplet can be of higher quality and have a higher boost on average, but it's also possible that the other chiplet will have the fastest core overall. I don't think any one person has sampled enough to make a general statement at this time.

 

[EXCellR8]

Flare X CL14 kit is on its way to RMA dept... which is of course all the way across the country. In the interim however, I scooped a cheap pair of 3600mhz Ripjaws just to mess around with--figuring they aren't going to operate at full speed on the older board but I kind of just want to be able to use the computer at this point since it's been completed. I miss the old days of memory kits just working seemingly no matter what you bought so long as it was compatible with the slot.

 

[Interitus]

My 16g set was 150ish, so not too bad given it being B-Die.

I think my kit was $280'ish when I bought it. Right at the peak of DDR4 inflation, lol. I did think about grabbing another 2x8 kit and running 4 sticks, but I haven't really seen any reports on what the upper limits of 4x8 setups are yet.

I got my gskill 3200 ram running at 3600 without much effort and good timings. Just make sure the infinity fabric clock is in sync with your RAM speeds. 1800/1800

Yep I've seen people get 3600 CL14 out of these kits at around 1.45v. Might just go that route. I don't really NEED 32GB... but the voices tell me I do

--EDIT--

Or.... these are much more sane prices. Not great, but hey it's B-Die.

https://www.newegg.com/p/pl?d=neo&N=100007611 50008476&name=Desktop Memory&isdeptsrh=1

 

[realibrad]

I think my kit was $280'ish when I bought it. Right at the peak of DDR4 inflation, lol. I did think about grabbing another 2x8 kit and running 4 sticks, but I haven't really seen any reports on what the upper limits of 4x8 setups are yet.



Yep I've seen people get 3600 CL14 out of these kits at around 1.45v. Might just go that route. I don't really NEED 32GB... but the voices tell me I do

--EDIT--

Or.... these are much more sane prices. Not great, but hey it's B-Die.

https://www.newegg.com/p/pl?d=neo&N=100007611 50008476&name=Desktop Memory&isdeptsrh=1

This set looks to be kinda amazing given the timings and the speeds.

https://www.newegg.com/g-skill-16gb...cription=neo&cm_re=neo-_-20-232-856-_-Product

That cheap set though, that cant be B-Die can it?

 

[Kenmitch]

If your want b-die and you can wait a week or so it's available for backorder for $139.99 on Amazon currently. Shows eta of the 1st. 3rd party seller, but fulfilled by Amazon.

Or.... these are much more sane prices. Not great, but hey it's B-Die.

https://www.amazon.com/gp/offer-listing/B06XFT7DF9/ref=dp_olp_all_mbc?ie=UTF8&condition=all

I bought the same kit from newegg a couple of days before the 3xxx series launch for $129.99 and mine took to 3600 CL14 with a couple of mouse clicks. I did have to go back in and set all the sub-timings to shave a few ns off the latency.

 

[Fanatical Meat]

Flare X CL14 kit is on its way to RMA dept... which is of course all the way across the country. In the interim however, I scooped a cheap pair of 3600mhz Ripjaws just to mess around with--figuring they aren't going to operate at full speed on the older board but I kind of just want to be able to use the computer at this point since it's been completed. I miss the old days of memory kits just working seemingly no matter what you bought so long as it was compatible with the slot.

I 100% agree and it is my main gripe with AMD. Intel doesn't seem to have the same memory issues.

 

[coercitiv]

I 100% agree and it is my main gripe with AMD. Intel doesn't seem to have the same memory issues.

Couldn't agree more, got to work on a i5 9400 lately and no matter what one does the RAM "just works" at a rock solid 2666Mhz. I miss the old days when all but a few CPUs were locked down to ensure stability and compatibility with those seemingly working memory kits. /s

If you want to ensure compatibility buy RAM configured to work at max validated MT/s for that chip.

Ryzen 1000 -> 2667
Ryzen 2000 -> 2993
Ryzen 3000 -> 3200

Limit the gain, limit the pain. But we like the pain, don't we? Even if it means 1-2% more performance, we jump in there like moths into flames. LATENCIEEES !?%#$!28@#!&*!~31 IRQL_NOT_LESS_OR_EQUAL

PS. this post was written using DDR4 4000 CL17 because I miss my blue Corsair DDR 400 dual channel kit and burning memtest on a CD-ROM for optical precision memory testing.

 

[EXCellR8]

Memory compatibility is a double bummer since once a kit is proven to work it's either unavailable within hours or days, or the price goes through the roof. In the past, I'd pick the maximum supported speed w/ tightest timings and just run with them. I've never put so much thought and consideration into memory until Zen came along lol

 

[DisEnchantment]

I pulled the trigger.
However the moment I placed my order, Mindfactory.de shifted the expected date to 14.08.2019. What a bummer.

I ordered the R9 3900X
Ordered 2 sticks of Trident Z RGB 8GB 3200 CL16. Will combine them with my Trident Z 3200 CL14 2x8 GB.

I will use my H110i 240 mm AIO.
I also have a Noctua AM4 CPU cooler.
I have an XFX RX 480 GTR. Awesome card. I wanted to get an XFX RX5700 XT AIB card, but I am not sure which one will be best for my compute use cases.

I love and support Open source. So I want AMD. Not sure about the Mother board.

I had to order a Samsung S10+ too while at it , so that I have something new to play with while waiting for the Ryzen 3900X, Damn you Mindfactory!

 

[Kenmitch]

There's a updated for 3xxx series guide for Ryzen Master.

https://www.amd.com/system/files/documents/ryzen-master-quick-reference-guide.pdf

 

[Fanatical Meat]

Memory compatibility is a double bummer since once a kit is proven to work it's either unavailable within hours or days, or the price goes through the roof. In the past, I'd pick the maximum supported speed w/ tightest timings and just run with them. I've never put so much thought and consideration into memory until Zen came along lol

Agreed

 

[Dave3000]

If you want to ensure compatibility buy RAM configured to work at max validated MT/s for that chip.

Ryzen 1000 -> 2667
Ryzen 2000 -> 2993
Ryzen 3000 -> 3200

Does that mean that if I buy a Ryzen 3000 series CPU I should be buying JEDEC spec'd 3200 with real DDR4-3200 chips and not memory that is factory overclocked and only DDR4-3200 through XMP if I want to ensure stability? I was thinking about buying a Ryzen 3900x and I'm wondering if I should buy some real JEDEC spec'd DDR4-3200 despite the higher latencies? Am I overthinking this XMP/JEDEC situation?

 

[Markfw]

Does that mean that if I buy a Ryzen 3000 series CPU I should be buying JEDEC spec'd 3200 with real DDR4-3200 chips and not memory that is factory overclocked and only DDR4-3200 through XMP if I want to ensure stability? I was thinking about buying a Ryzen 3900x and I'm wondering if I should buy some real JEDEC spec'd DDR4-3200 despite the higher latencies? Am I overthinking this XMP/JEDEC situation?

IMO get 3600 cl16, use XMP on it. Set dram voltage at 1.375. Thats what I have on all my Ryzens

 

[Dave3000]

IMO get 3600 cl16, use XMP on it. Set dram voltage at 1.375. Thats what I have on all my Ryzens

Does the XMP setting on that memory also add voltage to the IMC? I've come across a few memory kits in the past, including the one I'm currently using that add voltage to the IMC if using the XMP setting. Do Ryzen 3000's work fine with DDR4-3600 without adding additional voltage to the IMC?