AMD Ryzen 3000 Builders Thread

[lopri]

Silly question, how did you mount it? Stock AM4 side clips? Can you take a picture or two? lol

 

[Ravenfeeder]

@Markfw where did you find BIOS v5.61 for your x370 TaiChi? I can only see 5.60 on the ASRock site. I'm very interested how you get on with this as I have a x370 TaiChi, currently use a PS/2 keyboard and am planning on upgrading my r5 1600 to a r7 3700x.

 

[IndyColtsFan]

@Markfw where did you find BIOS v5.61 for your x370 TaiChi? I can only see 5.60 on the ASRock site. I'm very interested how you get on with this as I have a x370 TaiChi, currently use a PS/2 keyboard and am planning on upgrading my r5 1600 to a r7 3700x.

Repoman (I believe) found it here on post 5417:

https://www.overclock.net/forum/11-...rock-x370-taichi-overclocking-thread-542.html

 

[Christopher Bohling]

I guess it's ok.the game is probably
only using a few threads so a fairly high voltage may be ok. 1.45 under light load is again what AMD says is normal but above what The Stilt says was the max safe low load voltage. PBO is supposed to monitor current and make sure it's safe but seeing higher voltages on 7nm than I was comfortable using with games on my 2500k makes me nervous!. I tried a manual all core overclock and it seemed stable @1.3 4200. the only issue was that the voltage does not go down at idle when setting manual voltage.

I am.on a noctua d15 and in winter do my temps are very low.

AC Odyssey's actually a fairly heavy load, running 60%+ usage on all threads most of the time. I have decided to not worry about it for now lol.

 

[RichUK]

I've not followed this thread too closely, but it would appear the core OC results are broadly potato, no?

 

[maddie]

I've not followed this thread too closely, but it would appear the core OC results are broadly potato, no?

This is an international audience, yes?

 

[IEC]

I've not followed this thread too closely, but it would appear the core OC results are broadly potato, no?

4.5GHz all-core is currently the very best you can expect at reasonable voltage and that's on a golden 8-core chiplet (i.e. very good sample for an early-production R7 3800X). More typical results would be 4.3-4.4GHz. Per CCX overclocking can yield better results. YMMV.

A 4.5GHz Ryzen 3000 core matches or beats my golden i7-8700K OC (no AVX offset) in ST performance. I think AMD is *almost* there, despite the early-adopter BIOS issues. Once the 7nm process matures I would expect 4.5GHz+ to be more routinely achievable.

Potato on 4.5GHz (4491 actual) fixed clock AGESA 1.0.0.2 manual OC vs i7-8700K OC @ 5.1 (5.0 AVX):

 

[VirtualLarry]

Do you have any other AM4 boards to test the chip with?

I do. If I feel adventurous, I've got a Asus B450-F ROG STRIX ATX, currently with a 2700 @ 4.0Ghz under 240mm AIO WC, and a X370 Gaming Gigabyte ATX board I could throw it into (no WC on that one).

The B450-F ROG STRIX was kind of a jerry-rigged fit, it didn't square up, and I only have like 2-3 mobo screws in, so I hesitate to mess with that, now that it's working. The other one, I could replace the 1600 with the 3600.

Anyways, I tried re-pasting my 3600 and Gammaxx 400 last night, no change in temps.

I had previously put a thin line of paste on each of the four heatpipes, and had a clean CPU heatspreader. When I took it off, coverage was good, if anything maybe a little too much paste.

This last time, I cleaned off the heatsink with paper towel, and then cleaned off the CPU (maybe not in that order), and put down a few peas worth of paste on the CPU heatspreader, and spread them around with paper towel and then my finger, because the paper towel kept wiping it off, until I had a thin grey layer, then attached the heatsink. No change in temps, that I could see, so it wasn't the paste job I had done previously that was the problem (as I suspected).

 

[Terzo]

Question about cleaning thermal paste; there still seems to be a thin "stain" of paste even after cleaning with 70% alcohol. Is that going to be an issue for reapplying paste? Do I need to make sure it's 100% shiny?

I'm considering going to buy some acetone but I don't want to if it's not really necessary.

 

[UsandThem]

Question about cleaning thermal paste; there still seems to be a thin "stain" of paste even after cleaning with 70% alcohol. Is that going to be an issue for reapplying paste? Do I need to make sure it's 100% shiny?

I'm considering going to buy some acetone but I don't want to if it's not really necessary.

No, slightly normal discoloration from thermal paste is normal.

You'd likely have to lap the heat spreader to get it looking new.

 

[Terzo]

No, slightly normal discoloration from thermal paste is normal.

You'd likely have to lap the heat spreader to get it looking new.

Thanks! I ended up swapping my mobo with one that has the ryzen 3000 sticker on it. If it doesn’t boot properly i’m going to throw in the towel and take the system back to micro center to diagnose.

 

[B-Riz]

View attachment 8690View attachment 8691

Thank you! Am going to check my Mugen 2 box and see if I have the clips for that kind of mounting.

 

[krumme]

Plugged in a 3600 for a 2 year old 1600x in a asus b350m k mb. Some cheap corsair 3000 lpx ram. Plug and play. Standard xmp. Stock cooler.

Cpu seems far more responsive on web YouTube. Zippy office stuff. Ran overwatch at least 30% faster if not 40%. Dont know why. Havnt seen more stuff. Happy teen.

Will see if a cheap lean zen 4 in 2 years can plug in too.

 

[VirtualLarry]

Plugged in a 3600 for a 2 year old 1600x in a asus b350m k mb. Some cheap corsair 3000 lpx ram. Plug and play. Standard xmp. Stock cooler.

Cpu seems far more responsive on web YouTube. Zippy office stuff. Ran overwatch at least 30% faster if not 40%. Dont know why. Havnt seen more stuff. Happy teen.

That's basically been my experiences too, very "Zippy" overall. Also, very much so, with an all-core 4.0Ghz OC, which was pretty decent, and lowered my idle temps versus stock, but on all-core full-load (12 threads of PrimeGrid), temps went REALLY high. Not that I haven't been battling temps at stock, either.

Going to throw on a 240mm AIO WC next month, will update 3000 Builder's thread, I probably pulled a bad ticket in the silicon lottery. If I can tame it with WC, good, because at 4.0Ghz, this thing flies! But if I can't, I'm probably going to replace it, and maybe sell this one.

 

[B-Riz]

I do. If I feel adventurous, I've got a Asus B450-F ROG STRIX ATX, currently with a 2700 @ 4.0Ghz under 240mm AIO WC, and a X370 Gaming Gigabyte ATX board I could throw it into (no WC on that one).

The B450-F ROG STRIX was kind of a jerry-rigged fit, it didn't square up, and I only have like 2-3 mobo screws in, so I hesitate to mess with that, now that it's working. The other one, I could replace the 1600 with the 3600.

Anyways, I tried re-pasting my 3600 and Gammaxx 400 last night, no change in temps.

I had previously put a thin line of paste on each of the four heatpipes, and had a clean CPU heatspreader. When I took it off, coverage was good, if anything maybe a little too much paste.

This last time, I cleaned off the heatsink with paper towel, and then cleaned off the CPU (maybe not in that order), and put down a few peas worth of paste on the CPU heatspreader, and spread them around with paper towel and then my finger, because the paper towel kept wiping it off, until I had a thin grey layer, then attached the heatsink. No change in temps, that I could see, so it wasn't the paste job I had done previously that was the problem (as I suspected).

I would try the chip in the Asus board, if it has ComboPi 1.0.0.2, if it is still off, RMA time?

 

[StinkyPinky]

I'm happy if it just gets to 3200/cl16 as suggested. Most of the "advertised" speeds for ram is on intel platforms, they don't tend to play as nice on AMD boards.

However if its gets to 3200/16 fine then I will try and push it to 3600. 3200/cl14 or 3600/cl16 does seem to be the sweet spot for modern desktops, a few more frames is a few more frames.

To quote myself, this ram seems good stuff. I enabled XMP and it ran fine. So I bumped it to 3400/cl16 without changing anything else and it seems fine. Haven't really had time to look at it further but pretty confident 3600/16 should be realistic without too much bother

 

[IEC]

To quote myself, this ram seems good stuff. I enabled XMP and it ran fine. So I bumped it to 3400/cl16 without changing anything else and it seems fine. Haven't really had time to look at it further but pretty confident 3600/16 should be realistic without too much bother

Yup, that's why I suggested it. Plenty of people hitting 3600 CL16 timings on it (Micron E-die) and it's very affordable. Hard to beat the bang for the buck.

 

[VirtualLarry]

I would try the chip in the Asus board, if it has ComboPi 1.0.0.2, if it is still off, RMA time?

I've got my 3600 in my B450-F ROG STRIX ATX board right now, finally, under 240mm AIO WC. (Replaced my 2700, after flashing over the internet to 2406 (which Internet Flash said was newest), and then to 2501, via USB.)

(BTW, my Team Group 128GB flash drive, WAS NOT recognized in UEFI / Asus Flash 3. Had to revert to using one of my Adata all-rounders, the 16GB UV128 Yellow/Black drives, that I buy relatively cheap in bulk at Newegg.)

So far, so good. Got my NVMe RAID-0 still working. (*Had some trouble getting that to re-activate, had to disable CSM booting entirely, to get it to work.)

I haven't set XMP yet, I have some GSKill DDR4-3600 RGB TridentRGB RAM ("for AMD").

Right now, it just finished installing Win10 1903 64-bit, so I have to reboot now and let it do the reboot phase. Catch ya in a bit.

Edit: Ok, finished installing Win10 1903 on the ROG STRIX B450-F Gaming ATX PC, set XMP (3600), set PBO to "Enabled", booting into Win10. Ryzen Master is giving similar temps, similar clocks, similar core voltages.

Memory clock is 1800, Fabric Clock is 1600, guess I need to adjust FCLK to 1800 to match up. (Wonder what the inefficiency level is with those not 1:1.) Worst case, I'll drop RAM clock to 3200.

Temps with all 12 threads processing WCG tasks (mostly Int-based, I think, not sure though), temps were 78-81C on water, clocks were just shy of 4.0Ghz, Vcore was ~1.35V or so.

With one PrimeGrid task (PPS LLR), temps went up to 84-86C. I fear that when the WCG tasks complete, and I'm running nothing but PrimeGrid PPS LLR tasks, that temps will again hit 95C or higher. On water.

Edit: Yep. WCG tasks finished overnight last night, I woke up the PC, 44C idle (better than before in the other mobo at "stock").

So I fired up some PrimeGrid tasks. In no time flat, 95-96C. Boom. Even on water. Granted, this is a fixed 4000Mhz 1.365V (actually, 1.36875V, according to Ryzen Master) OC.

 

[lopri]

I think my 3700X is dead. All I tried to do was to set multiplier at x42 and enable only 2 cores in BIOS, and the system does not POST anymore. The fan starts at full speed when I power on the system then slows down, but that is about it. No USB, no screen, no nothing. I have tried:

1. Clear CMOS (including battery removal)
2. Different memory sticks (in all 4 slots in varying configurations)
3. A different PSU
4. A different video card

No other peripheral other than a USB keyboard is connected. Monitor is connected via an HDMI cable. If anyone has an idea I'd appreciate it. If anyone wants to try manual overclocking with cores disabled, well, consider yourself warned!

Edit: Parts (all of them except the CPU and the board are functional as far as I know)

3700X
Gigabyte Aorus X570 Elite (F4g BIOS, AGESA 1.0.0.3AB)
G-Skill Ripjaws 3200/CL14 (2x16GB)
ADATA XPG 2400/CL16 (2x8GB)
Geforce GT240
Radeon 290X
Seasonic X-Series 1050W
Corsair HX620

 

[lopri]

Does anyone know what Microcenter's return/replacement policy is? Also, do they test CPUs and/or motherboards if requested?

 

[DrMrLordX]

@lopri

Your board supports Q-flash. Have you tried updating/rewriting the UEFI that way?

 

[lopri]

No. I will try that.

 

[lopri]

Can I do that when I have no working keyboard?

 

[DrMrLordX]

Hmmmm I think so. Let me check:

http://download.gigabyte.us/FileList/Manual/mb_manual_x570-features.pdf

Looks like it, yes. Check page 6.

Also you have dual BIOS I think. Can you boot from the secondary UEFI?

Scratch that you may not have dual BIOS. I'm unsure.

 

[lopri]

Found the Q-flash button on the board. I will try that right now.