[shadowxaero]

Really? Newest if F5E.

Can you try F5C - http://download.gigabyte.us/FileList/BIOS/mb_bios_x570-aorus-master_f5c.zip

I have 4x8GB @ 4000 XMP no problem. Worked on all the F5x BIOS I have tried.

Glad you guys are working on improving the chipset fan curve. On the Master my chipset fan wont go under 3750 rpm, it is EXTREMELY annoying. On another note, I have a 4000 CL19 kit that I am running at 3733 CL16. For you to run 4000Mhz, do you also have the infinity fabric clocked at 2000Mhz or is it running at 1000?

Also any idea when the Xtreme boards will be available? The extra money would be worth not hearing this obnoxious high pitch fan noise...

 

[shadowxaero]

Update on fan -

I got a new test bios today. If everything checks out we will have 3 profiles and manual fan curve control. Manual control is new and why we have a new test BIOS. Also why I don't have a public beta yet. Once validated I'll have a beta bios for y'all to try.

(Very hasty test on my own system. Not scientific at all. Ymmv on rpm based on ambient Temps and load)

Silent - fan rpm 1600 idle / 1800 load.
Balance (default) - rpm 2000 idle / 2200 load.
Performance - rpm 2600 idle and load.

You are a Godsend. Any chance that beta bios will drop today at all?

 

[shadowxaero]

@GBT-MatthewH

So I have been having a lot of boost issues and I FINALLY got to the bottom of them. For my it related to memory clocks (I was able to replicate the issue multiple times once I figured it out. Not sure if this is a bios issue or issue with my particular 3900x.

Anyway, whenever I run my RAM at anything higher than 3566 my CPU will NOT boost NOR downclock. It will instead get stuck at 4.275Ghz. I have tested complete stock, with PBO and with Auto-Overclock. All works normally unless I run the RAM at speeds over 3566.

Do you have any knowledge of memory speeds effecting boost clocks? I am also on bios F5e and this happens with and without Ryzen Master installed.

I have also noticed when boost IS working correctly my CPU will boost higher with standard PBO. If I do PBO with the 200Mhz offset it will boost to lower clocks. CPU is cooled with a custom loop and temps are fine so not sure if it is related to temps.

Lastly, does your chip boost to 4.6Ghz at stock or with PBO? The only way I have gotten mine to boost to 4.6Ghz (a little over 4.6 actually) is with a 103 bclk.

 

[shadowxaero]

After reading your post I decided to try lowering my RAM to 3566 or 3533 but my PBO still will not go beyond 4.3

Also I have noticed if I touch anything related to PBO in the bios it will perform worse. ie CB15 single drops from 200 to 191-193 while the multi also takes a huge hit as well.

Probably an AGESA teething issue but the more people report back the better chance of the problems becoming resolved

For science, can you set bios to stock and see if your CPU boost correctly? If it does can you try setting RAM to say 3200 and trying it again? I am wondering if RAM indeed does effect boost and whether or not different chips are effected at different clocks.

If I enable PBO my multi core score goes up in CB15 but I take a hit in single core.

Stock - Multi Core: 3286 Single Core: 211
PBO - Multi Core: 3329 Single Core: 209

These fall within margin of error but the scores were pretty consistent across multiple runs.

 

[shadowxaero]

Stupid question: When i set up a raid 0 on my master and update the bios, is the raid than dead? Do i have to make a backup of the raid first and then format and recreate the raid with the new bios?

Not a stupid question at all. Your raid array will survive BIOS updates and you can even transfer your array to other AMD motherboards if you want. When you update your bios just re-enable raid.

Do note if your raid away and boot drive are separated (for example I have a striped array set up for storage and my OS is on a NVMe drive separate from me array) and you update your bios and boot into windows without first re-enabled raid, when you go to re-enable it you will blue screen with "boot drive inaccessible" errors. To fix this restart into safe mode, before booting into safe mode go into bios and enable raid, let windows boot into safe mode. Login and you should see you array, restart and everything should be working correctly.

TL: DR just make sure you don't forget to re-enable raid after your bios updates lol.

 

[shadowxaero]

Now that I have had my X570 Master and 3900X for a few days; I have a couple of questions mainly PBO behavior since that's all I have bothered to test as of now.

Any Ram/infinity clock under 3600mhz, I see boosts of up to 4.3 -4.55ghz on various cores (only tested 3200/3533 with 1600/1767 infinity clock). When I 3600/3733mhz with 1800/1867 Infinity Clock the cpu only seems to boost to a max of 4.25ghz; but seems to mostly maintain it on all of the cores. Now I am somewhat confused by all of this... PBO is an interesting beast. Coming from 8700K (5.2ghz)/9900K (5.1ghz) this is just far different; my last real run with Ryzen was a 1700 at launch.

Now, games play well on this CPU minus BFV which seems to have quite a few hiccups while playing. Just grasping for some info is all; most of my changes were only toward power limits in the PBO settings while I left most things on auto since I feel rather lost with settings right now.

I can confirm the same. I had this issue on my Aorus Master and I am still having it on the Extreme.

 

[shadowxaero]

Spoiler

PBO Tested by GN

This video should explain alot. For now I would recommend you just keep PBO off as there is no gain and possibly higher power consumption (True for all X570, not GIGABYTE specific).

My issue is present regardless of whether or not PBO is enabled or disabled. If I OC my RAM any higher than 3566 or the Infinity Fabric higher than 1767 the CPU gets stuck in the 4.25ish GHz range. It doesn't downclock or boost when this happens.

 

[shadowxaero]

I'm running into the same thing. 3900X, Master, F5g BIOS with F4-3600C17-16GTZ memory. 3532 is running better than 3600.



I'm also seeing weird behavior where it's not able to set CL 17 on the memory. I set memory voltage to 1.35 and XMP profile enabled. It detects CL 17 from the SPD. Auto and manually setting it to 17 actually runs CL18. If I set it to 16, it runs as 16.

This happens to me as well haha. When ever I try to run an odd CAS it sets the to even number above. So CAS 13 runs at 14, CAS 15 runs at 16 CAS 17 runs at 18. Having Geardown enabled is SUPPOSED to allow you to run at odd times but it doesn't.
@GBT-MatthewH

Have you heard of any other reports outside of the 5 or 6 people on this forum of the 3900x no longer boosting or downclocking when set to RAM and Fabric clocks of 3600/1800 and above? At this point it has to be some kind of bug as I can repeat the issue very easily. Maybe I can make a video or something showing you exactly what is happening? Again this occurs whether or not PBO is enabled.

 

[shadowxaero]

Is this only 3900X? Anyone else with another CPU? Just to be clear just go to start -> sleep then it won't resume? Does the debug LED say anything? What board?

Personally I only have 3600X right now to test, but if I can gather enough info to try and reproduce I can have HQ test.

I also have a 3900x. I have noticed that when my PC tried to go to sleep all my fans shots up to 100% and just stayed there. My RGB lights reset back to that factory orange and my PC become unresponsive. I haven since disabled sleep, but when I get home I will try again and report if any meaningful codes appear via debug LED. Do note this was on my Master board, I have the Xtreme now but still haven't tried putting it to sleep. Still I will test when I get home.

 

[shadowxaero]

Is anyone running Raid 0 on their X570? if so how do you set it up?

I am running raid 0 on two 660p Intel drives via the chipset. What are you having trouble with?

 

[shadowxaero]

I'm running an NVme drive for boot. But I have a pair of 256gb ssd that I want in Raid0 and and 2 480gb I want in Raid0 and I can't even get either pair to raid.

Are you getting any error messages are anything in BIOS when you try to raid them? Any screenshots you can share?

 

[shadowxaero]

@GBT-MatthewH I know this may be a niche use case but I am also having issues with my Gigabyte PCIe 4.0 SSD reaching the correct speeds. In my case I have my OS installed on the Gigabyte NVMe drive and I also have two Intel 660p drives connected through the chipset in Raid 0. I get the correct speeds if I disabled raid, but that sort of defeats the purpose of having two 2TB 660ps for my storage array.

I haven't been able to really find anyone else using NVMe raid 0 over the chipset while using a single NVMe drive for boot, so I am not sure if it is a bios issue or raid driver issue.

Anyway you can test this on your end?

 

[shadowxaero]

Sure, walk me through the setup -

SATA Mode: RAID
NVMe RAID: Enabled?
Anything else changed in BIOS?

Is it windows RAID or setup through AMD/BIOS?

I just did a quick and dirty enable RAID, enable NVMe raid, test. Got 5GB read. Granted I didn't create the array yet, so maybe that will change things.

Sure thing, in bios I have the following set.

SATA Mode: AHCI
NVMe RAID: Enabled
SATA Chipset Ports: Disabled

Just ran CrystalDiskMark and my Read is maxing at 3766.1MB/s. Write is hitting 4273.3MB/s which is close enough to the 4400 for me to say it is working as normal but the read is falling quite a bit short of 5GB.
Runnng Crystal Disk mark on my 660p array I seem to be having read speed issues as well. The two drives can only manage 2819.3MB/s read while write speeds are 3487.4MB/s.

Also another niche random issue I have been meaning to report. I have two Radeon VII's installed. On the x570 Xtreme (when I had the Master I also had this issue) the board is treating the second card at the "Primary" GPU. Changing the Primary graphics slot between PCIe slot 1 and PCIe slot 2 in the Bios doesn't seem to do anything.

 

[shadowxaero]

It appears that the early adopters of the new AMD chips are paying a heavy price for moving to the new boards and new chips. 141 pages of problems here and the other major X570 board mfg has 67 pages of problems. Frankly it appears that the answer is to stick with Intel or wait a few months until AMD and the board mfgrs can work things out. It seems like the system has a lot of promise but little real performance or utility at this juncture. Am I missing hundreds of happy campers whose boards and chips are working perfectly?

It it just growing pains with a new technology. You can bet Intel will have issues as well when 10nm desktop chips hit the market. But honestly most of my issues are small. No, things are not working 100% but to say little real performance is a stretch. I still have 12 cores and 24 threads that rip through all my daily workloads. My question to you would be, what is REAL performance and utility?

My 3900x performs exactly how it should. I have sorted out my boost issues so I can single core boost to 4.661Ghz and I sustain 4.25Ghz all core under heavy workloads. I am having some NVMe issues with speeds regarding raid but I am still getting close the 4GB read and 4.2GB write, I SHOULD be getting 5GB read but does 4GB read mean I lack real performance lol? I suppose I am having some PCIe addressing issues as well where for some reason my board is treating the GPU PCIe Slot 2 as the primary video card instead of the card I have in PCIe slot 1. But that doesn't hinder performance at all either.

A lot of people are experiencing boost issues and their CPU's are only getting 95% of the performance they COULD be getting, but that 95% is still well ahead of competing ships. So growing pains or not, it is still more advantageous to go AMD over Intel right now when considering OVERALL performance.

 

[shadowxaero]

I have the 3900x and the Master as well. How did you guys hit the advertised boost clocks? I only hit 4.5 on 1 core

which board and which bios version . how about cooling ? custom water cooling ?

I am running Aorus Xtreme, and I am water cooled with a custom loop.

In Bios I have my memory clock and timing manually set (not XMP). FCLK is manually set. Vcore is set to "Normal" and PBO is disabled. My BCLK is set to 103Mhz.
@GBT-MatthewH

Following up on our conversation of NMVe Raid using SSD's connected through the chipset slowing down PCIe 4.0 SSD performance (which is connected directly to the CPU), in Bios I have the following:

SATA Mode: AHCI
NVMe RAID: Enabled
SATA Chipset Ports: Disabled

Again the gigabyte SSD is only reading at 3800MB/s, well shy of the 5GB/s it is supposed to reach. If I benchmark without NVMe raid enabled I get the full speed.

Do note that from my experience, when I enable NVMe Raid and install Windows, even though my main SSD is not apart of the raid array it still uses the NVMe Raid drivers. If I disable NVMe Raid in bios, Windows will no longer boot. So when testing I did have to re-install windows to benchmark the gigabyte ssd without raid.

On another note, I have two Radeon VII GPU's installed. On the F3i bios, for some reason the second GPU is treated at the "Primary" card regardless of what is set in the bios.

The beta bios F3L fixes this issue but for some reason when I use this bios I have major issues with GPU performance. Normally my GPU runs around 2100Mhz (overclocked) but on F3L it for some reason the GPU clocks can barely sustain 1300Mhz and jumps around from 900 to 1300Mhz constantly. This happens regardless of if the card is OCed or stock. I have also tried re-installed the video drivers as well. Speaking of video drivers, when I run F3L and go to install drivers, I get that warning that the driver publisher could not be verified, which does not happen on F3i.

Now I know these problems do not sound like that should be connected to the bios at all, but I swear I must have spent like 3 hours last night flashing between F3i and F3L and was able to consistently repeat the issues above.