• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

AMD Ryzen 5000 Builders Thread

Page 40 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Makaveli

Diamond Member
Feb 8, 2002
4,284
474
126
@Dave3000

Offset did not do anything, still hit 80*C. Tried PPT Limit at 120, that worked to bring it down to ~74*C.

So the PPT limit is your best option if you have a warm to hot ambient temp and did not load the case up with fans.

I am going to fiddle with PPT some more, but may just end up running stock config. If AMD designed for 80*C limit, does not bother me.
Interesting find.

When I have it set to motherboard in the bios it sets a pretty high PPT limit.

ryzen master 2.PNG
 

Dave3000

Golden Member
Jan 10, 2011
1,047
37
91
Main timings are not going to do that much difference, hence I said optimized timings. Where the difference will show is doing complete main and sub-timing tweak, it has a large influence on benchmark like these. I would run it on my PC, but I'm busy till weekend, with little to no spare time in the week.
I've seen uplifts north of 20% by just setting all the memory timings manually on a good B-Die modules. Also, some motherboards have a BIOS setting for 3DMark03 specificaly.

What also matters is which vendor GPU you have and generation, as old 3DMark engine relies on specific DX functions no longer supported in hardware on newer cards, so there is a lot to the final performance.
I set my RAM speed to the default non-XMP setting in the BIOS which set it to 2666MHz (20-19-19--43-62). Still 163 fps at that point in Troll's Lair. However, underclocking my CPU does lower the fps at that point in the scene. So if slowing down my RAM makes no difference at that point, I don't think speeding up my RAM will either, not to mention that I don't want to run my RAM above it's advertised specs as I don't want to have an unstable system or risk data corruption. So no difference between the XMP-3200 (16-18-18-36-74) vs the default non-XMP 2666MHz (20-19-19-43-62) at 2:23-2:25 in Troll's Lair.

Update: I just found out that my motherboard sets the tRC to 74 for the DOCP setting although the XMP profile in the RAM says it should be 72 as I checked with CPU-Z and the Asus SPD Tool in the BIOS. Therefore, I manually lowered the tRC to 72 in the BIOS to match what the XMP profile states it should be and no change in fps at that point in Troll's Lair.
 
Last edited:

Paul98

Diamond Member
Jan 31, 2010
3,698
138
106
Trying to decide on which motherboard I should pickup for the 5900x I just ordered, looking in the 150-200ish range. Thinking one of those less expensive x570's, any suggestions, anything I should avoid or that are extra good for their price?
 

Makaveli

Diamond Member
Feb 8, 2002
4,284
474
126
Trying to decide on which motherboard I should pickup for the 5900x I just ordered, looking in the 150-200ish range. Thinking one of those less expensive x570's, any suggestions, anything I should avoid or that are extra good for their price?
Is this 150-200 in USD?
 
Last edited:

Makaveli

Diamond Member
Feb 8, 2002
4,284
474
126
Yes, I can go higher if there is a significant upgrade. But would rather spend the money elsewhere otherwise.
When I search in that price range I see these.

Aorus X570 Elite

Asus TUF Gaming X570-Plus

Asus TUF Gaming X570-Pro

Asus Prime X570-P
 

Noid

Platinum Member
Sep 20, 2000
2,247
84
91
MSI released the beta for mine yesterday.
My 3Dmark CPU score over 16k now.
( All Core 4.55 GHz )
However, Nvidia released some crappy drivers that tanked my total score.
IF1933 has fewer WHEA - but L3 drops below 1900 score ( still )
 

Attachments

Dave3000

Golden Member
Jan 10, 2011
1,047
37
91
My Asus B550-F Strix (Wi-Fi) motherboard's BIOS sets the TRdwr for Channel A to 8 and Channel B to 7 if using the DOCP profile. I'm using 32GB Ballistix Gaming DDR4-3200 memory. Should both memory channels be the same timing for the TRdwr? I set it manually to 8 so both memory channels are 8 but I'm not sure if both channels should be set to 7 or 8. I'm using the 1804 BIOS. I sometimes have been getting crashing to the desktop when playing Sunset Overdrive and the audio would start chopping up just before crashing to the desktop and I'm wondering if it's because of the different timing for both memory channels that the auto setting did for the TRdwr.
 
Last edited:

Paul98

Diamond Member
Jan 31, 2010
3,698
138
106
Well after some more research, I think I narrowed it down to the B550 PG Velocita, and the x570 PG Velocita. Right now leaning towards the x570, because of the extra SATA slots, wifi and bios flashback
 
Feb 25, 2011
16,565
1,327
126
Assembled my 5600X rig this week, upgrading from a 3570K @ 4.3GHz. Just an Asrock A520M-ITX motherboard and no overclocking. 32GB of DDR4-3200.

I had to update the BIOS on the motherboard, but I was able to do that from a thumb drive with the CPU installed anyway. (Motherboard recognized the CPU and would boot to BIOS, but wouldn't boot Windows. Updating to latest fixed it immediately.) Wasn't that long ago you needed to borrow an older CPU to do that. Living in the future is weird.

This thing is just scary. Honestly feels faster than the Geekbench score improvement would indicate. The old CPU was no slouch, generally, but the few things it did chug on (single threaded indie games with lots of mods) are smooth as silk now.

I'm getting intermittent SATA errors on my Steam drive that I wasn't getting before. Event Viewer is saying the drive is resetting, causes a 30 second hang. Infrequent, but will see a couple immediately after a reboot. I think it's probably the new SATA cables. Will swap them out for the old ones (which are ugly, though) and will probably be ok.

Windows 10 still doesn't include all the correct drivers - had to install the chipset drivers from Asrock.
 
  • Like
Reactions: VirtualLarry

Det0x

Senior member
Sep 11, 2014
496
584
136
Today i found out how much latency difference there really is between a 1CCD and 2CCD Zen3 CPU :)
(5600x + 5800x VS 5900x + 5950x)

I disabled one CCD on my 5950x to simulate a 5800x
1 CCD = 51.7 ns in aida64
2 CCD = 54.2 ns in aida64
1ccd vs 2ccd.png
Running 4x8gigs bdie memory and PBO CO 24/7 settings
 

HumblePie

Lifer
Oct 30, 2000
14,519
338
126
Well I got my 5950x installed. Didn't have time to tweak it really at all yet. Just plunked it in and ran 3dmark. Score was basically the same as my 3950x which I was expecting anyhow. Guess the next potential upgrades would be a nice 4TB nvme drive and maybe switching over to a 64GB B-die kit. Unfortunately the former I am waiting on a good deal for and the later is hard to find these days.

 
Last edited:

Det0x

Senior member
Sep 11, 2014
496
584
136
Well I got my 5950x installed. Didn't have time to tweak it really at all yet. Just plunked it in and ran 3dmark. Score was basically the same as my 3950x which I was expecting anyhow. Guess the next potential upgrades would be a nice 4TB nvme drive and maybe switching over to a 64GB B-die kit. Unfortunately the former I am waiting on a good deal for and the later is hard to find these days.

Disable SMT and you will get ~1500-2000 more points in the CPU test :)
 

VashHT

Platinum Member
Feb 1, 2007
2,597
210
106
So from what I'm seeing 80C or so maxing out all cores on a 5800X is pretty normal? If I run prime95 constantly it eventually levels out at about 80C, had me worried at first because my 2700X would barely get into the 70s with the same cooler.
 

DrMrLordX

Lifer
Apr 27, 2000
17,031
6,001
136
So from what I'm seeing 80C or so maxing out all cores on a 5800X is pretty normal? If I run prime95 constantly it eventually levels out at about 80C, had me worried at first because my 2700X would barely get into the 70s with the same cooler.
The 2700x only supported half-speed AVX2. All Zen2 and Zen3 support full-speed AVX2 which comes at a significant cost to heat and power. Add in the fact that you're looking at hotspot temps for Zen2 and Zen3 and you should not be surprised. A 5900x will run cooler than your 5800x at the same power draw.
 

Det0x

Senior member
Sep 11, 2014
496
584
136
Like i wrote in a other thread some days ago, i have changed back to Agesa 1.1.8.0 because both boost clocks and latencies are better on older bios without training wheels.

After spending a few days finetuning i have now found my new 24/7 memory settings which iam pretty happy with, considering i'm running 4x8GB sticks and my 5950x wont run WHEA-free above 1900/3800 :)

Included some numbers for MLC for those interested.
(have to say its much easier to get low latency with single CCD CPU like 5600x or 5800x)
1ccd-1.png

Both CCD's enabled (5900x/5950x)
2ccd-1.png

Stability testing in my bloaty windows install (lost 0.2ns compared to safemode)
Could have run 50 cycles or more with 1usmus cfg as these settings are 24/7 safe settings and dont fail.
bloaty windows-10.png
No WHEA-errors on any of the runs.
Think this is also the fastest L3 latency i have seen, all thanks to the much better singlecore boosting in agesa 1.1.8.0
 
Last edited:

Makaveli

Diamond Member
Feb 8, 2002
4,284
474
126
Like i wrote in a other thread some days ago, i have changed back to Agesa 1.1.8.0 because both boost clocks and latencies are better on older bios without training wheels.

After spending a few days finetuning i have now found my new 24/7 memory settings which iam pretty happy with, considering i'm running 4x8GB sticks and my 5950x wont run WHEA-free above 1900/3800 :)

Included some numbers for MLC for those interested.
(have to say its much easier to get low latency with single CCD CPU like 5600x or 5800x)
View attachment 41116

Both CCD's enabled (5900x/5950x)
View attachment 41117

Stability testing in my bloaty windows install (lost 0.2ns compared to safemode)
Could have run 50 cycles or more with 1usmus cfg as these settings are 24/7 safe settings and dont fail.
View attachment 41118
No WHEA-errors on any of the runs.
Think this is also the fastest L3 latency i have seen, all thanks to the much better singlecore boosting in agesa 1.1.8.0
Funny cause I found the 3602 bios to be better for me with latency and boost clocks. When I was on 3001 I was able to boost to 5050mhz then bios 3405 reduced that to 5025mhz and now 3602 has it boosting back to 5050mhz. And latency is 1ns less on the newer bios for me.
 

Det0x

Senior member
Sep 11, 2014
496
584
136
Funny cause I found the 3602 bios to be better for me with latency and boost clocks. When I was on 3001 I was able to boost to 5050mhz then bios 3405 reduced that to 5025mhz and now 3602 has it boosting back to 5050mhz. And latency is 1ns less on the newer bios for me.
Nice
Then you should be happy with your motherboard manufacturer as these Asus bios is getting more and more limp on the real/effective boostclocks, atleast in my findings (Agesa 1.2.0.1)

Can share what i wrote in the asus x570 crosshair viii hero wifi thread on a other forum.

Bios (ageisa) is the key

The newer ageisa's is getting worse and worse performancewise, with more and more clockstreeching and even worse latency all in the name of user friendliness, with hidden autocorrection for everything...
Before you could tweak everything right up to the maximum, but if you crossed that line your computer would crashed.
Nowadays with newer ageisa its no problem to cross this line, you just get (hidden) reduced performance, and no real way to tell without investing days and weeks into finetuning and finding the sweepspot.. But you will never get as good boosting as in the old ageisa's. ((check L3 latency in aida64 which is a good indicator for ST performance)newer ageisa is never close to 10ns)

I've permanently switched back to bios 3003 from 3302 for this very reason :) (yesterday infact)
Its much harder to tune and there is some memory training funkynes which takes some time getting used to, (need to coldboot on failed/new memory training) but when you get the hang of it, ageisa 1.0.8.0 have much better performance. (PBO CO tuning is also a different beast compared to todays newer easymode bios)

Last night i had this very discussion on a other norwegian forum where it ended with me flashing back to 3003 and showcasing what true singlethread 5050 mhz without clockstreeching looks like.
1615845492364.png
 

moshpit

Junior Member
Sep 15, 2009
12
9
81
Just got my new Ryzen build up and going last week and still futzing around in CTR 2.0 RC5 to find the best balance, but this thing runs WAY cooler than I was expecting according to most other users experiences, however I think I scored a really good chip.

Specs:
AMD Ryzen 7 5800X
Asus TUF Gaming X570-Plus
Corsair Vengeance Pro RGB 32Gb DDR4 3200mhz (4x8Gb)
Corsair Force MP600 M.2 PCIe 4.0 500Gb
Corsair H100i Elite Capellix
Corsair HX850i Platinum
Corsair Carbide 400C
Corsair HS-45 USB DAC + Headphones (ext. audio solution)
Corsair ML120 RGB fans 2x, EVGA 120mm HP fans 2x
Toshiba X300 6Tb
PNY Geforce RTX 3090 24Gb Revel XLR8
Viewsonic XG2703-GS 1440p@165hz IPS Gsync
Windows 10 Pro 20H2 build

Diagnostic on CTR is consistent every run and doesn't fluctuate at all. I'm getting REALLY good results overall.





So far am EXTREMELY happy. I have a very mild OC with a strong undervolt working together followed the advice of the diag to set P1 and P2 profiles since it's extremely consistent no matter how many times I re-run it, boot after boot. Temperatures never reach 70C and rarely exceed 60C unless running CB R23 (Prime95 and CPUz only get temps up to 57-ish), am sure could push a lot more overclock, but am just too happy where I'm at now. Nailed a Golden Sample and it runs like a champ.
 

Makaveli

Diamond Member
Feb 8, 2002
4,284
474
126
Just got my new Ryzen build up and going last week and still futzing around in CTR 2.0 RC5 to find the best balance, but this thing runs WAY cooler than I was expecting according to most other users experiences, however I think I scored a really good chip.

Specs:
AMD Ryzen 7 5800X
Asus TUF Gaming X570-Plus
Corsair Vengeance Pro RGB 32Gb DDR4 3200mhz (4x8Gb)
Corsair Force MP600 M.2 PCIe 4.0 500Gb
Corsair H100i Elite Capellix
Corsair HX850i Platinum
Corsair Carbide 400C
Corsair HS-45 USB DAC + Headphones (ext. audio solution)
Corsair ML120 RGB fans 2x, EVGA 120mm HP fans 2x
Toshiba X300 6Tb
PNY Geforce RTX 3090 24Gb Revel XLR8
Viewsonic XG2703-GS 1440p@165hz IPS Gsync
Windows 10 Pro 20H2 build

Diagnostic on CTR is consistent every run and doesn't fluctuate at all. I'm getting REALLY good results overall.





So far am EXTREMELY happy. I have a very mild OC with a strong undervolt working together followed the advice of the diag to set P1 and P2 profiles since it's extremely consistent no matter how many times I re-run it, boot after boot. Temperatures never reach 70C and rarely exceed 60C unless running CB R23 (Prime95 and CPUz only get temps up to 57-ish), am sure could push a lot more overclock, but am just too happy where I'm at now. Nailed a Golden Sample and it runs like a champ.
Your boost clocks look to be stock at 4.85Ghz. What is the mild OC you are using?

I also have a Golden sample.

5800X CTR Golden Sample2.png
 

moshpit

Junior Member
Sep 15, 2009
12
9
81
The mild OC is running all core 4.65Ghz. When I said mild, I mean mild, lol. 250mhz over normal all core speed is a very tiny OC. And even that seem pointless and I might just drop the OC and tighten up the vcore a tiny bit more. Stock speed at ultra low vcore and temps would be totally acceptable to me.

Stock boost is 4.8Ghz, not 4.85, for the record. the .05 is also overclocked by PBO, 50mhz is a lot more than a simple fluctuation :p I didn't really do that part other than just turning on PBO to auto. I don't know why it thinks I need the extra 50mhz, but I let it do it's thing...
 

Makaveli

Diamond Member
Feb 8, 2002
4,284
474
126
The mild OC is running all core 4.65Ghz. When I said mild, I mean mild, lol. 250mhz over normal all core speed is a very tiny OC. And even that seem pointless and I might just drop the OC and tighten up the vcore a tiny bit more. Stock speed at ultra low vcore and temps would be totally acceptable to me.

Stock boost is 4.8Ghz, not 4.85, for the record. the .05 is also overclocked by PBO, 50mhz is a lot more than a simple fluctuation :p I didn't really do that part other than just turning on PBO to auto. I don't know why it thinks I need the extra 50mhz, but I let it do it's thing...
On my system when I'm at stock settings I boost to 4.85Ghz not 4.8.

I have PBO2 on + 200mhz right now so my single core boost at 5.05Ghz.
 

moshpit

Junior Member
Sep 15, 2009
12
9
81
AMD lists boost clock for 5800X as 4.8Ghz, end of story. If you are seeing 4.85Ghz, you have PBO enabled. It's not stock at 4.85Ghz. AMD's list is very specific about what stock speed is. 50mhz is not a standard fluctuation for any motherboard. That's a half multiplier worth of speed...

If your motherboard is running the chip outside of stock, that's still not stock. It's overclocked in AMD's view. I'm kinda exact about that. Call a spade a spade, stock is stock by what the manufacturer specifies is stock. If some other part overclocks it automatically, that's still an overclock. It is a MILD overclock, but an OC none the less.
 

ASK THE COMMUNITY