AMD Ryzen 5000 Builders Thread

Page 35 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

B-Riz

Golden Member
Feb 15, 2011
1,482
612
136
Last edited:

Makaveli

Diamond Member
Feb 8, 2002
4,717
1,051
136
Well I got tired of waiting for my 5950x order

View attachment 38802



I used paypal so I already fully paid for it. B&H sent out a recent update that merchants, which includes them, won't likely get stock until July for this chip. I just replied I'll continue to wait for now. Still I saw AMD direct had the 5600x and 5800x for sale yesterday. Snag one of each before they instantly sold out again. I'll probably let me wife use the 5600x for now and I'll use the 5800x. When the 5950x comes in I'll pass the 5800x to the wife and do something with the 1700x, 5600x, and 3950x at some point.

July.. God damn.

I wouldn't have waited that long either.
 

undertaker101

Banned
Apr 9, 2006
301
195
116
Have you verified you are 1:1 ratio with your DDR setting ?
You will need to post more info.
Attach memory tab info from CPU-Z. Ok?
Your going to want to use IF1900 - if possible
Secondary timings make a world of difference, who knew? Lol
That reddit thread was spot on, only halfway done but much happier now :)
 

Attachments

  • 20210129_095259.jpg
    20210129_095259.jpg
    708 KB · Views: 20
  • Screenshot 2021-01-29 101214.png
    Screenshot 2021-01-29 101214.png
    556.9 KB · Views: 21
  • Like
Reactions: lightmanek

undertaker101

Banned
Apr 9, 2006
301
195
116
Your graphics is carrying your score....
Look at my CPU score ... you should have ALOT more ...
You, my friend have a 5900x, I have a lowly 5800x...guru3d got 11.9k on a 5800x on their tuned rig with CL16 3600 ram so mine seems in line given how sensitive it is to mem timings
 

Noid

Platinum Member
Sep 20, 2000
2,376
183
106
You, my friend have a 5900x, I have a lowly 5800x...guru3d got 11.9k on a 5800x on their tuned rig with CL16 3600 ram so mine seems in line given how sensitive it is to mem timings

top scores with your system

You will need a 3DMark user profile to see my link. ( login )
Can't see what memory these guys are using ...
( must look at individual results )
... lots of stuff that they dont add to compare ..
( esp. since I pay for this syht -- it's probably in the Advanced version )
 
Last edited:

undertaker101

Banned
Apr 9, 2006
301
195
116
I will try tuning my system a bit more but am not in the benching contest lol, just want my PC to work well per specs. Many 3080 FTW3s with the 450W bios score in the 19k graphics range, no point in me trying to get there with my Zotac card and practical difference is less than 5% outside of benchmarks like this.
PS: Looks like most of those 12k+ CPU score are running tuned 3800 MHZ+ low latency B-die kits, my Oloy isn't gonna get there...
 

Noid

Platinum Member
Sep 20, 2000
2,376
183
106
I hate to say this .. but your score is just slightly higher than my old 3800x ( at IF1900 )
Spent $$$ for my memory, but it's carried though for 3 different chips on this MB.
( level up ! )
You have made a great investment in the other components, why falter on memory ?
 

undertaker101

Banned
Apr 9, 2006
301
195
116
I hate to say this .. but your score is just slightly higher than my old 3800x ( at IF1900 )
Spent $$$ for my memory, but it's carried though for 3 different chips on this MB.
( level up ! )
You have made a great investment in the other components, why falter on memory ?
I am a 'value' hound. The 5800x was 400 before taxes, the x570 was $170 with 20% off during Amzn Warehouse Sales, the Zotac 3080 ,well was a 3080 :), I lucked into getting and wasn't bad at 800 with taxes back in early Nov. The Oloy ram was only $80 before taxes during NE's thanksgiving sale...you get the picture.
Also as that other reddit thread shows, TimeSpy is very very sensitive to the memory subsystem more so than almost any benchmarks..I doubt your 3800x does 650/6500 in ST/MT in CPU-Z benches (the memory tweaks above made no difference there).
 
Last edited:
  • Like
Reactions: Tlh97 and killster1

Justinus

Diamond Member
Oct 10, 2005
3,173
1,515
136
Heh, heh, heh. There must be some hidden difference either between motherboards executing PBO/CO or between CPU samples self-regulating how far they'll boost - Even emulating every setting @Det0x is using I cannot get my core to boost over 5000 MHz effective clock, 5000-5025 reported.

I imagine its probably more of the CPU self-regulation disallowing the higher boost tiers... or is it?
 

Makaveli

Diamond Member
Feb 8, 2002
4,717
1,051
136

undertaker101

Banned
Apr 9, 2006
301
195
116
Some final tweaking:

1611986586279.png

Happy with this for now. My CPU-Z ST/MT scores are always in the 640-650/6400-6500 range no matter the PBO settings so not going to worry about that at all...
Thanks to this thread I got off my butt and gain 1200 points in timespy cpu, that's not nothing :)
 

Makaveli

Diamond Member
Feb 8, 2002
4,717
1,051
136
66 ns for RAM. Per the TPU review, this board boosts quite conservatively so that may be why it's a couple % off from yours.
That sounds about right the only thing holding you back is the latency on your ram.

I was able to get this but only after applying Fast Profile from the Dram Calc this is at 3200 speed.

Since I have Samsung B die will eventually try at 3600 CL14 speeds which I believe should get me to < 60 ns.

Latency on 5800X3.PNG
 
  • Like
Reactions: lightmanek

undertaker101

Banned
Apr 9, 2006
301
195
116
That sounds about right the only thing holding you back is the latency on your ram.

I was able to get this but only after applying Fast Profile from the Dram Calc this is at 3200 speed.

Since I have Samsung B die will eventually try at 3600 CL14 speeds which I believe should get me to < 60 ns.

View attachment 38901
That's fine, getting 2-3% more performance isn't going to be worth the price of fancier ram and will definitely not improve my quality of life. 🤣
 
  • Like
Reactions: lightmanek

EXCellR8

Diamond Member
Sep 1, 2010
3,982
839
136
Haven't actually gotten any numbers with my 5600X, been largely trying to stay away from PC building as of late to be perfectly honest.
 

undertaker101

Banned
Apr 9, 2006
301
195
116
Here's mine on an X470 board with everything at defaults, except the RAM changed to 3400 MHz w/1.40v. I have a Noctua NH-D15 cooler.
View attachment 38917
Looks like going from 3600 CL18 To 3200 CL14 is worth about 3-4% performance wise in some benchmarks owing to latency, makes sense. Some of it may be due to single vs dual rank as well (mine is single rank), not sure if the folks getting 660/6600+ are running dual rank kits...
 
Last edited:

undertaker101

Banned
Apr 9, 2006
301
195
116
Tried a few things, changing PBO boost from 0 to 200, enabling/disabling PBO FMAX enhancer etc...looks like I have wringed what I can out of this. PBO FMAX Enhancer enabled and +200 seem to work best most consistently. 3402 beta bios.

Slightly OT but look at those beautiful scores from the quad channel X99 setup almost 6 years ago now!
 

Attachments

  • cachemem_5800x.png
    cachemem_5800x.png
    92.3 KB · Views: 24
  • cachemem.png
    cachemem.png
    196.4 KB · Views: 23

HumblePie

Lifer
Oct 30, 2000
14,667
440
126
For me, there is no way for me to justify the cost on some of the Samsung B-Die memory kits out there for some ePeen synthetic benchmark numbers that translate to nothing in real world usage. I have the G Skill Trident Neo Kits with the Hynix memory in them. 3600 CL 16-19-19-38. I can push it to CL 16-16-16-35 but not any further. Even then I get barely any score increase in my benchmarks. I bought my kits when they were $150 a pop and now the same kits go for $210. The Samsung kits are like $300+ish when last I checked. The price to performance gain for those isn't worth it to me or in my opinion anyone really. GOOD Memory prices for the ryzen builds are not dropping at all compared to other memory. They are going up in fact.
 
  • Like
Reactions: undertaker101