Ryzen Owners Zen Garden

[Count von Schwalbe]

I have a 5600X in my father's computer and it seems that it doesn't boost to the proper boost speed like it should. Even under a full Cinebench R23 test, the CPU barely rises above 65 degrees. I thought that they're supposed to boost until it reaches 95 degrees and then slow down.

Zen 3 vs Zen 4. The thermal density is lower and the heatspreader is thinner, so it reaches a frequency wall before a thermal one.

 

[trparky]

Zen 3 vs Zen 4. The thermal density is lower and the heatspreader is thinner, so it reaches a frequency wall before a thermal one.

Are you saying that the Zen 3's heatspreader is thinner?

 

[sethmatrix7]

My 7800X3D boosts to 5.05GHz on single cores, and 4.8GHz on the lot. Bang on spec. That's default with no PBO, or anything, I beleive?

Same on my 7800X3D. 5050Mhz on all cores with PBO 90 Level 4.

 

[trparky]

I wonder...

5600x settings per Zach_01
Default limits of 5600X are:
PPT: 76W
TDC: 60A
EDC: 90A

I wonder if it's running on those settings and that I have to tweak it with settings like this...

PPT: 90W
TDC: 75A
EDC: 100A

 

[Count von Schwalbe]

Are you saying that the Zen 3's heatspreader is thinner?

Yep. The PGA package is thicker than LGA, and they kept the total socket z-height the same. They did this by thickening the heatspreader. This made it less sensitive to heatpipe placement than Zen 3, but harder to cool overall.

 

[A Computer Guy]

I have a 5600X in my father's computer and it seems that it doesn't boost to the proper boost speed like it should. Even under a full Cinebench R23 test, the CPU barely rises above 65 degrees. I thought that they're supposed to boost until it reaches 95 degrees and then slow down.

You should be able to see boost happening for Zen3 in HWiNFO64 for light workloads, not all core workloads, and not all cores on light workload will hit the boost speed by default if I recall properly. PBO and curve optimizer tuning to get those max boosts on the weaker cores.

I wonder...

I wonder if it's running on those settings and that I have to tweak it with settings like this...

If the EDC is too high you kill your boost clock so you have to find a happy balance in PBO between all core performance and boost clock performance.

You might find a few of my posts starting here from my PBO experimentation interesting, although they are with a 5950x not 5600x.

5950x and PBO overheated something or busted custom loop?

Tiachi ordered. Now we wait. Found a video that suggest the extra 4 pin can be used to help supply PCIe power if your using a lot of PCIe devices. If it's there, plug it in.

www.techpowerup.com

Eventually a few posts down I found a hole where the board wasn't respecting my EDC limit so I stopped and I'm waiting until my new UEFI/BIOS comes out of beta to test again.

 

[sethmatrix7]

Just in case anybody tries it, AMD ryzen master "auto OC" with PBO 90 Level 4 lowers performance relative to just PBO Level 4 (7800X3D). Per Cinebench and 3dMark Time Spy.

 

[trparky]

Just in case anybody tries it, AMD ryzen master "auto OC" with PBO 90 Level 4 lowers performance relative to just PBO Level 4. Per Cinebench and 3dMark Time Spy.

How do you set it without "auto OC"?

 

[sethmatrix7]

How do you set it without "auto OC"?

Just leave Ryzen Master on default, PBO 90 Level 4 in Gigabyte B650 Aorus Elite Ax BIOS for me.

 

[AusWolf]

Screenshot dump:

I'd like to try and get my EXPO settings running, preferably at 6000MHz. I can find the DRAM Voltage tab, but it's on auto, and it doesn't click. There are other options I can switch to manual, but they don't allow it to change. Ideally, there ought to be manuals for this sort of thing, but of course, the manufacturers don't offer them, instead they give you a list of variables and dare you to fuck things up...

Of note, the DRAM voltage with the latest (safe voltage) BIOS is stuck at 1.1v. Far too low for the rated 1.35v for the 6000MHz setting. CPU SOC looks like 1.018v? VCore at 1.18v.

EDIT: System is rock stable at default BIOS, CPU runs fine, but memory is at 4800MHz. TBH, I'd probably not notice a difference at the faster RAM speeds.

With my 4070ti in my old system (3700X, 16GB 3200MHz) in PUBG (yes, I know), I was getting 100-120fps. Now, at the BIOS defaults, it's steady at 140fps (I game 4Hz below refresh rate).

View attachment 304795


View attachment 304796

View attachment 304797

View attachment 304798

That's strange! I can edit those with no problem (or at least I could on earlier BIOSes - I haven't tried on the latest one).

What's in "Advanced DRAM configuration"?

(I'd love to check myself, but I'm on a holiday, about 1500 miles away from my computer right now)

 

[sethmatrix7]

Anybody know where to find a benchmark showing XMP vs EXPO with very similiar if not the same settings on a 7800X3D? Love to see if there is some sort of performance gain

 

[Nordic]

I purchased a GIGABYTE B650 AORUS Elite AX, 7800x3d, and DDR5 6000mhz, Cas 30 memory. I will be making sure I am on the latest bios ASAP. How does overclocking a 7800x3d differ from Zen4?

 

[kapone32]

I purchased a GIGABYTE B650 AORUS Elite AX, 7800x3d, and DDR5 6000mhz, Cas 30 memory. I will be making sure I am on the latest bios ASAP. How does overclocking a 7800x3d differ from Zen4?

You don't. Just let it run and it will do everything you want from it.

 

[Nordic]

You don't. Just let it run and it will do everything you want from it.

Yes and no. From what I have been able to gather so far is that Curve Optimizer goes further than 30 to -30 on AM4.

 

[Klemc]

Anybody know where to find a benchmark showing XMP vs EXPO with very similiar if not the same settings on a 7800X3D? Love to see if there is some sort of performance gain

I'll follow answers on that...

 

[tabascosauz]

Yes and no. From what I have been able to gather so far is that Curve Optimizer goes further than 30 to -30 on AM4.

Did you mean "doesn't go further"? AM4 is coded with -30 as the hard limit. It's AM5 that has seemingly been expanded up to -50 in firmware, although obviously it doesn't look like silicon quality has caught up yet to make use of it.

The 7900x3d is on sale for $20 more than the 7800x3d. $20 for 4 more cores is quite a deal. Has the thread scheduling improved much in the last few months?

In the week before Prime Day CC had a deal for a $700cad (~$520usd) 7950X3D. Poor 7900X3D was stuck at the same "deal" price of $700cad 7800X3D remaining at MSRP, of course

Honestly, I think that's the only time I've been extremely tempted by AM5 pricing.

Even at just $20 more I wouldn't want to deal with having to set up all the Game Bar scheduling crap. Problem with relying so heavily on scheduling is that it's near impossible to easily tell if you always get the performance you paid for; it can clock admirably in HWInfo and give great R23 scores, but if scheduling is fucked up yet not fucked up enough to be obvious, you can't tell without deep testing. Look at how 2CCD Vermeer changed from Win10 to 11, and how long it took people to expose the 2CCD Raphael scheduling problem in Win11.

 

[Nordic]

Did you mean "doesn't go further"? AM4 is coded with -30 as the hard limit. It's AM5 that has seemingly been expanded up to -50 in firmware, although obviously it doesn't look like silicon quality has caught up yet to make use of it.

My phrasing could have been better. I saw in a video that they had a -35 setting on AM5. I should have said it as "AM5 Curve Optimizer goes further than 30 to -30 like on AM4." I didn't see the limit though.

It looks like there is also some asynchronous BCLK things to play with too. Optimizing the BCLK and Curve Optimizer together might get a bit tedious. It appears that you might need a positive Curve Optimizer setting on some cores to support a higher BCLK.

Even at just $20 more I wouldn't want to deal with having to set up all the Game Bar scheduling crap.

That is exactly why I chose the 7800x3d. I want single threaded gaming performance above all else. The 7800x3d does that best.

 

[sethmatrix7]

I'll follow answers on that...

I see little difference here https://www.techpowerup.com/review/team-group-t-force-delta-ddr5-6000-cl38-amd-2x-16-gb/7.html Would love to see with the X3D though.

I ended up resetting my BIOS after messing around a bit too much- so I had to go back and reset XMP, manually set 6000Mhz, PBO etc... Low and behold it then decided to run the CPU at 1.29 volts rather than the 1.25 before the reset. I ended up setting it to 1.2 volts and haven't had any issues. XMP still working at 6000 speed 32-40-40-70 CAS.

 

[Klemc]

I see little difference here https://www.techpowerup.com/review/team-group-t-force-delta-ddr5-6000-cl38-amd-2x-16-gb/7.html Would love to see with the X3D though.

I ended up resetting my BIOS after messing around a bit too much- so I had to go back and reset XMP, manually set 6000Mhz, PBO etc... Low and behold it then decided to run the CPU at 1.29 volts rather than the 1.25 before the reset. I ended up setting it to 1.2 volts and haven't had any issues. XMP still working at 6000 speed 32-40-40-70 CAS.

Yes, i used DOPC, and afetr setting 1.2 instead the 1.25 auto by BIOS, it was even better.

KF560C32RSK2-32

 

[TheLostSwede]

Looks like everythings working, so now I just need to order the rest of the bits once I'm back in Sweden.

https://valid.x86.fr/5ynj5y

 

[Klemc]

For my 5700G in a DeskMini, i :

set TDP tto45w,
disable CPUBoost,
enable XMP (3200)

So, is that right (fan profile silent )

Spoiler

Fan profile set to normal if turbo enabled that btw i don't need for web browsing anyway

 

[AusWolf]

Anybody know where to find a benchmark showing XMP vs EXPO with very similiar if not the same settings on a 7800X3D? Love to see if there is some sort of performance gain

Just use any benchmark you would normally use. Looking for a specific one to show a specific difference that you don't necessarily feel otherwise is pointless, imo.

 

[Count von Schwalbe]

Just use any benchmark you would normally use. Looking for a specific one to show a specific difference that you don't necessarily feel otherwise is pointless, imo.

I think he was asking for a review benchmark so he knows which kit to buy.

 

[AusWolf]

I think he was asking for a review benchmark so he knows which kit to buy.

Ah, I see! Well, I can test 4800 MHz vs 6000 MHz on Sunday, if that's OK.

As for XMP vs EXPO, I don't think there's much difference in performance, if any.

 

[sethmatrix7]

I think he was asking for a review benchmark so he knows which kit to buy.

Nope. Wondering about any actual difference between XMP and EXPO. TPU memory reviews seem to point to no difference.