Ryzen 7000 / Zen 4 Owner's Club

[ENTERPRISE]

3 hours ago, neurotix said:

I found a link on a Chinese site that had 1.5 something but clicking the link to download it gave a redirect to Baidu, and no download. I wouldn't trust it.

 

Ultimately, beggers can't be choosers. Unless you want to pony up $10 to 1usmus, use the one I just uploaded.

Thanks for the revised link. Will also see if I can just get the latest Hydra Pro latest version on a one off payment.

[kaliz]

6 minutes ago, ENTERPRISE said:

Thanks for your input, super helpful. I have 2133 FCLK stable and will test performance with RAM at 6400 to see how that plays out against performance I get now with 6600. 

 

Currently stability testing 2167 FCLK. I am able to set 2200 and do benchmarks but GSAT was unstable. I did increase SOC voltage to 1.273 but to no avail but may need to revise the VDDG IOD and CCD a little higher to see if that can help get 2200 FCLK stable.

 

 

So if you have 2167 stable, what you can do is:

 

set 6400 : 2133 ratio and play with bclk!

On my cheapo mobo i can boot 102.5 , and run it up to 103,25 in windows with it! For 6600 2200 fclk in 1:1 mode. 

 

[kaliz]

Edited October 28, 2023 by kaliz

[kaliz]

 

 just rebooted and set 102.5 bclk in bios for this!

Edited October 28, 2023 by kaliz

[kaliz]

btw guys do not bother with Hydra Pro, that is really a bad software. It tests with Cinebench single core ... and does a per core Curve Optimizer. So if it is finished after a long long time, you get extreme curves which will not work 100%

 

 

 

What i do to test my Curves is use this program : SMUDebugtool 1.3.4 , found it somewhere on overclocknet. It lets you play with PBO in windows. Then to test it, i run Ycruncher (something really heavy on the CPU and RAM) and lower the curve alltogether.

 

then start tweaking core by core. After it is done set the whole curve 2 or 4 points higher just to make sure. This isnt waterproof but 10x faster and way more reliable!

 

[kaliz]

[kaliz]

SMUDebugTool1399.rar - Google Drive

DRIVE.GOOGLE.COM

 

[J7SC_Orion]

The DDR5 RAM was already pretty well cooled with an Arctic P12 pst fan right above it in my 670E setup, but I was rummaging around in some older boxes and found this genuine GSkill RAM cooler with dual fans, complete with blue LEDs .  Believe it or not, it lowered RAM temps another 4 C - 5 C during longer RAM stress tests. Also, tightened tRDRD and tWRWR just a bit. The rest are still my 24/7 daily 8000 timings.

 

[ENTERPRISE]

18 hours ago, kaliz said:

 

I did a 8400C34-45-40-42-72-504 with GDM disabled and all tight timings on a Ryzen 5 7500F single CCX for topscore in Pyprime 2b for non X3D models, it really depends on the RAM and mobo. The AM5 CPU's themselves are capable of around DDR5-8800!! With a 7800X3D managed to run DDR5-8720 and thats about the max this 4 DIMM slot mobo can handle! Imagine a GENE!

 

 

I think 6400 with 2133 fclk in 1:1 mode is best for your setup!. That way it is all synced; 3:1:3 , eg.: 6400:2133:6400 in 1:1 mode. If your CPU is very capable, the next best setup will be 6600 with 2200 fclk, so its again synced for 3:1:3. 

 

So if your CPU is only capable of running 2133 fclk, i wouldnt even go 6600, but stay at 6400 to have that synchronization. You can test it for yourself ofcourse for speeds, but a AMD engineer said on social media last year that this is the way to do it..

 

The next option is to run in 1:2 mode, and to negate the latency penalty, you have to run above DDR5-7600. My preference is DDR5-8000 with either 2000 fclk (for 2:1:1) or 8000 with 2200 fclk for max throughput.

 

you can set VDDG IOD and CCD to 1,050v to runhigher fclk. max will be 1,15v for those 

 

So I attempted to run 1:1 with 6400 and I can boot into Windows, but as soon as I run a memory bench I end up with a BSOD. Do these settings make a difference to whether or not 1:1 can be stable? First image are my settings which I have used throughout my journey , I did actually try settings I saw on your ZenTimings screenshot (Second image) but this did not help. Any ideas ?

 

 

 

 

 

 

 

[kaliz]

you never want to use someone elses RTT's, ProcODT and drivestrenght. As it is board dependent and CPU dependant. Leave all those at auto specially procODT and RTTnom. 

it is because you have 2x32GB M-die and nother CPU and mobo, that my settings in that section are different.

 

the other timings are setpretty conservative, cant believe it wont work for you

 

Edited October 29, 2023 by kaliz

[kaliz]

looks like a voltage issue somewhere, if you bluescreen btw! try running TM5 Usmus and look what the errors do by lowering or upping the voltage (any voltage really)

 

[ENTERPRISE]

2 hours ago, kaliz said:

you never want to use someone elses RTT's, ProcODT and drivestrenght. As it is board dependent and CPU dependant. Leave all those at auto specially procODT and RTTnom. 

it is because you have 2x32GB M-die and nother CPU and mobo, that my settings in that section are different.

 

the other timings are setpretty conservative, cant believe it wont work for you

 

 

That is a fair shout, I re-watched a BuildZoid video and he recommends leaving those values to Auto unless you are really trying to get Gear Down Mode Disabled to work. I have set all to Auto and will re run some tests. 

 

2 hours ago, kaliz said:

looks like a voltage issue somewhere, if you bluescreen btw! try running TM5 Usmus and look what the errors do by lowering or upping the voltage (any voltage really)

 

 

I will check that out, do you have 1Usmus config file to hand ? I will double check voltages and use TM5 if that will help point me in the right direction as I would like to run 1:1 but something is bombing out.

 

This is what I am currently testing, just with respect to how far I can push the FCLK. This includes all the AUTO values for things like ProcODT etc etc. 

 

 

 

[ENTERPRISE]

So while 2167 FCLK appeared stable after stress tests, it evidently was not as Linpack Extreme 8GB chunk cycles, using the standard 5 cycle benchmark would yield in progressively dropping Gflops as shown below.  (Running at 6600 Memclock)
 

However the 2133 seems to be rock solid stable  or at least far more stable as after stress tests and then running Linpack, the Gflops through 5 cycles remained consistent.  (Running at 6600 Memclock)

 

 

So I will stick at 2133 which is still an improvement thanks to this latest BIOS release with AGESA 1.0.0.8. 

 

No that I have the FCLK Figured, I now need to see about getting 6400 memclock running at 1:1 with 2133 FCLK. Which should be as easy as setting UCLK=MEMCLK  but after setting it, I can get into Windows but any benches will cause BSOD. So I will have to see what voltages I need to adjust to get everything stable at UCLK=MEMCLK

[neurotix]

E, I would suggest just running 6600MHz/2100fclk as you said it was stable.

 

Higher fclk will be nice and fclk DOES affect memory bandwidth speed (when I tested 2033 vs 2100 I got 3GB/sec faster read bandwidth at 2100) but honestly being at 6600 will make more of a difference in bandwidth than raising your fclk will.

 

If you found settings that are stable and pass everything at 6600, I would kindly suggest sticking with those and printing off your ZenTimings blown up big to use to reenter your settings on BIOS update (I lose my saved profiles every time I update to a new bios and use the printoff to reenter all the settings, dunno if your AORUS board keeps them or not or if you can save bios settings to a profile)

 

The push for more is nice and you've earned my respect, but I learned the hard way with my B-Die DDR5 that it just would not do anything over 6200MHz stably. My current kit will not do anything over 7600MHz stably but I can bench at 8000MHz, wouldn't want to run it anyway because of risk of data/file corruption. So if you've found a sweet spot with 6600MHz I would stick with that.

[ENTERPRISE]

25 minutes ago, neurotix said:

E, I would suggest just running 6600MHz/2100fclk as you said it was stable.

 

Higher fclk will be nice and fclk DOES affect memory bandwidth speed (when I tested 2033 vs 2100 I got 3GB/sec faster read bandwidth at 2100) but honestly being at 6600 will make more of a difference in bandwidth than raising your fclk will.

 

If you found settings that are stable and pass everything at 6600, I would kindly suggest sticking with those and printing off your ZenTimings blown up big to use to reenter your settings on BIOS update (I lose my saved profiles every time I update to a new bios and use the printoff to reenter all the settings, dunno if your AORUS board keeps them or not or if you can save bios settings to a profile)

 

The push for more is nice and you've earned my respect, but I learned the hard way with my B-Die DDR5 that it just would not do anything over 6200MHz stably. My current kit will not do anything over 7600MHz stably but I can bench at 8000MHz, wouldn't want to run it anyway because of risk of data/file corruption. So if you've found a sweet spot with 6600MHz I would stick with that.

Hey bud. 

 

Yeah I agree. After I made my last post I decided to stop chasing my tail and be happy with what I have. I have settled on 6600 @ 2133 FCLK and the performance I am getting is great compared to EXPO (6000). 

 

I have routinely been saving my BIOS screenies of my settings so after a BIOS update I can just plug them back in

 

My board like yours wipes the OC profiles clean upon updates, I get why they do it. 

 

Now I will start looking into CPU CO in the next week or so and start that journey. Will go for all core OC first to give a baseline, then move into per core CO. Going to try and get the latest version of Hydra and try the other piece of software Kaliz recommended. 

 

Thanks!

 

 

[neurotix]

9 minutes ago, ENTERPRISE said:

Hey bud. 

 

Yeah I agree. After I made my last post I decided to stop chasing my tail and be happy with what I have. I have settled on 6600 @ 2133 FCLK and the performance I am getting is great compared to EXPO (6000). 

 

I have routinely been saving my BIOS screenies of my settings so after a BIOS update I can just plug them back in

 

My board like yours wipes the OC profiles clean upon updates, I get why they do it. 

 

Now I will start looking into CPU CO in the next week or so and start that journey. Will go for all core OC first to give a baseline, then move into per core CO. Going to try and get the latest version of Hydra and try the other piece of software Kaliz recommended. 

 

Thanks!

 

 

Start with an all-core CO of -20. That's what's recommended anyway.

 

Hydra is a little overwhelming and confusing at first, so here's how to use it: open it, you will probably see the first tab with a ton of sliders for CO. Go to the "diagnosis" tab and then click "diagnosis" or "run diagnosis" in the bottom right. Then wait for like an hour probably more because you have a 16 core chip. It will test with Cinebench single core CO offsets in the background, I think it also uses Prime95.

 

When it's done it'll have found your CO values and they should all have been automatically added to the first tab, and you can apply them.

 

That's how it was when I used it anyway but as I said, it didn't work for me using the values it found. I'd bsod pretty quickly after applying the CO values.

[J7SC_Orion]

After 'eclk' and stock bclk tuning, I did some more custom bclk and custom CO; single core on CCD0 VCache can now hit 5370...hasn't failed  single test yet but I don't want to push the CPU over the edge, either...

 

[Hurricane28]

3 hours ago, J7SC_Orion said:

After 'eclk' and stock bclk tuning, I did some more custom bclk and custom CO; single core on CCD0 VCache can now hit 5370...hasn't failed  single test yet but I don't want to push the CPU over the edge, either...

 

 

More voltage always does the trick.

[J7SC_Orion]

1 hour ago, Hurricane28 said:

 

More voltage always does the trick.

 

This is a 7950X3D, essentially voltage-locked 'upwards' - stork voltages in my earlier post (confirmed by HWInfo)

[Hurricane28]

3 hours ago, J7SC_Orion said:

 

This is a 7950X3D, essentially voltage-locked 'upwards' - stork voltages in my earlier post (confirmed by HWInfo)

 

Ah yes. Well, i actually undervolt my CPU to get more performance lol. 

 

I just set my curve optimizer to negative 22 for the voltage and its good to go.

I tried manual OC but its not worth it imo. 

[neurotix]

12 minutes ago, Hurricane28 said:

 

Ah yes. Well, i actually undervolt my CPU to get more performance lol. 

 

I just set my curve optimizer to negative 22 for the voltage and its good to go.

I tried manual OC but its not worth it imo. 

FYI hurricane, the X3D chips are voltage and ratio locked. We literally have no voltage or cpu core ratio options in our bios.

 

Only way to control it/overclock is through PBO or CO curves.

[neurotix]

Sorry for double posting. I haven't seen this mentioned anywhere else in this thread, or on the site.

 

Thought I'd clue you guys in to something you may not be familiar with.

 

1usmus has made two different Power Plans for use with Ryzen that may offer higher boost clocks and performance compared to the stock Windows high performance or Ultimate performance power plans and this is true for older platforms as well (which he made them for).

 

GitHub - neon-dev/Ryzen-Optimized-Power-Plan: Further optimized version of 1usmus' great power plan (https://www.techpowerup.com/review/1usmus-power-plan-for-amd-ryzen-new-developments/)

GITHUB.COM

---

Further optimized version of 1usmus' great power plan...

 

That is an older one. The most current version, you can grab here from TPU as well as read his documentation:

 

1usmus Power Plan for AMD Ryzen - New Developments

WWW.TECHPOWERUP.COM

---

Two weeks ago, we released the 1usmus Power Plan for AMD Ryzen processors, which received a ton of attention. Both Microsoft and...

 

When my 7900X3D was actually boosting on the second CCD and not being limited to 5100mhz under any actual workload (waiting for bios fix Asus!!!) , I noticed significantly higher boost speeds using the "Ryzen Optimized Balanced" power plan, which I think is the first link. I was only able to hit around 5300-5400 boosts in 3dmark11 and 3dm Vantage with the Windows high performance power plan and +200 PBO, with the Optimized Balanced power plan I was seeing 5650 on two cores in the 2nd CCD and 5525MHz on the rest. It bounced around a bit but never really went lower than 5400MHz.

 

Anyway, just thought I'd share. Hope this helps.

Edited November 3, 2023 by neurotix

[J7SC_Orion]

MannyX-ITA (OCN) released several custom versions specific to each Ryzen 7 type...I tried some of them, and they're quite good, with or without core-parking...but again, it also comes down to choosing the right ones, ie. single vs dual CCDs, and also regular CCD Ryzen 7K or the X3D ones.  I reverted back to my 'hi-po' regular Win 11 Pro plan for now.

 

[ENTERPRISE]

2 minutes ago, J7SC_Orion said:

MannyX-ITA (OCN) released several custom versions specific to each Ryzen 7 type...I tried some of them, and they're quite good, with or without core-parking...but again, it also comes down to choosing the right ones, ie. single vs dual CCDs, and also regular CCD Ryzen 7K or the X3D ones.  I reverted back to my 'hi-po' regular Win 11 Pro plan for now.

 

 

Do you have a link to those files?

[neurotix]

59 minutes ago, ENTERPRISE said:

 

Do you have a link to those files?

Agreed.