I totally agree with the post in the release thread that all the talk of results, overclocking, testing, etc. Should have a new thread. Here it is!

I'm not sure if the mods can or want to migrate posts from that thread to this one to clean that one up or if people would like I can probably do some copy pasta over here from there, but I don't have time right at the moment. I just thought I'd kick off the thread for everyone's use.
 
Some info on my results so far:

^this OC run was stable for memory, but not the CPU OC (intel XTU stability test failed after about 10 minutes or so). I need to do some more work to figure out why. I had monkeyed with settings in both BIOS and Intel XTU. I would like to find a set of happy settings in the BIOS so that I don't ever have to touch XTU and nothing is fighting each other or requiring to be set every time I boot up.
 
One of my CPU-z scores (733 SC, 7244 MC v2017.1) I think this is at the same settings as the aida-64 memory test above.
 
One of my Cinebench r23 runs (1674 SC, 16390 MC):

^this is at a fully stable setting. I believe it is essentially a 5.3GHz all-core with the memory at 3733MHz gear 1, 16-16-16-36, 1.4V.
 
Personally, I'm logging more information in my MODS Rigs post, but I can try to keep some of the more informational stuff specific to the Z590 FTW settings here to help the others with this board so they don't have to go looking for my specific post.

Here's what is stable. Going to 5.4GHz on two cores is still not stable. curious if I'll need a BIOS update or some actual voltage tuning for that because I'm currently just on "auto" for vcore.
 



 
Here's where we get during a few minutes of R23 multi-core. You can see that temps hit 70 so it drops to 5.1GHz. I see it go a little lower than that on a few cores sometimes too.
 
Sort of makes sense from a current-draw perspective, but the Vcore and Core VIDs drop quite a bit when under load. They are at their highest when mostly idle, but clocks are also higher at 5.3GHz when I'm not stressing the system so it could be the higher clocks too.

evgadaddio
I am having an odd issue with the OCing.

I am stable using AIDA-64 stress test (1.5 hours is the longest I have run it).
Memory is Gear-2, 4000, 1.4V  and CPU is at 5.3 for 2 cores and 5.2 for the rest.

The issue I have is that if I shut down the computer, it often will not start up with those settings and I need to do a bios reset.  Also sometimes has a similar problem coming out of sleep.
 
Without any OCing, I don't have any problems.
 
Any thoughts on what could cause this?

I sometimes have a similar issue, but have thus far assumed I'm doing something wrong. Sometimes when I load my settings in the BIOS, then get into Windows, it's at default, but with the boosts turned on (so 5.3GHz on two cores and 5.1GHz on the others). I then open the XTU app and manually set my 5.3GHz all-core by setting the package ratio-limit to 8-core workload: 5.3GHz (automatically sets all above to that speed too). I think it normally works right if I set the package-OC in the BIOS, but if I try to set the per-core OC in bios I don't think it takes and windows shows the 5.3/5.1 clocks.
 
Also, I noticed that in the XTU app, if you wanted 5.3 for two cores and 5.2 for the rest, and you wanted the 5.3 to happen on specific cores, you'd have to set both the package ratio-limit to 5.3 for two core workloads and 5.2 for the rest, then go into the per-core settings and manually set the clocks there too. Just setting the per-core clocks doesn't actually make the clock-speeds increase. You have to also set the package ratio limits. I'm not sure if the BIOS OC settings are the same way.
 
On a separate note, the last two times I booted up, the LEDs on the I/O shield remembered the settings I put in for them. So I didn't change anything, but it seems to be working correctly now...no idea. I did check the BIOS version on the USB drive is the same as what was already on it and the drivers are (as of last night anyway) the same as the ones on the website.
 
So now I just need to do some research into stable and safe voltages and see if giving a boost to my vcore, offset, vdroop, etc. can help me get to the 5.4GHz all-core (which is what the OC Robot tells me I should be able to do) or at least the 5.4GHz on two cores stable.
Edit: it sounds like my best bet might be to watch the voltages during the OC-robot run like a hawk and then maybe set that voltage manually. I still need to read about the usage of the offset voltage as I haven't had to use that before. Any advice there?

mine does not show :(

Nike_7688

Anassaidi
mine does not show :(

Can you be more specific? If you're talking about the LEDs on the I/O shield, you may need to turn them on via the ELEET X1 app. There's a separate thread here with a link to that download.

yup, i submitted a ticket a few weeks ago and said they are working on a new version, they released the version that works with our motherboards here:
https://forums.evga.com/E...erboards-m3296354.aspx

tuanmassey

rblaes_99
can you post your CPU/Voltage BIOS settings too?  

are you still having problems overclocking your CPU? is your robot OC working? when i did my robot OC i made sure all my fans were manually set to 100%. also put my AIO pump to 100%. after doing that, robot OC worked out for me. 

Can you post some of your results? BIOS settings or just tell us what you've had luck with for CPU clocks/voltages?

Those are the robot OC auto settings. The Stress test was in Non-AVX mode. If I try to stress test it with AVX2, the computer locks up while in BIOS on 5.4GHz. I have to put it down to 5.2GHz to be able to stress test on AVX2. This test was done with AIO pump and all fans at 100% speed. I'm a little confused as to why it says 5.4GHz but it only actually runs at 5.3GHz? can someone explain?

 
3DMARK results with Video card and CPU on this setting:
http://www.3dmark.com/spy/20300965
 
I think it's safer for me to daily run it at 5.2GHz on all cores since that's what the AVX2 CPU stress test likes it at. 
 

Thanks for posting all that! The OC Robot gives me the same settings. It sets the global multiplier at 5.3, but each core load multiplier at 5.4. This results in an all-core oc at 5.3GHz for me. What's odd is that if I manually set any core loads at 5.4, it automatically moves the global level to 5.4, and then I see cores at 5.4 in Windows. I can't manually set it the same way as the robot and I'm really not sure what it is doing there. I'll have to screen grab my oc robot and stress test to compare it to yours.

Can you change the avx2 enabled or disabled in the BIOS stress test? I don't remember seeing that option, but maybe I missed it.

Nike_7688
Thanks for posting all that! The OC Robot gives me the same settings. It sets the global multiplier at 5.3, but each core load multiplier at 5.4. This results in an all-core oc at 5.3GHz for me. What's odd is that if I manually set any core loads at 5.4, it automatically moves the global level to 5.4, and then I see cores at 5.4 in Windows. I can't manually set it the same way as the robot and I'm really not sure what it is doing there. I'll have to screen grab my oc robot and stress test to compare it to yours.

Can you change the avx2 enabled or disabled in the BIOS stress test? I don't remember seeing that option, but maybe I missed it.

when you stress test cpu in BIOS, i always have the option to choose non-avx or use avx2 for stress test.

tuanmassey

Nike_7688
Thanks for posting all that! The OC Robot gives me the same settings. It sets the global multiplier at 5.3, but each core load multiplier at 5.4. This results in an all-core oc at 5.3GHz for me. What's odd is that if I manually set any core loads at 5.4, it automatically moves the global level to 5.4, and then I see cores at 5.4 in Windows. I can't manually set it the same way as the robot and I'm really not sure what it is doing there. I'll have to screen grab my oc robot and stress test to compare it to yours.

Can you change the avx2 enabled or disabled in the BIOS stress test? I don't remember seeing that option, but maybe I missed it.

when you stress test cpu in BIOS, i always have the option to choose non-avx or use avx2 for stress test.

Cool. I think I only ran the stress test in BIOS once, so I couldn't remember what was there. I'm at work so I couldn't just go look lol. I plan on re-running the tools in BIOS and posting my results. I will do some manual voltage tuning too to see what I can accomplish towards running at 5.4GHz (or even getting 5.5 on one core good for benching). I noticed in the scatterbencher videos, they talk about V/F offset voltages at different setpoints. I don't think we have that functionality? It seems like it might be something in the ASUS and Gigabyte Tachyon boards. I wonder if it is just done "under the hood" on the EVGA side. I'd love some guidance on the vcore vs. vcore offset settings. For example, would it be good to run 1.5V vcore with no offset voltage, or 1.45 vcore with +.120V offset? No idea which is better or which to try first. I've been reading that you can use + or - offsets in various motherboards to basically either add or subtract voltage at load vs. idle, but you can also use a turbo offset voltage? I need a Haynes guide lol.
 
Also, I just realized your BIOS screenshots were pictures. Not sure you saw it, but you can press F12 for screen-shots if you have a USB drive in. It just saves them straight to the drive. That's where mine came from.

tuanmassey
 
I think you should throw your RAM into gear 1 and retest your 3dmark results. Even if it's at 3200MHz or 3333MHz. your CPU results should be much higher than that at 5.2GHz

Tuan:  this is me gear 1 @ 3333 x 5.2ghz.  much better score
 
NVIDIA GeForce RTX 3090 video card benchmark result - Intel Core i9-11900K Processor,EVGA Corp. Z590 FTW WIFI (3dmark.com)
 
i am 99% sure the first few BIOS updates will improve on this memory support business.

Nice! Now your latency is much lower in gear 1 vs gear 2 which is why the score is better. The sacrifice is the read and write speeds are a bit slower when running RAM at a slower speed but i think the overall performance is better in gear 1.
 
When in gear 2, your "memory controller" speed is reduced by 50% to be able to handle the extra RAM speed. If you were running something like 5333MHz in gear 2, it will be really close to the latency of your current 3333MHz gear 1 setting.
 
Ideally, with Z590 boards, you want to be running 3733MHz in gear 1 with the lowest timings to get the best performance. Just like how Nike_7688 has. He's the lucky one and scored with his RAM selection.. lol.
 
I wonder if the board cannot handle 64GB in total for gear 1 higher than 3333MHz. rblaes_99 - can you be a trooper since you have 4 sticks of 16GB, please try only two sticks, totalling 32GB and see how fast you can get RAM in gear 1? one stick in slot 2 and other stick in slot 4

Well, I tried to play with voltages a bit. This is what the 5.4GHz OC robot showed:

So I went to 1.45Vcore and +50mV offset. That's maxed out for "normal voltages". When I check HWiNFO with these settings, my voltages max out around 1.565 (which is a bit higher than the 1.455 from before, but I mostly monitored it at 5.3GHz). If I go to "extreme voltages", which I tried only briefly, it shows 1.6+ volts when I run the BIOS stress test (which was showing the same 1.548V as the OC robot when left on auto previously). So that made me hope that the 1.45V and +50mV settings were actually working as intended...
 
The BIOS stress test (non-avx) showed the same voltage, so I was aiming for that. I ran Cinebench with 5.4GHz on two cores and 5.3GHz on the rest (16909 MC, 1669 SC):

It ran the aida-64 stress test fine for a couple minutes (after doing both non-avx and avx stress tests a few minutes each in BIOS), but crashed when doing some memory testing with memtest64. It showed no memory errors though...What was weird was if I set it in ELEET X1, it was much happier than if I set it in Intel's XTU app. Also, the settings in BIOS don't seem to always carry into Windows for the core speeds. It seems like the per-core functions don't work right and the manual ratio-limits only work right if I just apply the OC Robot settings. So that's kind of odd. 
 
So my voltage bump was definitely more stable than before, as I could run several minutes of Aida64 system stability test, but it still isn't totally stable. If I go higher on voltage, I'm into the "extreme voltages" and I'm not sure I want to play with that too much until after I fix up my loop (I have some fittings to change out still) and check my CPU mount. I am in agreement that it seems we need a BIOS update with some stability and functionality improvements (especially in reference to settings not taking correctly). Currently, it seems I have found some settings I can run several benchmarks on, but not a full gamut of stability tests lol. I guess I'll "settle" for 5.3GHz all-core...which is still pretty awesome.

tuanmassey
Thanks for the info! Your water cooling setup is way beefier than mine so I won't be able to push those numbers lol. I'm going to wait for BIOS improvements before I OC my CPU any harder. 

I just lost a step on that front. Idk if it is the EK fluid just not being as efficient as distilled water, that there are still some air bubbles, or if I've not done as well as my original mount (tried both a different method with the pk-3 and the kpx), but I've lost some of the thermal performance today. Looks like it's running about 6-8° hotter under stress testing, which came with a drop in stability. I have tested a couple methods of setting clocks in BIOS, and none of them seem to stick when I get to windows. I can change them in ELEET X1 or Intel XTU, but that is definitely not preferable. All the memory clocks and voltages stay, but the multipliers don't. It just goes to 5.1GHz all-core when I boot to windows...every time.

tuanmassey
You have “ Intel adaptive boost technology” disabled in BIOS right? In advanced CPU configuration. If that’s enabled it will “boost” to 5.1GHz. Hopefully that’s the issue.

Hmm... I'll check. I think I had turned that on thinking I needed it.