I tried to play with the XTU app a bit more and it shows you your "strong cores" with Rocket Lake. I tried to set it so that those specific cores went to 5.4GHz and the rest to 5.3GHz. Stability test was not so good there. What seems odd is that it doesn't perform as you'd expect. When you set the package ratio settings to clock the CPU depending on how many cores are being used, you'd think that setting:
1-core workload: 5.4GHz
2-core workload: 5.4GHz
3-8 core workload: 5.3GHz
You'd see that SC-workloads go to 5.4GHz on a random core, and it would run 5.3GHz the rest of the time basically. I've tried to shut down everything else, but watching HWiNFO during Cinebench SC, I see one core get warm and the utilization go up, but it jumps to different cores every time CB starts a new block. This is normal, but I never see it go to 5.4GHz.
If I set the per-core OC so that it does
5.3GHz
5.3GHz
5.4GHz (strong core)
5.4GHz (strong core)
5.3
5.3
5.3
5.3
I thought you'd see one of the strong cores pick up the work for SC-workloads, but I had some stability issues here before I got too far into testing. Also, it looks like it locks them at that speed all the time (as opposed to downclocking when not busy). I haven't really messed with my voltages at all so far, I've just been running them at adaptive-auto. Right now I can run 5.3GHz all-core with zero issues. I can set it to run 3 cores at 5.4GHz in BIOS and 5.3GHz for the rest and it runs with zero issues, but when monitoring in Windows, it seems like none of the cores ever actually run at 5.4GHz. It seems a bit strange. It behaves the same exact way if I set the per-workload package OC in XTU (5.4 for 2-3 cores, 5.3 for the rest results in stabile performance, but it never does anything at 5.4GHz). it's only when I manually set in XTU package OC to all be at 5.4GHz, or do a per-core OC with 5.4GHz that I start running into stability issues. This could be voltage related. When I do stability testing, my temperatures max out in the low 70's. If I do a test with AVX instructions, they hit the low-to-mid 80's. On previous CPUz and EVGA motherboards, I've generally gone by Jacob and a couple others' recommendations for a safe 24/7 OC voltage. it was 1.3V for my 5820k and it never had an issue. I'm not sure what that number is for the 11900k, but it's higher than that. It is also higher than it was for 10900k's, so I may need to wait to see some guides from other people who are more experienced at this than I am. Maybe I need to re-watch some scatterbencher videos lol.
Also, there's this issue:
Nike_7688
I'm not positive my paste application was perfect. I went a little heavier than I normally would have in the past and I don't have a little spreader tool so I did a big blob and cranked it down hoping to spread out the PK-3. I honestly had several plans and when it came time to do the paste something in my brain just shut off when I realized the Pk-3 didn't come with a spreader like I thought it did and went "JUST DO IT TO IT" and a blob of paste came out on the center of the cpu lol.
-Yeah, I'm not convinced this is doing so well. This thing is a beast and a half, which I expected, but the temperature ramps seem to be a bit faster that I expected for some reason. This might just be the effect of ~240W hitting all at once when it cranks up the jams, but I might try and take the block off when I have to fix my loop anyway just to see how the mount was. I will be adjusting from there, and will either try to make a spreader tool from something I have laying around or I'll switch over to the KPx to see how it does. per the temperatures listed above, I don't think it's a
terrible mount, but I'm just not yet convinced that it is optimal. I'm not going to be until I take it off and look at it. Most pastes I think are fine with the puffy grain of rice approach or a round blob, but the PK-3 seems to really prefer the thin spread as it is very thick and I don't think it squishes well. The reviews I read seemed to show different results for it depending on the application method and because I
knew that, I'm a bit grumpy with myself that I just blobbed it anyway.
driam
Epic to see that you got it up and running! Also project name checks out, I like the black tubing and vibes. It reminds me of the lighting in submarines, like in the movies when they are on alert and things get red and dramatic.
You could use accents in the case or outside like nautical bulkhead lights mapped to the current RGB of your system. Of course smaller ones that can fit lol!
You could also make a cage around the reservoir to look like a bulkhead light as well:
This is a pretty cool idea. I'll have to think about that. I'm not sure how I could pull it off with the design of the exterior of the case, but maybe with some searching I'll find something that fits.
driam
I wonder if you could place some LEDs on one side of the reservoir that the liquid pours into, this could potently refract and give a cool lighting effect, though I think you did that already on in one of the pictures, hard to tell.
so I currently don't have liquid going into the top, it just goes into the left side of the front pump-top under the reservoir. The case lighting I have is just two LED strips that came with the case, one along the inside front edge and one along the top edge. It's sort of "front lighting", but the way it's hidden behind edges you don't see from anywhere you normally look at it, it sort of appears like "back-lighting" which I like. I prefer that to RGB-everything. I've been playing with the G-skill memory lighting tool, because I can't find a way through BIOS to make it work, but I've got the memory doing a very slow blue "breathing" effect. I also had the motherboard do that over the I/O shield, which you sort of see a glow of behind the GPU's radiator mounted to the back (left in pictures). I like this effect so far, but it does seem like I may need to open both g.skill's app and the ELEET X1 app to make them work right every time I boot the computer...which is annoying. If I could get those set up in BIOS, it would be significantly easier.
driam
There may be a chemical you can use that wont harm the PCB if it spills on it, but will break down the glue. A heat gun sounds terrifying to apply to a chip lol. You could twist it off, as I think pulling on it would place a lot of stress on the solder joints, but twisting would use force on a stronger axis, after maybe use a knife to carefully remove the rest.
I did some more digging and even asked Patriot customer support. They said it would probably damage the drive to get it off, and some reviews I found that took it off to try and do some standardized temperature testing said that they couldn't get all the adhesive off the chips from whatever thermal tape Patriot used no matter what they threw at it. There were a lot of people that said they almost broke or did break their SSD trying, so I'm thinking it just stays. Maybe when prices sort of normalize (it has to happen some day, right?), I'll buy a gen-4 NVMe drive and sell this one. I may also just do it whenever it is that I upgrade my GPU. It's interesting that I'm probably losing a pinch of FPS on my 2080 Super, but if I upgrade my GPU, it would be using 8x gen 4, which would be much faster than the 8x gen 3 I'm currently utilizing, which means maybe it wouldn't matter at all anyway...idk lol.
Edit:
I was looking at my CPU-Z validations and all but one of them don't show the GPU. They also don't show the memory brand/model. The only one that does show it is the one I did where I messed up the memory timings by accident.
test1,
test2,
test3,
test4. I wonder if something has gone awry in my BIOS settings after I tweaked so many things, causing the instability with two cores at 5.4GHz, which seemed to work fine previously. I may reset defaults, check the version (I don't think I even checked to see if the version on the USB stick was different to what was on there...but there isn't one listed on this site yet), and then apply my settings from scratch. May also re-install CPU-z.