Mr. Fox

No, with manual voltage. Auto voltage is out of control sky high and dangerous, and extremely hot.

It's totally normal. Are you capping your power limit? Every single 10th, 12th and 13th Gen CPU I have owned hits power draw numbers like this in Cinebench R23. So did the 5950X, at much lower clock speeds. At 6.0GHz all P-core sync is it over 600W.

Yes. With the IHS it will go to like 105°C and blue screen running the same BIOS settings in Cinebench R23. Anything above 57x all P-core was unstable in Cinebench R23 due to extreme heat before delid and bare die. This is on the system with the Nova 1080 radiator, no chiller and no 5-gallon reservoir, so it has to work a lot harder than our systems with 5-6 times more liquid.

No, look again. It is 1.270V under load. It goes up to 1.380V at idle, which is normal. Benchmate grabs the highest voltage, not the accurate voltage. (Benchmate is not as special as some like to pretend it is. But, it makes pretty screenshots for HWBOT, LOL.) My water is 29°C, so about 3-4°C hotter than ambient. I might have added the second screenshot before you refreshed the page, so maybe you did not see it. This is on the system with the Nova 1080 radiator, no chiller and no 5-gallon reservoir, so it has to work a lot harder than our systems with 5-6 times more liquid.

No. Totally normal. You've forgotten what it is like to run a system without chilled water. Voltage, power and temps are all MUCH higher even with a golden silicon sample. With chilled water even a mediocre CPU can do things a golden sample can't do if cooling with ambient water temperatures. I just ran this. My ambient temperature is 74°F (24°C). 561W and 92°C, and that's bare die. My AC and DC are fully tuned and run exactly the same value under load. Only using 1.270V under load.

Easily above 550W and over 100°C. Without even trying. Even with my SP117 CPU running the lowest possible borderline stable voltage. Ambient cooling (air or water) within an air conditioned room (70-73°F) won't get you anywhere. It's nice being able to use chilled water 24/7 if you have that ability, but that's not a normal real-world scenario. The best SP-rated CPU in the world won't run low power and low voltage if cooled using ordinary measures in normal ambient temperatures. As soon as the water in the loop normalizes it will become unmanageable and start breaking 100°C and chugging close to 600W with an all P-core 60x overclock. And, that's one of the things that makes it awesome. Simply because it can.

This isn't for points, just a down and dirty run to confirm overclocking ability of the Gigabutt Gaming OC GPU. The stock vBIOS is voltage knee-capped at 1.070V like all of the new cancer GPU firmware. *SMH* BUT, nothing that a quick flash to the Galax vBIOS can't fix. Now it has a less castrated 1.100V voltage limit. Before it would crash above about 2950MHz on core, but now 3000+ is doable. Memory has no problem with +1500 so far. Don't know if it can run +1700 like the Suprim can, but I will find out when I have time to test it on chilled water. So far everything I can see tells me this is a good GPU. Max and Average on clocks and voltage...

Congratulations. I have been signed up for 4090 Suprim for 6 months and signed up for 4080S Suprim stock alerts on launch day and have never received one. I am no longer looking for one, but I have not cancelled them. I just checked my email and nothing.

Yeah, there is not much logic or common sense left with things related to video cards today. It's crazy how many normally smart people suddenly become so stupid.

Maybe after the Chinese New Year has passed and people go back to work over there stock will become available. https://www.nowinstock.net/computers/videocards/nvidia/rtx4080super/ Bear in mind, the Strix GPUs are almost always grossly overpriced and they add insult to injury with the fact that things are hit or miss in terms of core and memory bin quality. Plus, ALL of the ASUS GPUs have an oddball proprietary 12VHPWR cable that is made upside-down compared to all other brands. You'll have to use the ugly 8-pin adapter or have an ATX 3.0 PSU cable that supports the stupid ASUS orientation. A different brand is a better option here for multiple reasons. ASUS is near the bottom of the list for me when it comes to GPUs. If I were going to compromise on the 12VHPWR cable, the TUF is the only one that is not grossly overpriced, and it usually overclocks as well as a Strix.

Some people will always find a reason to complain. Maybe they would prefer a melted GPU power socket. Even if I never had a problem it would suck having to worry about it. Not needing to think about it would be better.

I submitted my refund request for the recall. It felt strange breaking it and taking a photo, but that is easier and better than having to mail it back to them.

Thank you. Yeah, that really sucks. We can have or do anything we want as long as it is what our icky dick taters want for us.

Thank you. I agree. It does look fantastic, and it runs very cool with that massive heatsink. IMC is really solid, but I haven't taken time to tune the memory further. My MCSP is higher than the 13900KS in the white Apex, so I suspect I can. Yes, the GPU and RAM have smaller tubing diameter on this system. The RAM is deliberate and that way on both systems as I want more water flow to other things like CPU and GPU. The GPU is that way only because I ran out of larger tubing and haven't purchased any more. The Alphacool Eisenblock is super nice. Nicer than the Bykski GPU blocks, and heavier. It was super easy to assemble. Now it is time for bed. It's 2:47 AM, LOL. But, the work is done. CPU delidded and GPU blocked.

Thanks. Yes, 1.390V for 60x all P-core synced and Cinebench R23 stable is very good on ambient water. My ambient temperature is 74°F right now and I have a 1.5°C delta between the ambient temperatuer and the water temperature. The BIOS is showing I can raise the E-cores to 50x all core synced before the current voltage value is too low. I don't know if they will be stable at 50x synced, but I will try that when I get around to hooking it up to the chiller.

OK, did some more tuning. This is very good for a system with no chiller. With this SP117 I am able to run this with no thermal throttling. Delid and bare die helped tremendously. The SP117 did not behave as well as I had hoped before going bare die. After I get the Gigabyte 4090 blocked I will move the chiller over to this system and see how it does. Maybe can do 6.2GHz P-core and 4.9 or 5.0GHz on E-core with chilled water. I like the IceMan CPU and direct touch memory block a lot. Very well made products.

Cool (pun intended) this arrived early. It wasn't supposed to be here until Monday. So, I guess I know what I am going to be doing either later tonight or tomorrow.

I haven't changed any BIOS settings, just delidded and put the IceMan bare die block on. Running Cinebench R23 before with these settings was 102°C. So, now I can apply my profiles from before and tune from there for the better E-Core SP rating. (P-Cores are essentially the same.) Now the temps are... Thank you. The system with the chiller runs cooler even without the chiller running because it holds over five gallons and the other system holds less than one gallon. Yes, I use the same automotive antifreeze premix in both systems. I suspect the flow rates are similar in spite of having three more pumps on the chiller setup because it has about four times as much tubing, more fittings, including more than twice as many QDC fittings. All of those things make the pumps work harder and reduce flow rate. But, the flow rate doesn't need to be any higher at around 295-300 L/H. I think the system without the chiller having only two pumps and pushing 280 L/H is really good flow rate for only two pumps. So, are you saying I am an idiot for buying a Gigabyte GPU, but I just got lucky getting a GPU with the revised PCB? 😉 Congratulations. That sounds really nice. I hope your family is blessed with many years of love and happiness in your new home. I don't know why, but I have three now. I'm using the mean little dwarf for my work PC and have two to play with. Maybe I am just remembering the good old days with a home filled with PCs and sons that loved having frag matches with daddy on the home network. Maybe now that I have three I can convince a couple of them to come over and play sometime.

I decided to check while I was delidding the SP117 13900KS. Yes, it does have the improved PCB design.

Thank you for the compliment. No, those are just lines on the bottom of the case that lead to my external cooling system. The pumps, reservoir and 1080 (360x3) radiator are outside of the case. I have passthrough fittings in the empty bottom PCIe slot in the case. Both of my large desktops have the loop designed the same way. The one with the water chiller has a 5 gallon tank reservoir and five D5 pumps. The one in the photo has two D5 pumps. Strangely enough the flow rate is almost the same on both systems. You can buy it direct from EKWB and usually less than from Titan Rig or Performance-PCS unless you can grab one on sale or a good used one on eBay. https://www.ekwb.com/shop/ek-pro-manifold-2cpu-4gpu-acet There is a smaller version with fewer ports, but I prefer the larger one. Even though I do not use all of them, they are available and it allows me to space the fittings further apart (an empty plugged port between each fitting). On the other system I have it mounted to the floor of the case. I did not know there was an improved version. Thanks for the link. I will check the design when I take it out to install the waterblock. I have no idea which one it has. Either way I am not too worried because I never allow any GPU to hang without support. I'm kind of anal about that because I think it is harmful no matter how the PCB is designed.

Yes it is. I have one on each of my custom loop desktops. It makes things so much easier and neater without having components in a loop routed in series. I can disconnect just one component with QDC fittings without disturbing any other components or the loop itself. I can't imagine functioning without them any more. They seem like they are a bit on the spendy side, but when you consider it is a buy it once and use it forever purchase it's actually not very expensive.

I played COD Modern Warfare (2023) for just over 2 hours this evening with the GPU overclocked +100 core, +1400 memory, +100mV and pulling over 380W yet temps were perfect. Not even close to getting hot. Max temps were core 54.2°C, memory 58°C and hot spot 62.6°C. That's fantastic for an air cooled GPU with an overclock. Now I just have to pop the lid off of this SP117 CPU and run it bare die so I can clock it higher that 5.7GHz without it getting too hot under a severe load like Cinebench. Looks like with the maxed out settings and ray tracing candy cranked the GPU was using almost 15GB of vRAM.

I bet there are people that own an electric car that would go off the rails complaining about the 4090 price being too expensive, not knowing that replacing the battery pack in their EV turdmobile will cost them as much as you just spent on your house, or the equivalent of about seven 4090 GPUs and 1,000 times more harmful to the environment, even though going "green" may have been part of their irrational justification for wasting gobs of money on an unreliable mode of transportation.