not exactly a problem, just noticed that when PiTool is running my 4090 is running at full speed. close PiTool and it drops down to idle. don’t remember that happening with my 3080TI/3090 (although it could have but i didn’t notice it). wonder why this happens? bit strange.
That was always so. As soon as an image is transmitted, even if it’s just the Pimax logo, the card works.
But she doesn’t run particularly hot.
But as soon as SteamVR starts (at the same time), the big power consumption starts.
2 Likes
Just an update , I tried running the 4090 directly to the motherboard without the riser cable , and not change , but I did install the latest drivers that came out yesterday and it did seem to help the fps but still have micro stutters , a lot of them especially in contractors vr where my 3090 ran it very good no stutters , after doing some research apparently a lot of ppl are experiencing micro stutters with the 4090 and point to the windows latest update , it just really sucks paying so much for a 4090 with these problems we haven’t had with the 3090
2 Likes
tl;dr Gamers Nexus has definitively proven that the melted 12vhpwr connectors are rare (~0.05%) and are happening due to user error in not fully seating the connectors. They conclude the new connectors are safe. Just make sure they are fully seated and latched.
3 Likes
The analysis from Gamers Nexus involves extensive failure analysis using exotic equipment. Watch the video. This is no joke. Their case is strong.
They also warn that users may be inducing failures by repeatedly disconnecting and reconnecting their power connectors to keep checking for this problem until they make a mistake and actually cause it. Making concern about problems with the new power connector a self fulfilling prophesy.
Games Nexus also criticizes other news sources, particularly Jay and Igor, for their sensational and irresponsible reporting on this issue.
1 Like
Got a Zotac OC 4090 (only one I could find MSRP) and the performance is very good. The card has some design issues (for example, longer than it needs to be due to a plastic overhang that can’t be doing anything but taking up space for the sake of design ‘aesthetics.’ It runs nice and cool, but is also very loud (fans) compared to the Aorus 3080Ti I replaced it with. In terms of build-quality it’s way better than other Zotac cards I’ve installed in the past, but definitely a step-down from big names.
I believe I may have the micro-stutter problem, but only at 90hz. I think - I have not had time to deep-dive into the issue. But I’m seeing stutters in some titles even when I’m well below full GPU usage. I did not experience these on my 3080Ti unless I was right on the hairy edge. These stutters happen under fast movement and whether or not I’m running an OC at the time.
A note on 4090 the performance: Yes, it’s a huge uplift, especially in titles that ran very poorly. Fallout 4 VR is now very enjoyable at Normal FOV, 75hz, 100% SteamSS (or even a bit higher) where before I could barely manage 60fps at Small FOV. Lone Echo 2 also made large gains. Uplift in highly-optimized games is less, however.
IMHO, if you have a 3080Ti or better, you might consider waiting it out until the 4090Ti comes along, and the new AMD cards come to market and hopefully rewrite the price points a bit. The 4090 is good, yes, but it’s not good enough to justify the cost, not even at MSRP. It’s got, maybe, $999 worth of performance, not $1599+.
4 Likes
I think this is really going to depend on what games you play and how well your current GPU is performing on them. For me, the majority of my VR time is in VRChat which was always a struggle for my previous 3080 Ti. It also struggled with Green Hell, and to a lesser degree even After the Fall. A lot of the stuff I enjoy most is pretty heavy. And on VRChat especially, the uplift in performance with the 4090 is more than double, and makes a transformative difference.
So in my case, I feel like my 8KX is a new upgraded VR headset now. The 4090 is able to drive VRChat at 100% resolution on large FOV, and it is gloriously immersive especially in worlds with high photorealism.
EAC has blocked the use of FFR which crippled performance in VRChat with my previous 3080 Ti. I’d had to turn resolution down to 64%, making everything a bit blurry. Now the 4090 can just brute force it with no foveated rendering, and EAC doesn’t matter to me anymore.
But if I was playing, for instance, Contractors VR, Blade & Sorcery, Half Life Alyx, or VTOL VR most of the time, upgrading from the 3080 TI to the 4090 still improves things somewhat, but as you say, hard to justify $1600+ for that difference.
For me, however, the big uplift in the games I personally play most has been completely worth it.
2 Likes
finally someone made it melt
1 Like
New GpuZ version
1 Like
Here’s my summary of issues as they stand today for anyone considering buying a 4090 to use with a Pimax 8KX:
Melting 12vhpwr connectors: This issue has turned out to not actually exist. Melted power connectors were caused by users not plugging in the connectors all the way such that they were unlatched and eventually worked themselves back out. This has only happened for a very small percentage of users (0.04%). The same thing can happen with 8 pin PCIE power connectors if they are not latched.
Warnings not to bend the new power cables or else they will melt are also false. Care and handling of the new power connectors and cables is the same as it was for the old ones. They can be safely bent like any other power cable.
The entire concern about melting 12vhpwr connectors was manufactured by well meaning but irresponsible journalism.
Latency: Some users have reported issues with visual latency when turning their head sharply. This seems to have affected multiple VR headsets (not just Pimax) with the 4090 and doesn’t happen with prior 30 series nVidia cards. For most users, this seems to be resolved by uninstalling/disabling MSI Afterburner and/or fpsVR.
The issue reportedly affected the Varjo Aero much worse than the Pimax 8KX, and Varjo was able to release an update to resolve the issue. Pimax is looking into doing the same.
3 Likes
That fixes it for some users to perhaps most. There is a growing number of users whom are experiencing it without any monitoring software. A user has reported pimax support has now been able to duplicate it and are working on it.
1 Like
New Article
1 Like
To my reading, this article is nothing but spin. The title is clickbait. The “PCI standards body” suggested no such thing.
There’s a lawsuit that was launched at the height of the hysteria about melted power connectors, and apparently that lawsuit is still proceeding despite the wind being taken out of the sails by GN’s and nVidia’s failure analysis reports. This message from PCI-SIG is very likely a response to the lawsuit, basically just officially saying “keep us out of this”.
My expectation is that the 12vhpwr connector will become the standard despite the media and memes. If there’s anything that all the public testing in response to melting claims has been good for, it’s proving that this small, fragile looking power connector is actually very robust. Media BS about it will blow over and before too long people will hardly remember that there was a stink early on.
Actually it does indeed exist but they should have posted the full proper name as is in the Article vs the Article title.
I didn’t mean PCI-SIG doesn’t exist. I mean the title and subtext of the article imply that PCI-SIG is suggesting that there really is a problem that isn’t just user error and blaming nVidia for it. But the message that PCI-SIG has posted doesn’t actually imply any such thing at all. The article is intentionally insinuating that it does in order to try to squeeze a few more clicks out of this dead horse.
1 Like
It seems they might be indeed as the connector on the card side could have had better audit to ensure it is easier to fully insert. Now my ho is this is maybe more of a 50/50 issue as with previous pci-x and similar plugs usually give a satisfying click. I have akways been more concerned with the Molex connectors in pc which have often been poor quality.
Nvidia and there vendors are indeed doing the right thing replacing cards and other damaged components like the psu in some cased.
PCI-SIG is considering shortening the signal pins in order to add an ability to detect improper seating conditions and avoid melting the connector even then. That shouldn’t be taken as an admission that something is wrong with the new connector. Previous power connectors have not had such a feature and have always been subject to failures caused by not fully seating them. This would be a new feature that would make the new power connectors even safer than old power connectors have been. So it’s a good idea.
The companies involved here are handling this issue very well. nVidia could have just said it’s user error, so they won’t honor warranties. And PCI-SIG could have said this is the same as its ever been and this problem is rare, so we won’t try to improve it further to make it more idiot proof. I wish the media would let go of trying to vilify them over this thing and instead give them some props for how consumer friendly they’ve handled it.
Sometimes they do but not always. You’ve never been able to rely on hearing a click to determine that power connectors are fully seated.
I maintain that the old PCIE 8 pin power connectors that everyone now seems to think never ever had any faults (even though they melted, too) behave the same in these regards as the new power connectors. It’s just that the new connector has been put under a media magnifying glass which is still trying very hard to find even the most minor thing to cast blame upon. Most of this has blown over at this point, but there are a few out there that are still kicking this dead horse.
8K X 2075, 4090 Gigabyte Gaming OC, Win10Pro, Apex z690, 12900KF all waterchilled. Have problem with a lag in comparison with 3080ti.
1 Like