Yup, they mention the connector peaked at 70C with stock clocks and 80C under OC. Fun times.
From HUB's review of the MSI 5090 Suprim
Yup, they mention the connector peaked at 70C with stock clocks and 80C under OC. Fun times.You should have waited a few more seconds.
Yup, they mention the connector peaked at 70C with stock clocks and 80C under OC. Fun times.
How do you reconcile the ~70C measured in this review with the data you brought up a while ago? (emphasis mine)Not that I would OC a 600W card, but all the FIT connectors are rated for 105C operation with Tin connectors, and 125C with gold plated connectors, so either way still comfortably under spec.
I've seen testing on the connector, where 900W (skipping the GPU and using a PSU load tester) was pumped through it for half an an hour and the hot spot stabilized at about 52C. Which is fine, especially considering that's a 50% overload for half an hour (which would never realistically happen), and then pumped 1200W through it for 10+ minutes, then it reached something like 82C which is a good point to stop before you start melting things, but that's double the the continuous load rating, way beyond never going to happen.
How do you reconcile the ~70C measured in this review with the data you brought up a while ago? (emphasis mine)
To clarify my stance, I don't doubt this connector can work properly @ 600W in good conditions, but I have my doubts about what happens with samples that are close to manufacturing tolerance and operating under adverse conditions (high ambient temp, mediocre ventilation etc.). We'll see how it goes, the good thing is people are unlikely to use adapters of any kind, and this should help a lot.
Is this correct connector?The previous time, I was going from memory of a test I had seen of some small channel testing with a load tester. I have no idea why HWUB is getting higher temperatures with lower power.
But it is still comfortably within spec, and this time, I checked the Molex connector specs:
The whole issue with 12+4pin and the 12+4pin apologisms is that it is a gross mistake to pick a connector based on good conditions and just assume it will always work in good conditions and just blame the user or Canada if it doesn't. People have to bend the cable close to the connector's terminal in almost every card installation, and the PSU end that uses te same connector is usually hidden in cable mess, enclosed in the PSU tunnel where you may not even notice it starting smoking...How do you reconcile the ~70C measured in this review with the data you brought up a while ago? (emphasis mine)
To clarify my stance, I don't doubt this connector can work properly @ 600W in good conditions, but I have my doubts about what happens with samples that are close to manufacturing tolerance and operating under adverse conditions (high ambient temp, mediocre ventilation etc.). We'll see how it goes, the good thing is people are unlikely to use adapters of any kind, and this should help a lot.
Evidence (in the form of smoldering 12VHPWR connectors and molten plastic) would seem to indicate otherwise.All across industry, Cables/Connectors/switch have amperage rating and they are used right up to that amperage rating, because even at their full ratings, there is plenty of safety factor built in. That is how engineering works.
That is. Note the field temperature and life rating is 65°C for 10 years. Rule of thumb with wiring is every 10°C rise lowers lifespan by a factor of two. A gamer is obviously not going to run at 65°C 24/7, but at the same time 105°C isn't meant to be a long term operating temperature any more than you'd want to run an engine at redline constantly.
I kinda' ignored the PSU end because I was used to seeing the cables made for "older" PSUs that had the 12HPWR connector at the GPU end but used 2 (proprietary) 6-pin connectors at the other end. I just realized I need to pay extra attention if and when I buy any new PSU.and the PSU end that uses te same connector is usually hidden in cable mess, enclosed in the PSU tunnel where you may not even notice it starting smoking...
It's quite possible this specific microfit connector itself is a failure/outlier in this regard, basically being more fragile/less over-engineered than is usual, not having as much headroom as other designs.Evidence (in the form of smoldering 12VHPWR connectors and molten plastic) would seem to indicate otherwise.
Oh that's certainly the case! Definitely in the case of insufficient headroom. Which is what PCIe 8-pin fans have been arguing from the start.It's quite possible this specific microfit connector itself is a failure/outlier in this regard, basically being more fragile/less over-engineered than is usual, not having as much headroom as other designs.
It's quite possible this specific microfit connector itself is a failure/outlier in this regard, basically being more fragile/less over-engineered than is usual, not having as much headroom as other designs.
The two are measuring different things, so it's not directly comparable.It's more likely it just wasn't properly seated.
I found the original video with the Load tester. It only hit 50C at the full 600W, and it stabilized at about 82C with 900 Watts continuous:
Is this the first time that NV has acknowledged that the 4090/12VHPWR has "cable melt issues"?New article on this topic,
No, they acknowledged user error before, now they made changes to the connector to help their usersIs this the first time that NV has acknowledged that the 4090/12VHPWR has "cable melt issues"?
They have no qualms about badmouthing their "old" product to upsell their new product.Is this the first time that NV has acknowledged that the 4090/12VHPWR has "cable melt issues"?
Oh right, of course we can blame 12VHPWR now that they've abandoned it for something that's the same except for some shorter sense pins . . . (okay that's not all that's changed but still)No, they acknowledged user error before, now they made changes to the connector to help their users
Oh right, of course we can blame 12VHPWR now that they've abandoned it for something that's the same except for some shorter sense pins . . . (okay that's not all that's changed but still)
Even NV just admitted it was a connector problem. Oops!When the problem is connectors not being properly seated