News Seasonic’s next-generation Prime PSUs to will try to stop connectors from melting

[Admin]

Seasonic is equipping its next-generation Prime power supplies with sensors, firmware, and external monitoring to prevent overheating and melting of 12VHPWR and 12V-2×6 connectors.

Seasonic’s next-generation Prime PSUs to will try to stop connectors from melting : Read more

 

[TheyStoppedit]

Well this is nice, my data on my SSD/HDD will take the fall for the well being of my GPU. At least my GPU won't melt. Lol. This is so stupid. A GPU can be replaced. Data can not. It's not PSU manufacturers job to try and clean up a problem that NVidia created especially if we're gonna create one problem to solve another. What good is that? PSU manufacturers should be going to NVidia and telling them to be a man, own their mistakes and fix the problem at the root, not try and do a whole pile of electronic gymnastics to dodge the symptoms of a greater problem. This is getting more and more ridiculous by the second. A class action is needed here. This won't fix the problem. It might mask the symptoms at best. Someone needs to get through to NVidia and I think only a judge can at this point lol

 

[chaz_music]

Hmm. The correct solution is to put interleaved Buck converters on the GPU side for each pin. Make each pin current share directly through control. Simple fix and with modest cost delta.

Of course, it really would have been easy if they had just chosen a larger 2 pin connector to bring power in, like the Anderson Powerpole 45A series. Soldering those monsters on a PCB is a chore though.

 

[S58_is_the_goat]

Their solution still doesn't really fix the problem, an amp limit per pin is the solution, but it seems Nvidia is the one who should be implementing this. But seems Jensen doesn't want to spend a dime and needs another ivory back scratcher.

 

[hotaru251]

PSU makers shouldnt be the ones needing to try and fix issue of the GPU's connection.

GPU's in past had extra safety...don't reward them for taking em away and making others have to do it instead :|

 

[m3city]

I have an idea. NVidia, please Stop that 500W graphic bs, focus on reducing power draw. AMD keeps tdp of cpu's at generally same lev for over 10 years nów, while performance mulitiplied. Intel similarly, but their cpus skyrocketed its tdp i latest revisions. World does not need RT games at 4k, interpolated fps. 30 years ago game developers made games that worked. Nów we have this constant push for 'more power'. And melting connectors. And 1500W psu.

 

[John Kiser]

Well this is nice, my data on my SSD/HDD will take the fall for the well being of my GPU. At least my GPU won't melt. Lol. This is so stupid. A GPU can be replaced. Data can not. It's not PSU manufacturers job to try and clean up a problem that NVidia created especially if we're gonna create one problem to solve another. What good is that? PSU manufacturers should be going to NVidia and telling them to be a man, own their mistakes and fix the problem at the root, not try and do a whole pile of electronic gymnastics to dodge the symptoms of a greater problem. This is getting more and more ridiculous by the second. A class action is needed here. This won't fix the problem. It might mask the symptoms at best. Someone needs to get through to NVidia and I think only a judge can at this point lol

I get annoyed hy this sentiment that it is Nvidia at fault. You do know that Nvidia alone didnt design the connector and outside of the founders edition cards straight from Nvidia the AIB parnters sre responsible for the power connectors that they do.

Nvidia qasnine of the designers and ifnthe connector wasn't ready the overlying body who designed it should have been adamant that no-one use it for production products at all.

 

[Heiro78]

I get annoyed hy this sentiment that it is Nvidia at fault. You do know that Nvidia alone didnt design the connector and outside of the founders edition cards straight from Nvidia the AIB parnters sre responsible for the power connectors that they do.

Nvidia qasnine of the designers and ifnthe connector wasn't ready the overlying body who designed it should have been adamant that no-one use it for production products at all.

The 12VHPWR spec was adopted by PCI-SIG. Many companies are a part of it. Including Nvidia, Intel, AMD, Apple, even video card AIB partners like Asus, MSI, and Gigabyte. We don't know which companies worked on or voted in favor of 12VHPWR. Nvidia was the one to implement it into their video cards AND mandate it for the high end products. AMD capatalized on it in their marketing with their not using it. If you watch the video linked below from Actual Hardcore Overclocking, you'll see another reason why people blame Nvidia for the hardware failures versus AIB partners. They intentionally did not have the per wire protections that could hamper the issues. Companies like EVGA left the business due to disagreements with Nvidia. We can only speculate as to who truly is at fault with the 12VHPWR connector.

People have commented to me that Intel was the head developer of the spec, but again who knows. My personal head canon is that Nvidia didn't want to continue with the 8-pin connectors because it would require 4 plus on a card per it's spec (safety factor) to achieve 500+ watts of power.

View: https://www.youtube.com/watch?v=kb5YzMoVQyw

 

[aberkae]

I bought a thermal camera to mitigate risk. Not taking a chance with these promises.

 

[Heiro78]

I bought a thermal camera to mitigate risk. Not taking a chance with these promises.

You're kidding right? How much was it?

 

[aberkae]

You're kidding right? How much was it?

$299 for macro usb one . It's in transit. I can return it for a full refund after I use it too. Going to do some stress testing, oc, Undervolting and acoustic tweaking. Currently I have it at 450 watts power limit (pny oc 5090) at 3ghz and barely notice any significant performance loss with acceptable noise levels but want to see if it can be improved.

 

[ezst036]

Looks like Nvidia's cost cutting strategy is now a smashing success.

They have successfully outsourced power protection from their video cards to the PSU makers. It only took some melted GPUs, but......

What's a couple thousand melted high-end GPUs between friends in order to achieve an objective?

Now when the RTX 6xxx gets released, Nvidia can easily say that only the Seasonic Prime version 2026+ are supported for their GPUs and if not your warranty is void. <--- this is a much more likely scenario than Nvidia putting the shims back that it had on the 3090 cards.

Nvidia will make the customer pay. Why would Nvidia make Nvidia pay?

Heck I could even see Nvidia saying that only these Seasonic PSUs (once released) are considered warranty for the 5xxx series cards in the near future.

 

[beyondlogic]

Seasonic is equipping its next-generation Prime power supplies with sensors, firmware, and external monitoring to prevent overheating and melting of 12VHPWR and 12V-2×6 connectors.

Seasonic’s next-generation Prime PSUs to will try to stop connectors from melting : Read more

this is a band aid fix the real problem is nvidia not admiting these cards are trash out the gate

 

[YSCCC]

$299 for macro usb one . It's in transit. I can return it for a full refund after I use it too. Going to do some stress testing, oc, Undervolting and accusing tweaking. Currently I have it at 450 watts power limit (pny oc 5090) at 3ghz and barely notice any significant performance loss with acceptable noise levels but want to see if it can be improved.

My solution was just buy a 9070XT base model, slightly undervolt it and call it a day. I figured that spending $700 (MSRP in my local currency) for a base model is better than $1500 for a 5080, which I can upgrade next gen with the spare money if needed. And no worry of the stupid connector

 

[aberkae]

My solution was just buy a 9070XT base model, slightly undervolt it and call it a day. I figured that spending $700 (MSRP in my local currency) for a base model is better than $1500 for a 5080, which I can upgrade next gen with the spare money if needed. And no worry of the stupid connector

True but I was going froma 4090suprim liquid for 2.5 years that would have beena downgrade. In Oblivion remastered although the experience at 4k dlaa/frame gen ( lumen hardware) was playable now I am getting a much smoother experience. 9070xt is a great card but I wished AMD had a 5090 competitor which I would have paid for. Although I am not sure what connector they would have used. Maybe AMD is waiting for the beta testing phase to end to go all in.

 

[_D_D]

My personal head canon is that Nvidia didn't want to continue with the 8-pin connectors because it would require 4 plus on a card per it's spec (safety factor) to achieve 500+ watts of power.

4 decent 8-pin connectors should have enough safety factor for 600W IMO but maybe Nvidia are not stopping at 600W? What niggles me is that some people force these connectors side to side instead of straight push/pull and maybe causing contact deformation and degradation.

Looks like just the top 2 connectors have sense resistors?

 

[wwenze1]

I support self-correcting or "dumb" solutions. A single resistor or active load (transistor) of the appropriate size will automatically drop output voltage as current on that wire increases. If two wires have different output voltages, more current flows from the one with higher voltage until equilibrium is re-obtained due to output impedances.

 

[YSCCC]

4 decent 8-pin connectors should have enough safety factor for 600W IMO but maybe Nvidia are not stopping at 600W? What niggles me is that some people force these connectors side to side instead of straight push/pull and maybe causing contact deformation and degradation.

Looks like just the top 2 connectors have sense resistors?

This connector have been discussed to death already, yes the PSU side looks like they are is adding the shuts to do the power blanacing stuff, but the issue is that on the card side the load balancing is removed, so it either force shut down your whole PC instead of throttling your GPU if something goes wrong.

For the "some people force these connectors side to side instead of straight push/pull" thing, it is actually not a choice, since RTX 4000 series the cards are so tall they leave almost no room for cable to go straight, and cable management have beend pulling the PCI-E 8 pin connector for literally a decade sideways to tidy it up and it's not a problem. you can't get out a new standard which requires re-design of cases altogether don't you?

 

[greenreaper]

I disagree with the premise that a PSU should not step in to prevent a dangerous overcurrent situation. Data loss outside of unsaved data is unlikely - any modern file system should be able to handle a power outage, including NTFS. The thing to be worried about is your computer burning up - perhaps taking other things with it.

 

[sama02]

If only GPU makers had another non melting connector that have been used for decades and worked perfectly fine even for power load WAAAY ABOVE 1kw right !? Also it would be nice if that new and absolutely not unnecessary redesign could be monitored by the GPU maker who forced his bad opinion onto consumers, right!?

 

[sama02]

I disagree with the premise that a PSU should not step in to prevent a dangerous overcurrent situation. Data loss outside of unsaved data is unlikely - any modern file system should be able to handle a power outage, including NTFS. The thing to be worried about is your computer burning up - perhaps taking other things with it.

The issue is why is it suddenly necessary. It's all due to the 12 pin hpwr connector, so MAYBE, and I say MAYBE Nvidia should consider taking their loss and run back tail between legs toward mommy MOLEX and apologize for being a bad kid. Because really that redesign sucks and is 100% unnecessary.

 

[sama02]

This connector have been discussed to death already, yes the PSU side looks like they are is adding the shuts to do the power blanacing stuff, but the issue is that on the card side the load balancing is removed, so it either force shut down your whole PC instead of throttling your GPU if something goes wrong.

For the "some people force these connectors side to side instead of straight push/pull" thing, it is actually not a choice, since RTX 4000 series the cards are so tall they leave almost no room for cable to go straight, and cable management have beend pulling the PCI-E 8 pin connector for literally a decade sideways to tidy it up and it's not a problem. you can't get out a new standard which requires re-design of cases altogether don't you?

Then maybe and I know it sounds crazy but since NVIDIA caused both issues... They should back down and apologize like real adults should do. Also, the 5090's PCB isn't that big, so changing it's shape to be longer instead of taller is always an option and it has for benefit among others that the GPU would for once not even get longer since the cooler already reach that way past the PCB as it currently stands.

 

[sama02]

The 12VHPWR spec was adopted by PCI-SIG. Many companies are a part of it. Including Nvidia, Intel, AMD, Apple, even video card AIB partners like Asus, MSI, and Gigabyte. We don't know which companies worked on or voted in favor of 12VHPWR. Nvidia was the one to implement it into their video cards AND mandate it for the high end products. AMD capatalized on it in their marketing with their not using it. If you watch the video linked below from Actual Hardcore Overclocking, you'll see another reason why people blame Nvidia for the hardware failures versus AIB partners. They intentionally did not have the per wire protections that could hamper the issues. Companies like EVGA left the business due to disagreements with Nvidia. We can only speculate as to who truly is at fault with the 12VHPWR connector.

People have commented to me that Intel was the head developer of the spec, but again who knows. My personal head canon is that Nvidia didn't want to continue with the 8-pin connectors because it would require 4 plus on a card per it's spec (safety factor) to achieve 500+ watts of power.

View: https://www.youtube.com/watch?v=kb5YzMoVQyw

My personal head cannon is that Nvidia is wrong and they shouldn't push their mistakes onto consumers and power supply manufacturers.

 

[sama02]

True but I was going froma 4090suprim liquid for 2.5 years that would have beena downgrade. In Oblivion remastered although the experience at 4k dlaa/frame gen ( lumen hardware) was playable now I am getting a much smoother experience. 9070xt is a great card but I wished AMD had a 5090 competitor which I would have paid for. Although I am not sure what connector they would have used. Maybe AMD is waiting for the beta testing phase to end to go all in.

AMD do not currently seem to give a damn about 12 pin hpwr for the simple reason that this connector suck, solve absolutely 0 problem causes new ones, and is totally unnecessary.

Also they do not have a competitor for the 5090 because the market for "I need to compensate" cards is simply too small. (Per their own testimony)

 

[sama02]

Looks like Nvidia's cost cutting strategy is now a smashing success.

They have successfully outsourced power protection from their video cards to the PSU makers. It only took some melted GPUs, but......

What's a couple thousand melted high-end GPUs between friends in order to achieve an objective?

Now when the RTX 6xxx gets released, Nvidia can easily say that only the Seasonic Prime version 2026+ are supported for their GPUs and if not your warranty is void. <--- this is a much more likely scenario than Nvidia putting the shims back that it had on the 3090 cards.

Nvidia will make the customer pay. Why would Nvidia make Nvidia pay?

Heck I could even see Nvidia saying that only these Seasonic PSUs (once released) are considered warranty for the 5xxx series cards in the near future.

In fact I can't see a future where they don't. The only real choice we consumers have is to swallow that white liquid or buy another brand. Personally I don't like swallowing and pineapple juice doesn't help so... I no longer buy Nvidia's products, end of story.