When you buy an expensive piece of hardware, you want to know how to detect that it is here. And this is exactly what the new GPU-Z program fromTechPowerUpdoes. Furthermore, it also detects the notorious 16-pin 12VHPWR adapter and how it behaves, which might be quite handy, given the situation around Nvidia’s adapter that is supplied with some of the best graphics cards.
The new GPU-Z version 2.51.0 can correctly observe availability of Nvidia’s GeForceRTX 4080andRTX 4090in the system and, what is more important for the lucky owners of the best graphics cards for gaming, the sensor behind the 16-pin 12VHPWR connector. Now, the software is not going to tell you whether you plugged the adapter firmly and correctly, but what it is supposed to reveal is power distribution across different phases of your voltage regulating module.
This may well be quite handy for enthusiasts and overclockers, which is exactly the audience of the GPU-Z application.
The version history of the GPU-Z v2.51.0 is as follows:
- Added full support for NVIDIA GeForce RTX 4080
- Added BIOS save/upload support for NVIDIA GeForce RTX 4080 and 4090
- Fixed wrong transistor count on GeForce RTX 4090
- Added support for monitoring 16-pin power input
- Fixed missing memory temperature on GeForce 40 Series
- Fixed crash in Glenfly Advanced panel
- DLSS Scan in Advanced Panel no longer starts automatically and lets you select the drives to scan first
- When the "Stop" option is selected in DLSS Scan, properly indicate that the search has stopped
- The list of Vulkan extensions is now one entry per line
- The list of OpenCL extensions is now one entry per line and sorted alphabetically
- Fixed negative Gather Offsets range displayed as positive integer in Vulkan info
- Added support for NVIDIA GeForce RTX 3060 Ti GDDR6X, RTX 3060 (GA104-B), MX750 A, RTX A500 Laptop, RTX A4500 Embedded, Tesla T10, Quadro K5100M (GK104-B)
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Anton Shilov is a Freelance News Writer at Tom’s Hardware US. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.
More about gpus
Latest
12 CommentsComment from the forums
Nvidia's hardware is unable to monitor power delivery through the connector? How come the card, with such sophisticated power delivery system, cannot monitor itself.
With the connector heated up to the point of melting plastic there should've been some variance in current or voltage flow. No warning signals?Reply
InvalidError tommo1982 said:
Nvidia's hardware is unable to monitor power delivery through the connector? How come the card, with such sophisticated power delivery system, cannot monitor itself.
With the connector heated up to the point of melting plastic there should've been some variance in current or voltage flow. No warning signals?All pins are shorted together inside the connector and all pins are shorted together on the PCB, so there is no way to do anything on a per-pin basis without a PCB and connector re-design. At best, there could be an SMD thermistor near the he HPWR connector to measure 12V/GND plane temperature in the area.
Reply
nichrome InvalidError said:
All pins are shorted together inside the connector and all pins are shorted together on the PCB, so there is no way to do anything on a per-pin basis without a PCB and connector re-design. At best, there could be an SMD thermistor near the he HPWR connector to measure 12V/GND plane temperature in the area.Reply
nichrome Now Elwood Blues isn't the only one who can make white toast,dry.
Reply
InvalidError The_Git said:
What nonsense are you talking about all pins shorted together? Do you mean ground plane connection for the earth, because you can possibly make a claim like that for the power rails. Where do you get this drivel from.I mean the connector itself has a foil bus bar connecting all 12V pins together and all GND pins together while the FE cards at the very least have all 12V pins from the connector directly soldered to the 12V plane, same for the GND pins, which makes it impossible to monitor things on a per-wire/per-pin basis.
Reply
Wrss The_Git said:
What nonsense are you talking about all pins shorted together? Do you mean ground plane connection for the earth, because you can possibly make a claim like that for the power rails. Where do you get this drivel from.
Adding a thermistor would be the same as adding a SMD monitoring chip, that is the nature of PCB design and why they leave vias holes open, for testing and monitoring. Testing chips are easily added for simple one shot functions, you'd only be correct if you tried to add a boundary test system.Soon as you mentioned drivel I was reminded that I was reading it...
Adapter teardowns show the 6 current-carrying pins that are 12v are simply shorted with the 12 PCIe plug 12v pins. The other 6 pins that represent ground are similarly shorted. There are 4 additional sense/sideband pins that communicate to the graphics card what power it can draw, if any. These are not data pins; they do not communicate temperature. Unscrupulous adapter manufacturers leave those sense pins in an always-full-power state, while a proper adapter has a small IC that tells the GPU the power budget based on how many PCIe plugs are connected. If there's going to be a thermistor in there, it would be to decide when to tell the GPU to cut off all power.
Reply
InvalidError Wrss said:
If there's going to be a thermistor in there, it would be to decide when to tell the GPU to cut off all power.The place it would make the most sense to put a thermistor in would be right next to the row of 12V pins from the HPWR connector on the GPU's PCB since any heat from the pins would get conducted into the PCB's 12V plane running next and under the thermistor. If a pin dissipates enough power to reach 170+C and melt plastic, there should be enough heat conducted down the pin into the PCB to detect an unexpectedly high connector temperature there. Of course, that would require a PCB designed with a thermistor there.
Reply
TJ Hooker The_Git said:
Shorting power leads in this ways is simply bad design, and shows they are just paralleling power over multiple lines that are rated less than the 1 or 2 molex leads.... Great job intelThis is how virtually all internal PC power connectors work though. 4/8 pin CPU power connector, 6/8 PCIe power connector, they all have 2-4 12V parallel conductors that are usually shorted together on the motherboard/graphics PCB. The fact that the 12VHPWR adapter additionally shorts them together in the connector itself doesn't seem particularly significant.
And I don't think the spec explicitly calls for the conductors to be shorted in the connector. That's just how the Nvidia adapter was designed. Although, again, I don't think it matters much either way.
Reply
AkroZ Maybe he is speaking of well known Molex AMP four-pin Mate-n-Lok made in 1963.
https://upload.wikimedia.org/wikipedia/commons/3/31/Molex_female_connector.jpgImage and data from Wikipedia.�
Pin #ColorFunction1
Yellow+12 V2
BlackGround3
BlackGround4
Red+5 V
Funnily you can read that they have the same issue (arcing), this new connector just make it worse.Despite its widespread adoption, the connector does have problems. It is difficult to remove because it is held in place by friction instead of a latch, and some poorly constructed connectors may have one or more pins detach from the connector during mating or de-mating. There is also a tendency for the loosely inserted pins on the male connector to skew out of alignment. The female sockets can spread, making the connection imperfect and subject to arcing. Standard practice is to check for any sign of blackening or browning on the white plastic shell, which would indicate the need to replace the arcing connector. In extreme cases the whole connector can melt due to the heat from arcing.
Source: https://en.wikipedia.org/wiki/Molex_connector#Disk_drive
Reply
kyzarvs InvalidError said:
All pins are shorted together inside the connector and all pins are shorted together on the PCB, so there is no way to do anything on a per-pin basis without a PCB and connector re-design. At best, there could be an SMD thermistor near the he HPWR connector to measure 12V/GND plane temperature in the area.Would measuring resistance work? Surely if even one pin is shorting, the resistance will increase - anything over a perscibed level should alert & lower the max power through the connector?
Reply
