Nvidia gave us their internal tools… What could go wrong?

[Senzelian]

8 minutes ago, X-System said:

I don't have problem with LINK software, it works perfect, stable and I got all informations. I bought my HX750i in January 2020.

 

On Corsair website, I see it's last version 4.9.9.3 for almost 2 years (01/28/2019). So I use this version (I downloaded it in 2020)

That does look a lot better than the old version I had.
But I just tried it and of course it doesn't even work anymore, cause it doesn't even detect my PSU. Maybe the Link dongle broke. Who knows.

 

Can't believe that they haven't integrated this into iCue yet.

[Egg-Roll]

23 hours ago, GabenJr said:

When Nvidia asked us to attend a presentation on something “not graphics related”, we had no idea how much we’d love what they were about to show us…

Can we trust them for AMD products however? It's one reason why I wouldn't be buying something from a manufacture, they might skew their data points for their own products and damage their competitions. This video hasn't proven them playing nice nor playing nasty.

[X-System]

50 minutes ago, Senzelian said:

That does look a lot better than the old version I had.
But I just tried it and of course it doesn't even work anymore, cause it doesn't even detect my PSU. Maybe the Link dongle broke. Who knows.

 

Can't believe that they haven't integrated this into iCue yet.

I had installed iCue first (recommanded by Corsair...) and I don't have all informations from PSU on iCue (that sucks!). I found LINK software but you can't use LINK software because iCue installed... If you want to use LINK software, you must uninstall iCue first... So I unistalled iCue and installed LINK

[Nystemy]

6 hours ago, Senzelian said:

Hopefully someone is working on it. I bet that these would sell like hot cakes for $29 each or $50 in a bundle.

Hitting the 30 $ price point would be semi easy for the latency tester.

But doing it for the power monitor is a bit harder.
Just the PCIe 16x connector is a couple of dollars.
The power connectors can't really be bought in a one off quantity. (And at the minimum order quantity of 5540 they still cost 2 dollars each. (for the 8 pin right angle board mount connector.) _{(Now, I haven't done an exhaustive search, since there is no such thing as "PCIe power connector", it's mostly just an Industry standard (the whole "electronics industry", not just PCs and computers.) connector that is used in the PC industry , this means that there is literally thousands of different versions of the connectors, some pins are square, others have cut corners, and you can get any combination you want. And few sites have filters for where the notch should be, nor have correct product pictures. Ie, one will need to dig through data sheets until one finds the corresponding part number and then search for that. Long story short, the connectors are expensive. (Actual manufacturers just goes to Amphenol or Molex directly instead of the resellers...))}

 

So if one puts 6 of these connectors on the power measuring board, then we are already up at 12 dollars, then there is the PCIe slot too (About 0.5 USD each @ 1000 quantity.) Then one needs the PCB, a decent set of current shunts, and an amplifier, not to mention a mico.

I would estimate that the BOM cost of a product like this is above 20 USD, not to mention assembly cost of it.
Also, since non of the shunts are "precision components" nor is our amp, then we will need to calibrate it... (so add 1-3 USD for that alone)
Then there is shipping, payment processing, and overhead to cover the unavoidable manufacturing defects and lost shipments.
Then there is sales tax too.
And inventory costs to just having the product on a shelf in one's own warehouse.
So the power meter might end up costing closer to 50 USD alone to just cover the costs involved. (If one wants to make profits, then it needs to be an even higher price.)

And I would be surprised if the actual market for these is even above 5-12 K units total. (When including people that just buy it for "oh, this is cool!".)

And for everyone looking at that going, "but if I make 10$ each, then that is 50-120 K$!" Well, yes, but then you have the time costs of CADing the board, hunting down the components, making the firmware for it, then making a user friendly and "nicely polished" user interface for it too. (Now, making software and CADing boards is rather easy, but its still work.)

In the end.
If I were Nvidia, I would sell it for around 75-100 USD, since the main customer base is reviewers. (But I also suspect that there wouldn't be enough interest to cover their development costs if they sold it for 100. So I wouldn't be surprised if they go for 150. But I mostly suspect that they made this board to give reviewers the ability to check their claims of improved power efficiency without other system components getting in the way of things. (Though, then they instead have the problem that they make the validation tool, so can we trust it? I would say, since it just measures power, there isn't much for it to go one to detect what card it is working with, so it should be mostly oblivious, if it isn't, then Steve over at GN will call them out on it rather quickly. (After all, a current clamp meter still is decently accurate, so they shouldn't be off by more than 2-5% with that.)))

Reviewers honestly do have an interest in getting accurate power monitoring, as to measure power efficiency of a product. (And the main thing this product offers is the ability to measure current through the PCIe slot itself.)

Overclockers on the other hand already tap into the VRMs as is, they are likely rather uninterested in a device like this, since they mostly already have one. (And if the card's readouts are off by 2-15%, then one just takes that into consideration when running the card.) Not to mention that serious overclockers back in the day chopped of half the card to add on their own VRMs.... (So making one's own power monitoring is not really that much harder.)

To the average Joe, then this is nothing but a curiosity.

I could go and make it, but I do not suspect there to be an actual interest in it to cover the initial investment costs. (I in short don't feel like spending 10 grand on connectors...)

[nick name]

23 hours ago, HTMLmoron said:

What an absolute gamer move by Nvidia. They must know you're going to be testing the 3000s as well as the AMD Navis. That's a ton of faith they have in their products. This is incredible marketing. 

I wouldn't be surprised if NVIDIA knows AMD's new GPUs pull a lot of power and NVIDIA wanted reviewers to be able to show that in their reviews.  

[WereCatf]

2 hours ago, Egg-Roll said:

Can we trust them for AMD products however? It's one reason why I wouldn't be buying something from a manufacture, they might skew their data points for their own products and damage their competitions. This video hasn't proven them playing nice nor playing nasty.

The devices don't have any way of knowing what device's power-draw they are measuring, mate. You can't magically know what the current flowing through a PCIe power-connector whether that current is going into an NVIDIA-card, AMD-card, to a massive resistor or what else.

[xnamkcor]

47 minutes ago, WereCatf said:

The devices don't have any way of knowing what device's power-draw they are measuring, mate. You can't magically know what the current flowing through a PCIe power-connector whether that current is going into an NVIDIA-card, AMD-card, to a massive resistor or what else.

It has a tap on the PCI-e connector. It's not impossible to identify a PCI-e device.

[Kroon]

4 hours ago, Nystemy said:

The power connectors can't really be bought in a one off quantity. (And at the minimum order quantity of 5540 they still cost 2 dollars each. (for the 8 pin right angle board mount connector.) 

 

I know you wrote a disclaimer but your prising is WAY off.  I repair electronics on my spare time and a lot of that is broken 8 pin connectors on GFX.  I buy 100 of 8-pin PCB Female Header Through Holes for about $20 for whole all of them.  Order a huge batch and you get hen even cheaper and that is from an Swedish supplier and is not knows to be cheap.

 

Tested to buy 5 for $1.5 from AliExpress recently, not something I would use to sell to someone but usable for home and testing.

[Nystemy]

21 minutes ago, Kroon said:

 

I know you wrote a disclaimer but your prising is WAY off.  I repair electronics on my spare time and a lot of that is broken 8 pin connectors on GFX.  I buy 100 of 8-pin PCB Female Header Through Holes for about $20 for whole all of them.  Order a huge batch and you get hen even cheaper and that is from an Swedish supplier and is not knows to be cheap.

 

Tested to buy 5 for $1.5 from AliExpress recently, not something I would use to sell to someone but usable for home and testing.

Well, Elfa is known for their frankly odd prices to be fair. And their rapidly dwindling assortment of components...

I have also seen the AliExpress and even Ebay offerings too, but the quality of these is "questionable", so not really ideal for an actual product, but yes, good enough for home testing and hobby use.

Though, the 6 pin connectors are far cheaper in comparison to what I have found, and can actually be found in 1 off quantity.
But I would just buy extension cables from a reputable brand and chop in the board in between to be fair.

[Kroon]

5 minutes ago, Nystemy said:

Well, Elfa is known for their frankly odd prices to be fair.

 

No kidding!  Problem that their quality can be questionable sometimes.

 

Just checked with the company we use to make our PCB and then I would pay $0.4 per 8-pin connector but then soldering are included and a minimum of 1000 must be ordered.