*Official Benchmark Thread* - Post it here or it didn't happen :D

[Mr. Fox]

On 11/14/2022 at 8:57 AM, tps3443 said:

I’m using Alphacool 17Mkw thermalpaste on the IHS.

I got a tube of this with the Kryos waterblock and it did not work well at all for me. It seems way too thick. My load temps were about 10°C higher than they are with KPX or MX-4. With KPX and MX-4 I end up with an almost invisible layer of thermal paste on the IHS after mounting the block. After remounting twice with the Alphacool paste it remained very thick in comparison. I think it might work really well for a laptop and offer resistance to pump-out because of that.

 

On the first attempt, I tried using the mounting pressure to spread it from a line down the middle. Big mistake. Did not compress well enough. I spread it evenly across the IHS with a spatula after remounting and that helped, but the temps were still north of 102°C in Cinebench R23 (easily a 10°C temperature increase).

[Papusan]

13 hours ago, Papusan said:

Btw. Some wonder why... Hmm...why this megathread deleted from main page? Maybe Nvidia reps have asked the mods to hide it?😀 And yep, the mods for the nvidia reddit forum won't tell the truth. Hide it in between a loads of other more or less useless threads, is a smart move.

 

I remember when the smoking posts pop up first time on reddit. The lead mods closed them one by one in a damn fast speed. Said they added links to them in the mega thread. So No need to post in them. This way... None could post anything in any of them. Nice done.

 

12 hours ago, johnksss said:

No I get it, the mods over there are full on on the job to detract from the current situation of bad mouthing nvidia. Seen that in full effect, but what wasn't in full effect was a fully seated cable burning which has magically appeared after GN did forensicate testing. could be coincident could be not. Who knows at this point. And I'm sorry, just because people post stuff on the internet I'm just not jumping on board the Jim Jones juice train just because they say it's so.

Got the answer here...

 

 

Yeees.. Why pin the Nvidia disaster thread on top of nvidia reddit main page? Perhaps due bad publicity for Nvidia? Or is there another sensible reason for unpinning the mega thread? 

 

[johnksss]

5 minutes ago, Papusan said:

 

Got the answer here...

 

 

Yeees.. Why pin the Nvidia disaster thread on top of nvidia reddit main page? Perhaps due bad publicity for Nvidia? Or is there another sensible reason for unpinning the mega thread? 

 

I kind of was only talking about the "cable plugged in burning" part and the other stuff we already knew the answer since the beginning.😂

[Mr. Fox]

Just released... minutes ago.

 

[johnksss]

43 minutes ago, Mr. Fox said:

Just released... minutes ago.

 

And I still bet 0 people come forward with "it was my fault"😂

Nvidia also shut down the warranty, 3rd party cables, user error rumor mills down. So people can stop with the conspiracy theories now. And can start by taking the correct pictures of the side of the adapter instead of beating a dead horse with the tips being burnt pictures.

 

I also wonder if GN heard about the 1 guy with the "fully" seated cable issue? I went to see if that guy updated his pictures but nope, not as of yet. I wish they would have posted those owners blatantly lying about what happened and put them on blast, but that will never happen. But if I catch someone who submitted their cables saying it wasn't their fault is getting clowned! Next up...How to remove the crimp line. 🤣

[Etern4l]

1 hour ago, Mr. Fox said:

Just released... minutes ago.

 

 

You know it's bad when a 15 minute tutorial is needed to ensure a cable is properly plugged in.

[Mr. Fox]

I bet NVIDIA is wishing they would have left it alone. The push for the new 12VHPWR connection was another idiotic change for the sake of change, purely for the sake of padding someone's, or some group's, wallet. There was, and is, no need for this stupid revision. Another example of "if it ain't broke, don't fix it" being the best advice. In a bit of a sadistic way, I find it gratifying to see self-serving schemes like this turning to crap. Unfortunately, they won't learn anything from it.

[johnksss]

I think it was a good idea, just badly executed. And should have made it more idiot proof. Plugging in a single cable is far better than plugging in 2 to 4 cables. 

Quick search reveals plenty of power cable issues before this came about.

https://www.google.com/search?q=burnt+up+pcie+cables&rlz=1C1ONGR_enUS990US991&sxsrf=ALiCzsbmB0_u8QBJRb7ooANHDLykHsoILQ:1668804057082&source=lnms&tbm=isch&sa=X&ved=2ahUKEwjWmpbIy7j7AhWgLUQIHenxAzMQ_AUoAXoECAEQAw&biw=2037&bih=850&dpr=1.88

[Mr. Fox]

10 minutes ago, johnksss said:

I think it was a good idea, just badly executed. And should have made it more idiot proof. Plugging in a single cable is far better than plugging in 2 to 4 cables. 

Quick search reveals plenty of power cable issues before this came about.

https://www.google.com/search?q=burnt+up+pcie+cables&rlz=1C1ONGR_enUS990US991&sxsrf=ALiCzsbmB0_u8QBJRb7ooANHDLykHsoILQ:1668804057082&source=lnms&tbm=isch&sa=X&ved=2ahUKEwjWmpbIy7j7AhWgLUQIHenxAzMQ_AUoAXoECAEQAw&biw=2037&bih=850&dpr=1.88

Definitely could have been executed better. I think putting it in a really dumb location at the top center on the PCB and trying too hard to make it a little connector reflect poor planning. They could have done something beefier, like a shorter 12-pin version of the 24-pin motherboard cable. I think making the PCB 2/3 of the length of the GPU cooler created design issues for them that had a cascading effect that caused other issues. Having the cooler be 1/3rd longer than the actual GPU seems really limiting and  restrictive on engineering flexibility. Doing that accomplishes nothing useful for the end-user when the total package is massive. It would be different if the entire GPU were smaller. Then they could move the connector to the end of the PCB facing the opposite direction of the I/O panel. They could have also put it on the back of the PCB through the backplate with a 90° cable so everything would be neat and tidy.

 

[Etern4l]

21 minutes ago, Mr. Fox said:

Definitely could have been executed better. I think putting it in a really dumb location at the top center on the PCB and trying too hard to make it a little connector reflect poor planning. They could have done something beefier, like a shorter 12-pin version of the 24-pin motherboard cable. I think making the PCB 2/3 of the length of the GPU cooler created design issues for them that had a cascading effect that caused other issues. Having the cooler be 1/3rd longer than the actual GPU seems really limiting and  restrictive on engineering flexibility.

 

They knew the top-side location of the socket + crazy adapter was bad in the 3090 Ti already, hence the attempt to mitigate using the angled socket. Seems like with the Ada consumer cards, someone at Nvidia said "f.. it!" and just left it at the right angle. The workstation cards typically have the socket at the rear of the card. I guess more space that way in most cases, and no need for weird cable routing. Like you said though, the socket location options are limited by the shorter PCB relative to the cooler, to accommodate the rear fan. In theory board partners could have improved on this (unless they are forced to use Nvidia board designs verbatim), but I guess none of them did.

[Mr. Fox]

4 minutes ago, Etern4l said:

 

They knew the top-side location of the socket + crazy adapter was bad in the 3090 Ti already, hence the attempt to mitigate using the angled socket. Seems like with the Ada consumer cards, someone at Nvidia said "f.. it!" and just left it at the right angle. The workstation cards typically have the socket at the rear of the card. I guess more space that way in most cases, and no need for weird cable routing. Like you said though, the socket location options are limited by the shorter PCB relative to the cooler, to accommodate the rear fan. In theory board partners could have mitigated this (unless they are forced to use Nvidia board designs verbatim), but I guess none of them did.

Yeah, I think the "blow-through" fan thing is also a really stupid gimmick. I know it is subjective, but I think that also detracts from the aesthetic quality of the final product. I find the openings on the backplate to be very unattractive. I also do not like the opening for the four screws and X-bracket on the modern GPUs. I'd much prefer a clean backplate covering up that ugly mess. Overall, I think the back of the newer GPUs is very ugly-looking.

[Etern4l]

38 minutes ago, Mr. Fox said:

Yeah, I think the "blow-through" fan thing is also a really stupid gimmick. I know it is subjective, but I think that also detracts from the aesthetic quality of the final product. I find the openings on the backplate to be very unattractive. I also do not like the opening for the four screws and X-bracket on the modern GPUs. I'd much prefer a clean backplate covering up that ugly mess. Overall, I think the back of the newer GPUs is very ugly-looking.

 

So originally I thought this kind of cooling potentially makes sense in constrained space, and might enable side-by-side mounting of 2 3090/4090s 3 slots apart. Unfortunately I've seen a video claiming a 3090TI and a 4090 FEs seated 3 slots apart run very hot, so it's probably not as effective as the typical workstation/server blower designs which also have top-side and rear intakes. 

[Papusan]

7 hours ago, Mr. Fox said:

I bet NVIDIA is wishing they would have left it alone. The push for the new 12VHPWR connection was another idiotic change for the sake of change, purely for the sake of padding someone's, or some group's, wallet. There was, and is, no need for this stupid revision. Another example of "if it ain't broke, don't fix it" being the best advice. In a bit of a sadistic way, I find it gratifying to see self-serving schemes like this turning to crap. Unfortunately, they won't learn anything from it.

Even more stupid. AIC partners can put the 12VHPWR connection whatever way they want. Upside down or opposite. See here.... ASUS TUF 4090 (The latch will come on top of the card)  is the card who have seen most smoke and fire. Nvidia's 4090 FE the oppsite. And one thing for sure... I don't think there have been sold loads more of the TUF cards from Asus than the FE cards from Nvidia or other brands 4090 models.. 

 

 

Edit The TUF is overrepresented regarding fire and smoke. About 1/3 of all the affected cards is branded TUF only on reddit's mega fire thread. Is this just a coincidence or just bad luck. Pick whats fits. Or just add in what buyer groups will go for the custom Asus card. But more adult people prefer the Nvidia design. So maybe there is not a coincidence that you have seen less of the burned up 4090 FE cards? Food for thoughts.

 

Btw. I don't think CableMod and other custom cable mod vendors are that happy about the smoke and melting finds. If people can learn how to attach Nvidia's cable spaghetti adapter correctly, there will be less fire and less need for custom cables. CableMods vs nvidia = 0-1 🙂

[Mr. Fox]

I am glad that I have no interest in purchasing a GPU right now.

It feels nice that I can say, "Nah, I'm good... no thanks."

[Etern4l]

33 minutes ago, Mr. Fox said:

I am glad that I have no interest in purchasing a GPU right now.

It feels nice that I can say, "Nah, I'm good... no thanks."

 

Will be interesting to see if the market ever normalizes such that consumers can expect to actually get more value from next generation graphics cards. It's possible that it will take some technical breakthrough that will dethrone the current technologies and/or players.

 

For the moment though, Ampere and perhaps Radeons look like better value for most people in practical terms.

[tps3443]

7 hours ago, Mr. Fox said:

I got a tube of this with the Kryos waterblock and it did not work well at all for me. It seems way too thick. My load temps were about 10°C higher than they are with KPX or MX-4. With KPX and MX-4 I end up with an almost invisible layer of thermal paste on the IHS after mounting the block. After remounting twice with the Alphacool paste it remained very thick in comparison. I think it might work really well for a laptop and offer resistance to pump-out because of that.

 

On the first attempt, I tried using the mounting pressure to spread it from a line down the middle. Big mistake. Did not compress well enough. I spread it evenly across the IHS with a spatula after remounting and that helped, but the temps were still north of 102°C in Cinebench R23 (easily a 10°C temperature increase).

Their 16M/kw thermal paste is much softer than the 17M/kw. They are literally (2) different pastes. 
 

I noticed bad temps at first too though with the 17mk stuff, so terrible that I immediately switched back to that TF7 paste, then after delidding etc, I gave the Alphacool past another chance. And this is what happen. I tightened my waterblock in cross pattern to where it was really really snug on each corner and good to go. I ran my PC, checked my waterblock after a few minutes and it wasn’t that snug anymore lol. I was able to tighten a substantial amount more after it all warmed up. Then I saw the temps collapse on me. 

[tps3443]

15 hours ago, Mr. Fox said:

I am glad that I have no interest in purchasing a GPU right now.

It feels nice that I can say, "Nah, I'm good... no thanks."

 

Same. I’m good. If I were to buy something right now, it would be exchanging my DDR5 7200 for the DDR5 7600 “Since they pretty much cost the same” and the worst urge is wanting to buy a Asus Z790 Apex really bad lol. Even though I don’t need either. 

[Talon]

19 hours ago, johnksss said:

I think it was a good idea, just badly executed. And should have made it more idiot proof. Plugging in a single cable is far better than plugging in 2 to 4 cables. 

Quick search reveals plenty of power cable issues before this came about.

https://www.google.com/search?q=burnt+up+pcie+cables&rlz=1C1ONGR_enUS990US991&sxsrf=ALiCzsbmB0_u8QBJRb7ooANHDLykHsoILQ:1668804057082&source=lnms&tbm=isch&sa=X&ved=2ahUKEwjWmpbIy7j7AhWgLUQIHenxAzMQ_AUoAXoECAEQAw&biw=2037&bih=850&dpr=1.88

 

100% agree.

 

After using this cable on a couple 40 series cards, I've had zero issues with the cable other than being  semi difficult to plug in and unplug. I mentioned that a couple weeks ago when I got my cablemod cable I noticed how difficult it could be to just seat the cable fully. And look and what came out, it actually is people failing to ensure a properly seated cable. In my opinion, people building computers and no less responsibility to ensure it's done properly than say someone installing a new lighting fixture. If you manage to do that incorrectly you could just as easily burn down your home. Should those people be allowed to turn around and sue the company that sold the lighting fixture? But Nvidia and the cable spec should have made this easier IMO. 

[tps3443]

@Talon

 

I think I remember you saying a while back you ran one of those carbonaut thermal pads, or something similar. Those thin dry thermal pads that are reusable and used instead of thermal paste on the CPU. I bought one to try. How do they work? 

[Papusan]

7 minutes ago, tps3443 said:

@Talon

 

I think I remember you saying a while back you ran one of those carbonaut thermal pads, or something similar. Those thin dry thermal pads that are reusable and used instead of thermal paste on the CPU. I bought one to try. How do they work? 

Waste of money if you want best coolig performance. But here you go. See the whole video if this pads is of interest

 

[tps3443]

8 minutes ago, Papusan said:

Waste of money if you want best coolig performance. But here you go. See the whole video if this pads is of interest

 

I found some for $1 dollar lol. So I’m curious to try it out and see how bad it really is. So from what I have researched mushing them down as thin as possible is what creates a really good contact. So I’m gonna try and flatten them out first. And then mount them in place for further flattening. I have seen them perform within a few degrees of even Kryonaut. I’m just curious is all. 

[Etern4l]

Looks like this is just for quick CPU tests/heatsink tests to avoid messing around with the paste.

[electrosoft]

17 hours ago, Mr. Fox said:

I am glad that I have no interest in purchasing a GPU right now.

It feels nice that I can say, "Nah, I'm good... no thanks."

 

2 hours ago, tps3443 said:

 

Same. I’m good. If I were to buy something right now, it would be exchanging my DDR5 7200 for the DDR5 7600 “Since they pretty much cost the same” and the worst urge is wanting to buy a Asus Z790 Apex really bad lol. Even though I don’t need either. 

 

Ditto I'm losing steam real quick. I've lost my zest for even a 13900k and I sold my BNIB Classified Z690. I listed my KPE 3090ti at a, "Price i'm willing to sell" point (~$1950) but if it doesn't sell *shoulder shrug*. I have completely lost interest in the 4090 and the only thing that has me curious is a 7900XTX atm on the higher end but if I end up riding out my 3090ti I'm am more than fine with that. I have contemplated selling off my Strix D4 + mem and swap in a KP Z690 bundle since I still have this set of 2x16GB A die heatsinked DDR5 modules on the shelf to play with.

 

Right now, I have a used  12400 coming in and I'm picking through some budget (sub $100) boards along with all the other components on hand  to build out another PC I am going to equip with my A380 to set up for my daughter. That's the next plan of attack for myself. I have both a mATX/itx and full ATX case on hand.

 

Still haven't had a chance to test the A770 @Mr. Fox sorry.

 

 

 

 

[Mr. Fox]

11 minutes ago, Etern4l said:

Looks like this is just for quick CPU tests/heatsink tests to avoid messing around with the paste.

Carbonaut won't even work for that. It is too fragile. You have to use a generic (Panasonic) sheet or the IC Diamond option if you want it to hold up to remounts.

 

I have not tried Carbonaut (nor do I plan to waste time doing so) but I tried the Panasonic sheets from Digi-Key and IC Diamond had me test the product they offer before it went to market and both were absolutely horrible. Worst thermals I had ever seen before... to the point of making my overclocked system unusable.

[Talon]

58 minutes ago, tps3443 said:

@Talon

 

I think I remember you saying a while back you ran one of those carbonaut thermal pads, or something similar. Those thin dry thermal pads that are reusable and used instead of thermal paste on the CPU. I bought one to try. How do they work? 

 

They work great for quick CPU test, for example just making sure the system is up and running, checking an SP score, etc. No mess, not wasting tons of cash on expensive paste. You could easily use them 24/7 if not running high clocks/voltage and benching. Like if you were just to build a gaming system, you 100% use the thermal pad and not have overheating issues. But for best temps and highest perf, of course good paste is the way to go.