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

[Mr. Fox]

Follow up after further investigation... It was user error at first, but overall really no improvement. That is very puzzling, but maybe the cold plate didn't have enough effort in the design. It's OK, but so far kind of a disappointment. Not sure if I am going to keep it or not. Will play with it some more before I decide. Still kind of shocking that the temps are not better since so much less material is present between the water and the die.

 

 

[Papusan]

3 hours ago, Mr. Fox said:

Well, I installed the Supercool direct die block and my temps are a little bit higher than they were with the copper IHS. The installation was pretty straightforward. The CPU was already delidded, so there wasn't much effort involved. To rule out the possibility of user error on my part, I am going to remove it and check to confirm the liquid metal application was adequate and test once more before I decide if I should go back to using an IHS. This would be the first time ever in my experience that direct die cooling did not result in an impressive improvement. I am impressed with the Supercool 10th Gen direct die block, so I am hoping this is a result of user error and not a flaw in 12th Gen product design.

 

Edit: I had previously done the ILM washer mod so I will also remove them to rule out the possibility that is a factor.

Have you tested the contact point/pressure between die and the cold plate with only a layer of LM on the die? Maybe a pressure/contact problem. 

[Mr. Fox]

2 hours ago, Papusan said:

Have you tested the contact point/pressure between die and the cold plate with only a layer of LM on the die? Maybe a pressure/contact problem. 

I suspect it is either a contact issue or something with the fins on the water side of the cold plate isn't good enough. That is what I meant by "didn't have enough effort in the design" in my previous post.

 

I showed this in the video. I might need to lap the die side of the cold plate to increase the contact pressure. I did not have adequate coverage, but you can see where the edge of the die contacted the bare cold plate. Had the fit been well it would have squished the liquid metal across the area that is bare metal. It is also only in the lower left corner in the photo. It should have been visible the full length of the die. Adding more liquid metal helped, but there are a couple of cores getting a lot hotter than the rest with the bare die block. Core temps were very even and close to the same on all cores/threads with the copper IHS.

[Papusan]

18 hours ago, Mr. Fox said:

I suspect it is either a contact issue or something with the fins on the water side of the cold plate isn't good enough. That is what I meant by "didn't have enough effort in the design" in my previous post.

 

I showed this in the video. I might need to lap the die side of the cold plate to increase the contact pressure. I did not have adequate coverage, but you can see where the edge of the die contacted the bare cold plate. Had the fit been well it would have squished the liquid metal across the area that is bare metal. It is also only in the lower left corner in the photo. It should have been visible the full length of the die. Adding more liquid metal helped, but there are a couple of cores getting a lot hotter than the rest with the bare die block. Core temps were very even and close to the same on all cores/threads with the copper IHS.

I checked that Direct Die kit from Supercomputer. From what I can see from the pict the fins looks quite ok. So most likely a contact issue. But not so sure you'll get equal good results as you got with your previous kit for the older Cpu with an perfect coldplate.  It need radical better results to match that.  Have you seen others results for the kit for 12th gen? 

 

------------------------------------------------------------------------------------

 

The bot is down as usual. Second time within 2 days. So I couldn't post the new scores from my old stuff. But I'll grab an 3rd and 4th place in 3DM Cloud Gate and 3DM11 Entry when I get it posted🙂

 

 

 

https://www.3dmark.com/cg/4892136 - https://hwbot.org/submission/4996606_papusan_3dmark___cloud_gate_geforce_gtx_1070_77715_marks

 

https://www.3dmark.com/3dm11/14977975 - https://hwbot.org/submission/4996602_papusan_3dmark11___entry_geforce_gtx_1070_42748_marks

 

And will get an 6th place in 3DM 11 Performance. But got an good Physics score https://www.3dmark.com/3dm11/14978010 - https://hwbot.org/submission/4996599_papusan_3dmark11___performance_geforce_gtx_1070_28615_marks

 

The others on the bot top ranking use TESS and LOD tweaks (Not valid on 3DM). Hence I'm  alone on the top on Futuremark leaderbord for 1070. Better than nothing, LOOL

 

Not bad with an AIO.... (Splave benched with LN2 and 7GHz and only gained an 7.6% lead)🙂 https://www.3dmark.com/hall-of-fame-2/3dmark+11+physics+score/version+1.0.132

 

[Mr. Fox]

4 minutes ago, Papusan said:

I checked that Direct Die kit from Supercomputer. From what I can see from the pict the fins looks quite ok. So most likely a contact issue. But not so sure you'll get equal good results as you got with your previous kit for the older Cpu with an perfect coldplate.  It need radical better results to match that.  Have you seen others results for the kit for 12th gen? 

Maybe next weekend I will take it apart again and lap the bottom of the cold plate to increase pressure on the die. The recessed area on the CPU side of the cold plate is probably deeper than it should be.

[Mr. Fox]

On 5/12/2022 at 7:23 PM, Mr. Fox said:

- The 6950XT is a joke from the NVIDIA playbook that smells like nothing more than the equivalent of a 3090 Ti scam from the red side of the aisle

LOL... I just watched a video where Roman basically said the same thing. AMD is pulling an NVIDIA.

[Papusan]

24 minutes ago, Mr. Fox said:

LOL... I just watched a video where Roman basically said the same thing. AMD is pulling an NVIDIA.

They added a clock speed cap for the memory. Darn disgusting.

 

And they can make better cooling if they want. This is Jokebook design decisions behaviour. Aka make it half-baked then improve a bit on the refresh. Yep.....

[Mr. Fox]

17 hours ago, Papusan said:

They added a clock speed cap for the memory. Darn disgusting.

 

And they can make better cooling if they want. This is Jokebook design decisions behaviour. Aka make it half-baked then improve a bit on the refresh. 

Hey kids! Here's our bestest GPU in the whole wide world. Buy this one.  But, please please hurry. Prices might not stay this high forever and we need stupid people to act quickly.

...after a little while... 

Spoiler

 

Ha ha, faked you out. Guess what, kids?! It's not really our bestestest GPU. It's this one. It has the cooling and clock speeds that you deserved before when you let us screw you the first time, but you didn't get it because we needed you to come back and let us milk you for more. But, that's OK. It's cheaper now. But, hurry. Prices might not stay this lower forever and we need stupid people that opened their wallet last time to do it again before the smart people discover it's the same product made the right way on the second time around.

[johnksss]

 

[johnksss]

Moved up a tiny little bit.☺️

[Papusan]

38 minutes ago, johnksss said:

Moved up a tiny little bit.☺️

Yep, there is still headroom for improvements. At least if we continue bench with Win 7🙂 with Win 10/11 it will be a huge failure. But it will be hard grabbing 1th place without better cooling, HeHe. And as always, nice scores bro John. 

[johnksss]

7 hours ago, johnksss said:

Moved up a tiny little bit.☺️

 

7 hours ago, Papusan said:

Yep, there is still headroom for improvements. At least if we continue bench with Win 7🙂 with Win 10/11 it will be a huge failure. But it will be hard grabbing 1th place without better cooling, HeHe. And as always, nice scores bro John. 

The first picture was to show you we had the same Physics score.... Again.🙂

 

I do have better cooling, SS and LN2 pot.🙂 but that's not how we would grab first place. The only reason that is in sight is because most of the hard core crowd has basically retired from the sport.....

 

Thanks brother. Your scores are always impressive as well.

 

[aldarxt]

10 hours ago, Mr. Fox said:

Hey kids! Here's our bestest GPU in the whole wide world. Buy this one.  But, please please hurry. Prices might not stay this high forever and we need stupid people to act quickly.

...after a little while... 

  Hide contents

 

Ha ha, faked you out. Guess what, kids?! It's not really our bestestest GPU. It's this one. It has the cooling and clock speeds that you deserved before when you let us screw you the first time, but you didn't get it because we needed you to come back and let us milk you for more. But, that's OK. It's cheaper now. But, hurry. Prices might not stay this lower forever and we need stupid people that opened their wallet last time to do it again before the smart people discover it's the same product made the right the second time around.

Hey I remember that song, We used to sing it in the U.S.M.C. only it was spelled different  F-U-C - K-E-D  A-G-A-I-N  it even had a chorus

[Mr. Fox]

1 hour ago, johnksss said:

 

The first picture was to show you we had the same Physics score.... Again.🙂

 

I do have better cooling, SS and LN2 pot.🙂 but that's not how we would grab first place. The only reason that is in sight is because most of the hard core crowd has basically retired from the sport.....

 

Thanks brother. Your scores are always impressive as well.

 

I think I am going to probably be joining them. The hardware manufacturers and Redmond Retards have effed things up enough that it's not as fun as it used to be, and I'm getting sick of having to deal with their dumbass bullcrap. That's on top of the fact that their trash is overpriced to the point that it offers no value.

[johnksss]

25 minutes ago, Mr. Fox said:

I think I am going to probably be joining them. The hardware manufacturers and Redmond Retards have effed things up enough that it's not as fun as it used to be, and I'm getting sick of having to deal with their dumbass bullcrap. That's on top of the fact that their trash is overpriced to the point that it offers no value.

This sport was never about value unfortunately. And there has never really been value in buying high end parts or being first to buy them either. Ergo most people buy cheap and over clock to run on par with stock higher end parts for far less money which would be the value goal for most.

 

Right now I do it more so at my leisure than in serious competition. This sport is far to expensive with zero (Self and world recognition fall under a different category) return on ones money. There was never a good return on ones money from day one.....

 

 

[Reciever]

In other news, I just purchased 5x Lamptron CFP30 since they were 7 USD each on that TitanRig website. They handle 120w each and would do well for controlling the fan speed of for a few of the GPU's as I am replacing the the GPU fans with 3x 92mmX38mm 106CFM fans. Had to rewire them but thankfully due to the connector and terminated end it was relatively simple task to complete.

 

Also finished rinsing off my heatsinks in the shower 🙂

[aldarxt]

3 hours ago, Mr. Fox said:

I think I am going to probably be joining them. The hardware manufacturers and Redmond Retards have effed things up enough that it's not as fun as it used to be, and I'm getting sick of having to deal with their dumbass bullcrap. That's on top of the fact that their trash is overpriced to the point that it offers no value.

Here is a link to exactly the thorn in your side,  I go to this site just to check on what windows update is doing to the PCs in the field and if something catches my eye but this https://www.askwoody.com/2022/ewaste-or-usable-week-2/

reminded me of what you have been saying for well close to a decade!

[electrosoft]

I didn't realize the shroud on the KPE 3090ti was going to be solid metal for the first time. This should look sleek and ergonomically pleasing...

 

[Papusan]

53 minutes ago, electrosoft said:

I didn't realize the shroud on the KPE 3090ti was going to be solid metal for the first time. This should look sleek and ergonomically pleasing...

 

Yep, important to show up some improvements/news for the buyers so close into next gen cards…. Only 2 weeks and we are in June. I could understand it if this was done for a new gen cards. But not a refreshment on a soon 2 years old graphics chips/architecture. This reminds me too much of AMDs game bundle to sell out their stocks of cards to stop the price drop.

 

https://www.notebookcheck.net/Nvidia-GeForce-RTX-4000-series-could-be-launched-earlier-than-initially-anticipated.619820.0.html

 

[Mr. Fox]

9 hours ago, johnksss said:

This sport was never about value unfortunately. And there has never really been value in buying high end parts or being first to buy them either. Ergo most people buy cheap and over clock to run on par with stock higher end parts for far less money which would be the value goal for most.

 

Right now I do it more so at my leisure than in serious competition. This sport is far to expensive with zero (Self and world recognition fall under a different category) return on ones money. There was never a good return on ones money from day one.....

 

 

I never expected anything worth having to be cheap. I didn't expect everything to be overpriced garbage though. Expensive stuff that is good is good. We expect to pay more for good stuff, because it's better. I just don't expect to pay a lot for trash. I don't like trash even when it's free, LOL. (Crippled is one definition of trash, but the list of what belongs in that bucket is massive.) And, I don't expect to pay excessive prices for everything regardless of whether or not it's any good. The good stuff is excessively overpriced and severely out of scope with what you receive in exchange for your money, and a lot of the "high end" stuff is crippled now. 

 

I farted around with the Supercool direct die block for a few hours today and decided it's more trouble than it is worth. I am pretty sure if I spend enough time getting the cold plate to fit right it would work great, like the one does on my 10900K, but I got tired of sucking the water out of it with a vacuum cleaner. After the third time I was like, "Fox, what the hell are you doing? And, why?" I can put up with inconvenient design when it works well and produces pleasing results, but I'm not seeing that. It should have just worked right out of the box. The temps were not horrible, but they should have been better than with an IHS and they were not. Not good enough to justify the hassle, that's for sure.

 

I did not try sanding down the CPU side of the cold plate because the tabs or "wings" on the sides where the ILM presses down are already extremely thin and if I remove some of the material they might end up being too thin and bending under pressure. Plus, when you buy something special, you expect it to work right. I am losing interest in modding things other people have made that are supposed to be awesome (which is why you buy it in the first place) in order to get the results the product was expected to produce when it was purchased. (See my last post above about "trash" LOL.)

 

I think I am going to order one of those ILM replacements and call it good since I am not benching much any more. If someone comes up with a bare die frame for 12th Gen that allows me to use a normal water block, I'll grab one if and when that happens.

 

I lapped the stock IHS and put it back together with KPX on the top side and I'm done for now. Considering it's not liquid metal on both sides of the IHS, the temps are actually very good. As usual (for me) the stock IHS works better than the aftermarket copper IHS on a desktop. The aftermarket IHSes have only been better than stock for me on a laptop.

 

7 hours ago, aldarxt said:

Here is a link to exactly the thorn in your side,  I go to this site just to check on what windows update is doing to the PCs in the field and if something catches my eye but this https://www.askwoody.com/2022/ewaste-or-usable-week-2/

reminded me of what you have been saying for well close to a decade!

Yup, and now more than ever, Macro$lop is going full Hitler on things. They already have WAY TOO MUCH control on what people can do with their "rented" computer tech and it's only going to get worse going forward.

[Papusan]

23 hours ago, johnksss said:

Moved up a tiny little bit.☺️

Yep, if you grind on what you have there is often some headroom for improvements. Sometimes you improve some of the sub tests and lose on others. Then back you go. Get everything better is what drives you forwards. This is more as looking on the same old film several times. But it has to be a film you enjoy🙂

 

Broke +39000 in 3DM 11 Physics but this isn't valid for 3Dmark. It’s what it is. 

 

This one is valid for Hall of Fame, but I didn't get the best total score...

 

Grabbed 3rd place in 3DM11 Entry for GTX 1070 with a "huge margin" of 4 points 🙂

https://hwbot.org/submission/4997025_papusan_3dmark11___entry_geforce_gtx_1070_42914_marks

 

https://hwbot.org/submission/4997028_papusan_3dmark11___performance_geforce_gtx_1070_28669_marks

[ryan]

 

 

my best yet 12265 or something...wonder how it stacks up

[Mr. Fox]

4 hours ago, ryan said:

 

 

my best yet 12265 or something...wonder how it stacks up

Shut down all of the programs that are open and kill all of the unneeded Windoze bloat Services, set Cinebench.exe to run in  realtime priority and your best yet score will be even higher. All of the junk that is running slows things down measurably.

[Papusan]

7 hours ago, ryan said:

 

 

my best yet 12265 or something...wonder how it stacks up

Use the bot or Notebookcheck info page to check where it stack up. You have still headroom for improvements. Can add this into bro Fox post.... Use an Macrium backup and save your current OS setup and then test out different Windows versions. All depends on how much time you are willing to put into it 🙂 

 

https://hwbot.org/hardware/processor/core_i7_11800h/ - https://www.notebookcheck.net/Intel-Core-i7-11800H-Processor-Benchmarks-and-Specs.525586.0.html

[Mr. Fox]

3 hours ago, Mr. Fox said:

Shut down all of the programs that are open and kill all of the unneeded Windoze bloat Services, set Cinebench.exe to run in  realtime priority and your best yet score will be even higher. All of the junk that is running slows things down measurably.

 

1 hour ago, Papusan said:

Use the bot or Notebookcheck info page to check where it stack up. You have still headroom for improvements. Can add this into bro Fox post.... Use an Macrium backup and save your current OS setup and then test out different Windows versions. All depends on how much time you are willing to put into it 🙂 

 

https://hwbot.org/hardware/processor/core_i7_11800h/ - https://www.notebookcheck.net/Intel-Core-i7-11800H-Processor-Benchmarks-and-Specs.525586.0.html

This will also be helpful. Apply the "Extreme" profile and manually add bloat Services to the list as you deem appropriate. The defaults for the Extreme profile will improve performance without any tweaking.

 

https://www.sordum.org/8637/easy-service-optimizer-v1-2/ 

Also use the Chris Titus Windows debloat script and run that, remove Cortana, OneDrive and any of the other worthless trash that you can live without. All of that worthless rubbish steals CPU clock cycles and makes your computer run slower.

 

https://christitus.com/debloat-windows-10-2020/

 

I highlighted what I run. I disable updates, but you might prefer to allow security updates only. (That doesn't matter much to me.)