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

[tps3443]

52 minutes ago, Rage Set said:

@Mr. Fox

 

I'm sorry I've been missing in action. Did you get the warranty straight with the 2080 TI? Lots to read through. (If you ever need me, you have my number, just call or text me bro).

Long time no see. How are you? 

[Mr. Fox]

2 hours ago, tps3443 said:

I’m really happy with my 13900KF. I think

its a keeper for sure. I’m really glad I upgraded. Massive improvement over the 11900K. 
 

Looking at how these compare to 12th Gen it is just amazing! 
 

PS: DDR5 7200 will be here tomorrow Hynix A die.
 

@Mr. Fox

 

Delidding your 12900KS only builds that confidence for the 13900K delid. I think 6Ghz all core is very doable once delidded. Anyways, no matter how many times we’ve delidded, after that 11900K incident yeah I’d be nervous about Delidding too, I tore my 1st 11900K up pretty bad ripped of (16) small SMD’s. Confidence is put in check real quick after something like that with these small fragile components. But, I’m very glad it was successful on your 12900KS. Did you heat it up first? Or just pop the IHS cold?
 

Maybe you did too many good delids, and with the 11900K you’re like pshhhh walk in the park, I got this, delid # 398 coming right up, easy peasy lemon squeezy, I can do this with eyes closed, one hand tied behind back, “Crunch” oops. 😄

I delidded the first 12900K, the second 12900K and now the 12900KS. I heated the first one but not the next two. It did not seem to help anything other than making it more difficult to handle. Just a heads up, in all three cases the delid causes minor damage to the 12th gen IHS and will probably do the same to the 13th gen. The first one (that I heated) got the top surface slightly deformed the top contact surface and I had to lap it to make it flat. It works flawlessly. On the second 12900K and the 12900KS the little "wing" that the ILM presses on received minor scarring where the delid tool presses against it. In both cases I easily fixed it by sanding the bottom surface where the rubber seal makes contact.  This also accomplished cleaning off all of the old rubber gasket at the same time as fixing the burr on the wing.

I mention this so it does not come as a surprise. In all cases the delid tool screws were very tight, but it seems like the manufacturing is off a little bit on the tolerances. All previous gen Rockit delid tools popped the IHS off without a blemish. 

 

I had definitely grown complacent after doing about 20~ CPU delids and finding the process to always be quick, easy and painless. The 11900K disaster made me approach the process with much greater caution and care.  The delay in delidding the 12900KS was more being pure lazy than worry about how it would turn out. Of course, having a chip with a high  silicon quality is a good reason to be extraordinarily cautious. I used to delid all CPUs before installing them for the first time because I viewed it as a waste of time and thermal paste to install it before delidding. I have never seen a CPU that doesn't benefit from it, but I am more inclined to see what the temps are with the solder first and evaluate the potential benefit of a delid  against the risk.

2 hours ago, Rage Set said:

@Mr. Fox

 

I'm sorry I've been missing in action. Did you get the warranty straight with the 2080 TI? Lots to read through. (If you ever need me, you have my number, just call or text me bro).

Hi bro. We have missed you and I was planning to call you to make sure you were still happy and healthy. I suspect you have been busy with the new job, etc. I have not done anything yet on the 2080 Ti. It is not registered in my name, but I am assuming it might have a little bit of time left on the warranty. I was going to try to fix it, but it might make more sense to put the hybrid thing back together and RMA. Can you check the status for me and let me know how much (if any) time is remaining on the warranty and if we can arrange a cross-ship at my expense? I did not want to bother you with this since I suspect you have been very busy. Plus, I didn't want the warranty question to appear to be the only reason I would call to check on you.

[Papusan]

Another GeForce RTX 4090 16-pin Adapter Bites The Dust

published about 1 hour ago

 

Great balls of fire!!!

 

https://www.tomshardware.com/news/another-geforce-rtx-4090-16-pin-adapter-bites-the-dust

 

I wonder how long it will take before we see a statement on status from Nvidia. 
 

Nvidia launched an investigation into the matter when the first case surfaced, and with good reason, since meltdowns are starting to become widespread. The user feedback shows that the problem may reside with the design of the 12VHPWR power connector. Bending the cables too close to the connector seemingly causes some terminals to loosen up, leading to uneven mating.

[tps3443]

11 minutes ago, Papusan said:

Another GeForce RTX 4090 16-pin Adapter Bites The Dust

published about 1 hour ago

 

Great balls of fire!!!

 

https://www.tomshardware.com/news/another-geforce-rtx-4090-16-pin-adapter-bites-the-dust

 

I wonder how long it will take before we see a statement on status from Nvidia. 
 

Nvidia launched an investigation into the matter when the first case surfaced, and with good reason, since meltdowns are starting to become widespread. The user feedback shows that the problem may reside with the design of the 12VHPWR power connector. Bending the cables too close to the connector seemingly causes some terminals to loosen up, leading to uneven mating.

They should’ve just done (3) 8pin pci-e plug style on the 4090’s, I imagine it would have been okay right? I’ve pulled over 800+ watts with furmark for no other reason than just to see how high the power usage can go. My plugs are bent kind sideways too.
 

No issues at all.

[Papusan]

1 hour ago, tps3443 said:

They should’ve just done (3) 8pin pci-e plug style on the 4090’s, I imagine it would have been okay right? I’ve pulled over 800+ watts with furmark for no other reason than just to see how high the power usage can go. My plugs are bent kind sideways too.
 

No issues at all.

Nvidia's 12-pin Power Connector-plug is designed too small and weak. And Nvidia went with 6 power pins due the small delicate GPU PCB design (3x 8-pins power connectors have 9 power pins). So do your math and find out how many power pins there is if you use 4 of those 8-pin PCIe connectors. The new 12-pin Power Connector should have been 50% bigger and nvidia should have offered 2 (some AIB partners don't follow Nvidia's design and mount the connector upside down) 90 degres angeled adapters, outside offering the regualr adapter they offer in the package. 

 

3x 8-pin PCIe power connectors would never been approved for more than 450W (525w if we count in 75w from the  PCI Express x16 slot). Because thats the specs . Hence Nvidia neeeded an custom power connector. But they screwed up the design of the connector. We would never ever have seen graphics cards validated for more than 450/525W if Nvidia speced the new cards with 3x 8-pin PCIe power connectors. A custom Gpu power connector is actually a great idea if the connector has been designed/engineered properly from the beginning (bigger, more reliable and with more power pins).

 

Without the custom nvidia connector we would never seen Neptune 4090 with 630W vbios. Because nvidia would never  ever made 4090 FE cards with 4  8-pins power connectors. And we both know why nvidia released 4090 Fe with same max TDP as most of the custom made AIB cards. 

[tps3443]

@electrosoft

 

More MSI FORCE2 results and SP rating.  I feel like the force rating is not very consistent so far.

 

 

@Papusan

 

I believe the Molex spec for a single 8 pin is 324 watts. But due to crap PSU’s existing in the world, the ATX spec for a 8 pin pci cable is 150 watts. 

[Mr. Fox]

20 hours ago, Talon said:

Stock 13900K undervolted with 420mm AIO. Quiet fan curve. I wonder how low I can get my vmin, this is my SP106 chip. Finally messing around with it a bit more now that the BIOS is fixed. 

I flashed v2.03 on my Dark KPE last night and it is working fine with the 12900K. Something about it did not play nice with my Winduhz 11 OS. All USB function was lost. Windows 10 LTSC 2019 and 2021, and Windows 7 were fine, but I had to clean install Winduhz 11 for some reason. I spent several hours trying to fix it with no success. I finally gave up.

 

CPU and memory seem to function flawlessly. Not sure what the heck was wrong with the cancer OS. That was super weird, but it is weird and crummy by any other measurement anyhow.

[Reciever]

30 minutes ago, Papusan said:

Nvidia's 12-pin Power Connector-plug is designed too small and weak. And Nvidia went with 6 power pins due the small delicate PCB design (3x 8-pins power connectors have 9 power pins). So do your math and find out how many power pins there is if you use 4 of those 8-pin PCIe connectors. The new 12-pin Power Connector should have been 50% bigger and nvidia should have offered 2 (some AIB partners don't follow Nvidia's design and mount the connector upside down) 90 degres angeled adapters, outside offering the regualr adapter they offer in the package. 

 

3x 8-pin PCIe power connectors would never been approved for more than 450W. Because thats the specs . Hence Nvidia neeeded an custom power connector. But they screwed up the design of the connector. We would never ever have seen graphics cards validated for more than 450W if Nvidia speced the new cards with 3x 8-pin PCIe power connectors.

Then you know, make it 4?

[Papusan]

1 hour ago, tps3443 said:

@Papusan

 

I believe the Molex spec for a single 8 pin is 324 watts. But due to crap PSU’s existing in the world, the ATX spec for a 8 pin pci cable is 150 watts. 

The PSU's quality can be ok but they cut down on the wire gauge thickness. More metal increase prices and thinner wires make it easier for (less savy) people to route the cables properly in their chassis. Smaller chassis the worse. Should have been at least 16 gauge thickness on all (PCIe) cables. 15 thickness would be more proper (higher safety margin/higher current). 

 

 

Nvidia investigating reports of RTX 4090 power cables burning or melting theverge.com
https://www.theverge.com/2022/10/25/23422349/nvidia-rtx-4090-power-cables-connectors-melting-burning

 

I wonder how much this will affect the supply of 4080 and more of the 4090 cards coming weeks. Maybe the new new sibling of 3060 will help nvidia and it's partners to keep up sales of Nvidia graphics in this difficult  time. 

 

The Nvidia fire crack continue... So appropriate with some fire music.

 

Update 10/25/2022 4:30 pm PT

A fourth case has allegedly surfaced on a YouTube video(opens in new tab) from the Bang4BuckPC Gamer channel. The user allegedly purchased a prebuilt gaming PC from CyberPowerPC with a GeForce RTX 4090. The power connector for the graphics card seemingly melted, and the user returned the PC to CyberPowerPC. We've reached out to the system builder for comment.

 

[Reciever]

youtube comments as a source for news? 

[Papusan]

28 minutes ago, Reciever said:

youtube comments as a source for news? 

This is  topic brother. They sniff up news about Nvidia and fire everywhere. What next?

 

And AMD use it for all it's worth. Scott Herkelman confirm that AMD do as Intel and scrapped this new Nvidia 12/16 pin "fire cracker connector".

 

It’s a surprisingly concrete and technical statement about an upcoming GPU design, still more than a week out from its announcement on November 3rd. And to play devil’s advocate, it’s unlikely that Nvidia’s RTX 4090 issues played a part in AMD’s decision—it’s simply too late in the design process for AMD to switch course.

 

[Reciever]

13 minutes ago, Papusan said:

This is  topic brother. They sniff up news about Nvidia and fire everywhere. What next?

Not if Youtube comments are a source for news imho. 

 

We skipped reddit comments and straight to YouTube lol

 

That channel perplexes me, maybe because I dont much time on it but bang4buck and 13900k/4090 probably isnt what most think of, more power to him/her though hope it does well (youtube channel I mean) 

[Talon]

2 hours ago, Mr. Fox said:

I flashed v2.03 on my Dark KPE last night and it is working fine with the 12900K. Something about it did not play nice with my Winduhz 11 OS. All USB function was lost. Windows 10 LTSC 2019 and 2021, and Windows 7 were fine, but I had to clean install Winduhz 11 for some reason. I spent several hours trying to fix it with no success. I finally gave up.

 

CPU and memory seem to function flawlessly. Not sure what the heck was wrong with the cancer OS. That was super weird, but it is weird and crummy by any other measurement anyhow.

 

So strange as I had no issues at all with my BIOS update and I updated to 2.02 while others had it brick thier machines. EVGA LOL. At least the BIOS is good. 

[electrosoft]

4 hours ago, tps3443 said:

@electrosoft

 

More MSI FORCE2 results and SP rating.  I feel like the force rating is not very consistent so far.

 

 

 

True, we need to figure out the parameters used to ascertain the rating vs Asus because if an SP101 is coming back 149 and Sugi's SP114 is a 108 something needs to be established that is consistent considering @johnksss SP99 came back at 136....their testing criteria either needs work or is much different than SP.

[Mr. Fox]

10 hours ago, electrosoft said:

 

True, we need to figure out the parameters used to ascertain the rating vs Asus because if an SP101 is coming back 149 and Sugi's SP114 is a 108 something needs to be established that is consistent considering @johnksss SP99 came back at 136....their testing criteria either needs work or is much different than SP.

I suspect it is a new work in process with a lot of learning and data gathering that needs to happen before it can be viewed as accurate or reliable. It needs to be accurate or reliable before it can be related to ASUS SP ratings. 

 

Another example of where an industry standard would benefit both the seller and the buyer and improve the outcome for all involved. When we see things like this remain proprietary it really diminishes the value it can add. Building a reliance on ASUS SP ratings is also problematic.

 

It can also be misleading because it is sometimes not meaningful compared to the user experience. It better to be disappointed finding out later that an SP rating is low on a CPU that runs like a top than it is to have a CPU with a good SP rating that is mediocre, and we have seen both of those things.

[Mr. Fox]

Here is some interesting information about A-die versus M-die timings and the impact of performance. A-die is only better when it can handle good timings, but apparently sometimes it does not.

 

 

[Mr. Fox]

Block installed. Man, what a huge improvement.

[Rage Set]

15 hours ago, tps3443 said:

Long time no see. How are you? 

 

I'm good man, how are you? I was eventually going to make my way back here to see how you guys compete.

 

 

15 hours ago, Mr. Fox said:

I delidded the first 12900K, the second 12900K and now the 12900KS. I heated the first one but not the next two. It did not seem to help anything other than making it more difficult to handle. Just a heads up, in all three cases the delid causes minor damage to the 12th gen IHS and will probably do the same to the 13th gen. The first one (that I heated) got the top surface slightly deformed the top contact surface and I had to lap it to make it flat. It works flawlessly. On the second 12900K and the 12900KS the little "wing" that the ILM presses on received minor scarring where the delid tool presses against it. In both cases I easily fixed it by sanding the bottom surface where the rubber seal makes contact.  This also accomplished cleaning off all of the old rubber gasket at the same time as fixing the burr on the wing.

I mention this so it does not come as a surprise. In all cases the delid tool screws were very tight, but it seems like the manufacturing is off a little bit on the tolerances. All previous gen Rockit delid tools popped the IHS off without a blemish. 

 

I had definitely grown complacent after doing about 20~ CPU delids and finding the process to always be quick, easy and painless. The 11900K disaster made me approach the process with much greater caution and care.  The delay in delidding the 12900KS was more being pure lazy than worry about how it would turn out. Of course, having a chip with a high  silicon quality is a good reason to be extraordinarily cautious. I used to delid all CPUs before installing them for the first time because I viewed it as a waste of time and thermal paste to install it before delidding. I have never seen a CPU that doesn't benefit from it, but I am more inclined to see what the temps are with the solder first and evaluate the potential benefit of a delid  against the risk.

Hi bro. We have missed you and I was planning to call you to make sure you were still happy and healthy. I suspect you have been busy with the new job, etc. I have not done anything yet on the 2080 Ti. It is not registered in my name, but I am assuming it might have a little bit of time left on the warranty. I was going to try to fix it, but it might make more sense to put the hybrid thing back together and RMA. Can you check the status for me and let me know how much (if any) time is remaining on the warranty and if we can arrange a cross-ship at my expense? I did not want to bother you with this since I suspect you have been very busy. Plus, I didn't want the warranty question to appear to be the only reason I would call to check on you.

 

I have never registered that card and if I did, I've deleted it from My Products page on evga.com. The card was purchased on 2/4/19. I can send you the PDF of the invoice via e-mail. You should be able to register it. I still have my 3090's and 3080 TI Hybrid (still brand new in box) if you or anyone else is interested. 

 

I'm doing well. Life gave me some lemons and I'm making the best lemonade ever. I appreciate you all. I've gained brothers for life, even the ones I disagree with. 

 

I'm going to take brother @Papusan approach to hardware. But I am going to be around more often again. I'm thinking about the 13900K. Lord knows I have about four Z690 boards I can use with it, lol. 

 

[Mr. Fox]

1 hour ago, Rage Set said:

 

I have never registered that card and if I did, I've deleted it from My Products page on evga.com. The card was purchased on 2/4/19. I can send you the PDF of the invoice via e-mail. You should be able to register it.

Thank you. Yes, please email that to me. It may be out of warranty since 3 years have passed. If memory serves me correctly the standard warranty is 3 years, but I would need to check. I could still register it and find out that way, too.

[Papusan]

19 hours ago, Papusan said:

The way Nvidia think the chassis door (glass or metal) is meant to be used....🙂 

The way Nvidia prefer you install your brand new 4090 graphics cards. Big gain once again for Nvidia (less prone for fire and reduced chance for increased RMA costs), HaHa

 

While Nvidia investigate the fire and smoke... Very impotant to offer graphics cards that don't catch fire, and to keep up profits and at same time reduce inventory of older chips.
ZOTAC launches GeForce RTX 3070 Ti with GA102-150 GPU
https://videocardz.com/newz/zotac-launches-geforce-rtx-3070-ti-with-ga102-150-gpu

 

A card called Zotac GeForce RTX 3070Ti-8G6X Apocalypse GOC is a new model based on existing GA104 design. Both cards have an identical triple-fan cooler and, from what we can see from official pictures, a similar board design as well. This upgraded model features NVIDIA’s largest GA102-150 GPU, but only with 6144 CUDA cores. This means that as many as 4608 cores were disabled on this Ampere processor.

 

@Rage Set Nice see you back. Hope all is well 🙂

5 hours ago, Mr. Fox said:

Block installed. Man, what a huge improvement.

Nice improvements bro Fox.

[Ashtrix]

12 hours ago, electrosoft said:

 

True, we need to figure out the parameters used to ascertain the rating vs Asus because if an SP101 is coming back 149 and Sugi's SP114 is a 108 something needs to be established that is consistent considering @johnksss SP99 came back at 136....their testing criteria either needs work or is much different than SP.

 

I think this massive SP variance is probably because of E cores nonsense and their sharing of powerplane with P cores. And the fact how the new 13th gen i7 13700K is basically 12th gen i9 12900K now as they added more E cores, and this is for entire stack, 13600K also got E bump. Plus anything lower than 13600K is Alder Lake on 13th gen lol.

 

9th, 10th, 11th did not have this much massive delta in SP rating. For MSI I do not know even a bit to think about so it's a total blacked out zone. The only thing which is certain is SP ratings is ASUS own big brain algorithm which reads from VID factory values and thus estimates their OC capabilities and etc on the side panel in the BIOS. However do note 'sometimes' ASUS adjusts them as the BIOS gets updates etc.

 

ASUS solution is far superior to MSI esp given the fact how ASUS shows the factory VID + gives the approx values of Vcore and shows to us that is required atop the the clock / cache ratios (side note - Cache ratio got nerfed with 10th gen later on and 11th gen because Ring was too hard on 14nm++ and causes issues. ADL was start and it did not improve. RPL Intel explicitly mentioned that their Ring is very fast now as it operates at 5GHz it's a much tuned product than ADL ever was)

 

Maybe MSI doesn't store these values in their BIOS code and integrate, that is why probably their ratings are not same and variances... Also ASUS offers V_Latch on their Apex and up series (however Hero also has SP) and that helps them to read exact Vcore values at die-sense directly without the added socket sense LLC impact just ASUS engg at play with E nonsense and binning .. It's all just speculation from my side.

[tps3443]

2 hours ago, Papusan said:

The way Nvidia prefer you install your brand new 4090 graphics cards. Big gain once again for Nvidia (less prone for fire and reduced chance for increased RMA costs), HaHa

 

While Nvidia investigate the fire and smoke... Very impotant to offer graphics cards that don't catch fire, and to keep up profits and at same time reduce inventory of older chips.
ZOTAC launches GeForce RTX 3070 Ti with GA102-150 GPU
https://videocardz.com/newz/zotac-launches-geforce-rtx-3070-ti-with-ga102-150-gpu

 

A card called Zotac GeForce RTX 3070Ti-8G6X Apocalypse GOC is a new model based on existing GA104 design. Both cards have an identical triple-fan cooler and, from what we can see from official pictures, a similar board design as well. This upgraded model features NVIDIA’s largest GA102-150 GPU, but only with 6144 CUDA cores. This means that as many as 4608 cores were disabled on this Ampere processor.

 

@Rage Set Nice see you back. Hope all is well 🙂

Nice improvements bro Fox.

Hey, now Corsair and others needs to release a “Fire extinguisher for ATX cases” it’ll be sold as an accessory with included mounting, and it has 120/140 mounting holes. It can bolt on the back 120 slot of a case for quick access just in case your RTX4090 catches fire.

Miniature fire extinguisher.

[Papusan]

3 hours ago, tps3443 said:

Hey, now Corsair and others needs to release a “Fire extinguisher for ATX cases” it’ll be sold as an accessory with included mounting, and it has 120/140 mounting holes. It can bolt on the back 120 slot of a case for quick access just in case your RTX4090 catches fire.

Miniature fire extinguisher.

I wonder if this size is too big for the package....🙂 

 

Maybe 3080@20GB can stop the 4090 fire? Nvidia could save up 4090 dies and make 4090 Ti with improved power connectors/adapters and release it sooner than later.  

 

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

 

Newegg: Refund Issued for RTX 4090 Box Filled With Metal Weights tomshardware.com
https://www.tomshardware.com/news/newegg-responds-to-metal-in-rtx-4090-box

 

A Newegg representative told Tom's Hardware via email that it refunded a customer that claimed to have received blocks of metal instead of an Nvidia Geforce RTX 4090. The company also claims that banning customer accounts is a "normal" part of a fraud investigation process.

[Talon]

@Mr. FoxIt seems USPS came through a day early, shes out for delivery. 😀

[Mr. Fox]

32 minutes ago, Talon said:

@Mr. FoxIt seems USPS came through a day early, shes out for delivery. 😀

My eyes are peeled and I am waiting with baited breath. 😄