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

[Papusan]

7 hours ago, Papusan said:

A new one. Nvidia is in deaper and deaper water. Can Jensen swim? My best guess...  Class action or recall of cards/in best case sending out new adapters. Nvidia's investigation team can't keep up with this pace. One per day is +60 before Christmas Eve. And there isn't many 4090 cards out there. 

 

The horror has a face – NVIDIA’s hot 12VHPWR adapter for the GeForce RTX 4090 with a built-in breaking point

Summary and conclusion

The overall build quality of the included adapter for the GeForce RTX 4090, which is distributed by NVIDIA itself, is extremely poor and the internal construction should never have been approved like this. NVIDIA has to take its own supplier to task here, and replacing the adapters in circulation would actually be the least they could do. I will therefore summarize once again what has struck those involved (myself included) so far:

• The problem is not the 12VHPWR connection as such, nor the repeated plugging or unplugging.
• Standard compliant power supply cables from brand manufacturers are NOT affected by this so far.
• The current trigger is NVIDIA’s own adapter to 4x 8-pin in the accessories, whose inferior quality can lead to failures and has already caused damage in single cases.
• Splitting each of the four 14AWG leads onto each of the 6 pins in the 12VHPWR connector of the adapter by soldering them onto bridges that are much too thin is dangerous because the ends of the leads can break off at the solder joint (e.g., when kinked or bent several times).
• Bending or kinking the wires directly at the connector of the adapter puts too much pressure on the solder joints and bridges, so that they can break off.
• The inner bridge between the pins is too thin (resulting cross section) to compensate the current flow on two or three instead of four connected 12V lines.
• NVIDIA has already been informed in advance and the data and pictures were also provided by be quiet! directly to the R&D department.

 

Some heads has to roll at nvidia's engineering team. Because this adapter is a fire hazard and should have never seen the light.

 

What with the 12VHPWR connector on the GPU PCB? Hot running connector on the adapters side won't make it more reliable in the length. Plastic will wear out faster from the added heat (even if you replace it in time - before the smoke, melting and fire).

[Ashtrix]

29 minutes ago, Papusan said:

 

The horror has a face – NVIDIA’s hot 12VHPWR adapter for the GeForce RTX 4090 with a built-in breaking point

Summary and conclusion

The overall build quality of the included adapter for the GeForce RTX 4090, which is distributed by NVIDIA itself, is extremely poor and the internal construction should never have been approved like this. NVIDIA has to take its own supplier to task here, and replacing the adapters in circulation would actually be the least they could do. I will therefore summarize once again what has struck those involved (myself included) so far:

• The problem is not the 12VHPWR connection as such, nor the repeated plugging or unplugging.
• Standard compliant power supply cables from brand manufacturers are NOT affected by this so far.
• The current trigger is NVIDIA’s own adapter to 4x 8-pin in the accessories, whose inferior quality can lead to failures and has already caused damage in single cases.
• Splitting each of the four 14AWG leads onto each of the 6 pins in the 12VHPWR connector of the adapter by soldering them onto bridges that are much too thin is dangerous because the ends of the leads can break off at the solder joint (e.g., when kinked or bent several times).
• Bending or kinking the wires directly at the connector of the adapter puts too much pressure on the solder joints and bridges, so that they can break off.
• The inner bridge between the pins is too thin (resulting cross section) to compensate the current flow on two or three instead of four connected 12V lines.
• NVIDIA has already been informed in advance and the data and pictures were also provided by be quiet! directly to the R&D department.

 

Some heads has to roll at nvidia.

 

Was about to post it. This guy did a good job in showing the absolute pile of trash quality that Nvidia shoved hard. Esp with the quality of the soldering job, the pictures show the total horror as the title shows. Cheaped out garbage the high guage wires are going into this narrow thin sheet and crammed tight with low tolerance. RIP.

 

Below image with cut adapter to show the insides and the soldering for the cable. I bet the metal is thin and not up to the mark.

 

 

Even though he says the cable design / formfactor is fine I do not see it is a good replacement. Cablemod is explicitly making 90 degree plug for this, if the connector design was not a dumpster standard like this there would never have been any requirement.

 

Heck when we plug the 8Pin power lines to CPU on mobo sockets we kink the cables hardcore even if they are 4Pinx2, like literally bending all ways possible and we never have any sort of failures on that connector. The 8 Pin standard is really made solid bulletproof and does it's job very well. And both ASUS and GB have metallic shields for those CPU connectors...

 

None of the companies tried to even put something like this to reinforce the socket and adapters.. shame.

 

 

This new 12VHPWR is a joke. Nvidia made it even worse. No wonder AMD is totally skipping this BS for a standard form factor which is tried and tested.

 

"If it ain't broke, don't fix it"

[Mr. Fox]

Well, it is looking like the 12900KS is going to be a really nice upgrade for Banshee, my work PC. It's running nice and cool with the delid considering it is shut up in an enclosed case and the RAM run stable with this CPU at 4400 16-18-18-36 1T just as it did with the 13900K.

 

[Mr. Fox]

Time for bed. More tuning in the days ahead.

[Mr. Fox]

[Etern4l]

6 hours ago, Ashtrix said:

This new 12VHPWR is a joke. Nvidia made it even worse. No wonder AMD is totally skipping this BS for a standard form factor which is tried and tested.

 

"If it ain't broke, don't fix it"

 

Hmm, makes me wonder if this 12VHPWR or Nvidia. Have there also been any issues related to 12VHPWR cables from the likes of Corsair, BeQuiet, CableMod etc.

 

Are partners forced to use this new "Nvidia standard", or will we potentially see a reversal of the "12VHPWR Only!" policy, especially on lower powered boards?

 

[Etern4l]

19 minutes ago, Mr. Fox said:

 

Impressive! Is this long-term stable (by that I mean at least 24h without issues or crashes)?

[Mr. Fox]

4 hours ago, Etern4l said:

 

Impressive! Is this long-term stable (by that I mean at least 24h without issues or crashes)?

It hasn't even been installed 24 hours yet, but it seems stable so far. I generally do not tune my benching system for that though. I generally tune for highest possible benchmark scores and do not need it to be able to run Prime95 for hours. I still have a lot of learning and tuning to do.

---

This is an accomplishment for the Strix D4 mobo. Rare to see an ADL DDR4 system capable of doing this.

 

[tps3443]

5 minutes ago, Mr. Fox said:

It hasn't even been installed 24 hours yet, but it seems stable so far. I generally do not turn my benching system for that though. I generally tune for highest possible benchmark scores and do not need it to be able to run Prime95 for hours. I still have a lot of learning and tuning to do.

---

This is an accomplishment for the Strix D4 mobo. Rare to see an ADL DDR4 system capable of doing this.

 

 

Have you tried maybe 4000~4266 in Gear 1? 

 

 

 

 

[Etern4l]

23 minutes ago, Mr. Fox said:

It hasn't even been installed 24 hours yet, but it seems stable so far. I generally do not turn my benching system for that though. I generally tune for highest possible benchmark scores and do not need it to be able to run Prime95 for hours. I still have a lot of learning and tuning to do.

---

This is an accomplishment for the Strix D4 mobo. Rare to see an ADL DDR4 system capable of doing this.

 

 

In that case it's probably not far off stable state, very nice.

[tps3443]

I got a private message from CENS he wants to buy my 13900KF and LN2 it apparently. 

 

 

 

 

 

[jaybee83]

21 hours ago, tps3443 said:

I got a private message from CENS he wants to buy my 13900KF and LN2 it apparently. 

 

 

 

 

 

 

haha ure turning heads, congratz! are you gonna take him up on his offer? might wanna ask what they have in mind with "something nice" from EVGA. too bad they dont make GPUs anymore 😄

 

aw man, wish i could join u guys in CPU and RAM OC here, had a loooot to catch up on esp regarding Ryzen 7000 tuning. but im now slowly getting an idea of what settings matter and what approaches would be promising. in any case, currently started off with tuning the 4090, easiest and quickest to do for now 😛 got +1550 Mhz stable for 24/7 on the vRAM (= 24.1 Gbps), now turning my focus on the core. lets see what we get!

 

really happy with the STRIX OC vbios so far, lets me downclock and force way lower limits than before for desktop usage. also not bumping into power limits anymore, which is nice! 🙂 lets see if it stays that way once i start bumping up the core.

[ryan]

alot of heavy hitters here at NBR.

 

I'm scoring 12k in R23 I have no idea why a chip the nearly same size is scoring 41k. you would think better is like 20 percent not 100 percent better at least concerning newer(2 gens)

[Etern4l]

37 minutes ago, ryan said:

I have no idea why a chip the nearly same size is scoring 41k. you would think better is like 20 percent not 100 percent better at least concerning newer(2 gens)

 

10 instead of 14nm process combined with 16 new E-Cores. Ridiculed by many, without them AMD would be reigning supreme now. I guess in 13th gen they also may have further optimised the P-Cores by completely removing AVX-512 (I am not sure, just a guess on how they made space for double the E-cores and more cache).

[tps3443]

42 minutes ago, ryan said:

alot of heavy hitters here at NBR.

 

I'm scoring 12k in R23 I have no idea why a chip the nearly same size is scoring 41k. you would think better is like 20 percent not 100 percent better at least concerning newer(2 gens)

My 13900KF overclocked is literally 300% faster than a stock 11900K in R23 MT. Pretty crazy. 
 

 

Great chips for sure. I’m testing mine at 5.5Ghz all cores, HT enabled, E-Cores enabled, and it’s peak power is 250 watts during R23 still scoring 40,660. I’m losing nothing in terms of performance and it brings the 13900K down near the AMD 7950X in terms of power levels. 

Good AIO could handle that. Definitely an improvement over last Gen.

[Papusan]

50 minutes ago, ryan said:

alot of heavy hitters here at NBR.

 

I'm scoring 12k in R23 I have no idea why a chip the nearly same size is scoring 41k. you would think better is like 20 percent not 100 percent better at least concerning newer(2 gens)

You need to compare amount cores, clock speed and what power budget the Cpu have. You can't compare Intel chips tuned to be in Jokebooks. 

 

A sidenote.... 12th and 13th gen have same IPC, so... All boils down to castrating the chips so it can be able to be used in thin and flimsy

 

13 minutes ago, Etern4l said:

 

10 instead of 14nm process combined with 16 new E-Cores. Ridiculed by many, without them AMD would be reigning supreme now. I guess in 13th gen they also may have further optimised the P-Cores by completely removing AVX-512 (I am not sure, just a guess on how they made space for double the E-cores and more cache).

From what I remember he have 12800H aka 10nm chips (12th gen).

[Mr. Fox]

4 hours ago, tps3443 said:

Have you tried maybe 4000~4266 in Gear 1? 

Not with the 12900KS. Haven't had time to experiment, and since it is my work computer less of a recreational priority for me.

 

CENS has BIOS overclock profiles in the firmware on the Dark mobos. It's worth a lot more than market value based on the silicon grade. I would ask him what he has in mind for the price rather than give him a price based on what you think is fair. When you are selling something rare and in demand, the pricing should reflect that.

3 hours ago, Etern4l said:

In that case it's probably not far off stable state, very nice.

Yeah, the DDR4-4400 overclock is totally stable. I let it run for TM5 2.5 hours last night with no error. It requires a fan blowing on it just like DDR5. A fan blowing on it keeps the temps 38-40°C. Without the fan it hits 50°C within a few minutes and starts throwing errors in massive quantity, eventually ending in a BSOD.

[electrosoft]

4 hours ago, Mr. Fox said:

It hasn't even been installed 24 hours yet, but it seems stable so far. I generally do not turn my benching system for that though. I generally tune for highest possible benchmark scores and do not need it to be able to run Prime95 for hours. I still have a lot of learning and tuning to do.

---

This is an accomplishment for the Strix D4 mobo. Rare to see an ADL DDR4 system capable of doing this.

 

 

4 hours ago, tps3443 said:

 

Have you tried maybe 4000~4266 in Gear 1? 

 

 

 

 

 

1 minute ago, Mr. Fox said:

Not with the 12900KS. Haven't had time to experiment, and since it is my work computer less of a recreational priority for me.

 

CENS has BIOS overclock profiles in the firmware on the Dark mobos. It's worth a lot more than market value based on the silicon grade. I would ask him what he has in mind for the price rather than give him a price based on what you think is fair. When you are selling something rare and in demand, the pricing should reflect that.

Yeah, the DDR4-4400 overclock is totally stable. I let it run for 2.5 hours last night with no error. It requires a fan blowing on it just like DDR5. A fan blowing on it keeps the temps 38-40°C. Without the fan it hits 50°C within a few minutes and starts throwing errors in massive quantity, eventually ending in a BSOD.

 

 

Yeah, I definitely want to see what it does in G1. 🙂

 

 

[Etern4l]

8 minutes ago, Papusan said:

From what I remember he have 12800H aka 10nm chips (12th gen).

He said the gap between the CPUs in question is 2 gens so unless there was a little math fail, we would be talking 11th vs 13th.

[Papusan]

8 minutes ago, Etern4l said:

He said the gaps is 2 gens so unless it was a little math fail, we are talking 11th vs 13th.

11th gen Mobile is on 10nm. 12th and 13th gen is 10nm+++++++

 

Edit. And older mobile on 10nm have a lot less cores, so.

[electrosoft]

3 hours ago, tps3443 said:

I got a private message from CENS he wants to buy my 13900KF and LN2 it apparently. 

 

 

 

 

 

 

 

Heh....

 

Remember when I said your chip was a gooder and an SP108+ and stood by the assessment? 🙂

 

I'm glad you got a rockin' CPU bro! You have good, positive energy and an almost giddy love of hardware that is refreshing to see. You've been rewarded with a monster 11900k and now a stand out 13900k.

 

CENS clearly sees the same characteristics of your CPU and sees some major potential in it.

 

Congrats! 🙂

 

 

[Etern4l]

15 minutes ago, tps3443 said:

Great chips for sure. I’m testing mine at 5.5Ghz all cores, HT enabled, E-Cores enabled, and it’s peak power is 250 watts during R23 still scoring 40,660. I’m losing nothing in terms of performance and it brings the 13900K down near the AMD 7950X in terms of power levels.

 

Nice, just to be clear: is that stock voltages or undervolted?

[Mr. Fox]

Brother @tps3443 is either living right or struck a deal with the devil on the 11900K and his 13900KF. Two lucky strikes in a row. Only he could tell us which one is more accurate. 😉 

[Etern4l]

3 minutes ago, Papusan said:

11th gen Mobile is on 10nm. 12th and 13th gen is 10nm+++++++

 

OK, I am still assuming we meant to compare apples to apples (as opposed to mobile vs desktop etc) i.e. vs 2 gens back: the 14 nm i9-11900K 🙂

[tps3443]

5 minutes ago, Etern4l said:

 

Nice, just to be clear: is that stock voltages or undervolted?

Its undervolted. I’m kinda slowly dropping it off. Just to find a good stable baseline for 5.5Ghz all cores. 
 

It’s less power than my last CPU and so much faster.