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

[Mr. Fox]

2 hours ago, tps3443 said:

 

You are absolutely right! I turned my fan off and they started erroring of out within 45 seconds with no fan. I saw the temps hit 44C then maybe 45C and errors started popping up left and right. I BSODed before I could even close the HCI Memtest. They were already under load for 1hr+ prior to removing the fan.

 

I reduced voltage to 1.400V and temps are even lower now. And stable in HCI Memtest have not tried again without a fan. But, I’m gonna just leave the fan for sure 😂
 

DDR5 needs a fan like our CPU’s need thermal paste LOL. 
 

I may consider water cooling my ram. And then remove the fan to prevent all that dust build up. DDR5 is awesome though! I love it. Very fast. Very stable “With a fan” 🤣

Anyways, I still think they’re special. Only because they are Hynix A-Die and awesome so far. 

 

I have the same issues with DDR4, but they are less obvious running at roughly half the clock speed. It takes more stress to overheat the DDR4, but it still happens.  A lot of people don't believe me when I tell them their memory is overheating at 45-50°C and that is why their system is unstable, not because the memory is defective. At such a low temperature it seems hard to believe, but it is totally accurate.

 

TeamGroup Delta memory was really good, but because I decided to go with aftermarket heat sinks and water cooling having anything with RGB makes no sense. The RGB LEDs also do not help matters. They generate heat of their own that contributes to the thermal management problem, and if I am going to cover up the LEDs with thick aluminum plates there is no  point in having it. Given that I don't need or use XMP profiles, going with generic naked green PCB was the most intelligent move, and it costs less.

 

If you decide to remove the heatsinks on the memory you have, let me know and I will explain the super-easy way how to get them off without causing any damage. If you try to peel them off chances are very good you will destroy the memory. I know two people who tore memory chips off their PCB because the adhesives they use to attach them are stronger than the components they are attached to.

[tps3443]

6 minutes ago, Mr. Fox said:

I have the same issues with DDR4, but they are less obvious running at roughly half the clock speed. It takes more stress to overheat the DDR4, but it still happens.  A lot of people don't believe me when I tell them their memory is overheating at 45-50°C and that is why their system is unstable, not because the memory is defective. At such a low temperature it seems hard to believe, but it is totally accurate.

 

TeamGroup Delta memory was really good, but because I decided to go with aftermarket heat sinks and water cooling having anything with RGB makes no sense. The RGB LEDs also do not help matter. They generate heat of their own, and if I am going to cover up the LEDs with thick aluminum plates there is no  point in having it. Given that I don't need or use XMP profiles, going with generic naked green PCB was the most intelligent move, and it costs less.

 

If you decide to remove the heatsinks on the memory you have, let me know and I will explain the super-easy way how to get them off without causing any damage. If you try to peel them off chances are very good you will destroy the memory. I know two people who tore memory chips off their PCB because the adhesives they use to attach them are strong than the components they are attached to.

I didn’t realize DDR5 was so sensitive. I know DDR4 could be, but not a BSOD like that 30-45 seconds after taking the fan off. Usually Windows with DDR4 would keep running for me, and just spit errors out in memory testing. Anyways, I’m glad I know now. 
 

Fan is working great, but I will eventually watercool them in the coming weeks.
 

I just copied Team Group’s Hynix A-Die 7600 timings from their new sticks that are not even released yet, and it’s very stable so far!

 

DDR5 7600 CL36-46-46-81 @ 1.400V. I imagine 7800-8000 is possible though, haven’t tried yet. 
 

 

[Mr. Fox]

18 minutes ago, Mr. Fox said:

Anyways, I still think they’re special. Only because they are Hynix A-Die and awesome so far. 

I agree with that part. It's not the brand, it is the silicon that makes them special. And, the fact that A-die is overpriced and uncommon also makes them special. At some point in the near future it may become very ordinary like Samsung B-die was. B-die was and still is special to the extent it was better than all other DDR4 options, but not special in terms of being rare or unique among enthusiasts.

 

I will grab some A-die Hynix later on, after the cost:benefit ratio makes more sense. I definitely want it, but right now I would view it as being a frivolous purchase compared to M-die I already have that overclocks almost as well, but not quite. If I didn't already own M-die then it would make sense to pay extra for A-die.

[electrosoft]

41 minutes ago, Clamibot said:

 

I don't think I'm gonna get better results than this from my super 10900K unless I move the CPU back into my X170 and ghetto rig up some extreme cooling to the CPU part of the unified heatsink. This CPU requires less voltage in my X170 for the same clock speeds when running it at normal temperatures. Gigabyte's motherboards consistently require more voltage than other motherboard brands to keep the CPU stable at any given speed.

 

Imagine a 10900K running at 5.7, maybe even 5.8 GHz in a Clevo X170. Now that would be a sight to behold! I already got it running at 5.5 GHz on semi-normal cooling in this laptop. Time to move it back.

 

 

I saw the same issues across two Gigabyte Z590 boards. More Vcore required to achieve the same results as my Asus Prime and Hero and MSI Z590 boards. Gigabyte was the worst of the bunch by far. I had similar issues with a Z490 Gigabyte board too. I won't be using them again.

 

 

[tps3443]

4 minutes ago, electrosoft said:

 

I saw the same issues across two Gigabyte Z590 boards. More Vcore required to achieve the same results as my Asus Prime and Hero and MSI Z590 boards. Gigabyte was the worst of the bunch by far. I had similar issues with a Z490 Gigabyte board too. I won't be using them again.

 

 

When I was trying to bin for decent 11900K, I was testing them in a Z490 Gigabyte/Aorus motherboard. It was a joke.. lol. I had to test them all again on my EVGA Z590 Dark. Which managed to finally give me some consistent data on each chip.

I don’t think I could ever go for another Gigabyte motherboard again. Honestly, now I’m a MSI fan! Who would have thought that! 

[Mr. Fox]

I have owned one A$$Rock mobo and one Gigabutt mobo, and that was enough to cure me of wanting anything to do with either of the brands. Maybe that is too harsh and throwing the baby out with the bath water, but I work hard for my money and purchasing high-end products always represents a personal sacrifice. I don't appreciate it when I discover that I spent my money on a piece of trash, and I am generally reluctant to forgive it and allow for second chances. Sometimes I do, but it is because there is something compelling about the item that makes giving a second chance something that might benefit me, not because I am an overall nice guy and want to give them another chance to earn my trust. I am a super nice guy that extends forgiveness too easily when it comes to relationship with individuals, but I harbor resentment a long time when the other party is a business. Probably because I know they don't actually give a damn about me. That being the case, they don't deserve the second chance. They're not worthy of me or my money.

[Papusan]

The future is less of P-cores. Will AMD follow Intel and offer less performance cores and offer 75% of the cores as baby cores as well? Don't games need more than 6 real cores in the future or/will small E-cores be better for gaming? The transition to trash cores continue... And why go down to 6 P-cores then add 2 more for the refresh on same socket? A nice way to milk/offer an performance upgrade path? Remove cores, then add them back to offer something new. Almost as Dell did with their gaming laptops @Ashtrix+++

 

Based on the leaked slides, it looks like Intel's 14th Gen Meteor Lake-S Desktop CPUs will take a step back and reduce the number of P-Cores while retaining the number of E-Cores. The lineup listed in the charts includes five different SKU configurations with the top variants featuring up to 22 cores in a combination of 6 P-Cores and 16 E-Cores. The P-Cores on the Meteor Lake CPUs are based on the brand-new Redwood Cove architecture while the E-Cores will utilize the Crestmont design.

Intel Meteor Lake-S CPUs To Feature 6 Performance Cores, Arrow Lake-S With 8 P-Cores On LGA 1851 Socket

 

 

And more about Nvidia's disaster.

Jon Gerow, aka Jonny Guru (Director of Research Development @ Corsair), has posted a write-up about the 12VHPWR connector. 

 

 

SO, YOU’RE GOOD WITH THE 12VHPWR CONNECTOR?

Unfortunately, as of this writing (10/29/2022) Nvidia’s included adapter uses solder, some terminal I’ve never seen before (it has two splits instead of one) and I don’t know what else. Which turns an otherwise excellently designed connector into a disaster.

 

IS THAT IT?
Probably not.  Let’s consider this a living document as we find out more and more about the new 12VHPWR connector and its potential issues.

[Etern4l]

During a recent prosumer mobo purchase research it became clear it's either MSI or Asus these days. MSI won the argument on robustness, features, and memory stability (immediately evidenced by the max RAM speeds they officially support, in addition to supportive reviews). The main feature lacking is the frequently quoted SP rating lol, but then of course we have the Force 2 score to despair over 😉

Has worked a treat for me.

[tps3443]

4 minutes ago, Mr. Fox said:

I have owned one A$$$Rock mobo and one Gigabutt mobo, and that was enough to cure me of wanting anything to do with either of the brands. Maybe that is too harsh and throwing the baby out with the bath water, but I work hard for my money and purchasing high-end products always represents a personal sacrifice. I don't appreciate it when I discover that I spent my money on a piece of trash, and I am generally reluctant to forgive it and allow for second chances. Sometimes I do, but it is because there is something compelling about the item that makes giving a second chance something that might benefit me, not because I am an overall nice guy and want to give them another chance to earn my trust. I am a super nice guy that extends forgiveness too easily when it comes to relationship with individuals, but I harbor resentment a long time when the other party is a business. Probably because I know they don't actually give a damn about me. That being the case, they don't deserve the second chance. They're not worthy of me or my money.

I kinda started feeling that way over the EVGA motherboards lately. little issues here and there. Nothing too crazy. But after using this MSI Unify X. I’m not missing anything. You don’t even have to set the bclk to over 100+ to maintain a solid even CPU frequency. Not sure why EVGA motherboards have a weak or low bclk frequency. That’s just a minor nit pick, primarily the debug LED’s failing was a major feature I have been

upset over on all of my EVGA motherboards ever owned. 
 

Before even giving it a chance, I was seriously considering the Z690 Dark, or Z790 Dark. But I swear this MSI Unity X board is too good. 
 

Now, when buying future motherboards, it’s probably gonna be a first option to look at. I’m comfortable with the bios, and

confident enough to buy another one. Obviously I’d buy another EVGA board too, but only if it was a decent value.

[Etern4l]

5 minutes ago, Papusan said:

Based on the leaked slides, it looks like Intel's 14th Gen Meteor Lake-S Desktop CPUs will take a step back and reduce the number of P-Cores while retaining the number of E-Cores. The lineup listed in the charts includes five different SKU configurations with the top variants featuring up to 22 cores in a combination of 6 P-Cores and 16 E-Cores. The P-Cores on the Meteor Lake CPUs are based on the brand-new Redwood Cove architecture while the E-Cores will utilize the Crestmont design.

Intel Meteor Lake-S CPUs To Feature 6 Performance Cores, Arrow Lake-S With 8 P-Cores On LGA 1851 Socket

 

Doesn't make sense. How are they proposing to sell those cut down "next gen" CPUs? Lower price?

[Papusan]

6 minutes ago, Etern4l said:

 

Doesn't make sense. How are they proposing to sell those cut down "next gen" CPUs? Lower price?

High core counts processors and incresed IPC for baby cores. Remember AMD Ryzen mainstream will most likely be stuck on with 16 cores. Or they are forced to do as Intel to keep being competitive on price. Bigger cores will always be more expensive. 

[Etern4l]

6 minutes ago, Papusan said:

High core counts processors

 

? The leak suggests they are looking to reduce max core count to 22C / 28T in 14th gen. This smells of the 11th gen fiasco.

[Papusan]

On 10/30/2022 at 8:57 PM, Etern4l said:

 

? The leak suggests they are looking to reduce max core count to 22C / 28T in 14th gen. This smells of the 11th gen failure.

Remove features to add them back later is also Microsofts way do it. This way you will feel you get something back if you upgrade🙂

 

Moving on, we have the 15th Gen Arrow Lake-S Desktop CPUs which will bring back the 24 cores that we get on Raptor Lake CPUs today. The Arrow Lake-S top die will utilize up to 24 cores which will be a combination of 8 Performance Cores and 16 Efficiency Cores.

 

In the older days we was stuck with 4 cores in a long time. Now we will be stuck with a flood of Baby-cores.

 

So in short... Intel's new Tick/Tock process: -2 Cores/+2 Core😀

 

Edit. Who moderate Nvidia's reddit forum? Nvidia themself?

[Mr. Fox]

42 minutes ago, tps3443 said:

Before even giving it a chance, I was seriously considering the Z690 Dark, or Z790 Dark. But I swear this MSI Unity X board is too good. 

I loved the MSI Unify-X. But, I sold it to you and kept the EVGA board. I guess that tells you which one I think is better of the two. The trouble with ASUS and MSI (most of the others, too) is their products are only good when they work. When they stop working, you regret having given them your money. They'll treat you like a pile of dung. No worries with EVGA. A great product backed by a not-great company gives me pause. Unify-X is a great product. I do not think of MSI as being a great company.

[Etern4l]

9 minutes ago, Mr. Fox said:

I loved the MSI Unify-X. But, I sold it to you and kept the EVGA board. I guess that tells you which one I think is better of the two. The trouble with ASUS and MSI (most of the others, too) is their products are only good when they work. When they stop working, you regret having given them your money. They'll treat you like a pile of dung. No worries with EVGA.

 

You pay a significant premium for this, so depending on the usage patterns realized reliability of MSI/Asus, the better customer care may or may not be worth it (for me EVGA was out right away with just one PCIe slot on the Kingpin). For a hardcore OCer likely to fry things, paying the premium might make sense (but such customers are generally terrible business for EVGA).

 

At this point, the overall EVGA situation needs to be taken into account as well, unfortunately. Looks a bit like they are closing shop or scaling down significantly. No Z790 mobo for instance, their flagship PSU SuperNova T2 is marked as EOL by some retailers, no new model is advertised etc.

[johnksss]

 

Damn, might have to drop igor down a few notches on my list...LOL

[Mr. Fox]

26 minutes ago, Etern4l said:

 

You pay a significant premium for this, so depending on the usage patterns realized reliability of MSI/Asus, the better customer care may or may not be worth it (for me EVGA was out right away with just one PCIe slot on the Kingpin). For a hardcore OCer likely to fry things, paying the premium might make sense (but such customers are generally terrible business for EVGA).

 

At this point, the overall EVGA situation needs to be taken into account as well, unfortunately. Looks a bit like they are closing shop or scaling down significantly. No Z790 mobo for instance, their flagship PSU SuperNova T2 is marked as EOL by some retailers, no new model is advertised etc.

No, not really compared to other equivalent product. EVGA is actually not only better, but more affordable than the top end MSI and ASUS motherboards, that are grossly overpriced and very unreliable. You can't compare apples to oranges.  I'm just not OK with paying $700+ for a motherboard that takes 4 to 6 weeks (or longer) to get replaced under warranty when it fails. Companies that handle things that way suck more than words can say.

1 hour ago, tps3443 said:

I kinda started feeling that way over the EVGA motherboards lately. little issues here and there. Nothing too crazy.

The only EVGA product that I never liked was the Z590 Dark. It was also the only EVGA product I ever owned that had issues. On the basis that your most recent experience was with Z590 I would have to agree. Z590 was kind of an abortion for Intel in general.

[Clamibot]

1 hour ago, Papusan said:

The future is less of P-cores. Will AMD follow Intel and offer less performance cores and offer 75% of the cores as baby cores as well? Don't games need more than 6 real cores in the future or/will small E-cores be better for gaming? The transition to trash cores continue... And why go down to 6 P-cores then add 2 more for the refresh on same socket? A nice way to milk/offer an performance upgrade path? Remove cores, then add them back to offer something new. Almost as Dell did with their gaming laptops @Ashtrix+++

 

Based on the leaked slides, it looks like Intel's 14th Gen Meteor Lake-S Desktop CPUs will take a step back and reduce the number of P-Cores while retaining the number of E-Cores. The lineup listed in the charts includes five different SKU configurations with the top variants featuring up to 22 cores in a combination of 6 P-Cores and 16 E-Cores. The P-Cores on the Meteor Lake CPUs are based on the brand-new Redwood Cove architecture while the E-Cores will utilize the Crestmont design.

Intel Meteor Lake-S CPUs To Feature 6 Performance Cores, Arrow Lake-S With 8 P-Cores On LGA 1851 Socket

 

 

And more about Nvidia's disaster.

Jon Gerow, aka Jonny Guru (Director of Research Development @ Corsair), has posted a write-up about the 12VHPWR connector. 

 

 

SO, YOU’RE GOOD WITH THE 12VHPWR CONNECTOR?

Unfortunately, as of this writing (10/29/2022) Nvidia’s included adapter uses solder, some terminal I’ve never seen before (it has two splits instead of one) and I don’t know what else. Which turns an otherwise excellently designed connector into a disaster.

 

IS THAT IT?
Probably not.  Let’s consider this a living document as we find out more and more about the new 12VHPWR connector and its potential issues.

 

Oh joy! Intel has gone full retard as well! Who's next?

 

It's like every company whose products I like are one by one going full retard in an effort to make me not want to buy anything from them anymore.

[Mr. Fox]

5 minutes ago, Clamibot said:

 

Oh joy! Intel has gone full retard as well! Who's next?

 

It's like every company whose products I like are one by one going full retard in an effort to make me not want to buy anything from them anymore.

Easy enough to understand. One retard begats another. When the vast majority of the people buying your products are as stupid as a box of rocks, you can get away with a lot of unacceptable behavior and you don't have to be the sharpest tool in the shed to know that they're not about to go an extra mile for anyone. The modern approach is to do just barely enough to be acceptable, then back the dial up one notch short of success. But, call it a success because the sheeple will believe you.

[Etern4l]

11 minutes ago, Mr. Fox said:

No, not really compared to other equivalent product. EVGA is actually not only better, but more affordable than the top end MSI and ASUS motherboards, that are grossly overpriced and very unreliable. You can't compare apples to oranges.  I'm just not OK with paying $700+ for a motherboard that takes 4 to 6 weeks (or longer) to get replaced under warranty when it fails. Companies that handle things that way suck more than words can say.

 

Right away there are some problems. First of all, it's often difficult to find exactly equivalent motherboards across brands. For instance, what is this "entry level" $630 single PCIe slot Kingpin equivalent to? I can't see anything in MSI offering for instance.

If you are saying that the $860 Dark Kingpin is equivalent to the likes of MSI Godlike, then I will take your word for it, since I haven't looked at those boards.

Second of all, do we have any independent reliable (sic!) data on motherboard reliability? I guess not.

 

Anyway, hope EVGA manages to pull through and continue to set industry standards on warranty. More customer choice is always better.

[Mr. Fox]

On 10/30/2022 at 2:28 PM, Etern4l said:

Second of all, do we have any independent reliable (sic!) data on motherboard reliability? I guess not.

My independent experience is reliable enough for me. 100% of my expensive enthusiast ASUS motherboards have experienced catastrophic failures under warranty, including a fire and one that killed the best binned 10900KF I have ever seen when it failed. In every case it took those losers more than a month to replace it. The CPU "wasn't their problem" so that was an even more expensive motherboard when that loss is counted. Coincidentally, I have replaced every ASUS motherboard that failed with an EVGA motherboard that never failed.

 

The closest thing I could identify as an EVGA motherboard failure is the Q-code LED burning out on the Z590 Dark, which was a common issue. EVGA replaced it under warranty swiftly, with no hassles, questions or insinuations that I did something that caused it (like ASUS). I dialed their phone number, spoke to a person (a Californian with English as his native language) and explained the situation, and had an RMA number and pre-paid shipping label the same day. I sold it because it was worthless for Windows 7 (ACPI incompatibilities). Good luck getting anything that remotely resembles acceptable customer service from another brand. 

 

I will say that my inexpensive ASUS gamerboy motherboards (Prime and Strix) have been reliable. Only their enthusiast-grade products have been unreliable, which seems totally counter-intuitive. Other than functional limitations for severe overclocking, which they are not intended for, I have had a great experience with those less expensive mid-range products.

 

My Z690 Dark was $399. I didn't buy it when it was new to market. Early adopters always pay more than things are worth. That applies to products at each price point. EVGA has great sales. My 3060 Ti FTW3 was $200 less from them than I could buy it anywhere else. 

 

I do hope the rumors that they are going out of business are false because they are the only brand I have confidence in. I was sad they stopped building GPUs for selfish reasons. I have purchased their GPUs almost exclusively, and now I will have to settle for something inferior with another brand. I do not believe anything Jensen has to say on the subject is accurate because he lies about everything. I am glad they stopped building GPUs for the reasons EVGA stated they have, and I would believe them before I believed anything Jensen would say. He is going to say whatever he thinks make him and NVIDIA look better, and that's his job. He probably wouldn't keep his job if he were completely honest. 

[Etern4l]

10 minutes ago, Mr. Fox said:

My independent experience is reliable enough for me. 100% of my expensive enthusiast ASUS motherboard have experience catastrophic failures under warrant, including a fire and one that killed the best binned 10900KF I have ever seen when it failed. In every case it took those losers more than a month to replace it. I have replaced every ASUS motherboard that failed with an EVGA motherboard that never failed.

 

The closest thing I could identify as an EVGA motherboard failure is the Q-code LED burning out on the Z590 Dark, which was a common issue. EVGA replaced it under warranty swiftly, which no hassles, questions or insinuations that I did something that caused it. I dialed their number, spoke to a person and explained the situation, and had an RMA number and pre-paid shipping label the same day. I sold it because it was worthless for Windows 7 (ACPI incompatibilities). Good luck getting anything that remotely resembles acceptable customer service from another brand. 

 

I will say that my inexpensive ASUS gamerboy motherboards (Prime and Strix) have been reliable. Only their enthusiast-grade products have been unreliable, which seems totally counter-intuitive. Other that functional limitations for severe overclocking, (which they are not intended for,) I have had a great experience with those less expensive mid-range products.

 

My Z690 Dark was $399. I didn't buy it when it was new to market. Early adopters always pay more than things are worth. That applies to products at each price point. EVGA has great sales. My 3060 Ti FTW3 was $200 less from them than I could buy it anywhere else. 

 

I do hope the rumors that they are going out of business are false because they are the only brand I have confidence in. I was sad they stopped building GPUs for selfish reasons. I do not believe anything Jensen has to say on the subject is accurate because he lies about everything. I am glad they stopped building GPUs for the reasons EVGA stated they have, and I would believe them before I believed anything Jensen would say. He is going to say whatever he thinks make him and NVIDIA look better, and that's his job. He probably wouldn't keep his job if he were completely honest. 

 

I crossed Asus off pretty quickly too due to the Hero Z690 issue and their handling of it. Their prosumer board, forgot the name - Creator something, is clearly more flimsy and less performant than the MEG ACE. 10GbE was the only thing it had going for it, and that too is apparently based on some dodgy chipset. 

 

My concerns about EVGA have nothing to do with what Jensen was saying, unfortunately. I wanted to buy their PSU due to the legendary reliability - suddenly went out of stock/EOL. 

[Reciever]

I have been having some issues getting the memory past 3000Mhz on the z490. Though to be honest I think its mostly out of my ignorance lol. Last time I did some overclocking from the sBIOS, was probably on the evga x58 micro

[Mr. Fox]

6 minutes ago, Reciever said:

I have been having some issues getting the memory past 3000Mhz on the z490. Though to be honest I think its mostly out of my ignorance lol. Last time I did some overclocking from the sBIOS, was probably on the evga x58 micro

The same memory that I always ran at 4000+ or other modules? The XMP profile doesn't work as a starting point?

[Reciever]

30 minutes ago, Mr. Fox said:

The same memory that I always ran at 4000+ or other modules? The XMP profile doesn't work as a starting point?

Same modules. I let them run default to get the CPU OC to 51 first then start working on the memory.

 

Then switched it to XMP 1, no POST, XMP 2, no POST. 

 

Perhaps I have the wrong DIMM slots populated, checking now. Looks like A1B1.

 

From what I am reading here, looks like I should move the DIMMS over 1 position each to A2/B2?