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

[Mr. Fox]

7 hours ago, Clamibot said:

 

Solid proof of why high speed, low latency RAM does in fact matter a lot in gaming. Even with the amount of 3D vcache these X3D CPUs have, it's still possible to exhaust that cache many times over in some games. This is exactly the reason Intel CPUs scale so well with fast memory; it helps mititgate the shortcomings of the cache on Intel's CPUs vs AMD's X3D CPUs (in regards to cache size that is).

Intel cores run at MUCH higher clock speeds stock, overclock better and the memory latency is lower, along with data transfer speeds that are much higher. Or, should we say used to. Then came the Intel Core Ultra chiplet abortion and flushed it all down the toilet. Like a marathon runner that has a decent lead in a race deciding to stab himself in the calf with an ice pick just for the hell of it. 

[electrosoft]

12 hours ago, Clamibot said:

 

Solid proof of why high speed, low latency RAM does in fact matter a lot in gaming. Even with the amount of 3D vcache these X3D CPUs have, it's still possible to exhaust that cache many times over in some games. This is exactly the reason Intel CPUs scale so well with fast memory; it helps mititgate the shortcomings of the cache on Intel's CPUs vs AMD's X3D CPUs (in regards to cache size that is).

 

Agreed. There is a widely held misnomer that tuning isn't really required with X3D and that is a fallacy especially when the cache is exhausted/saturated.

 

World of Warcraft has been shown to really benefit from tight timings vs bandwidth over and over when things are said and done.

 

Of course if you can have both then why not? 🙂

 

4 hours ago, Mr. Fox said:

Intel cores run at MUCH higher clock speeds stock, overclock better and the memory latency is lower, along with data transfer speeds that are much higher. Or, should we say used to. Then came the Intel Core Ultra chiplet abortion and flushed it all down the toilet. Like a marathon runner that has a decent lead in a race deciding to stab himself in the calf with an ice pick just for the hell of it. 

 

*Sigh* hopefully the chips act and the fact Intel can't sell their foundries along with the removal of Pat along with 18a in the wind will get them back on track.

 

@Talon How has the 4090 testing been going with the 9800X3D vs 14900KS vs 285k?

 

 

[tps3443]

Getting ready to test the new Xeon w3175X. This thing was brand new and sealed. Just opened it up. 
 

Shall the best Xeon win! This will be easy though, because I know my chip well by this point. 
 

[Mr. Fox]

45 minutes ago, tps3443 said:

Getting ready to test the new Xeon w3175X. This thing was brand new and sealed. Just opened it up. 
 

Shall the best Xeon win! This will be easy though, because I know my chip well by this point. 
 

---

For the AMD owners here... give me a shout if you need to do this. Similar to the issue of getting a new Intel ME cancer payload with a BIOS update, it doesn't go back to the old AMD SMU with a flash downgrade, but there is a way of doing it. (Got the info from Svet at MSI forum and have tested it three times with success.)

[Clamibot]

Since oddly Cinebench R15 refuses to use both the P cores and E cores on my 14900KF at the same time, I ended up doing a P core only run and an E core only run by manually setting processor affinity each time to those respective core groups.

 

The P core only run is the one in orange and the E core run is the one in brown. Looks like 8 P cores is more powerful than 16 E cores.

[Papusan]

58 minutes ago, Clamibot said:

Since oddly Cinebench R15 refuses to use both the P cores and E cores on my 14900KF at the same time, I ended up doing a P core only run and an E core only run by manually setting processor affinity each time to those respective core groups.

 

The P core only run is the one in orange and the E core run is the one in brown. Looks like 8 P cores is more powerful than 16 E cores.

 

Hmmm. Should work. Is this done with stock clocks? If not then try default clock speed.

[Clamibot]

2 hours ago, Papusan said:

 

Hmmm. Should work. Is this done with stock clocks? If not then try default clock speed.

 

I just tried that and got the same behavior. Cinebench R15 was adamant that it only ran on one set of cores. I can get it to run on either P cores or E cores, but not both types at the same time.

[electrosoft]

The king is dead! Long live the king!

 

Boxed up my Liquid 4090 to ship off to the new owner tomorrow so I snagged a Ventus 3x 4070 Super locally from best buy to hold me over for the ~58 days. The Liquid X gave me almost two years of rock solid service and I can honestly say it has been my most favorite graphics card to own even more than the Kingpin 3090. Just a beauty of a card.

 

I'm still holding on at 4k Ultra in WoW but barely. I had to begrudgingly turn off RT and there are some wicked dips into the 50s even with RT off so I suspect I'll be inching that Ultra 10 down to a consolation 8 sooner than later. 🙂

 

**Bring on the 5090!!**

 

So adorable..... 🤣

 

No coil whine. It might be the quietest gpu I've own ever. Closest would be the KPE 3090ti. That was ridiculously quiet.

 

WoW went from being rarely being pegged at 99% GPU to almost all the time 99%.

 

 

[Mr. Fox]

4 hours ago, Clamibot said:

Since oddly Cinebench R15 refuses to use both the P cores and E cores on my 14900KF at the same time, I ended up doing a P core only run and an E core only run by manually setting processor affinity each time to those respective core groups.

 

The P core only run is the one in orange and the E core run is the one in brown. Looks like 8 P cores is more powerful than 16 E cores.

I have never had that issue before. Are you using one of the saved BIOS profiles or testing your own? You should be getting like 43-44K easily in CBR23 with one of saved profiles unless it is just getting too hot. Unless there is some kind of error elsewhere causing Cinebench to malfunction.

https://hwbot.org/submission/5546221

https://hwbot.org/submission/5546469

  

27 minutes ago, electrosoft said:

The king is dead! Long live the king!

 

Boxed up my Liquid 4090 to ship off to the new owner tomorrow so I snagged a Ventus 3x 4070 Super locally from best buy to hold me over for the ~58 days. The Liquid X gave me almost two years of rock solid service and I can honestly say it has been my most favorite graphics card to own even more than the Kingpin 3090. Just a beauty of a card.

 

I'm still holding on at 4k Ultra in WoW but barely. I had to begrudgingly turn off RT and there are some wicked dips into the 50s even with RT off so I suspect I'll be inching that Ultra 10 down to a consolation 8 sooner than later. 🙂

 

**Bring on the 5090!!**

 

So adorable..... 🤣

 

WoW went from being rarely being pegged at 99% GPU to almost all the time 99%.

That iTX motherboard and 4070 look SO TINY in that big Enthoo case, LOL.

 

[tps3443]

10 hours ago, Clamibot said:

 

I just tried that and got the same behavior. Cinebench R15 was adamant that it only ran on one set of cores. I can get it to run on either P cores or E cores, but not both types at the same time.

Cinebench R15 was always troublesome with Intel 13/14 gen chips. It was always the hardest to run of R20/R23/R24. If you can run R15 then the dang thing can run Ycruncher with the same settings. That’s just how it is with the P/E core chips using R15. Sometimes TVB can cause issues, sometimes you may need it on/off. 
 

Just know R15 is tough on these chips. If I can run R23/R24 with 1.225V completely stable. Then that means R15 will need 1.250V+ to run lol. If you wanna get an Intel 13/14th gen stable of it can pass R15, it’ll pass anything. 

[Clamibot]

10 hours ago, Mr. Fox said:

I have never had that issue before. Are you using one of the saved BIOS profiles or testing your own? You should be getting like 43-44K easily in CBR23 with one of saved profiles unless it is just getting too hot. Unless there is some kind of error elsewhere causing Cinebench to malfunction.

https://hwbot.org/submission/5546221

https://hwbot.org/submission/5546469

  

That iTX motherboard and 4070 look SO TINY in that big Enthoo case, LOL.

 

 

I'm using the BIOS profile that was on the motherboard when I received the parts. Also I'm benchmarking using Cinebench R15.

 

1 hour ago, tps3443 said:

Cinebench R15 was always troublesome with Intel 13/14 gen chips. It was always the hardest to run of R20/R23/R24. If you can run R15 then the dang thing can run Ycruncher with the same settings. That’s just how it is with the P/E core chips using R15. Sometimes TVB can cause issues, sometimes you may need it on/off. 
 

Just know R15 is tough on these chips. 

 

Looks like it'll still serve as a good stability test for me then. I've always used Cinebench R15 to test my overclocks. If they pass a benchmark, they're pretty much always stable in games.

[Clamibot]

10 hours ago, Mr. Fox said:

I have never had that issue before. Are you using one of the saved BIOS profiles or testing your own? You should be getting like 43-44K easily in CBR23 with one of saved profiles unless it is just getting too hot. Unless there is some kind of error elsewhere causing Cinebench to malfunction.

https://hwbot.org/submission/5546221

https://hwbot.org/submission/5546469

  

That iTX motherboard and 4070 look SO TINY in that big Enthoo case, LOL.

 

 

I'm using the BIOS profile that was on the motherboard when I received the parts. Also I'm benchmarking using Cinebench R15.

 

Also funnily enough, my system feels significantly more responsive and snappy with the E cores disabled. I haven't done any official benchmarks for this, but it just feels faster.

 

1 hour ago, tps3443 said:

Cinebench R15 was always troublesome with Intel 13/14 gen chips. It was always the hardest to run of R20/R23/R24. If you can run R15 then the dang thing can run Ycruncher with the same settings. That’s just how it is with the P/E core chips using R15. Sometimes TVB can cause issues, sometimes you may need it on/off. 
 

Just know R15 is tough on these chips. 

 

Looks like it'll still serve as a good stability test for me then. I've always used Cinebench R15 to test my overclocks. If they pass a benchmark, they're pretty much always stable in games.

[tps3443]

11 minutes ago, Clamibot said:

 

I'm using the BIOS profile that was on the motherboard when I received the parts. Also I'm benchmarking using Cinebench R15.

 

Also funnily enough, my system feels significantly more responsive and snappy with the E cores disabled. I haven't done any official benchmarks for this, but it just feels faster.

 

 

Looks like it'll still serve as a good stability test for me then. I've always used Cinebench R15 to test my overclocks. If they pass a benchmark, they're pretty much always stable in games.

It’s strange because it is quite the opposite for other chips. Using my 3175X I can run R15 at 5.0Ghz probably lol. But R23 is gonna hit limits at 4.7-4.8Ghz range. So there is something with the instruction used on a 13/14th gen chip that makes R15 the toughest of them all. 
 

 

@electrosoft @Mr. Fox I really had to take a close look to even see a motherboard inside that big case at all!!! Wow. Where’s it at? 🤣

[electrosoft]

51 minutes ago, tps3443 said:

It’s strange because it is quite the opposite for other chips. Using my 3175X I can run R15 at 5.0Ghz probably lol. But R23 is gonna hit limits at 4.7-4.8Ghz range. So there is something with the instruction used on a 13/14th gen chip that makes R15 the toughest of them all. 
 

 

@electrosoft @Mr. Fox I really had to take a close look to even see a motherboard inside that big case at all!!! Wow. Where’s it at? 🤣

 

Everything has shrunk!

 

Asrock Z790i Lightning + SP109 14900KS + TG 8200@8400 sticks + 990 Pro 2TB + MSI 3x Ventus 4070 Super

 

CPU went from gasping for breath to waiting for the 4070 Super to catch up. 🤣

 

@Mr. Fox and @tps3443 on a serious note, I'm looking for the best bang/bunk (aka cheap but performs relatively well) AM4/5 / Socket 1700 CPU block and D5 pump/reservoir combo. Thinking of pulling the AC LF 420 and putting a 420 + 360 Rad on the bottom loop in the case.

 

 

[Mr. Fox]

8 hours ago, electrosoft said:

 

Everything has shrunk!

 

Asrock Z790i Lightning + SP109 14900KS + TG 8200@8400 sticks + 990 Pro 2TB + MSI 3x Ventus 4070 Super

 

CPU went from gasping for breath to waiting for the 4070 Super to catch up. 🤣

 

@Mr. Fox and @tps3443 on a serious note, I'm looking for the best bang/bunk (aka cheap but performs relatively well) AM4/5 / Socket 1700 CPU block and D5 pump/reservoir combo. Thinking of pulling the AC LF 420 and putting a 420 + 360 Rad on the bottom loop in the case.

 

 

The Heatkiller Pro IV CPU waterblocks are amazing. I actually like them as much as the Optimus CPU blocks. My favorite D5 pump/res units are the XSPC Photon (getting hard to find) and the Alphacool units. Both have thick glass tubes, which is the only way to fly. Steer clear of acrylic reservoir tubes as they crack and scratch too easily. The XSPC Photon glass tubes are insanely thick glass and molded with threads on each end that literally screw onto the pump body and end cap on the Photon tubes. They cannot separate over time and leaks are not even a thing. Likewise, the Alphacool Eisbecher Aurora have 4 metal rods to keep the glass tube sandwiched between the top and pump body.  Avoid any that simply rely on rubber o-ring compression like rigid tubing compression fittings. They are bad about leaking and internal loop pressure pushing the top off the tube (Primochill D5 pump/res for example).

XSPC Photon D5

Alphacool Eisbecher Aurora D5

 

[Papusan]

22 hours ago, Clamibot said:

 

I just tried that and got the same behavior. Cinebench R15 was adamant that it only ran on one set of cores. I can get it to run on either P cores or E cores, but not both types at the same time.

 

Try another OS version. Remember the stupids at Microsoft campus have forgot how to make a good working OS. 

 

 

Second stupid modern tech for the day... 

 

LOOL. Now you  need an QVL power supply list. Modern tech strikes again. Corsair's R&D director (PC power supplies) Jonny Guru just shaking his head. 

 

Corsair's power supply R&D director discovered that some graphics cards do not support 300W 16-PIN cables?

 

Yup, nothing is better than loads of 8 pin connectors on your PSU so you can power up any of your modern Nvidia graphics card. No need for an stupid designed native cable with 4+12 pin power connector. More 8-pins connectors, the better. And even more important if you run older GPUs or newer Radeon cards. Many modern so called high end PSU's have all to few of them. 

[Mr. Fox]

16 minutes ago, Papusan said:

Try another OS version. Remember the stupids at Microsoft campus have forgot how to make a good working OS. 

I failed to notice that Brother @Clamibot was having issues specifically with Cinebench R15. I hardly ever run R15 so that might be why I was not connecting the dots. I use R20 and R23 95% of the time. I agree the problem is most likely Windoze being a garbage OS and not the CPU or Cinebench. And, R15 is probably old enough that the stooges at Micro$lop don't care if it works correctly with their modern garbage OSes.

[Papusan]

27 minutes ago, Mr. Fox said:

I failed to notice that Brother @Clamibot was having issues specifically with Cinebench R15. I hardly ever run R15 so that might be why I was not connecting the dots. I use R20 and R23 95% of the time. I agree the problem is most likely Windoze being a garbage OS and not the CPU or Cinebench. And, R15 is probably old enough that the stooges at Micro$lop don't care if it works correctly with their modern garbage OSes.

 

Microsoft can't even make a good OS for processors without stupid E-cores and hybrid mess. And they change power plan settings or power scheduler to their likings to fit their own ARM garbage. All is about power efficiency. Maybe they think half the processors shoud be shut down in work flow. Yup... Disgusting.

 

See. Their changes in the OS is an intended move. All they care about is that it works as intended with their own surface Joke Books.

 

Apple make own OS for tablets. Same also for their desktops and smartphones. Microsoft want it "all in one". Of course it will fail.  

 

[Mr. Fox]

21 minutes ago, Papusan said:

 

Microsoft can't even make a good OS for processors without stupid E-cores and hybrid mess. And they change power plan settings to their likings to fit their own ARM garbage. All is about power efficiency. Maybe they think half the processors shoud be shut down in work flow. Yup... Disgusting.

Maybe we can add to our expressions of disdain for who they are and how they do things by calling them Micro$heist or Micro$cheiße. The older Micro$lop just doesn't seem to be as derrogatory or contemptuous as they deserve.

 

 

[Papusan]

12 minutes ago, Mr. Fox said:

Maybe we can add to our expressions of disrespect for who they are by calling them Micro$heist or Micro$cheiße. The older Micro$lop just doesn't seem to be as derrogatory as they deserve.

 

Microsoft can even make Windows OS worse than it is now. Just start up with phones again. We haven't yet seen the utter bottom from Redmond campus. So it is.......

 

 

[tps3443]

I’m testing the 10900K LTX SP106 currently. And it does really well so far. Or at least I think so. 
 

I’m running all-cores at 5.0Ghz@138 watts max with an AIO during R15 Cinebench.  This cooler doesn’t seem all that great, or maybe it’s not working properly. 🤷‍♂️ Temps are peaking at 92c package still. I’m not all that impressed with the subzero so far, unless I’m using it wrong. I may have to check the mount as well. 
 

This seems kinda toasty for sub 140 watts. 
 

@Clamibot Nevermind. I found the software download url in the manual! It’s working now. Holy crap works so MUCH BETTER… 😃

 

 

[Clamibot]

5.8 GHz P Core only all core run in Cinebench R15:

 

6 GHz P core only single core run in Cinebench R15:

 

Shadow Of The Tomb Raider Benchmark with the 14900Kf at 5.8 GHz all core:

 

I can do 240 fps raw on this system, mwahaha!

[tps3443]

8 minutes ago, Clamibot said:

5.8 GHz P Core only all core run in Cinebench R15:

 

6 GHz P core only single core run in Cinebench R15:

 

Shadow Of The Tomb Raider Benchmark with the 14900Kf at 5.8 GHz all core:

 

I can do 240 fps raw on this system, mwahaha!

Are you using Windows 10? That’s probably why your Cinebench scores are low then and not using the E-Cores. I had that issue before. You need a newer Windows 10, or preferably Windows 11. I remember this happen to me when I first upgraded to 13900K on launch. 

[Clamibot]

18 minutes ago, tps3443 said:

I’m testing the 10900K LTX SP106 currently. And it does really well so far. Or at least I think so. 
 

I’m running all-cores at 5.0Ghz@138 watts max with an AIO during R15 Cinebench.  This cooler doesn’t seem all that great, or maybe it’s not working properly. 🤷‍♂️ Temps are peaking at 92c package still. I’m not all that impressed with the subzero so far, unless I’m using it wrong. I may have to check the mount as well. 
 

This seems kinda toasty for sub 140 watts. 
 

@Clamibot anything I need to do to enable this CM ML360 subzero is there a command center for it?  it’s all hooked up correctly. PCIe 8 pin, USB connection, and all of that. 

 

 

 

Yes, the cooler sucks until you enable the TEC. There are 2 different control center applications you can use for it. You can either use the Intel official Cryo Cooling software: https://www.intel.com/content/www/us/en/download/715177/intel-cryo-cooling-technology-gen-2.html

 

Or you can download the open source modded version that doesn't contain the stupid artificial CPU model check, which allows the TEC to work with any CPU: https://github.com/juvgrfunex/cryo-cooler-controller

 

I prefer the open source version.

 

Once you get either one of these applications set up along with the required drivers to make them work, just enable the TEC and watch your CPU go brrrrr (literally, the CPU can get pretty cold).

 

If you use the Intel official version, I'd recommend setting it to Cryo mode for daily driving, and enable unlimited mode only when benchmarking in short bursts.

 

If you use the open source version, setting the Offset parameter to positive 2 is equivalent to running Cryo mode on the Intel official software, and setting that Offset parameter to a negative number is equivalent to running Unlimited mode on the Intel official software.

 

Once you enable the TEC, the cooler is good for a CPU power draw up to about 200 watts before you start thermal throttling. I also used liquid metal between the CPU die and IHS, and also between the IHS and the cooler coldplate nozzle.

 

Note, DO NOT use liquid metal between the IHS and coldplate on the MasterLiquid ML 360 Sub Zero if your IHS is pure copper, use nickel plated IHSes only! If you use a pure copper one like I did, the liquid metal will weld both copper surfaces together over time with the TEC active as the colder temperatures from the TEC coldplate and the heat from the CPU accelerate the absorption of the gallium from the liquid metal into both copper surfaces, and eventually bonds the copper surfaces together.

 

Alternatively, you can pre treat both copper surfaces with liquid metal to prevent this from happening. Stupid me, now I can't get the cooler off my golden 10900K. It really needs a repaste.😭

 

I was able to do 5.4 GHz all core in games on this thing using the TEC, 5.6 GHz single core boost on my particular chip. It can hold all core speeds over 5.4 GHz indefinitely as long as the temperature is kept below 140°F (60°C).

[Clamibot]

8 minutes ago, tps3443 said:

Are you using Windows 10? That’s probably why your Cinebench scores are low then and not using the E-Cores. I had that issue before. You need a newer Windows 10, or preferably Windows 11. I remember this happen to me when I first upgraded to 13900K on launch. 

 

I'm using Windows 10 22H2. I'm specifically using the WindowsXLite version for maximum performance.

 

Cinebench R15 is able to use the E cores, just not at the same time as the P cores. I can make it use either core types, just not at the same time.