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

[Mr. Fox]

50 minutes ago, electrosoft said:

Those are fantastic results AND you gained a bin on your 4090 as a result even at better temps! Where did you source your PTM7950? A lot of counterfeit sources on the market at the moment unfortunately.

Thank you. Yes, I was pleased with the result. I have purchased it twice from this Amazon seller. 

 

https://www.amazon.com/gp/product/B0D8W47P21

 

I can't prove whether it is genuine or not, but it works like it is genuine. I used it on the air cooler installed on the X79 system I gave to my son-in-law because of its long-term durability and when I pulled it back from 4.7GHz all core to 4.5GHz all core the temps were better than I expected them to be. Still not as good as 4.7GHz all core with the EK CR360 Nucleus Dark AIO, but still reasonable. About 85°C max core temp running Cinebench R23 at 4.5GHz all core with the air cooler.

 

[tps3443]

1 hour ago, electrosoft said:

 

Segmenting them into two classes of loading asset stutter (SSD stutter, texture loading/shifting stutter), they are a similar problem but I see where you're coming from.

 

But it does make me even more curious about using a P5800X would eliminate the problem and if so then using one in other games to see if the SSD is partially the culprit. The problem I can think of is even with a P5800X, if you're running very high refresh gaming (let's say 60fps or greater), even a P5800X won't be able to compensate. You would have to find the sweet spot to lock your fps to fluidly and seamlessly traverse from one location to the next. Faster the random I/O,/ms the higher the fps you can lock.

 

This is still the basics and dream of M$ DirectStorage but you need a competent SSD/technology to implement it properly.

 

@tps3443 you need to order one and see where the sweet spot falls. You could also inversely continue to lock your fps lower and lower till you find the sweet spot with your current SSD and the problem goes away.

 

 

Next NEXT NEXT generation then hmmm? 🤣

 

 

Those are fantastic results AND you gained a bin on your 4090 as a result even at better temps! Where did you source your PTM7950? A lot of counterfeit sources on the market at the moment unfortunately.

 

 

Honestly, even just a cheaper Optane 905P will probably fix it, people swear by them fixing the skips in games, just because how brutally fast that random access is and not relying on a cache to be fast. There is a video somewhere of a guy demonstrating this problem in “Kingdom Come deliverance” and some other game, one of the worst games for this loading issue. He was riding through on his horse and GPU usage was dropping like crazy using a regular M.2, while on the Optane it stayed steady and I think this was a much older Optane 900p, ever since seeing this I believed that it would fix the issue.
 

I will order a 905p, and see if it fixes the problem. 😂 I’m going to just install the drive and copy the Black Myth game to it. Fingers crossed. I’m about 98% confident it won’t drop GPU usage anymore. 
 

If it does fix it with just a 905P which is only Gen3 NVME , I will dedicate getting a P5800X/P5801X (400/800gb) for just games I play 🙂 lol. Those are a little pricey though. But worth it if Xpoint really does resolve this. 

[Mr. Fox]

9 hours ago, Mr. Fox said:

I saw the hotspot temp creeping up on one of my 4090 GPUs and decided to do something about it.

 

@tps3443 you had asked about PTM7950 for a desktop CPU and it's not great. But it does work very well for GPUs.

 

See my "before" GPU temps below.

 

 

I used PTM7950 (pad) on the 4090 core. The thermal paste (Alphacool Apex) that was on it was very dried out and that explains the core temps being a little bit elevated. But, the hotspot was what concerned me most. When I took it apart I saw all of the thermal pads were nearly cut in half. The block came with 1.0mm pads. I replaced them with 0.5mm pads to improve die contact pressue and everything across the board improved. Core, hotspot and memory junction temperatures.

 

I found an easy way to remove the film on the PTM7950, which is always a bit tedious and often rips the pad. Put it in the freezer, then cut it and peel off one side. Stick it to the GPU core, then put the GPU in the freezer for about 10 minutes. Then the top film peels off easily without sticking or ripping the paper thin PTM7950 pad. I stick Scotch tape to the film and find that it peels the film off effortlessly with no damage to the PTM7950 pad. The melting point is so low and the pad so thin that you need it very cold and hard to stop the film from sticking to it and tearing up the pad.

 

I used these 12.8 W/mK thermal pads. https://www.amazon.com/dp/B09DC772PR 

 

Now look at the GPU temps. Huge improvement. Hotspot alone is 15°C cooler now, with a 10°C delta rather than a 17°C delta compared with the core temperature.

 

 

Now that my office has heated up to 82°F (28°C) I thought I would check this again. Looks like it's still fine. I am going to call this fixed. No more hotspot concern. Good to go.

[electrosoft]

9 hours ago, tps3443 said:

Honestly, even just a cheaper Optane 905P will probably fix it, people swear by them fixing the skips in games, just because how brutally fast that random access is and not relying on a cache to be fast. There is a video somewhere of a guy demonstrating this problem in “Kingdom Come deliverance” and some other game, one of the worst games for this loading issue. He was riding through on his horse and GPU usage was dropping like crazy using a regular M.2, while on the Optane it stayed steady and I think this was a much older Optane 900p, ever since seeing this I believed that it would fix the issue.
 

I will order a 905p, and see if it fixes the problem. 😂 I’m going to just install the drive and copy the Black Myth game to it. Fingers crossed. I’m about 98% confident it won’t drop GPU usage anymore. 
 

If it does fix it with just a 905P which is only Gen3 NVME , I will dedicate getting a P5800X/P5801X (400/800gb) for just games I play 🙂 lol. Those are a little pricey though. But worth it if Xpoint really does resolve this. 

 

Agreed, if this can help with TS, I am real curious to see how this would help out the chunk/fps drop in WoW in similar situations. Where does asset loading stutter via SSD end and asset loading stutter via texture loading in memory begin?

 

I saw a few 800GB models for right around 1k but if it really helps smooth out game play I'd contemplate making it my game drop for select games that benefit from it especially WoW and raiding.

[tps3443]

3 minutes ago, electrosoft said:

 

Agreed, if this can help with TS, I am real curious to see how this would help out the chunk/fps drop in WoW in similar situations. Where does asset loading stutter via SSD end and asset loading stutter via texture loading in memory begin?

 

I saw a few 800GB models for right around 1k but if it really helps smooth out game play I'd contemplate making it my game drop for select games that benefit from it especially WoW and raiding.

Yep absolutely! Plus you pretty much can’t kill one, so deleting and installing games is ideal. They can be written and erased. nearly and unlimited amount of times, it is exactly what they are designed for, handling some ridiculous amount of numbers and data in realtime all day long for 5+ years straight. They do use a lot of power compared to traditional storage. The P5800X uses about 20-25 watts of power. I have seen people use over 30 watts with a P5801X on the Chinese YouTube videos. They also get pretty warm when under load. Not like these little baby 2 watt M.2 drives we have 🤭

[Prema]

[\SPOILER]

[Papusan]

Enjoy🙂

 

 

And the day have come.... From worse to worse for computer enthusiasts (water cooling) in Norway. Now we have only one real computer shop here home, but for the most part offer AIOs, bling bling and gaming gears for gamer-kids. And a couple of pc web shops where you have to wait for ordered parts 2-4 weeks. The best one is now out of business. This is just sad☹️  The enthusiast shop Digital Impuls is bankrupt

[electrosoft]

Looks like Intel gets some uplift in spots, but overall this is shaping up to be a monster win for AMD gaming. Those are some insane uplifts in pockets and an overall 10% on average uplift with no regression in performance elsewhere.

 

BS about 9000 vs 7000 aside (shame AMD), this is shaping up to be a rather juicy update and looking forward to the overall Intel v AMD gaming suite comparisons.

 

 

[Papusan]

8 hours ago, electrosoft said:

Looks like Intel gets some uplift in spots, but overall this is shaping up to be a monster win for AMD gaming. Those are some insane uplifts in pockets and an overall 10% on average uplift with no regression in performance elsewhere.

 

BS about 9000 vs 7000 aside (shame AMD), this is shaping up to be a rather juicy update and looking forward to the overall Intel v AMD gaming suite comparisons.

 

 

 

Here's the test with Windows 11 23H2 (new released Windows update KB5041587 fix for Ryzen gaming performance). Not as good for the Ryzen 9 9950X.

 

I really hope for the Red fanboys that Microsoft don't have to withdraw some of the performance gain later, because Ryzen chips is at risk become more vulnerable to security issues from this new optimization patch. Nothing will last forever🙂

Zen 5 tested with Windows Update KB5041587

 

 

 

[tps3443]

29 minutes ago, Papusan said:

 

Here's the test with Windows 11 23H2 (new released Windows update KB5041587 fix for Ryzen gaming performance). Not as good for the Ryzen 9 9950X.

 

I really hope for the Red fanboys that Microsoft don't have to withdraw some of the performance gain later, because Ryzen chips is at risk become more vulnerable to security issues from this new optimization patch. Nothing will last forever🙂

Zen 5 tested with Windows Update KB5041587

I am seeing some nice gains on 23H2 and 24H2 for the Ryzen guys. So I am trying this with my 14900KS system now.

[Mr. Fox]

34 minutes ago, Papusan said:

 

Here's the test with Windows 11 23H2 (new released Windows update KB5041587 fix for Ryzen gaming performance). Not as good for the Ryzen 9 9950X.

 

I really hope for the Red fanboys that Microsoft don't have to withdraw some of the performance gain later, because Ryzen chips is at risk become more vulnerable to security issues from this new optimization patch. Nothing will last forever🙂

Zen 5 tested with Windows Update KB5041587

It is unfortunate and inappropriate that hardware performance would be contingent or reliant upon the OS. Just the thought of that sucks and makes me angry. Windows needs to stick to being an interface with the hardware and a platform for running other software, and nothing more than that.

[tps3443]

6 minutes ago, Mr. Fox said:

It is unfortunate and inappropriate that hardware performance would be contingent or reliant upon the OS. Just the thought of that sucks and makes me angry. Windows needs to stick to being an interface with the hardware and a platform for running other software, and nothing more than that.

I have given up on it. I used to not update and stick to one old build. But it's always worse performance for me. I am going to test 24H2. I am going to assume it will probably provide a faster 14900KS. There is always a trade-off it seems. 

[tps3443]

The minimum FPS is better. I could never finish Heaven with 61.8 min before. The max is also slightly better. I ran this test a gazillion times before updating Windows 11, and I never could hold 60+ min through the entire test before. This test is great for checking your GPU OC is stable and smooth on the frame times. Because any overclock too high on the GPU, will harm your minimum frame rate in Heaven. So, if you have a low min FPS in Heaven benchmark, your GPU OC is too high for it to manage smooth frame times. 
 

Anyways, here is before and after. Same everything besides new windows update. 
 

Both are on bios Z790 Apex A04 (9901) 8600c36-49-49-49-63 (262k tREFI)

 

[Mr. Fox]

2 hours ago, tps3443 said:

I have given up on it. I used to not update and stick to one old build. But it's always worse performance for me. I am going to test 24H2. I am going to assume it will probably provide a faster 14900KS. There is always a trade-off it seems. 

I am still working on transitioning to Linux. I do not find the direction Micro$lop is headed to be acceptable to me and I am approaching the point where I would rather just do without some things than compromise and embrace their feces. The reason each release of Winduhz sucks worse than the one before is because the zombie sheeple put up with it and it is only going to get worse. I think ripping the band-aid off and moving on is a better idea. We often misuse the word "hate" as an adjective to express disagreement or disapproval for something, but I don't think it is an embellishment or improper use of the term to describe my opinion of Micro$lop at this point. I find it hard to hold anything other than contempt for them and I no longer care whether they remain solvent and successful because I don't need them and view them as being expendable. I use it for work because I have to, but Windoze isn't important to me anymore.

[tps3443]

@electrosoft Can you run this with your 7800X3D system? It is 720P Low Settings. Using Windows Update 24H2

[Talon]

4 hours ago, tps3443 said:

I am seeing some nice gains on 23H2 and 24H2 for the Ryzen guys. So I am trying this with my 14900KS system now.

 

I've been testing 24H2 ever since news of the performance uplift for AMD and some for Intel. I put it on an external USBC drive/install. No issues were noted and I picked a tad bit of performance in some apps and games. Gears 5 was the only one that showed major improvements for me so far. But overall the OS just felt snappier, smoother animations/transitions. It just felt less laggy compared to 23H2. I just installed it on my internal drive, did a long needed OS wipe this afternoon. Getting everything up and running now. 

[tps3443]

11 minutes ago, Talon said:

 

I've been testing 24H2 ever since news of the performance uplift for AMD and some for Intel. I put it on an external USBC drive/install. No issues were noted and I picked a tad bit of performance in some apps and games. Gears 5 was the only one that showed major improvements for me so far. But overall the OS just felt snappier, smoother animations/transitions. It just felt less laggy compared to 23H2. I just installed it on my internal drive, did a long needed OS wipe this afternoon. Getting everything up and running now. 

My OS is getting a bit old as well. Mine is right at 8 months old. Probably dragging around a lot of extra processes and services. 

[Tenoroon]

11 hours ago, electrosoft said:

Looks like Intel gets some uplift in spots, but overall this is shaping up to be a monster win for AMD gaming. Those are some insane uplifts in pockets and an overall 10% on average uplift with no regression in performance elsewhere.

 

BS about 9000 vs 7000 aside (shame AMD), this is shaping up to be a rather juicy update and looking forward to the overall Intel v AMD gaming suite comparisons.

 

 

Am still on Win10 21H2. I hope this performance hitch is only in newer Win11 versions. I refuse to move forward with this already garbage OS!

I've been reinstalling Windows all day today on my G14 because once it connects to the internet, it auto-installs all of ASUS's bloat and garbage... I have to install everything I need offline and then disable a bunch of stuff within Windows before I can connect to the internet "safely." And this is still 21H2. What a catastrophe... ASUS and Microsoft are never seeing a dollar from me ever again. Too much hassle.

I know people here have advised me against it, but I will be trying Gentoo within the next couple of days on my desktop. My goal is to move over as fast as possible as I'm sick of Windows. I've got friends who can assist me, and a lot more free time than I thought I would have here at University, so I am hoping to be proficient with it before the school year ends.

[chew]

On 8/26/2024 at 2:10 PM, Mr. Fox said:

I am not sure that Jay or the two Steves actually know what they are doing. If they are trying to test things on BIOS defaults like an average gamerboy, that will always suck no matter what brand of CPU. It seems clear the new gen Ryzen is a different product than the one it is replacing. It's too bad they are still pairing an inferior CCD with a better one instead of passing the inferior silicon down to one of cheaper CPUs. That drove me nuts with the 5950X and I felt the same way about it that Jufus the Dufus expressed.

 

Maybe Brothers @Raiderman or @chew will get one and show us what a 9950X really looks like since they both know how to properly tune a non-X3D Ryzen CPU.

At best you will validate it. At worst you will invalidate it. Either way you will have an insanely fast obsolete SSD that costs more that it is worth. 🤣

 

been there done that already oced all skus.

 

Jufes being jufes.

 

6gig on ambient delidded. Is it april 1st? 

 

Maybe i suck but not even happening at not blow em up volts on 9950 x even at -35.

 

one avx 12 app cuts cb23 down by 200mhz and cb23 doesn't run at 6.0.....

[Mr. Fox]

2 hours ago, chew said:

Maybe i suck but not even happening at not blow em up volts on 9950 x even at -35.

No you do not. Quite the opposite. Sounds like 9950X might though.

[Clamibot]

I'm still on Windows 10 LTSC V1809. It's held up incredibly well and there are minimal compatibility issues with new software at the moment. As far as I can see, just Unreal Engine 5 titles and Starfield so far.

 

I have so many games I can play through that the incompatibility with a few newer titles doesn't matter, and I'll only be updating if I get a performance uplift or if performance remains the same. I've benchmarked Windows 11 vs Windows 10 LTSC 1809. I get about a 20% uplift in framerate with LTSC V1809 vs Windows 11 across most of my games.

 

I don't know how people are getting better performance in Windows 11 vs Windows 10. My benchmarking says otherwise. To be fair, they're probably comparing against standard Windows 10 rather than LTSC, but even then I wouldn't expect Windows 11 to perform better.

[Mr. Fox]

15 minutes ago, Clamibot said:

I don't know how people are getting better performance in Windows 11 vs Windows 10.

Yeah, that has me stumped as well. Especially when it comes to CPU performance. It's so bloated and burdened with worthless crap, processes and services that serve no legitimate purpose as far as I'm concerned.

[tps3443]

1 hour ago, Clamibot said:

 

I don't know how people are getting better performance in Windows 11 vs Windows 10. My benchmarking says otherwise. To be fair, they're probably comparing against standard Windows 10 rather than LTSC, but even then I wouldn't expect Windows 11 to perform better.

Windows 11 24H2 performs really well. I'm too obsessed with performance as is. What CPU are you using anyways? 

[Clamibot]

2 hours ago, tps3443 said:

Windows 11 24H2 performs really well. I'm too obsessed with performance as is. What CPU are you using anyways? 

A 10900K, both on my desktop and laptop.

[Mr. Fox]

5 hours ago, Clamibot said:

I don't know how people are getting better performance in Windows 11 vs Windows 10.

 

4 hours ago, Mr. Fox said:

Yeah, that has me stumped as well. Especially when it comes to CPU performance. It's so bloated and burdened with worthless crap, processes and services that serve no legitimate purpose as far as I'm concerned.

Dumbed-down caterers to the dumbest of the lowest common denominators. A one-size fits all hat that only actually fits people with a shrunken head. The don't actually care about Windoze anymore, or the people that use it. They just want to sell you their other crap that only runs on Winduhz.

---

Jay did a nice job on this video and it should be helpful to noobs. His description of the qualities and behaviors of the thermal pastes he compared matches my experiences with each of them.