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

[electrosoft]

14 hours ago, Mr. Fox said:

Sweet! Nice to see Intel opening a can of whoopass. I wonder if there will be a 290K Plus? That would be the one I would want. It would need to be a 9950X beater in all core workloads.

Unless something changes in a big way with Zen 6, my next build will most likely be a return to Intel. Will see what happens when it happens and go from there. I think the 9950X is a very good CPU, but I do not like how many hard-coded functional limitations there are with AMD CPUs and the X3D CPU features don't really offer me anything important to me since I am not particularly obsessed with gaming.

 

They opted to skip the slated 290k Plus and just stick with the 250k and 270k. Seeing as the 270k is basically a 285k but depending on bins will be better and the 290k was literally the 285k just pushed a bin up, Intel actually did the right thing this time coming in at killer price points on both the 250k and 270k.

 

Technically, 285k has TVB but it is a non starter really for us.

 

And just like that, the 9600x drops to ~$182 a few days before the 250k launches.....

 

Mobile wise, I'm curious to see what 290hx plus brings to the table as it is rumored to have 10-15% more performance than the 275/285hx variants but that will come down to the supporting rig around the engine as always. Difference between the 275hx in my Acer Neo 16s and the 275hx in my Alienware 18 is night and day.

---

 Zen 6 is rumored to have a 24 core variant though and if we get that, double X3D AND what is rumored to be significant fabric interconnect latency improvements, it could be a killer chip too. 

 

But Nova is going to be a brand new architecture with its own 3d caching system and I am expecting absolutely killer performance out of it too. Unlocked with absolutely no boundaries and a ton of cores, it has pulled ~600-700w in testing. Sounds right up your alley @Mr. Fox!

 

 

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

 

Apple's new M5 Max chip is a decent little bump CPU wise but GPU wise it is a pretty substantial upgrade....

 

M1->M5 performance progression in just the Macbook 16" Max models. I imagine the slated M5 Ultra is going to be a beast especially now with M5 the CPU and GPU are own their own, individual modules with Apple having solved the high speed interconnect issues and allowing the same shared memory. This bodes well for future configs too for really building out Macs with multiple GPUs.

 

AI performance is absolutely through the roof. I suspect a lot of M5 Ultra boxes are going to be snapped up for AI work.

 

CPU:

 

 

GPU Raster:

 

 

GPU RT:

 

 

[Reciever]

So now I'm looking at one of the following 

 

7800 xt, 7900 xt, gre, 9070, xt, possible 7900 xtx

Also looking to pick up an x570-ace for dual GPU scenario in tandem with losses scaling. 

 

Only reason for entertaining a second 7900 xtx is in the faint hope I can get one with better memory overclocking. Seems to be where people are able to hit gold. 

 

At that point I'll pick up a waterblock for the 5800 X3D also, I'll make it easier on my fingers digging into the system time and again. 

 

Also tried booting some of the Gpu's I have but don't seem to be porting lol just my luck

[win32asmguy]

On 3/24/2026 at 11:31 AM, electrosoft said:

Mobile wise, I'm curious to see what 290hx plus brings to the table as it is rumored to have 10-15% more performance than the 275/285hx variants but that will come down to the supporting rig around the engine as always. Difference between the 275hx in my Acer Neo 16s and the 275hx in my Alienware 18 is night and day.

 

I think its supposed to be 30x NGU and D2D, 40x Ring. My 275HX in the Hydroc G2 can do that easily with Premamod, but the real advantage comes from memory tuning. I have seen a few post sub-80ns AIDA64 scores, but I find heat is a big limitation on combined load stability so realistically for me 85ns is the best it can do on air with reasonable ambient temperatures.

 

I think the A51 18 could do better temperature wise with the memory as the modules sit on the side opposite of the vapor chamber. You would have to use Smokeless UMAF to modify timings and test. Sadly max MT would be somewhat limited compared to even a mediocre desktop board as you cannot adjust VDD2 up beyond 1.1v (except on the Hydroc G2 with Premamod).

[Mr. Fox]

2 hours ago, Reciever said:

So now I'm looking at one of the following 

 

7800 xt, 7900 xt, gre, 9070, xt, possible 7900 xtx

Also looking to pick up an x570-ace for dual GPU scenario in tandem with losses scaling. 

 

Only reason for entertaining a second 7900 xtx is in the faint hope I can get one with better memory overclocking. Seems to be where people are able to hit gold. 

 

At that point I'll pick up a waterblock for the 5800 X3D also, I'll make it easier on my fingers digging into the system time and again. 

 

Also tried booting some of the Gpu's I have but don't seem to be porting lol just my luck

My vote would be for the 9070 XT. So far it is AMD's best GPU in terms of performance.

[Reciever]

1 hour ago, Mr. Fox said:

My vote would be for the 9070 XT. So far it is AMD's best GPU in terms of performance.

Forgive me I typed it all out on my phone at break but this second GPU would strictly be just for Frame Gen mule under lossless scaling, my 7900 XTX would still be my raster, at 3Ghz its pretty good for me 24/7. I dont care about power savings, but I do concern myself with the 12VHPWR which is why Im dancing around upgrading to something "modern".

This is all just so I have 4K120 comfortably in recent single player titles. I prefer the idea of offloading frame gen to a second GPU over having it done on the same one.

 

Also looking at the IOCrest M.2 to 10Gbps adapter also and lastly the GC-Titanridge TB3 card for the last slot.

I think this would make for the best "everything I can think of" daily driver for a while.

 

Welp, scratch the ACE. Looks like Asus never fixed Curve Optimizer in the BIOS so it doesnt actually work lol, Back to the drawing board.

 

The Crosshair VIII Dark Hero looks like a good replacement choice.

[Mr. Fox]

8 hours ago, Reciever said:

Forgive me I typed it all out on my phone at break but this second GPU would strictly be just for Frame Gen mule under lossless scaling, my 7900 XTX would still be my raster, at 3Ghz its pretty good for me 24/7. I dont care about power savings, but I do concern myself with the 12VHPWR which is why Im dancing around upgrading to something "modern".

This is all just so I have 4K120 comfortably in recent single player titles. I prefer the idea of offloading frame gen to a second GPU over having it done on the same one.

 

Also looking at the IOCrest M.2 to 10Gbps adapter also and lastly the GC-Titanridge TB3 card for the last slot.

I think this would make for the best "everything I can think of" daily driver for a while.

 

Welp, scratch the ACE. Looks like Asus never fixed Curve Optimizer in the BIOS so it doesnt actually work lol, Back to the drawing board.

 

The Crosshair VIII Dark Hero looks like a good replacement choice.

One of the best things about owning a 9070 XT is, unless you do something stupid in terms of choosing the wrong GPU, you do not have to give a second thought to the garbage 12VHPWR connector. Having the legacy 8-pin PCIe connector is, in my mind, the most compelling reason to own one. Thankfully, hardly any of them use 12VHPWR connectors. The few that do deserve to be ostracized for their exceptionally poor judgment and lack of regard for people that purchase their products.

[electrosoft]

15 hours ago, Reciever said:

So now I'm looking at one of the following 

 

7800 xt, 7900 xt, gre, 9070, xt, possible 7900 xtx

Also looking to pick up an x570-ace for dual GPU scenario in tandem with losses scaling. 

 

Only reason for entertaining a second 7900 xtx is in the faint hope I can get one with better memory overclocking. Seems to be where people are able to hit gold. 

 

At that point I'll pick up a waterblock for the 5800 X3D also, I'll make it easier on my fingers digging into the system time and again. 

 

Also tried booting some of the Gpu's I have but don't seem to be porting lol just my luck

 

Offloading frame gen to a secondary card via which method?

 

In regards to just generically picking amongst that group, it would be the 9070xt but depending on how you're offloading and executing FG, it would come down to potentially best bang for the buck too.

 

14 hours ago, win32asmguy said:

 

I think its supposed to be 30x NGU and D2D, 40x Ring. My 275HX in the Hydroc G2 can do that easily with Premamod, but the real advantage comes from memory tuning. I have seen a few post sub-80ns AIDA64 scores, but I find heat is a big limitation on combined load stability so realistically for me 85ns is the best it can do on air with reasonable ambient temperatures.

 

I think the A51 18 could do better temperature wise with the memory as the modules sit on the side opposite of the vapor chamber. You would have to use Smokeless UMAF to modify timings and test. Sadly max MT would be somewhat limited compared to even a mediocre desktop board as you cannot adjust VDD2 up beyond 1.1v (except on the Hydroc G2 with Premamod).

 

Did you ever give Smokeless a try on your Alienware 18 or Dell Max 18 before sending them back? How much flexibility do you have on your Lenovo? I really wish they had an 18" version of the Hydro. 😞

 

I'm content with my Alienware 18 275hx/5070ti primarily because it was free so bang for buck is undetermined since buck == 0; 🤣 I haven't had much time to play around with it lately though.

 

I've been plowing through Midnight at 4k Ultra settings and I think I can safely say we've reached end game with the 5090. Numerous spots can get into the mid to high 80s utilization but nothing so far can cap it out. This shows a 5080 or 4090 would still hit walls in some spots due to the high utilization spots on the 5090

 

As I'm sure you've experienced, Silvermoon, like Dorongal is an instant player physics frame killer....

 

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

 

AMD officially dropping the 9950X3D2 April 22nd...... I don't think this is going to do much of anything for gaming overall with the same fabric limitations which will be greatly enhanced in Zen 6 but you never know.....

 

I'll be keeping an eye out on this one as I skipped the 9850X3D as it seemed a silly bin bump that can't even sustained its rated speeds in many scenarios vs the 9800X3D which is rock solid.

 

 

 

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

 

Official confirmation of the 290k being scrapped (which we knew already)

 

270k officially available but showing up for $349.99 not $299.99 shipped and sold by Newegg and Amazon. Microcenter also has it listed for $357.....

 

Shenanigans!

 

Of course it is popping up in combo deals already as Newegg (and soon Micro$enter) tries to bundle and move stagnant hardware....

 

I'm going to hold off and let the air clear a bit as this SP83 265k is doing well enough atm....

 

---

 

Cheapest GPU prices:

 

5060 8GB........$349.99

5060 16GB......$699.99 (?)

 

5060ti 8GB.....$379.99

5060ti 16GB....$539.99

 

5070...............$629.99

 

5070ti.............$913.99

 

5080...............$1289.99

 

5090...............$3799.99 (You have TWO kidneys....you'll be ok!)

 

9060xt 8GB...$329.99

9060xt 16GB.$439.99

9070..............$599.99

9070xt...........$699.99

 

 

 

[win32asmguy]

7 minutes ago, electrosoft said:

Did you ever give Smokeless a try on your Alienware 18 or Dell Max 18 before sending them back? How much flexibility do you have on your Lenovo? I really wish they had an 18" version of the Hydro. 😞

 

I'm content with my Alienware 18 275hx/5070ti primarily because it was free so bang for buck is undetermined since buck == 0; 🤣 I haven't had much time to play around with it lately though.

 

Yeah, UMAF did work with both the A51 18 and Dell Pro Max 18 Plus. Sometimes it would freeze or fail to apply settings though so you have keep trying if something did not stick. Just be aware that with any later bios update Dell may write-protect the EFI setup variable region which effectively disables modifying any hidden settings.

 

With the Legion 9i SREP worked on the shipping bios but none of the updates. NGU and D2D can be adjusted but on this one memory timing changes always fail to train. Its basically stuck at 5200MT CL38 Dual Rank (the 64GB Kingston Fury kit). M-die also worked at 5600MT but timings were much worse than the Kingston kit.

 

There will be an 18 inch Uniwill chassis but this year it only gets up to 5070 mobile. Next year its supposed to have better GPU options according to Prema, so Eluktronics should end up carrying it.

[Reciever]

10 hours ago, electrosoft said:

 

Offloading frame gen to a secondary card via which method?

 

In regards to just generically picking amongst that group, it would be the 9070xt but depending on how you're offloading and executing FG, it would come down to potentially best bang for the buck too.

 

Lossless Scaling allows using a second GPU for Frame Generation. 

I tested it initially with the 3090Ti (raster) and 1080Ti as the frame gen card. I was able to get stable FPS with Monster Hunter Wilds running native at 3440x1440 and from a perception standpoint it was smooth. 

 

This was running 3090Ti PCIe 3 x16 and 1080Ti PCIe 3 x4 which is pretty limiting for this type of scenario. 

 

I would like this for smoothing out 4K120 until all the new stuff comes out. 

 

Ideally you run (for 4K120) at minimum PCIe 4 x8/x8. You also want to set a framerate that doesnt cause the GPU to sit at full tilt, so if 110 is your max average then set the framerate for 90, let the mule carry it to 120 via dynamic framerate target. It also has the standard 1/2/3 multipliers also.