Poll: Ryzen 7950X3D, 7900X3D, 7800X3D

[humbug]

I must admit I’m not up to speed with Curve Optimiser etc. Im so used to Intel and only Intel I’m a little apprehensive. Hopefully some up to date guides will be posted soon.

The Curve Optimiser is a great tool, its not difficult to use, you can tune each core individually like i have in the screenshot below or a one for all global setting in about 3 clicks in the BIOS.

Its really clever tech, Standard Undervolting you're only affecting the power consumption at the top end, Curve Optimiser tunes the power curve right across the load range.

Its worth watching the whole video here but i have time imprinted it to the important bit.

Doing this my CPU has gone from 130 to 100 watts and gained 6% performance in R23, more in games because the load is not so high and the curve knows it has more headroom.

What’s New with Precision Boost Overdrive 2

Check out how the new Precision Boost Overdrive 2 can improve your AMD Ryzen™ 5000 Series processor’s single-thread and multithread performance through new f...

youtu.be

 

[uscool]

Doing what and at what resolution?

Sounds great doesn't it but
Probably at 720p in some outlier game when FPS is already probably high to notice in real usage

 

[Hostile_18]

Single digit fps increase unless you game at 240p.. basically lol.

 

[Armageus]

I predict there will not be a single game that is faster on 7900X3D than on 7800X3D

I predict there will not be a single properly threaded task that is faster on 7800X3D


The 7900X3D isn't just for gaming though - it's the best of both worlds, maximum gaming performance due to the extra cache, whilst retaining 16 cores/32 threads for non-gaming related productivity tasks.

(Whereas on the previous platform you had to choose between 5950X if you wanted max productivity vs 5800X3D if you wanted best outright game performance)

 

[J.D]

Yeah, it's simple Harshell. It's just a matter of dialing in what you want and then running C23. If I could be botherd I'd probably do -10, -15, -20, -25 then -30 and see what gives me the best score which in total should take about an hour all in. If you want to halve that time, you could just do -10, -20, -30 and run C23.

 

[jigger]

unfortunately that doesn't apply since L3 cache of different CCD can't be accessed. So most a game thread will have fast access to is still 96MB. Theoretically the other CCD with 32MB would keep all the background apps, but that is a very small effect.

I predict there will not be a single game that is faster on 7900X3D than on 7800X3D

I disagree, and it will apply. If you can run everything within cache the performance gain will be substantial and games that run over 16 threads could benefit a lot.

 

[Deleted member 258511]

The Curve Optimiser is a great tool, its not difficult to use, you can tune each core individually like i have in the screenshot below or a one for all global setting in about 3 clicks in the BIOS.

Its really clever tech, Standard Undervolting you're only affecting the power consumption at the top end, Curve Optimiser tunes the power curve right across the load range.

Its worth watching the whole video here but i have time imprinted it to the important bit.

Doing this my CPU has gone from 130 to 100 watts and gained 6% performance in R23, more in games because the load is not so high and the curve knows it has more headroom.

What’s New with Precision Boost Overdrive 2

Check out how the new Precision Boost Overdrive 2 can improve your AMD Ryzen™ 5000 Series processor’s single-thread and multithread performance through new f...

youtu.be

This will come in handy. Thanks very much. It certainly doesn’t look anywhere near as time consuming as I’m used to.

Yeah, it's simple Harshell. It's just a matter of dialing in what you want and then running C23. If I could be botherd I'd probably do -10, -15, -20, -25 then -30 and see what gives me the best score which in total should take about an hour all in. If you want to halve that time, you could just do -10, -20, -30 and run C23.

Sounds simple enough. Fingers crossed I don’t see any smoke

 

[Aretak]

I dont see the need for these 7000 series X3d chips when you can get the 5800X3D for about £280 with far lower platform costs.
Works great even with my 4090.

The 7800X3D will be notably faster in games which like the 5800X3D. Hardware Unboxed tested Factorio and the gap from the 5800X3D to the (simulated) 7800X3D was almost 25%. The main game I play is another 5800X3D-loving title, so I'd expect a similar boost over it there too. The 7000 series also has AVX-512, which I want for emulation purposes. Plus I'm just kind of tired of AM4. I've been using it for almost six years now and my current 5800X for almost two. I'm sick of looking at the ugly mirrored RGB lighting on the X570 MEG ACE and the cheap placeholder kit of DDR4 I bought in 2019 and could never justify replacing. Baby want new toys, basically.

 

[humbug]

This will come in handy. Thanks very much. It certainly doesn’t look anywhere near as time consuming as I’m used to.


Sounds simple enough. Fingers crossed I don’t see any smoke

If you do get one and you do get stuck messing with it there are plenty of people here with a lot of experience willing to help.

But honestly from a user perspective its no different to undervolting an Intel CPU, its what its doing with your inputs that's different.

 

[rare]

I hope this delay to the 7800 allows for better stock than today

 

[huggz123]

Got my order in at 10:06 but no shipped email yet Is my order doomed?

 

[Hotwired]

I hope this delay to the 7800 allows for better stock than today

It's going to sell out basically instantly no matter what.

Flip a coin over whether "price gouging" or "garbage stock" is the villain of the day.

 

[Deleted User 456458]

Ryzen 9 7950X3D Undervolt & Overclock to 5900 MHz With Crosshair X670E Extreme | SkatterBencher #56

In today's video, I overclock the AMD Ryzen 9 7950X3D Raphael CPU up to 5900 MHz with the ASUS ROG Crosshair X670E Extreme motherboard and EK custom loop wat...

www.youtube.com

For those into tuning, this should serve as a good rerference.

 

[Bencher]

Single digit fps increase unless you game at 240p.. basically lol.

I still have hope that people, one day, will understand that those 240p results will actually translate in 4k results once you upgrade your GPU down the line. The reason the 5800x 3d is still doing great with a 4090 is exactly because it was doing well at "240p" with a 3090. If you are looking at 4k results then you would end up buying a 5600x, since its as fast at less than half the money. Then you buy your new shiny 4090 and go like "Oh ****" cause of the insane GPU bottleneck

Spoiler

 

[humbug]

I still have hope that people, one day, will understand that those 240p results will actually translate in 4k results once you upgrade your GPU down the line. The reason the 5800x 3d is still doing great with a 4090 is exactly because it was doing well at "240p" with a 3090. If you are looking at 4k results then you would end up buying a 5600x, since its as fast at less than half the money. Then you buy your new shiny 4090 and go like "Oh ****" cause of the insane GPU bottleneck

Spoiler

This... i get why low res testing may seem like a joke to some but its the right way to do it if what you want to know is how the CPU's you're looking at actually stack up.

Proper testing matters....

Edit: yes i know i highlighted the 5900X but its the same as the 5800X.

I want to show you something.

in 2020 i had <£500 for a CPU, i wanted it to be as fast as i could get for that money, 5900X and 5950X was out of reach, for £450 or £500 respectively my choices were 5800X or 10900K.

My first port of call is always Hardware Unboxed, (TechSpot) the first TechSpot slide below just looks GPU bound to me, so i went to AnandTech, they test resolutions as low as 480P, in their testing showed the 5800X way ahead of the 10900K, at £450 it was £50 cheaper, so i bought that.

Fast forward to faster GPU's and the second TechSpot slide shows that now in their testing also the 5800X is indeed faster, had i not seen the AnandTech review i would not have seen that, i could have ended up with the 10900K given its above the 5800X in the first slide, even if only by a very margin of error i could have thought with all other data not available to me that could have mattered.
As it happens i saved £50 and got the faster CPU, because of proper testing by AnandTech.

AMD Zen 3 Ryzen Deep Dive Review: 5950X, 5900X, 5800X and 5600X Tested

www.anandtech.com

 

[Rup]

The 7800X3D will be notably faster in games which like the 5800X3D. Hardware Unboxed tested Factorio and the gap from the 5800X3D to the (simulated) 7800X3D was almost 25%. The main game I play is another 5800X3D-loving title, so I'd expect a similar boost over it there too. The 7000 series also has AVX-512, which I want for emulation purposes. Plus I'm just kind of tired of AM4. I've been using it for almost six years now and my current 5800X for almost two. I'm sick of looking at the ugly mirrored RGB lighting on the X570 MEG ACE and the cheap placeholder kit of DDR4 I bought in 2019 and could never justify replacing. Baby want new toys, basically.

Yes fully understand the unneccessary upgrade itch, went from a 3080 to a 4090 when didnt really need to
Also depends on your monitor, need a high res high fps one to justify it. Eg my 5800x3d pretty much maxes my 4k 120 hz OLED in the games I play.
So a 7000 series x3d wouldnt make a visible difference.

 

[Bencher]

This... i get why low res testing may seem like a joke to some but its the right was to do it if what you want to know is how the CPU's you're looking at actually stack up.

Proper testing matters....

Edit: yes i know i highlighted the 5900X but its the same as the 5800X.

It's not only that, it's also that most reviews don't and cannot test a whole game. They have a suite of 50 games and they quickly go through them - which makes sense. There are games that have insanely CPU heavy areas where you can see big differences between cpus even in higher resolutions.

I remember techspot tested my R5 1600 back then, he showed an average of 90+ fps on AC Origins. I was like, cool, let's slap a 1080ti to this beast. Then I started playing the game, I got to Alexandria, and in the whole city I was dropping to 50-55. So yeah....averages and reviews can be misleadingg

 

[humbug]

It's not only that, it's also that most reviews don't and cannot test a whole game. They have a suite of 50 games and they quickly go through them - which makes sense. There are games that have insanely CPU heavy areas where you can see big differences between cpus even in higher resolutions.

I remember techspot tested my R5 1600 back then, he showed an average of 90+ fps on AC Origins. I was like, cool, let's slap a 1080ti to this beast. Then I started playing the game, I got to Alexandria, and in the whole city I was dropping to 50-55. So yeah....averages and reviews can be misleadingg

Very true, i have to tho... in same vain, i had a Ryzen 1600, a 4690K before that.

The first image is with that 4690K, the second with the Ryzen 1600. It also depends on the game, i mean notice the 100% load on the 46900K and the 33 FPS.... unplayable, 85 with the 1600 and its not over stressed.

 

[malachi]

AMD are making up for the slow sale of AM5. Complaints about expensive motherboards and DDR5 RAM. DDR5 isnt really expensive anymore and the uptake of 7950x3D and people waiting for the 7800x3D. AMD are making up for the loss now.

 

[steverebo]

Still no reviews of the 7900x 3d that's pretty telling of what to expect