• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

AMD presents Zen 5

winjer

Gold Member






v8MRKVv.jpeg


USGBXZs.jpeg


Oz1bvWF.jpeg


KjfGGrP.jpeg
 
Last edited:

Allandor

Member
So not really on par with zen4d at least in games.
Btw, are those the "old" 14900k benches or the new with a much lower power delivery?
 

Leonidas

AMD's Dogma: ARyzen (No Intel inside)
So not really on par with zen4d at least in games.
Btw, are those the "old" 14900k benches or the new with a much lower power delivery?
Yeah, there is a good chance Zen5 will lose to Zen4 3D. Zen4 3D seems to be faster in at least 3 of the games tested. Not enough data to say for sure though.

They hamstrung the Intel Machine by using Intel Defaults, which caps performance of any all-core tests. They also tested with AMD optimized DDR speeds. They also tested with a 7900 XTX and muddied the water with enabling Smart Access Memory on both machines. I've seen results online where Raptor Lake even loses performance with Smart Access Memory enabled. They hid all of this in their end-notes.
 
Last edited:

winjer

Gold Member
So not really on par with zen4d at least in games.
Btw, are those the "old" 14900k benches or the new with a much lower power delivery?

At this point, it's possible to tell, as we only have benchmarks from AMD.
The latest HuB test with the 14900K vs 7800X3D put them at a 5% difference.
While the limited sample of games results from AMD, puts the 9950X at a 13% advantage over the 14900K.
So by extrapolating results, the 9950X is 8% faster than the 7800X3D.

But this has to be taken with a pinch of salt, as the sample of games is limited and we don't know if the bench specs that HuB and AMD used, are a match.

8RK4Mix.jpeg
 

Leonidas

AMD's Dogma: ARyzen (No Intel inside)
So by extrapolating results, the 9950X is 8% faster than the 7800X3D.
Only someone who didn't read (or understand) the end-notes could come to such a conclusion.

Bo8WUEi.png


AMD muddied the waters big time.
 
Last edited:

winjer

Gold Member
Only someone who didn't read (or understand) the end-notes could come to such a conclusion.

LP06Pru.png


AMD muddied the waters big time.

How about you state the end-notes that pertain to the 14900K vs 9950X:

5 Testing as of May 2024 by AMD Performance Labs on test systems configured as follows: AMD Ryzen 9 9950X system: GIGABYTE X670E AORUS MASTER, Balanced, DDR5-6000, Radeon RX 7900 XTX, VBS=On, SAM=On, KRACKENX63 vs. similarly configured Intel Core i9-14900KS system: MSI MEG Z790 ACE MAX (MS-7D86), Balanced, DDR5-6000, Radeon RX 7900 XTX, VBS=On, SAM=On, KRAKENX63 {Profile=Intel Default} on the following applications: 3DMarkDandia, Blender, Cinebench, GeekBench, PCMark10, PassMark, ProcyonOffice, ProcyonPhotoEditing, ProcyonVideoEditing. System manufacturers may vary configurations, yielding different results. GNR-01.
 

Leonidas

AMD's Dogma: ARyzen (No Intel inside)
Wtf amd, where you get this? 9950x vs 7700x
AMD official site. I posted the wrong pic by the way, updated the post with the 14900K info, they actually used 14900KS, but they power limited it, limited it to DDR5 6000, and enabled SAM, which has been shown to not really help Intel CPUs or in some cases even loses performance with Intel.

The 13% average over 6 games over the 14900K can basically go straight in the trash.

9950X vs. 7700X was for their 16% IPC test.
 
Last edited:
That 9700X with a TDP of 65W might just be the most efficient CPU of the year.
At least, until the X3D variants show up.

wWQu0I5.jpeg
Man Sony Could take the Desktop Parts for the PlayStation 6 the 12 Core worked on clocked down a bit could bit 70 watts graphics 200 Watts rest of system 30 and you would have an amazing 300 Watt Playstation 6
 

winjer

Gold Member
AMD official site. I posted the wrong pic by the way, updated with the 14900K info, they actually used 14900KS, but they power limited it, limited it to DDR5 6000, and enabled SAM, which has been shown to not really help Intel CPUs or in some cases even loses performance with Intel.

The 13% average over 6 games over the 14900K can basically go straight in the trash.

9950X vs. 7700X was for their 16% IPC test.

where did you see the info that AMD limited the power for the 14900KS.

Also, you might want to remember that the 14900KS is faster than the 14900K. 6.2 Ghz vs 6Ghz boost.

Memory above 6000 Mbps does very little for the AMD or Intel performance. Both can reach 7200Mbps. In some cases, even 8000Mbps.
But to make things even, they both have to have the same memory speed.
Even with tuned memory, we are talking about going from an average of 223 fps at 6000MBps, to 228 fps at 7200Mbps.

xEfUWEl.png
 
Last edited:

Leonidas

AMD's Dogma: ARyzen (No Intel inside)
where did you see the info that AMD limited the power for the 14900KS.
Intel Defaults limits 14900KS to only 253 watts.

They did some MT tests (Blender). I've seen reviews where 14900KS pulls another 120 watts in this test at stock.

Also, you might want to remember that the 14900KS is faster than the 14900K. 6.2 Ghz vs 6Ghz boost.
That's really nice of AMD, after doing 3-4 other things which could hamstring performance.
 
Last edited:

winjer

Gold Member
Intel Defaults limits 14900KS to only 253 watts.

They did some MT tests (Blender). I've seen reviews where 14900KS pulls another 120 watts in this test at stock.

That's really nice of AMD, after doing 3-4 other things which could hamstring performance.

So you have nothing to prove your claim that AMD power limited the 19400KS.
Or that they did anything to hamstring performance.

At this point you are just making stuff up.
 

Leonidas

AMD's Dogma: ARyzen (No Intel inside)
So you have nothing to prove your claim that AMD power limited the 19400KS.
Or that they did anything to hamstring performance.

At this point you are just making stuff up.
End-notes. Intel Defaults.

That's 253 watts. Even 14900K CPU alone pulled 283 watts in Blender in some reviews.
 

winjer

Gold Member
End-notes. Intel Defaults.

That's 253 watts. Even 14900K CPU alone pulled 283 watts in Blender in some reviews.

The default that Intel made all motherboard manufactures comply with, so their CPUs would no crash and degrade.

BTW, remember when you said that your 13600K, with the new Intel power specs, was running cooler and boosting faster. So it as an advantage.
Funny how now, you say it's a disadvantage.
 
Last edited:

Leonidas

AMD's Dogma: ARyzen (No Intel inside)
The default that Intel made all motherboard manufactures comply with, so their CPUs would no crash and degrade.
The default that AMD selected to give themselves a 56% lead in Blender.

That's a lot less egregious than AMD using the 7900 XTX + Smart Access Memory combination on the gaming benchmarks though.

That "13% average" uplift has no meaning.
 

winjer

Gold Member
The default that AMD selected to give themselves a 56% lead in Blender.

That's a lot less egregious than AMD using the 7900 XTX + Smart Access Memory combination on the gaming benchmarks though.

That "13% average" uplift has no meaning.

Stop pretending that the Intel defaults are something that AMD choose.
There is a reason why it's called the Intel defaults. It's because it was mandated by Intel.
And the AMD CPUs are using the AMD defaults.
So they are both using the official defaults for both CPUs.

And both the 14900Ks and the 9950X have SAM/ReBar enabled.
 

MayauMiao

Member
Should go for AMD Zen 5 or wait for Intel Arrowlake?

Itching to refresh my ancient PC that I could barely get games running.
 

Leonidas

AMD's Dogma: ARyzen (No Intel inside)
And both the 14900Ks and the 9950X have SAM/ReBar enabled.
And why do you think AMD used it? Because it makes them look better, and it makes Intel look worse. They used to mention this stuff in the slides but nowadays they just hide it in end notes. Slimy stuff.

Should go for AMD Zen 5 or wait for Intel Arrowlake?

Itching to refresh my ancient PC that I could barely get games running.
Best to wait, Zen5 might not even beat Zen4 3D.
 

Leonidas

AMD's Dogma: ARyzen (No Intel inside)
Rebar works well in both Intel and AMD platforms. You are just making random excuses.
It absolutely does not work the same on both platforms. The uplift is different. AMD gains more with the 7900 XTX + Smart Access Memory combination.

AMD wouldn't have set those paramaters and then hide it in the notes if this wasn't the case. It was selected to help them reach that "13% average over 6 games". If you beleive otherwise you must be really credulous.
 

winjer

Gold Member
It absolutely does not work the same on both platforms. The uplift is different. AMD gains more with the 7900 XTX + Smart Access Memory combination.

Do you have any evidence for that?

The reality is that SAM is just a marketing term for Rebar, a tech that both AMD and Intel have support, on their CPUs and GPUs.
Both, Intel and AMD always recommend turning this feature ON.
 

Silver Wattle

Gold Member
It absolutely does not work the same on both platforms. The uplift is different. AMD gains more with the 7900 XTX + Smart Access Memory combination.

AMD wouldn't have set those paramaters and then hide it in the notes if this wasn't the case. It was selected to help them reach that "13% average over 6 games". If you beleive otherwise you must be really credulous.
SAM/Rebar is a requirement for Intel GPU's to function correctly, SAM/Rebar is basically the standard of all new PC set ups, to test without it would be idiotic.
 

Leonidas

AMD's Dogma: ARyzen (No Intel inside)
SAM/Rebar is a requirement for Intel GPU's to function correctly, SAM/Rebar is basically the standard of all new PC set ups, to test without it would be idiotic.
Intel GPUs yes, but they tested with a 7900 XTX.

A GPU that basically no one uses. Every reveiw will use the fastest GPU to test gaming performance, 4090, and later, 5090 once it is released.

7900 XTX + SAM gives AMD an advantage.

There is a reason they didn't put in a 4090 with ReBAR, they wouldn't have gotten "13% over 6 games average".

Every reviewer will use a 4090, and later a 5090, cards which are faster than the 7900 XTX to better test gaming performance, and ones which won't exploit the 7900 XTX + Smart Access Memory advantage on AMD CPUs.
 

winjer

Gold Member
Intel GPUs yes, but they tested with a 7900 XTX.

A GPU that basically no one uses. Every reveiw will use the fastest GPU to test gaming performance, 4090, and later, 5090 once it is released.

7900 XTX + SAM gives AMD an advantage.

There is a reason they didn't put in a 4090 with ReBAR, they wouldn't have gotten "13% over 6 games average".

Every reviewer will use a 4090, and later a 5090, cards which are faster than the 7900 XTX to better test gaming performance, and ones which won't exploit the 7900 XTX + Smart Access Memory advantage on AMD CPUs.

Show any evidence of that. Any.
 

poppabk

Cheeks Spread for Digital Only Future
Oh no, Leonidas trying to find fault, or make some up, with anything AMD.....how suprising.
Looking forward to his next "Why NVIDIA cards are actually too cheap" or similar crap thread.......
Are you trying to imply that Leonidas isn't completely impartial. He always evaluates the CPU's and always picks Intel because they are always the bestest and AMD always stink.
 

MayauMiao

Member
And why do you think AMD used it? Because it makes them look better, and it makes Intel look worse. They used to mention this stuff in the slides but nowadays they just hide it in end notes. Slimy stuff.


Best to wait, Zen5 might not even beat Zen4 3D.
There is no way to answer that question now.
At this point, we only have AMDs numbers. No independent reviews, until July.
And we have nothing about Arrow lake to compare.

I guess my new PC will have to be a Christmas gift.
 

Panajev2001a

GAF's Pleasant Genius
Man Sony Could take the Desktop Parts for the PlayStation 6 the 12 Core worked on clocked down a bit could bit 70 watts graphics 200 Watts rest of system 30 and you would have an amazing 300 Watt Playstation 6
PS6 is likely going to be based on a successor of the AI 300 series. Same thing for Steam Deck… Steam Deck 2 will go for a custom low end version of it (a successor of it maybe) and PS6 will go with a custom high end version (maybe based on the next or next-next generation so something between AI 400 and AI 500).
 
Top Bottom