• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Zen 5 rumours

That would be crazy MT performance from the 9950X in CB24. For example my 7970X gets 3644 pts. So it would 82% of the power with half the number of cores.

Better to wait for more evidence.

In R24 its 46% faster MT than a 7950X, 2052 vs 3000
42% faster in ST, 120 vs 170.

IMO that's too high, i'm thinking 20%, Mores Law insists on 13 to 15% and thinks that's good, its less than from Zen 2 to Zen 3 and Zen 3 to Zen 4. He also thinks Intel absolutely will be +40%, IMO he has some bias there.... He has a habit of taking anything Intel employees tell him as absolute fact, no matter how many times it turns out wrong.

But 40% is not impossible or unpresidented for AMD.
 
Last edited:
Also so... he has AMD insiders telling him sub 20%.

The thing about AMD is they always sandbag, they hate Intel knowing the truth.
Intel on the other hand are always seeking attention, trying to look relevant, trying to look like they are still the X86 leaders, when their boasting doesn't pan out they simply move on to doing the same thing with the next generation, and again, and again....

MLID is sucked in by all of that ^^^^ every single time.
 
Last edited:
Who said IPC, though? All that table shows is higher score. How is that achieved is a different matter - it could be higher clocks, higher power use, IPC, or (most likely) a combination of all that. I highly doubt it would be 40%+ pure IPC increase. Plus, IPC using which parts of the CPU is another matter, not less important.
 
Last edited:
The thing about AMD is they always sandbag, they hate Intel knowing the truth.
Or, if leaks come from engineering department, they might know all the downsides and it's a very averaged performance number, amongst all types of computation. In reality it could be close to 50% in CB, 5% in many games, 20% in office software, etc. etc. And then it will come down to what type of workload one looks at and they will get different performance changes comparing to current gen. Point is, it's impossible to say how much faster one CPU is comparing to another without looking at specific workloads.
 
Last edited:
Some more interesting tidbits...

  • Zen 5 (4nm) Monolithic Design
  • Up To 12 Cores In Hybrid Config (Zen 5 + Zen 5C)
  • 32 MB of Shared L3 cache
  • 35% Faster CPU Versus Phoenix at 50W
  • 16 RDNA 3+ Compute Units
  • 128-bit LPDDR5X Memory Controller
  • XDNA 2 Engine Integrated
  • ~25 TOPS AI Engine
  • 2H 2024 Launch (Expected)
 
Last edited:
If this is true and its a big if, then Intel are kerfuied, Pat is banking on his TSMC adventure being his Sandy Bridge 2 moment, to put AMD back in the box.

Maybe AMD took his insistent boasting about Panther Lake seriously and said no _____ off watch we'll do next.

This is why AMD don't constantly scream about what they are doing for the next 5 years, are in fact utterly silent on everything.
 
Last edited:
Last edited:
The way AI performance is measured now is all over the place

Intel for example is saying its next gen mobile system, called lunar lake, will be 100 tops.. way more than AMD's 25 right? The way Intel calculates is a 100 is they take the NPU performance then added the CPU cores performance, then add the GPU core performance and say hey guys it's a 100tops when you take the entire systems components into account

And then other companies like AMD just give you the NPU number, the 25 excludes the CPU core and GPU core AI performance and the other problem is there can be variance in the measurement of tops - for example some use INT4, some use INT8 etc, so there is no standard of what "1 TOPS AI performance" is
 
The way AI performance is measured now is all over the place

Intel for example is saying its next gen mobile system, called lunar lake, will be 100 tops.. way more than AMD's 25 right? The way Intel calculates is a 100 is they take the NPU performance then added the CPU cores performance, then add the GPU core performance and say hey guys it's a 100tops when you take the entire systems components into account

And then other companies like AMD just give you the NPU number, the 25 excludes the CPU core and GPU core AI performance and the other problem is there can be variance in the measurement of tops - for example some use INT4, some use INT8 etc, so there is no standard of what "1 TOPS AI performance" is

This is typical Intel tho, i don't know what the TOPS performance is if you include GPU and CPU neural emulation along with the NPU, i have 400 TOPS to drive Copilot, at 250 watts... and that's the point, an NPU to drive Copilot is supposed to be done at a couple of watts, an NPU is a dedicated limited instruction architecture specifically designed to accelerate a neural net at very low power, that's all it does, its not capable of doing anything else.

Anyone can lump together any bit of hardware capable of AI, a 5 year old CPU is capable of it, and call that your TOPS throughput, but its bastardising the whole point of AI capable chips driving AI capable operating systems, When people do that for marketing reason which only sound good to the informed boasting bigger numbers that are complete BS is when it all starts to get silly, and trust Intel to start making something that is perfect sense become silly to look bigger.

AMD measure the NPU only, as do Nvidia, Apple, Qualcomm... its the only thing actually used for OS level AI.
 
Last edited:
Any ideas yet if the IHS is going to be the same?

Literally just delidded and direct die cooled my 7950x3D and I'm wondering if the tool to do this, and the direct die block I have on my CPU, will be compatible with Zen 5 :confused:
 
This is, allegedly, Zen 5 (Ryzen 9000) looks the same to me.

5iUF4yg.jpeg
 
5800x3D still going strong, but being paired with a 4090 I imagine I'll be changing come year end

I have the same CPU and am hoping to get something in the 4080/XTX to 4090 performance class next gen. While a CPU upgrade would be nice, as I'm playing at 4K (on my TV) and hoping to nab a 1440p UW at some point I'm going to try and wait it out until a new platform to give myself an upgrade path, rather than jump in at the end of a generation.

That's assuming I'm still playing enough to want more performance; between Helldivers 2, Jedi Survivor and Cyberpunk I'm playing loads right now, but that typically drops off a bit when the weather improves, and I'm not sure there's much being released that will garner my interest in the next year or so.
 
I noticed he is now also view botting, I wonder when he started doing that sneaky ****

If you visit his website, it has his latest YouTube video embedded and automatically starts playing, meaning that random clicks to his website and Google traffic bots visiting that site will automatically trigger his YouTube video view counts to go up
If that's just a YouTube video about graphics cards, imagine what they do in politics...........
 
Yes, we know its around 45 TOPS + because AMD and Microsoft have been on stage stating that and mocking Intel, the existing 8700G, (Hawkpoint) already has 39 TOPS.
yes, but there is a difference between NPU TOPS and total-system TOPS.

i.e. Hawkpoint
NPU = 16
CPU = ~3
GPU = ~20

so when i say i don't believe the WCCFTech article, it's because i don't believe strix point has just 25 TOPS from the NPU (half that of strix halo).

specifically - i believe that both products use the same XDNA2 NPU at similar clocks, and both will achieve ~45 TOPS from the NPU.
 
Back
Top Bottom