These guys have no idea what they are talking about.

[Bencher]

With a username of Bencher i expected a more varied testbench. Maybe ask mods to change your name to Cinebencher?

If its not winning in cinebench it's not winning in anything. Other less threaded tasks will show even bigger differences. My testing of 1 bench is - believe it or not - more varied than the testing offered by the people claiming zen 3 is more efficient

I don't understand why people cant just accept and admit the facts. Anyways, lets keep dodging so we don't have to accept the facts cause they show our favorite multi billion $ company eating dust. Peace

 

[Cyber-Mav]

Intel's Alder Lake is slower than AMD's Cezanne Zen 3 at 45W TDP

We are currently getting more laptops with the new Intel Alder Lake processors. A direct comparison is not always easy due to different TDP settings, but the two devices Alienware x14 and the Lenovo Yoga Slim 7 Pro 14 now give us the chance for that - with the better end for AMD.

www.notebookcheck.net

More info there

 

[Vulture]

Intel's Alder Lake is slower than AMD's Cezanne Zen 3 at 45W TDP

We are currently getting more laptops with the new Intel Alder Lake processors. A direct comparison is not always easy due to different TDP settings, but the two devices Alienware x14 and the Lenovo Yoga Slim 7 Pro 14 now give us the chance for that - with the better end for AMD.

www.notebookcheck.net

More info there

6000 series is even more efficient, and 7000 series will have many of power saving features that 6000 have, no doubt that AMD is leading in efficiency side, while ADL is terrible at it. RPL will be even worse because it is based on the same node while having even more clock, cache and cores.

 

[Vince]

If its not winning in cinebench it's not winning in anything. Other less threaded tasks will show even bigger differences. My testing of 1 bench is - believe it or not - more varied than the testing offered by the people claiming zen 3 is more efficient

I don't understand why people cant just accept and admit the facts. Anyways, lets keep dodging so we don't have to accept the facts cause they show our favorite multi billion $ company eating dust. Peace

Problem is it's not really eating dust and there isn't a lot in it. Arguably at a given lower TDP Zen walks all over ADL, you can run a Zen3 core at like 2w a core. Not sure how well that works out on ADL. ADL power and therefore efficiency curve is probably a bit higher so at higher clocks and power draw ADL takes some efficiency back. Swings and roundabouts my friend and to be honest not many people probably care enough. It's like me saying my 280w tdp cpu is more efficient. at 240w my Zen part would lay waste to ADL in cinebench. In fact you could OC the balls off of your ADL and still wouldn't match that zen chip in cinebench.

Which is also likely why arguing around a fixed clock that suits the efficiency curve of ADL over Zen or vice verse isn't all that smart and ends up in an endless back and forward where nobody is really right.

 

[Bencher]

Problem is it's not really eating dust and there isn't a lot in it. Arguably at a given lower TDP Zen walks all over ADL, you can run a Zen3 core at like 2w a core. Not sure how well that works out on ADL. ADL power and therefore efficiency curve is probably a bit higher so at higher clocks and power draw ADL takes some efficiency back. Swings and roundabouts my friend and to be honest not many people probably care enough. It's like me saying my 280w tdp cpu is more efficient. at 240w my Zen part would lay waste to ADL in cinebench. In fact you could OC the balls off of your ADL and still wouldn't match that zen chip in cinebench.

Which is also likely why arguing around a fixed clock that suits the efficiency curve of ADL over Zen or vice verse isn't all that smart and ends up in an endless back and forward where nobody is really right.

I can run at whatever wattage you desire, core for core zen 3 stands absolutely no chance against golden cove in efficiency. It's not even a contest. The difference is that huge that I doubt if even zen 5 can actually tie it. Zen 4 of course stands no chance whatsover.

Of course taltking doesn't get us anywhere, pick the wattage you think is best for zen 3 and run a cbr23 with 8 zen 3 cores. Ill match your wattage with 8 gc cores and let the results speak themselves. Right?

 

[Bencher]

Intel's Alder Lake is slower than AMD's Cezanne Zen 3 at 45W TDP

We are currently getting more laptops with the new Intel Alder Lake processors. A direct comparison is not always easy due to different TDP settings, but the two devices Alienware x14 and the Lenovo Yoga Slim 7 Pro 14 now give us the chance for that - with the better end for AMD.

www.notebookcheck.net

More info there

Lets keep changing the topic just so we can dodge and not admit the facts. Your link doesnt test GC cores, the 12700h is a 6+8 configuration. Im talking about the actual GC cores, why can't you just admit it? How much of a fanboy can someone be as to try to pretend what he already knows is true isn't? This is insanity

 

[Vince]

I can run at whatever wattage you desire, core for core zen 3 stands absolutely no chance against golden cove in efficiency. It's not even a contest. The difference is that huge that I doubt if even zen 5 can actually tie it. Zen 4 of course stands no chance whatsover.

That's the point at 240w there are many more efficient options and larger core counts than ADL. We are not in a core for core comparison world, not really. At 240w and above there is nothing in the stack that even comes close in terms of efficiency to EPYC/TR, nothing, not even close. On low tdp parts zen pulls out ahead as can offer more full size cores at lower TDP's. Architecturally zen is a more efficient architecture overall.

 

[Vulture]

Intel's Alder Lake is slower than AMD's Cezanne Zen 3 at 45W TDP

We are currently getting more laptops with the new Intel Alder Lake processors. A direct comparison is not always easy due to different TDP settings, but the two devices Alienware x14 and the Lenovo Yoga Slim 7 Pro 14 now give us the chance for that - with the better end for AMD.

www.notebookcheck.net

More info there

Here is another article: https://www.world-today-news.com/ci...up-with-zen-3-but-amd-is-much-more-efficient/

 

[gpuerrilla]

Yes ^^^


Its a shame he didn't do that for the original RTX 3050 review, that's the one that counts.

Opethian (or maybe I) may have missed something then. I do recall watching the videos (over 4 months back) they output post nvidia spat (where they fell out with them) to suddenly recommend the geforce cards purely because they can do ray tracing. I think the 6600 has been recommended by Steve, but maybe all I got incorrect was the model, maybe the XT variant was overlooked by the 3060 which is why I find it funny he is now not promoting the RT and instead switching back to value for money. If you want to be pedantic he actually says buy a used 570 for most of the low end, but now the prices have adjusted globally suddenly RT isnt the deal maker now its the price..

 

[Bencher]

That's the point at 240w there are many more efficient options and larger core counts than ADL. We are not in a core for core comparison world, not really. At 240w and above there is nothing in the stack that even comes close in terms of efficiency to EPYC/TR, nothing, not even close. On low tdp parts zen pulls out ahead as can offer more full size cores at lower TDP's. Architecturally zen is a more efficient architecture overall.

Yes obviousy higher core count CPUs are more efficient. That is not the point. The point is, some people (i dont wanna name them) in this forum claimed that the zen 3 cores are more efficient than the GC cores. Thats what im trying to see but everyone keeps dodging... Can you run a CBR23 with 8 zen 3 cores and share the results? Then we can go from there.

 

[Vince]

Yes obviousy higher core count CPUs are more efficient. That is not the point. The point is, some people (i dont wanna name them) in this forum claimed that the zen 3 cores are more efficient than the GC cores. Thats what im trying to see but everyone keeps dodging... Can you run a CBR23 with 8 zen 3 cores and share the results? Then we can go from there.

it gets complicated though I think that's the problem, it doesn't matter how efficient an architecture is if you can't scale the fastest bit. I bet you if I had both here I could find an area on the curve where Zen 3 is significantly more efficient and areas where ADL is significantly more efficient. My rig is a 3960x so zen 2 cores but if I had a 5000 series part here I would happily do it without bias. Although my ability to tune zen is epically bad and it sounds like your ability to tune ADL might be pretty good. As a core to core comparison ADL is likely the faster core at a given tdp but there is a bigger picture to all this and that's simply the ability to scale so offering more cores at any given TDP. There is a reason why at the price point there isn't core count parity.

 

[Bencher]

it gets complicated though I think that's the problem, it doesn't matter how efficient an architecture is if you can't scale the fastest bit. I bet you if I had both here I could find an area on the curve where Zen 3 is significantly more efficient and areas where ADL is significantly more efficient. My rig is a 3960x so zen 2 cores but if I had a 5000 series part here I would happily do it without bias. Although my ability to tune zen is epically bad and it sounds like your ability to tune ADL might be pretty good. As a core to core comparison ADL is likely the faster core at a given tdp but there is a bigger picture to all this and that's simply the ability to scale so offering more cores at any given TDP. There is a reason why at the price point there isn't core parity.

Well I agree with all of that but users here are claiming that zen 3 cores are more efficient than GC cores, which I can't begin to describe how false that is. Intel is at least 2 generations ahead when it comes to their P core efficiency. I betcha zen 4 will still lose in that department, and most likely zen 5. I dont doubt your abilities, but you wouldnt find an area where the zen 3 cores are more efficient, cause there isnt one. If anyone bothered to share a CBR23 result with 8 zen 3 cores youd freak out by the actual difference. No wonder no one dares to share any, they wanna keep pretending

And it makes sense, golden cove is a huge core in terms of die space. And that's exactly the problem with scaling, it makes no financial sense for Intel to scale this thing in the mainstream space. They will end up with a huge expensive CPU. 16P cores would be the most efficient CPU on planet Earth, but the actual cost of that thing would be enormous. They are saving it for the hedt and server workspace where the money is. E cores are great for multithreaded performance but they lack a lot in efficiency compared to similarly clocked P cores.

 

[Vince]

Well I agree with all of that but users here are claiming that zen 3 cores are more efficient than GC cores, which I can't begin to describe how false that is. Intel is at least 2 generations ahead when it comes to their P core efficiency. I betcha zen 4 will still lose in that department, and most likely zen 5. I dont doubt your abilities, but you wouldnt find an area where the zen 3 cores are more efficient, cause there isnt one. If anyone bothered to share a CBR23 result with 8 zen 3 cores youd freak out by the actual difference. No wonder no one dares to share any, they wanna keep pretending

And it makes sense, golden cove is a huge core in terms of die space. And that's exactly the problem with scaling, it makes no financial sense for Intel to scale this thing in the mainstream space. They will end up with a huge expensive CPU. 16P cores would be the most efficient CPU on planet Earth, but the actual cost of that thing would be enormous. They are saving it for the hedt and server workspace where the money is. E cores are great for multithreaded performance but they lack a lot in efficiency compared to similarly clocked P cores.

You are mixing up fast and efficient... 16p cores would consume enormous amounts of power, would it be fast? yea it would, but efficient... it doesn't matter how fast a core is in the HEDT/server space without other aspects such as density. Fast <> efficient. a single ADL core at 3w for example would lose to a 3 zen cores of milan/genoa in the same power envelope.

 

[Bencher]

You are mixing up fast and efficient... 16p cores would consume enormous amounts of power, would it be fast? yea it would, but efficient... it doesn't matter how fast a core is in the HEDT/server space without other aspects such as density. Fast <> efficient. a single ADL core at 3w for example would lose to a 3 zen cores of milan/genoa in the same power envelope.

A 16P core would consume the exact amount of power it's power limit dictates. The same as any other CPU. I dont even understand why it would ever possibly consume more. 16p cores at 240w (same as the 12900k) would wipe the floor with the 8+8 configuration of the 12900k at 240w. So what exactly do you mean "would consume enormous amount of power".

Im pretty confident GC core would still wipe the floor with zen 3 at 3w. Of course we cant test it cause no one with zen 3 dares to, really makes you wonder why

 

[Vince]

A 16P core would consume the exact amount of power it's power limit dictates. The same as any other CPU. I dont even understand why it would ever possibly consume more. 16p cores at 240w (same as the 12900k) would wipe the floor with the 8+8 configuration of the 12900k at 240w. So what exactly do you mean "would consume enormous amount of power".

Im pretty confident GC core would still wipe the floor with zen 3 at 3w. Of course we cant test it cause no one with zen 3 dares to, really makes you wonder why

Ok let me put it this way so we can understand. Why in the server space do you think Zen/Epyc is the clear choice in terms of performance/density? Why has the 12900k got gracemont "efficiency cores" if the P cores are so efficient? Why does this 16P core configuration not exist? The answer to all of these questions is because golden cove isn't efficient. It's fast, 100% it is but it isn't efficient. It's big and as a core consumes too much power to scale in a workable package. It's a Ferrari, not a tesla.

Edit to add to this:

Efficiency is calculated by more than just this core at this power envelope, you absolutely have to consider things like interconnect power draw and scalability, die size (which for an idea sources randomly found on the web put them at, zen3 core 3.24mm² and golden cove approx 7mm² while gracemont is roughly the same as zen2 at 2.78mm) So many factors to a cores efficiency but for its size and power envelope golden cove is very much not efficient, for that matter compared to zen3 gracemont isn't either.

 

[Bencher]

Ok let me put it this way so we can understand. Why in the server space do you think Zen/Epyc is the clear choice in terms of performance/density?

Cause Intel doesn't exist anymore, they don't have any products there

Why has the 12900k got gracemont "efficiency cores" if the P cores are so efficient? Why does this 16P core configuration not exist? The answer to all of these questions is because golden cove isn't efficient. It's fast, 100% it is but it isn't efficient. It's big and as a core consumes too much power to scale in a workable package.

But you are just wrong. The e cores are WAY less efficient than the Pcores at same clockspeeds. I mean I can test it for you, the ecores are absolutely atrocious in everything, BUT performance per die space. That's what they are good at. The proper comparison is 1P cores vs 4E cores, since a P core takes the die space of 4 ecores.

Again, as everyone before you, you keep saying it's not efficient, but no one provides some 8 core zen 3 numbers so we can compare....

 

[Cyber-Mav]

Someone lock them both to 65w and run various benches.

 

[Bencher]

Someone lock them both to 65w and run various benches.

Well that's what I asked from the guy with the 5800x who made the claim that zen 3 is more efficient but he completely dodged. Im pretty sure he knows he was wrong

@Vince Here are 8GC cores at 25w

 

[Cyber-Mav]

I need to download cb23 and run some tests. I still got old cb20.

Also isnt gracemont the new atom core which is a considerable jump over the previous tremont core?
How do E cores fair on their own in the benchmark and how much power do they pull on their own?

 

[Bencher]

I need to download cb23 and run some tests. I still got old cb20.

Also isnt gracemont the new atom core which is a considerable jump over the previous tremont core?
How do E cores fair on their own in the benchmark and how much power do they pull on their own?

The ecores are way worse than the pcores at same clockspeeds in terms of efficiency. They only regain some efficiency if you run both at like very low wattages and clockspeeds, which you would never do on a desktop. Most people keep
thinking that the e cores are efficient, but it's not the case. P cores are way more efficient, the problem is intel clocked them ultra high so yeah - obviously they consume a truckload of wattage at 4.9ghz all core speeds, as every other cpu does.


Ill run a cb20 for you at 25 watts.