AMD talks next-gen Zen 4 CPUs, Ryzen 7000, Socket AM5, and more

[Strawman]

15K@35 watts makes 428 points per watt.

AMD Ryzen 6800U scores 14,231 @ 28 watts. That makes 508 points per watt. Your 12900K gets smoked by AMD laptop CPU on both performance and power consumption. I won't even bother to say how much it trashes your 12900K @35W on other tasks...

According to the reviews....it doesn't? From what I've seen it scores less than 10k at 28watts. Can you point me somwehere that says otherwise, or are you just shilling for amd again?

Are you seriously telling me that a 6800u scores the same as a 5800x while the latter consumes like more than 3 times the wattage??? Is the 5800x that freaking terrible, or are you just mistaken?

 

[Endymio]

Not even close, I didn't talk about his position at all, so by definition it cannot be a strawman. But, don't worry about it, you can try agian in the future

So a "Strawman" argument is not a strawman argument? Ok.

 

[Strawman]

So a "Strawman" argument is not a strawman argument? Ok.

 

[HardReset]

According to the reviews....it doesn't? From what I've seen it scores less than 10k at 28watts. Can you point me somwehere that says otherwise, or are you just shilling for amd again?

Just searched for results and found that one. Perhaps that result was flawed. Also found another that is over 14K. You can easily find those results. Problem with mobile CPU benchmarks is that TDP value and/or cooling may be something else than maximum.

Are you seriously telling me that a 6800u scores the same as a 5800x while the latter consumes like more than 3 times the wattage??? Is the 5800x that freaking terrible, or are you just mistaken?

Possible since...

1. 5800X is 7nm, not 6 nm
2. 5800X has IO die that is much less effective than monolithic design. I was never fan of chiplet design on desktop CPU.
3. 6800U was built power consumption in mind

 

[Strawman]

Just searched for results and found that one. Perhaps that result was flawed. Also found another that is over 14K. You can easily find those results. Problem with mobile CPU benchmarks is that TDP value and/or cooling may be something else than maximum.

Possible since...

1. 5800X is 7nm, not 6 nm
2. 5800X has IO die that is much less effective than monolithic design. I was never fan of chiplet design on desktop CPU.
3. 6800U was built power consumption in mind

I googled 50 results, it barely exceed 10k in a couple, in the rest it was below 10k. Please, show me a link where it actually gets 14k at 28watts. Else you are just shilling, as is the usual.

And anyways, you are admitting that the DESKTOP 12900k absolutely smokes zen 3 in efficiency. Yeah, thats my point

 

[HardReset]

I googled 50 results, it barely exceed 10k in a couple, in the rest it was below 10k. Please, show me a link where it actually gets 14k at 28watts. Else you are just shilling, as is the usual.

How hard it is actually search for:

Ryzen 6800U 14,231

Or Ryzen 6800U R23 14***

It's not that hard, believe me.

For so called "reliable" results I found were around 10K but either TDP was under 20W OR it was not specifically told. That's the problem about mobile CPU benchmarks. Different TDP, different cooling etcetc.

And anyways, you are admitting that the DESKTOP 12900k absolutely smokes zen 3 in efficiency. Yeah, thats my point

Probably smokes on CinePlay but on many other tasks don't. Also higher performance almost always mean lower efficiency if architecture is somewhat same.

 

[Strawman]

How hard it is actually search for:

Ryzen 6800U 14,231

Or Ryzen 6800U R23 14***

It's not that hard, believe me.

For so called "reliable" results I found were around 10K but either TDP was under 20W OR it was not specifically told. That's the problem about mobile CPU benchmarks. Different TDP, different cooling etcetc.


Probably smokes on CinePlay but on many other tasks don't. Also higher performance almost always mean lower efficiency if architecture is somewhat same.

Please, tell me what other app you want me to run and let's compare it to zen 3. I promise you, 12900k will smoke ANYTHING including the m1 in efficiency.

Also, yeah, as you just figured, the 6800u does not get 14k at cbr23, LOL. So yeah, I think you should take back the "smoking the 12900k part".

 

[HardReset]

Please, tell me what other app you want me to run and let's compare it to zen 3. I promise you, 12900k will smoke ANYTHING including the m1 in efficiency.

Using default settings? I don't think so.

Also mobile CPU testing is usually done on "hot" chip, not cold one. I doubt your 12900K has any cooling problems at 35W since it's desktop computer. That makes comparison pretty hard.

Not to say, you could tweak other CPUs also. Cinebench probably could run on lower voltage etc etc. Since you also didn't run on default settings, your claim is pretty hard to prove. And I don't bother to disprove it either.

Also, yeah, as you just figured, the 6800u does not get 14k at cbr23, LOL. So yeah, I think you should take back the "smoking the 12900k part".

Perhaps until I find some reliable benchmarks...

 

[Strawman]

Using default settings? I don't think so.

Also mobile CPU testing is usually done on "hot" chip, not cold one. I doubt your 12900K has any cooling problems at 35W since it's desktop computer. That makes comparison pretty hard.

Not to say, you could tweak other CPUs also. Cinebench probably could run on lower voltage etc etc. Since you also didn't run on default settings, your claim is pretty hard to prove. And I don't bother to disprove it either.

A tweaked 5950x gets half my score at 35w. Zen 3 can't compete in absolute efficiency cause of the IOD. It's just a fact. I can almost match a stock 5950x score at just 100watts (I score 24900).

It can only compete in efficiency when you are running alderlake at 5ghz all core, in which case, yeah, sure, every CPU is inefficient at that point.

 

[HardReset]

A tweaked 5950x gets half my score at 35w. Zen 3 can't compete in absolute efficiency cause of the IOD. It's just a fact. I can almost match a stock 5950x score at just 100watts (I score 24900).

It can only compete in efficiency when you are running alderlake at 5ghz all core, in which case, yeah, sure, every CPU is inefficient at that point.

Zen3 desktop chips were not meant to be driven at 35W. You could also tweak M1 to take 100 watts and then blame for it being ultra inefficient.

Zen3 desktop CPU's were never meant to be efficient so that's not real surprise.

 

[Strawman]

Zen3 desktop chips were not meant to be driven at 35W. You could also tweak M1 to take 100 watts and then blame for it being ultra inefficient.

Zen3 desktop CPU's were never meant to be efficient so that's not real surprise.

Im pretty sure alderlake were never meant to be efficient either...but here we are

 

[AlaskaGuy]

AMD Corrects Socket AM5 for Ryzen 7000 Power Specs: 230W Peak Power, 170W TDP (Updated)

Pump up the power

www.tomshardware.com

 

[HardReset]

AMD Corrects Socket AM5 for Ryzen 7000 Power Specs: 230W Peak Power, 170W TDP (Updated)

Pump up the power

www.tomshardware.com

Intel launched CPU actually consumes that much, AMD's future CPU released sometime May Perhaps consume that much.

Big difference.

If you don't get it, just look at LGA1151 fiasco.

 

[AlaskaGuy]

Intel launched CPU actually consumes that much, AMD's future CPU released sometime May Perhaps consume that much.

Big difference.

If you don't get it, just look at LGA1151 fiasco.

230W

 

[dalpilot]

I have been an AMD supporter even through the "dark year" of FX-6000 and 8000 CPU's. I was rewarded with the Zen 1 and AMD B450 motherboards. It was an amazing jump from FX-8300 to Ryzen 5 1600. I'm not a big FPS gamer (more RTS which also requires some pretty good CPU power if you don't want to wait forever for "AI Computer" moves in 1st person mode) but I do a LOT of video conversion - from rippred DVDs/Blu-Rays to recorded series from Netflix,etc. (We hate credits and commercials) and the Ryzen 5-1600 was a big leap forward for those tasks. I also jumped at the tail end sales of the Ryzen 5-3600 before it was discontinued and intend to use the Ryzen 5-1600 as my B/U computer (I have an extra MB (B-450), extra RAM (8 GB DDR-2600 that I kept when I upgraded to 16 GB of DDR-3000 RAM (in two sticks for dual channel compatibility) and an "extra" RX-470 when I picked up a "cheap" RX-570 on sale. I have all kinds of SSDs and regular HDs to build my B/U computer. And all of my videos are backed up (one on and one off site) so easily transferred if my primary computer goes on the blink. I am thinking I may do the same if the Ryzen 5 5600 goes on sale at its end of life for below $200 to get the final extension of the AM4 socket. It's been a great ride and to me a reward for helping keep AMD in business all those long years.

 

[Shadowboxer]

E-cores are a waste of silicon on desktop parts

If you genuinely believe this then you have fundamentally misunderstood how modern chip designs work. You can fit 4 e cores in the space of one performance core on a die. They use less power too.

Maybe you should write to all the chip manufacturers, AMD included and tell them that they are wasting their silicon? Practically every chip company is working on hybrid solutions.

 

[HardReset]

If you genuinely believe this then you have fundamentally misunderstood how modern chip designs work. You can fit 4 e cores in the space of one performance core on a die. They use less power too.

And they also perform worse on things important on desktop. And who cares about die size at all? AMD could fit 8 Zen3 cores on 81 mm2. Around half of that are caches.

Only reason Intel had to include E-cores is high power consumption of P-cores. That's only reason.

Feel free to tell why Intel is not and won't be using using E-cores on servers (Alder Lake based) where efficiency and multi core performance is important.

Maybe you should write to all the chip manufacturers, AMD included and tell them that they are wasting their silicon? Practically every chip company is working on hybrid solutions.

Yeah, AMD is working on hybrid solution. According rumours AMD will use Zen4 architecture with their "E-cores", not any Atom crap like Intel.

 

[yRaz]

If you genuinely believe this then you have fundamentally misunderstood how modern chip designs work. You can fit 4 e cores in the space of one performance core on a die. They use less power too.

Maybe you should write to all the chip manufacturers, AMD included and tell them that they are wasting their silicon? Practically every chip company is working on hybrid solutions.

well maybe I fundamentally misunderstand the fundamentals

 

[Strawman]

And they also perform worse on things important on desktop. And who cares about die size at all? AMD could fit 8 Zen3 cores on 81 mm2. Around half of that are caches.

Only reason Intel had to include E-cores is high power consumption of P-cores. That's only reason.

Feel free to tell why Intel is not and won't be using using E-cores on servers (Alder Lake based) where efficiency and multi core performance is important.

Didnt we discuss this before? Didn't we agree that if you turn down power limits Intel is way more efficient right now? 15k cbr23 @ 35 watts, beats the crap out of zen 3.

E cores are saving DIE space. Alderlake cores are very wide, that's why they have that increased IPC, which means they take way more die space.

FYI, intel managed to fit an 18 core monolithic die at 14nm, and you are telling me they can't make 16 at 10nm cause of power consumption? Ohkay

 

[HardReset]

Didnt we discuss this before? Didn't we agree that if you turn down power limits Intel is way more efficient right now? 15k cbr23 @ 35 watts, beats the crap out of zen 3.

For some reason, Intel decided to release CPU's that are much more inefficient.

E cores are saving DIE space. Alderlake cores are very wide, that's why they have that increased IPC, which means they take way more die space.

And who cares? Why is saving die space so important? Alder Lake 8 core chip is not even large. So why would Intel be concerned about saving die space? On servers where more cores and efficiency is more important than on desktop, Intel plans Not to use efficiency cores. That essentially proves they are waste of die space on desktop.

FYI, intel managed to fit an 18 core monolithic die at 14nm, and you are telling me they can't make 16 at 10nm cause of power consumption? Ohkay

Of course I claim. Just look how much power 8 P-cores consume. Double that figure and think again. Of course they Could but...

Those 14nm dies are not Alder Lake but Skylake. And Skylake cores are much cooler than Alder Lake ones.

 

[Strawman]

For some reason, Intel decided to release CPU's that are much more inefficient.

And who cares? Why is saving die space so important? Alder Lake 8 core chip is not even large. So why would Intel be concerned about saving die space? On servers where more cores and efficiency is more important than on desktop, Intel plans Not to use efficiency cores. That essentially proves they are waste of die space on desktop.

Of course I claim. Just look how much power 8 P-cores consume. Double that figure and think again. Of course they Could but...

Those 14nm dies are not Alder Lake but Skylake. And Skylake cores are much cooler than Alder Lake ones.

Saving die space is important for yields. You realize its a monolithic chip, right? Sure they can build a huge die with 16 p cores, but then the actual cost would be stupendous.

What do you mean its not even big? Its freaking huge, probably at the limit of what is cost efficient to produce. The zen 3 dies are less than half the size, lol.

Sir, does the 5950x consume double as much power as the 5800x? No? Then why the heck do you think a hypothetical 16p core intel would? You do understand they can limit the cpu so it consumes as much power as they want to consume, RIGHT? You do realize that the 12700 has twice the pcores + ecores yet it has the same tdp as the 12100? So wtf are you talking about???????

Its pointless talking to you honestly, your amd glasses are just blinding you.

 

[HardReset]

Saving die space is important for yields. You realize its a monolithic chip, right? Sure they can build a huge die with 16 p cores, but then the actual cost would be stupendous.

No it wouldn't. 11900K is larger than Alder Lake 8+8. And considering how big GPU are, answer is very simple.

For comparison:

Zen3 chiplet: around 80mm2
Ryzen 9 6900HS: 208 mm2
6900XT GPU: 520mm2

And that 6900XT is GPU only, not whole card.

What do you mean its not even big? Its freaking huge, probably at the limit of what is cost efficient to produce. The zen 3 dies are less than half the size, lol.

Becaus aroudn 200mm2 is not big at all. If around 200mm2 is big, then what is Nvidia's 6080 GPU, 628 mm2.

And too expensive? Even assuming very poor 50% yields and around 300mm2 die size (die size may be larger but yields much better too), that makes around 100 chips per wafer. Assuming wafer cost 10K$, that makes 100 dollars per chip. Still very profitable.

Once again, you have absolutely no clue what you are talking about.

Sir, does the 5950x consume double as much power as the 5800x? No? Then why the heck do you think a hypothetical 16p core intel would? You do understand they can limit the cpu so it consumes as much power as they want to consume, RIGHT? You do realize that the 12700 has twice the pcores + ecores yet it has the same tdp as the 12100? So wtf are you talking about???????

Its pointless talking to you honestly, your amd glasses are just blinding you.

I didn't say anything about clock speeds, did I? Of course they can limit power consumption but why put more cores and then lower clock speeds so much it's effectively pointless?

TDP is not power consumption. Never have been.

How about learning basics next time?

 

[Strawman]

No it wouldn't. 11900K is larger than Alder Lake 8+8. And considering how big GPU are, answer is very simple.

On 14 nms that are 10 years old now ,having like 99.9% yields...seriously dude?

 

[Strawman]

Zen3 chiplet: around 80mm2
Ryzen 9 6900HS: 208 mm2
6900XT GPU: 520mm2

And that 6900XT is GPU only, not whole card.

Becaus aroudn 200mm2 is not big at all. If around 200mm2 is big, then what is Nvidia's 6080 GPU, 628 mm2.

Nvidia cards are already cut down chips dude....you keep on going with this arent you?

 

[Strawman]

I didn't say anything about clock speeds, did I? Of course they can limit power consumption but why put more cores and then lower clock speeds so much it's effectively pointless?

TDP is not power consumption. Never have been.

How about learning basics next time?

HOW can any company DOUBLE the core count while keeping the clockspeeds and the consumption steady? DUDE, remove your glasses, you absolutely make no sense. Yes, OBVIOUSLY if they wanted to keep the clockspeeds they would have to increase the power consumption, but so does AMD. Is 5950x useless? Cause the clockspeeds are WAY lower than the 5800x in all core workloads. So, wtf are you talking about again?