Ryzen 9 7900 at 44 Watts = defeats Threadripper 1950x in Cinebench.

[tugrul_SIMD]

And its without reaching 50 Celcius.

12 cores vs 16 cores: 1.33x IPC

Cinebench r23 17400 vs 16300: 1.07x IPC

3GHz(44W) vs 3.6GHz: 1.2x IPC

70% higher IPC than the first Zen architecture. But at 1/4 power.

 

[fevgatos]

Color me unimpressed. 12900k @ 33w scores over 15k

 

[kapone32]

Color me unimpressed. 12900k @ 33w scores over 15k

Why do you always?

 

[GreiverBlade]

Color me unimpressed. 12900k @ 33w scores over 15k

well ... 4 core more same number of thread, 60w tdp more and cost 549 chf versus 410.80 chf for me ... the R9 7900 would still be the winner for me
(okay okay there is a running promotion at 489chf for the 12900k but still 78.20 chf ... i could get something else with that price difference )

so, the op was compairing Ryzen versus threadripper first gen, an innocent and interesting point of view on how AMD improved over the year and you come out with a "stick" waving about Intel doing 2.4k less at 11w less?

tbf even if the 12900k reach or equal the R9 7900 at 44w, it would be logical right? but given the other differences around ... even you should reckon that the R9 7900 is good, but you will not

 

[pavle]

Color me unimpressed. 12900k @ 33w scores over 15k...

Please, do test it at 44W so we can see what's what.

 

[fevgatos]

Why do you always?

I just posted a result, what's wrong with that? Should I be impressed that a newer more expensive cpu gets a worse score while consuming more watts? Wtf, I'm at a loss

well ... 4 core more same number of thread, 60w tdp more and cost 549 chf versus 410.80 chf for me ... the R9 7900 would still be the winner for me
(okay okay there is a running promotion at 489chf for the 12900k but still 78.20 chf ... i could get something else with that price difference )

so, the op was compairing Ryzen versus threadripper first gen, an innocent and interesting point of view on how AMD improved over the year and you come out with a "stick" waving about Intel doing 2.4k less at 11w less?

tbf even if the 12900k reach or equal the R9 7900 at 44w, it would be logical right? but given the other differences around ... even you should reckon that the R9 7900 is good, but you will not

I think you are confused. The 7900 at 35w scores 12k. The 12900k at 33w scores 15k.

 

[AlwaysHope]

So comparing a hybrid architecture with a non hybrid architecture & then arguing over a few watts of diff in performance?
Please, get a life.

 

[GreiverBlade]

I just posted a result, what's wrong with that? Should I be impressed that a newer more expensive cpu gets a worse score while consuming more watts? Wtf, I'm at a loss


I think you are confused. The 7900 at 35w scores 12k. The 12900k at 33w scores 15k.

and i think you did read crossway ... because i did not mention that and also the 12900k, even tho it's older, is more expensive as i mentioned (discounted is almost alright but still)

well the 13900k is even more expensive

the OP was also not mentioning Intel and the result from 1st gen threadripper to current gen Ryzen 9 is, indeed, impressive.

also 12k at 35w and 5.4k more at "just" 9w more is also impressive, thanks for your second input

 

[fevgatos]

So comparing a hybrid architecture with a non hybrid architecture & then arguing over a few watts of diff in performance?
Please, get a life.

You are right, I'm sorry guys, I'm really impressed with those 7900 numbers, I've never seen anything like it, insane efficiency

 

[GreiverBlade]

again missing the point ... the OP was showing AMD to AMD

 

[Beginner Micro Device]

AMD to AMD

AMD always loses at this match-up. AMD is completely IMBA and should be purged for the sake of sanity!

 

[Dr. Dro]

AMD always loses at this match-up. AMD is completely IMBA and should be purged for the sake of sanity!

That's normal. With AMD, you lose. AMD always loses, even when it wins.

And its without reaching 50 Celcius.

12 cores vs 16 cores: 1.33x IPC

Cinebench r23 17400 vs 16300: 1.07x IPC

3GHz(44W) vs 3.6GHz: 1.2x IPC

70% higher IPC than the first Zen architecture. But at 1/4 power.

In practice it should be quite a bit more than 70% over the 1800X. I was amazed at the 5950X in Eco mode being faster than my previous 3900XT in PBO at Cinebench, I'm glad to see Zen 4 keeping up the trend.

 

[Outback Bronze]

again missing the point ... the OP was showing AMD to AMD

Just show him how efficient the 7800X3D is in gaming then.

To OP yes. Major difference in gen arch.

 

[Beginner Micro Device]

AMD always loses, even when it wins.

AMD is like a 5D chess player who never ever read the lore of 4D chess whatsoever. That's why their moves are so genius and inhumanely silly at the same time.

 

[Dr. Dro]

AMD is like a 5D chess player who never ever read the lore of 4D chess whatsoever. That's why their moves are so genius and inhumanely silly at the same time.

Wow, you actually saw through it clear as crystal, that's exactly what I meant

 

[fevgatos]

Just show him how efficient the 7800X3D is in gaming then.

To OP yes. Major difference in gen arch.

It's 22% more efficient than my stock 12900k, not bad, not great either

Star Wars Jedi: Survivor Benchmark Test & Performance Analysis

Well check this video, it's a 12900k. Look at the CPU usage and tell me what you think would happen if I turned off those ecores? Buh, there are lot of broken beyond fixing games. Spiderman works better (and I mean, 20-25% better) with HT off. None of that makes sense, on a hardware level...

www.techpowerup.com

 

[Marcus L]

It's 22% more efficient than my stock 12900k, not bad, not great either

Star Wars Jedi: Survivor Benchmark Test & Performance Analysis

Well check this video, it's a 12900k. Look at the CPU usage and tell me what you think would happen if I turned off those ecores? Buh, there are lot of broken beyond fixing games. Spiderman works better (and I mean, 20-25% better) with HT off. None of that makes sense, on a hardware level...

www.techpowerup.com

From TPU's own review and over a suite of 13 games it's almost 100% more efficient (49w Vs 98) put your pecker away and take it to an Intel thread if you want to wave your e-peen around

 

[fevgatos]

From TPU's own review and over a suite of 13 games it's almost 100% more efficient (49w Vs 98) put your pecker away and take it to an Intel thread if you want to wave your e-peen around

I'm not, he asked, I replied.

EG. Efficiency is measure at same wattage. Anything else is just a faulty way of doing it.

 

[Marcus L]

I'm not, he asked, I replied.

He asked cause you have thread crapped since the OP

 

[fevgatos]

He asked cause you have thread crapped since the OP

I didn't, I posted once and people quoted me 19 times just because I wasnt impressed. I'm sorry, I should have been impressed, I should have politely asked my jaw to drop to the floor.

 

[Vya Domus]

Anything else is just a faulty way of doing it.

According to who, you ? Utter nonsense as usual.

EG. Efficiency is measure at same wattage.

No it's not. Efficiency is a ratio that you obtain from calculating work / energy expenditure, it's completely irrelevant whether or not it's at the same wattage because the thing that you're measuring is a ratio between two units of measurement.

100 fps / 40W is the exact same efficiency as 50 fps / 20W or 5 fps / 2W.

 

[Kyan]

I didn't, I posted once and people quoted me 19 times just because I wasnt impressed. I'm sorry, I should have been impressed, I should have politely asked my jaw to drop to the floor.

You really don't make any effort to be more pleasant to talk to. The thread title and first post was just putting some comparison of older vs newer amd cpu, and as the first answer to that statement, you put an intel result. What was the point ? If you where comparing older intel of roughly same age as the threadripper and the 12900k, ok that's a valid comparison and it can be compare with this amd evolution comparison.

 

[fevgatos]

According to who, you ? Utter nonsense as usual.


No it's not. Efficiency is a ratio that you obtain from calculating work / energy expenditure, it's completely irrelevant whether or not it's at the same wattage because the thing that you're measuring is a ratio between two units of measurement.

100 fps / 40W is the exact same efficiency as 50 fps / 20W or 5 fps / 2W.

According to common sense. Voltage doesn't scale linearly with performance, therefore the more you push a cpu the less efficient it will be.

If you don't compare at same wattages then you end up at stupid conclusions. For example at stock the 5950x is more efficient than the 7950x, but that's only because the latter has a way higher power limit. So if you are looking for the most efficient cpu, the right choice would be the 7950x because at same wattages it blasts the 5950x into oblivion in terms of performance, and therefore efficiency

 

[A Computer Guy]

And its without reaching 50 Celcius.

12 cores vs 16 cores: 1.33x IPC

Cinebench r23 17400 vs 16300: 1.07x IPC

3GHz(44W) vs 3.6GHz: 1.2x IPC

70% higher IPC than the first Zen architecture. But at 1/4 power.

The Threadripper still wins in supported number of PCIe lanes. I miss the old days where you had at least 5 useable slots at any time in a normal PC.

 

[Vya Domus]

According to common sense.

This is according to math not common sense, now I am starting to understand why you're wrong every time you post something.

Voltage doesn't scale linearly with performance, therefore the more you push a cpu the less efficient it will be.

If you don't compare at same wattages then you end up at stupid conclusions.

Once again you do not know math or physics. It's the power which does not scale linearly with voltage it has nothing to do with performance, that doesn't make any sense.

Wattage in ICs is a function of voltage, frequency and capacitance. There is simply no way to know if 2 CPUs at the same wattage are both operating within their optimal efficiency curve because the frequency and voltages are going to be different, as well as the manufacturing process. Energy efficiency does not scale linearly with voltages either because it's a function of power, which isn't linear. This means that trying to measure efficiency at arbitrary wattages does not offer any useful information whatsoever to anyone, expect you it seems.

You think that if you equalize the wattage that means both CPUs are always operating optimally but you're wrong, because you don't know the math behind how that works.