Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)

[Vattila]

Except for the details about the improvements in the microarchitecture, we now know pretty well what to expect with Zen 3.

The leaked presentation by AMD Senior Manager Martin Hilgeman shows that EPYC 3 "Milan" will, as promised and expected, reuse the current platform (SP3), and the system architecture and packaging looks to be the same, with the same 9-die chiplet design and the same maximum core and thread-count (no SMT-4, contrary to rumour). The biggest change revealed so far is the enlargement of the compute complex from 4 cores to 8 cores, all sharing a larger L3 cache ("32+ MB", likely to double to 64 MB, I think).

Hilgeman's slides did also show that EPYC 4 "Genoa" is in the definition phase (or was at the time of the presentation in September, at least), and will come with a new platform (SP5), with new memory support (likely DDR5).

What else do you think we will see with Zen 4? PCI-Express 5 support? Increased core-count? 4-way SMT? New packaging (interposer, 2.5D, 3D)? Integrated memory on package (HBM)?

Vote in the poll and share your thoughts!

 

[uzzi38]

Could you be more delusional? AMD it's not done yet beating Intel and it's only gonna get worst for Intel.

Nah, the future is one of close competition, thankfully. The more competition, the better, and thankfully I don't think we're going to be let down in that regard.

Well, I'm talking about the desktop market when I say that, I don't want to take any bets on server.

 

[AtenRa]

I think you are arguing semantics. Technically you should not be even allowed to say "x percent less time is x percent faster", because what does that even mean? What exactly is faster? Its not the time to do the task, because time cant go "faster". Thus its the processor, which as you yourself agree, is 46 percent faster.
Thats the only conclusion to this and i dont see any point in further arguing about it. But whatever floats your boat.

I believe the conception of LESS TIME = faster is not understood here.

Try to put this under the context of

LESS TIME = faster
and
MORE TIME = slower


ps. I said that the performance (instructions/sec) of the ZEN 4 is 46% higher vs 12900K in order to finish the benchmark at 31% less time , I didnt agree that ZEN 4 in the Blender benchmark is 46% faster.

 

[uzzi38]

Let's back to topic. I don't want to join Zen3 vs Golden, 5950x vs 12900k debate...

DDR5-only factor makes the speculation about Zen4 even more complicated.... these comments/questions below are what I'm wondering too....

https://twitter.com/x/status/1529220869976883210

https://twitter.com/x/status/1529229372971880448

DDR5 won't have any benefit in Cinebench, not sure about Blender off the top of my head either. It will have an impact on things like the Adobe suite though, at least certain applications like Premiere should see a gain.

 

[Paul98]

In the benchmark "faster means less time", it has nothing to do with speed.


In the benchmark, "less time = faster" and "more time = slower"

2s = 50% faster vs 4s

4s = 100% slower vs 2s

we are only using time, and less time is faster , more time is slower

start with 4s and reduce it's time by half(50%) what do you get?
start with 4s make it 100% faster what do you get?


Like in AMD's case if it's a 31% reduction in the time it takes, it's 1/.31 faster or 45% faster

 

[maddie]

In the benchmark "faster means less time", it has nothing to do with speed.


In the benchmark, "less time = faster" and "more time = slower"

2s = 50% faster vs 4s

4s = 100% slower vs 2s

we are only using time, and less time is faster , more time is slower

Pythagoras is turning in his grave.

 

[ondma]

No you cannot! Really, what the heck is going on?! You're bending Math as if it's an urban dictionary.

Something that takes 50% less time is 100% faster, with no exception. Either you understand this and reconsider your position or we have no common ground to communicate on.

Lets say a task takes 10 min. 50% less time is 5 min. The new time to complete the task is (10 - 5) = 5 min. So it is faster by 5 min compared to the original 10 min, so is that not 5/10 or 50% faster, compared to the original task? That seems the only logical comparison to me. The denominator is the original time.

 

[IntelUser2000]

Are you two so seriously bad at math to not understand this?

If you take the context with time, then less time = faster and more time = slower

In that context of time, you can say that finishing the benchmark at 31% less time equals 31% faster that the competition.
After all, you need 31% less time to finish the same job and that equals for that CPU to be 31% faster as of using 31% less time.

Same applies for 50% less time, you just finished the same job at half the time or 50% faster in the context of time.

ps. I said that the performance (instructions/sec) of the ZEN 4 is 46% higher vs 12900K in order to finish the benchmark at 31% less time , I didnt agree that ZEN 4 in the Blender benchmark is 46% faster.

Lets say a task takes 10 min. 50% less time 5 min. The new time to complete the task is (10 - 5) = 5 min. So it is faster by 5 min compared to the original 10 min, so is that not 5/10 or 50% faster, compared to the original task? That seems the only logical comparison to me. The denominator is the original time.

So let me ask you this. One CPU takes 100 seconds to complete a task, and another CPU takes 0 seconds to complete a task. How much faster is the second CPU?

According to you two, the second CPU is 100% faster in time.

Second question: Car A is speeding at 100km/h. Car B is crawling at 1km/h. How much faster is Car B?

 

[Timmah!]

I believe the conception of LESS TIME = faster is not understood here.

Try to put this under the context of

LESS TIME = faster
and
MORE TIME = slower


ps. I said that the performance (instructions/sec) of the ZEN 4 is 46% higher vs 12900K in order to finish the benchmark at 31% less time , I didnt agree that ZEN 4 in the Blender benchmark is 46% faster.

If the performance of the CPU is 46 percent higher, than the processor is 46 percent faster. Do you want to claim 46 percent better performing processor is 31 percent faster? Surely not...


"we are only using time, and less time is faster , more time is slower"

This is the exact place, where you are wrong. If you are talking faster/slower, you are talking SPEED. Thus, you are not only "using time".

 

[Hitman928]

Lets say a task takes 10 min. 50% less time 5 min. The new time to complete the task is (10 - 5) = 5 min. So it is faster by 5 min compared to the original 10 min, so is that not 5/10 or 50% faster, compared to the original task? That seems the only logical comparison to me. The denominator is the original time.

As I said before, try doing it your way and calculate how much time the faster version would be for a 100% speedup. You won't get an answer that makes sense.

 

[IntelUser2000]

A light will travel at 300,000kmh, that is, three hundred thousand kilometers per hour.

You are walking at a pace of 3km/h. How much faster is the light compared to you?

I know some people will have the time of their lives trying to get this. The basics are lost once they are out of high school.

 

[coercitiv]

4s = 100% slower vs 2s

we are only using time, and less time is faster , more time is slower

There is no such thing as 100% slower when talking about time, the same way as there is no such think about 100% lighter when talking about mass, or 100% smaller when talking about volume.

So it is faster by 5 min compared to the original 10 min, so is that not 5/10 or 50% faster, compared to the original task?

I can eat a hamburger in 10 mintes, you can eat one in 5 minutes. How much faster are you at eating hamburgers?

PS: our relative speed in space is negligible.

 

[Abwx]

......

Also, 4 seconds is double the time of 2 seconds, that means that 4 seconds is 100% SLOWER vs 2 seconds.

If a CPU take 2s instead of 4s its is 2x faster, that is 100% faster.

It also require 0.5x the time, that is, 50% less time because the other CPU is 50% slower, that is, it run at 50% of the faster device speed.

 

[tamz_msc]

Sigh. Are we still talking about CPUs in this thread?

 

[nicalandia]

Sigh. Are we still talking about CPUs in this thread?

We are, but there are posters here that can't comprehend basic elementary math. They will not stop bickering about if it's 31% or 40% Faster on a Blender Render

Are you two so seriously bad at math to not understand this?

Ca We Please Stop with this nonsense? It's Painlessly obvious that you can't comprehend relativistic Space/Time measurements.

 

[biostud]

Sigh. Are we still talking about CPUs in this thread?

I donno, but for some reason I remember that you could upgrade a 386 with a math co-processor and that the first Pentium P60 had a calculation error....

 

[nicalandia]

I donno, but for some reason I remember that you could upgrade a 386 with a math co-processor and that the first Pentium P60 had a calculation error....

So if I added 2 + 2 on a Pentium P60 I was going to get 5?

 

[Timorous]

So if I added 2 + 2 on a Pentium P60 I was going to get 5?

Yes for large values of 2.

 

[biostud]

So if I added 2 + 2 on a Pentium P60 I was going to get 5?

The Pentium FDIV bug is a hardware bug affecting the floating-point unit (FPU) of the early Intel Pentium processors. Because of the bug, the processor would return incorrect binary floating point results when dividing certain pairs of high-precision numbers

Pentium FDIV bug - Wikipedia

en.m.wikipedia.org

 

[ondma]

Are you two so seriously bad at math to not understand this?






So let me ask you this. One CPU takes 100 seconds to complete a task, and another CPU takes 0 seconds to complete a task. How much faster is the second CPU?

According to you two, the second CPU is 100% faster in time.

Second question: Car A is speeding at 100km/h. Car B is crawling at 1km/h. How much faster is Car B?

If the performance of the CPU is 46 percent higher, than the processor is 46 percent faster. Do you want to claim 46 percent better performing processor is 31 percent faster? Surely not...


"we are only using time, and less time is faster , more time is slower"

This is the exact place, where you are wrong. If you are talking faster/slower, you are talking SPEED. Thus, you are not only "using time".

I see what you are saying, but essentially one metric is looking at rate of performance, while the other is looking at time to complete the task. So you can say it is 46% faster at rate of performance, but the time to complete the task is only 31% less. If it makes you feel better, I will re-phrase as so;
The rate of performance is 46% faster, but the time saved, compared to the original time, is only 31%. (Mathematically, it comes down to whether the calculation uses the smaller or larger number for the denominator.)

@ Inteluser: BTW, your first two examples are absurd. No cpu can complete a task in 0 seconds (the rate of performance would be infinite), and Car B is not faster than car A.

 

[Schmide]

rise over run away, if you keep doubling down your rate increases exponentially, there is a limit to this convergence

 

[Thibsie]

I hope AMD has an in between socket for Threadripper and workstation parts with Zen 4. Using a giant Epyc IO die and a giant SP5 sized socket seems very wasteful.

SP6 - Epyc

 

[PJVol]

You cannot say 31% less time is 46% faster, the performance of the CPU is 46% higher and that translates in to finishing the benchmark at 31% less time of 31% faster.

https://twitter.com/x/status/1528719223786192896

It's funny you quoted that tweet with the same flawed logic.
There's a statement of
zen4 31% faster in Blender (based on time elapsed)
zen4 ~40% faster in Cinebench (based on CB score)

The thing is Cinebench score is calculated from the time spent rendering the scene.

Example: My 5600Х СB R23 two runs, PBO+200 and Fmax capped:
* 4850Mhz: 12440 finished in 67 sec
* 4000Mhz: 10936 finished in 75 sec
Based on time elapsed 1st run takes 17% less time and the score is 20% higher.

Q: How much faster uncapped CPU in this bench?

 

[majord]

If blender benchmark was measured in pixels per second or something , there wouldn't be so many confused people.

lets assume the blender renderd imaged had 1000 pixels (obviously it's more but does'nt matter ,you'll get the same answer regardless )


207s to render 1000 pixels = 1000 / 207 = 4.83 pixels/second.

297s to render 1000 pixels = 1000 / 297 = 3.37 pixels / second

4.83 / 3.37 = 1.43x = 43% faster

if anyone wants to still insist it's 31% faster, they'll have to also argue that every gaming performance review in the history of gaming performance reviews is wrong, and redo them all.. since they all use FPS , and calculate percentage the same way.

Oh, and you'll still be wrong..

 

[Hans de Vries]

So, there are 3 buses. Which bus came last?
The correct answer = ?


(total confusion)

 

[nicalandia]

So, there are 3 buses. Which bus came last?
The correct answer = ?

View attachment 62109
(total confusion)

I just reported you for being Off Topic and adding more confusion to this thread.