News Intel's Core i9-13900KS Rips Ryzen 9 7950X In Early Benchmark

[alceryes]

I thought we already agreed that a 105W TDP ryzen auto overclocks to 142W...
Intel at least states that 253W is the max boost, AMD keeps the 142W number as hidden as possible.

No. Ryzen 5950X limited to 105W (or 142W PPT) is without PBO enabled. This chip draws over 180W TDP not PPT when stressed with PBO enabled. If you want to test for yourself open up Ryzen Master both with and without PBO enabled. You'll see the 142W PPT limit there when PBO is disabled.

So, again (despite my 105/142W mix up) we have higher memory speeds on the 13900k/12900k and we have the 13900k and 12900k allowed to boost all the way up to 253W and 241W TDP. This is compared to a 5950X that is not allower to boost (limited to 105W TDP) and with slower 3200MT/s memory.

 

[SunMaster]

No. Ryzen 5950X limited to 105W (or 142W PPT) is without PBO enabled. This chip draws over 180W TDP not PPT when stressed with PBO enabled. If you want to test for yourself open up Ryzen Master both with and without PBO enabled. You'll see the 142W PPT limit there when PBO is disabled.

Unless restricted my 5950x draws up to 270 watt sustained PPT when doing a test like Cinebench.

 

[alceryes]

Unless restricted my 5950x draws up to 270 watt sustained PPT when doing a test like Cinebench.

True. But insane PPT can be produced for both the 13900k and 5950X with an LN₂ pot.

My want was more to compare apples to apples to give a representation of what to expect in a normal systems builder environment.

 

[TerryLaze]

No. Ryzen 5950X limited to 105W (or 142W PPT) is without PBO enabled. This chip draws over 180W TDP not PPT when stressed with PBO enabled. If you want to test for yourself open up Ryzen Master both with and without PBO enabled. You'll see the 142W PPT limit there when PBO is disabled.

So, again (despite my 105/142W mix up) we have higher memory speeds on the 13900k/12900k and we have the 13900k and 12900k allowed to boost all the way up to 253W and 241W TDP. This is compared to a 5950X that is not allower to boost (limited to 105W TDP) and with slower 3200MT/s memory.

Precision boost is the normal turbo and is allowed under warranty, precision boost override is not and will void your warranty.
Max turbo power is allowed under warranty.
Both CPUs are boosting as much as the companies allow without the user loosing the warranty and both are using the fastest ram that keep your warranty intact.

 

[alceryes]

Precision boost is the normal turbo and is allowed under warranty, precision boost override is not and will void your warranty.
Max turbo power is allowed under warranty.
Both CPUs are boosting as much as the companies allow without the user loosing the warranty and both are using the fastest ram that keep your warranty intact.

So waranty is the deciding factor for you then. Okay.
It seems that, in your world, an apples to apples comparision can never be realized.

 

[blkspade]

All the games are made for consoles and consoles have a shared pool of ram between cpu and gpu...even if this isn't directly the issue it's still the issue.
Rebar improves the data transfer between ram and vram I guess more cache does the same.

Think about it, you really think all the CPU benchmarks ALL fit completely into the normal cache? Because otherwise we would have huge increases in any CPU benchmark that doesn't fit into normal cache as well.

Most CPU benchmarks are prolonged work where the end result is accumulated a over some relatively long period of time. There is so much its design to do per clock cycle, and the entirety of that work is much larger than the cache. A cinebench run for example isn't done until the whole thing is complete over many seconds. It scales with threads and each thread is working on something different from its cores local L2 cache. It takes so long for each tile to complete that the next bit of data is long queued up and just waiting. No matter how much L3 you have. Which is why it also doesn't scale much with memory bandwidth or latency.

For a game however the CPUs entire principal data set (draw call) is literally a single frames worth of information that it's trying to get to the GPU as fast as possible. The CPU is compiling the details to tell the GPU how to draw a frame, and the GPU's ability to draw each successive frame faster is limited by the CPU's ability to tell it to. Which is in milliseconds. At that speed the prior frame can be out of the way faster than system RAM can put the next batch of data in L3. More L3 allows repeating data to remain more present, while also accommodating more unique data.

Rebar is a memory addressing function over PCIe, which is not incredibly different from when we switched from 32bit for 64bit. That's about improving asset loading which is separate work from the Draw calls. The Draw calls do have to act on the data in the frame buffer so it helps to be able to move larger chunks at a time.

The thing about consoles is that they target a fixed frame rate. Which if its 60FPS is equal to ~16.7ms per frame. Not all games are design for console, but if your port of choice has an unlocked frame rate then the CPU has to turn in its work at much faster rate to keep the much stronger GPU fed to maintain a much higher desired minimum.

 

[TerryLaze]

So waranty is the deciding factor for you then. Okay.
It seems that, in your world, an apples to apples comparision can never be realized.

It's the deciding factor for a company selling their product.
Are you ok with a company lying to you about performance, showing you overclocked but not covering it?
What would be the reason of them showing you performance that they don't cover under warranty?
If you want to see overclocking comparisons you will have to wait for 3rd party.

For a game however the CPUs entire principal data set (draw call) is literally a single frames worth of information that it's trying to get to the GPU as fast as possible. The CPU is compiling the details to tell the GPU how to draw a frame, and the GPU's ability to draw each successive frame faster is limited by the CPU's ability to tell it to. Which is in milliseconds. At that speed the prior frame can be out of the way faster than system RAM can put the next batch of data in L3. More L3 allows repeating data to remain more present, while also accommodating more unique data.

So in fewer words it's about how fast the data can go from system ram to the gpu...