3 hours ago, odourboy said:

The cache size is about 2% of the ram required to load MSFS

while code/data in the cache is being accessed at maximum speed, another chunk of instructions and data can be loaded simultaneously in the background from slower main memory. no large application like any flight simulator can/needs to be loaded 100% into memory. why else does the x3D version execute MSFS faster than its non v-cache siblings despite 300 MHz slower clock speeds? 

the reason the X3D v-cache enabled Ryzens perform so exceptionally well compared to other games: in flight simulation there are less sudden extreme events or change of objects being loaded, no sudden opponents or warriors showing up out of nowhere which would require sudden loading of new data, texture and code. code instructions are a lot more predictable, hence less cache misses, hence more cache efficiency. main memory speed is less important the larger the cache and the more effective the use of the cache is. the bad news: AMD was 6 months late to market. what is 15-25% fps increase when you just got 100% performance jumps from Nvidia's frame generation?

unless you have a real old CPU, and if you already have a RTX 4000 with frame generation you can wait and see what RTX 8000 or Intel's Raptor Lake refresh will bring in the near future. new AMD Ryzen 7900x3D CPU, mobo and DDR 5 RAM will cost similar to a RTX 4000.

Edited March 1, 20233 yr by turbomax

9 hours ago, turbomax said:

while code/data in the cache is being accessed at maximum speed, another chunk of instructions and data can be loaded simultaneously in the background from slower main memory. no large application like any flight simulator can/needs to be loaded 100% into memory. why else does the x3D version execute MSFS faster than its non v-cache siblings despite 300 MHz slower clock speeds? 

the reason the X3D v-cache enabled Ryzens perform so exceptionally well compared to other games: in flight simulation there are less sudden extreme events or change of objects being loaded, no sudden opponents or warriors showing up out of nowhere which would require sudden loading of new data, texture and code. code instructions are a lot more predictable, hence less cache misses, hence more cache efficiency. main memory speed is less important the larger the cache and the more effective the use of the cache is. the bad news: AMD was 6 months late to market. what is 15-25% fps increase when you just got 100% performance jumps from Nvidia's frame generation?

unless you have a real old CPU, and if you already have a RTX 4000 with frame generation you can wait and see what RTX 8000 or Intel's Raptor Lake refresh will bring in the near future. new AMD Ryzen 7900x3D CPU, mobo and DDR 5 RAM will cost similar to a RTX 4000.

I have a 10700k with 4090. My system is very cpu limited even with frame generation at big hubs it stutters. Would you suggest me to upgrade 5800x3d for budget option? 

2 hours ago, spitzer45 said:

Would you suggest me to upgrade 5800x3d for budget option? 

hard to say. you would still need a new mobo from which you could not later upgrade to a 7800x3D/7900x3D. you might want to wait for april when AMD launches the 7800x3D which is $ 250 less than the 7950x3D, but you would then also need new DDR 5 memory for the new AM 5 platform.

while GamersNexus didn't test MSFS I would refer to their F1 racing test instead, that should give you a comparable environment as far as cache utilization goes. the tests were run at only 1080p which dramatizes the fps gains, on the other hand without 4090 frame generation.

7950x3D was on average 25% faster than 5800x3D in gaming @ low resolution 720p/1080p.

here a better test @720p with MSFS from pcgameshardware.de. @ 4K the results will most likely be closer to each other.

Edited March 1, 20233 yr by turbomax

I'm using 4k resolution so Maybe we should see how 7800x3d will perform. so waiting for a month is better idea. 

Edited March 1, 20233 yr by spitzer45

I'm still waiting for the real time benchmark showing MSFS 13900K 117FPS and the 7950X3D + 48% =173FPS and not 124FPS at 1080P which is used to test CPU performance,  

47 minutes ago, G-RFRY said:

I'm still waiting for the real time benchmark showing MSFS 13900K 117FPS and the 7950X3D

at 4K I expect performance to be very similar, except that the AMD 7900x3D uses less power. it seems @ 4K we are GPU bottlenecked again 😊

22 minutes ago, turbomax said:

at 4K I expect performance to be very similar, except that the AMD 7900x3D uses less power. it seems @ 4K we are GPU bottlenecked again 😊

Your also GPU bottlenecked at 4K that`s the best your GPU with that CPU processor can do a 4090 is said to be using only 60-70% of its capabilities. When they bring out a 6.5ghz CPU then you will get better performance from a CPU dependent title, but don`t hold your breath.

Edited March 1, 20233 yr by G-RFRY

You guys keep saying that at 4K, the comparisons aren't valid because you're GPU bottleneck. But my experience with the 4090, that's just not the case in the settings I typically use (ultra, tlod 400, rt traffic and weather). The 4090 eats it up while barely breaking a sweat. Main thread limited all the time.

Edited March 1, 20233 yr by odourboy

38 minutes ago, odourboy said:

The 4090 eats it up while barely breaking a sweat. Main thread limited all the time.

how do fps change if you switch from exclusive fullscreen 4K to exclusive 1080p full screen (Alt Enter) without Framegeneration?

1 hour ago, turbomax said:

how do fps change if you switch from exclusive fullscreen 4K to exclusive 1080p full screen (Alt Enter) without Framegeneration?

I'll have to get back to you on this.

need to test 2 different scenarios

1) extreme CPU heavy:  highest traffic at most complex airport, aircraft and weather

2) the opposite thereof

similar to this, but as mentioned with maximum and minimum CPU-heavy scenarios.

 

Edited March 1, 20233 yr by turbomax

4 hours ago, turbomax said:

how do fps change if you switch from exclusive fullscreen 4K to exclusive 1080p full screen (Alt Enter) without Framegeneration?

I just did this test and inexplicably, I actually got one or two FPS higher for 4K vs 1080p. I'm clearly incapable of providing reliable test results. 😬

imagine what fps you would get at 10K ! 😊

perhaps a rounding error, at least you are not GPU limited then.

Edited March 1, 20233 yr by turbomax

55 minutes ago, turbomax said:

perhaps a rounding error, at least you are not GPU limited then

No sir. In both cases the developer pop-up indicated main thread limited. 

37 minutes ago, odourboy said:

the developer pop-up indicated main thread limited. 

that's what you get from upgrading to an RTX 4090 😀. we will always be limited by one or the other.

Edited March 1, 20233 yr by turbomax