Question 5600g vs 5700X3D

[m3city]

Hi,

I'm considering and upgrade like in subject and went through reviews here on TH and some YT side by side comparisons. The reviews praised the X3D for performance boost in games and that's what I'm looking for. I already got some after I added a discrete Quadro P2000 that works well enough for me. It is a substantial upgrade from iGPU (although 5600g is also good for 1080 in games I play).

Two questions based on yt videos:
- one reviewer said that X3D will provide more performance with high end GPU card, less with medium card (I assume P2000 fits to this category) - true or false?
- it struck me that in comparisons like the one linked, FPS is more or less 50%more, but GPU power draw is also 50% up, while CPU is not taxed at max, lets say 50-80%, similar both for 5600g and 5700X3D. What happens here?

I call anything above 60fps as a performance reserve for future, but where does it come from in case of this comparison - X3D cache influence or simply higher utization of GPU?

View: https://youtu.be/aQ-ki4S0NE0

 

[Lutfij]

You're also losing out on PCIe gen with the APU. In the literal sense, you're not comparing an apple to an apple. In 2024, seeing how the AM4 platform has been in the market for a while, I'd get the 5700x3D as you have a discrete GPU. The APU's were made for very budget oriented builds...something higher than an office PC or even in an HTPC.

You've forgotten to mention the make and model of your motherboard and the rams used since Ryzen needs them both to perform well.

I'm considering and upgrade
Why are you considering an upgrade? What sort of tasks do you currently tax the system with?

 

[CountMike]

Hi,

I'm considering and upgrade like in subject and went through reviews here on TH and some YT side by side comparisons. The reviews praised the X3D for performance boost in games and that's what I'm looking for. I already got some after I added a discrete Quadro P2000 that works well enough for me. It is a substantial upgrade from iGPU (although 5600g is also good for 1080 in games I play).

Two questions based on yt videos:
- one reviewer said that X3D will provide more performance with high end GPU card, less with medium card (I assume P2000 fits to this category) - true or false?
- it struck me that in comparisons like the one linked, FPS is more or less 50%more, but GPU power draw is also 50% up, while CPU is not taxed at max, lets say 50-80%, similar both for 5600g and 5700X3D. What happens here?

I call anything above 60fps as a performance reserve for future, but where does it come from in case of this comparison - X3D cache influence or simply higher utization of GPU?

View: https://youtu.be/aQ-ki4S0NE0

There is no comparison between those 2 CPUs, they are not in same bracket. 5600G is even sower than 5600x.
5700x3D is way better for gaming, 2 cores and 4 threads more and that extra cache memory that does real number on most modern games.
x3D cache has nothing to do with GPU utilization which depends on game graphic settings. Up to 1080p resolution @60Hz CPU works more so there 5700x3D would help more. GPU power consumption may go up because with 5600G it's utilized less but good cooling can soak that up.

 

[m3city]

@Lutfij

I did forget about PCI-E thing, but a quick look indicates it doesn't matter?
In regard to mainboard - it is Gigabyte B450m DS3H with 32GB memory, some basic 3600MHz. But as I'm keeping those as they are, it should not also matter - I mean, if there is an upgrade from X3D then it will be regardless of ram.

Upgrade due to games like Stormworks: Searsch and Rescue that is either poorly optimised or uses CPU a lot for physics. It does not run as smooth as I would expect after adding that Quadro P2000.

It still puzzles me where this GPU power draw increase from the first video comes from - while CPU utization is same for most tested titles. As if 50% FPS increase came from GPU only. Are those yt tests flawed in some sense or I'm missing sth out. @CountMike I read your response, but still don't get it.

Pci-e 3 v4:

View: https://www.youtube.com/watch?v=aMOxX1IvfIw

 

[tennis2]

It still puzzles me where this GPU power draw increase from the first video comes from

You can be CPU or GPU limited in performance.

In the case of the video (I just watched the first 20 seconds in RDR2) they're clearly CPU limited (using an RTX4090) on the 5600G. So by giving more CPU horsepower, the GPU is able to crank through/out more FPS, hence having to work harder.
Note the GPU usage @ 0:20, 50% on the 5600G vs 84% on the 5700X3D. But you're also seeing almost x2 the FPS.

Quick synopsis (VERY simplified)

• CPU figures out what each frame is supposed to look like (think of this as a rough sketch, location of frame contents, etc)
• CPU sends that rough sketch to the GPU (draw call)
• GPU renders the frame (makes it photorealistic) and sends it out to the monitor.

Now:

• You're always capped by how many FPS a CPU can send to the GPU (you can't exceed the max FPS of the CPU for a given game, no matter how low you crank the resolution/details), BUT...
• The higher the resolution/detail/etc (image quality) you require from the GPU, the longer it takes to draw out each frame. So, with higher resolution/quality settings, the GPU could end up taking longer to draw each frame (think of this as GPU FPS) than it takes for the CPU to send the next one (call this CPU FPS).
• If the GPU can finish drawing a frame before the CPU sends the next one, the GPU has to wait. This is idle time. FPS is limited by the CPU.
  • We typically look at GPU utilization (%) to figure out where the wait is happening.
• If the CPU is ready to send the next draw call before the GPU finishes the one it's working on, the CPU has to wait. FPS is limited by the GPU.
  • CPU utilization is harder to monitor because its also a product of thread utilization (ie in the video you linked at 0:20, you could have all ACTIVE threads at 100% utilization, but if you're only using 9 out of 12 threads, you're going to see 75% utilization)

As explained above, a general rule of thumb, you typically look at GPU utilization and shoot for >80% by adjusting video settings in-game. This gives maximum image quality without hurting FPS too much. Getting to/over 100% is pushing into GPU-limited scenario, which isn't necessarily bad, as long as FPS is still acceptable.

 

[tennis2]

It does not run as smooth as I would expect after adding that Quadro P2000

The quadro P2000 isn't really optimized for gaming. That being said, this is a 2017 GPU that uses 75W. Not going to be a gaming powerhouse.

You're looking at being roughly equivalent to a 1050Ti in the chart below (bottom right corner).

More generally, on pricing the 5700X3D is considerably more $ than a 5600G. However, the 5600G isn't really a good value for discrete GPU use. They charge a bit more because the 5600G has a serviceable iGP.
If you're using this as a jumping off point and will be replacing your P2000 with a higher end GPU sometime in the near future, go 5700X3D. If $$ is limited, the cost savings of the 5600G could help accelerate you toward a GPU upgrade that would leave you with a more balanced system. You generally want to shoot for a [gaming] system with GPU costing.....1.5x - 2x the cost of the CPU. With a 5700X3D and a P2000 (GTX1050Ti) you're upside down on that model currently.

Keep in mind, I'm using a 5600G + 3060Ti. And I find it perfectly acceptable. Sure, I'm leaving performance on the table in some situations, but I also bought my 5600G back around launch for $125 when the 5600X was $300.

 

[beyondlogic]

Hi,

I'm considering and upgrade like in subject and went through reviews here on TH and some YT side by side comparisons. The reviews praised the X3D for performance boost in games and that's what I'm looking for. I already got some after I added a discrete Quadro P2000 that works well enough for me. It is a substantial upgrade from iGPU (although 5600g is also good for 1080 in games I play).

Two questions based on yt videos:
- one reviewer said that X3D will provide more performance with high end GPU card, less with medium card (I assume P2000 fits to this category) - true or false?
- it struck me that in comparisons like the one linked, FPS is more or less 50%more, but GPU power draw is also 50% up, while CPU is not taxed at max, lets say 50-80%, similar both for 5600g and 5700X3D. What happens here?

I call anything above 60fps as a performance reserve for future, but where does it come from in case of this comparison - X3D cache influence or simply higher utization of GPU?

View: https://youtu.be/aQ-ki4S0NE0

Things to consider

Is cooling you may need a better cooler to run a 5700x3d chips run hot as there 105w parts a 7 heatpipe cooler I would recommend for it but amd recommends a aio.

I personally would not recomend the 5600g

But the vanilla Ryzen 7 5700x should be more then enough at 1080p

 

[tennis2]

Is cooling you may need a better cooler to run a 5700x3d chips run hot as there 105w parts a 7 heatpipe cooler I would recommend for it but amd recommends a aio.

I'd bet a "typical" 4-5 heatpipe 120mm tower cooler could handle a 5700X3D just fine.

Remember that going "bigger" doesn't mean you have to spend lots of $$. Something like the ID-Cooling A620 Pro SE = $30. Or the Thermalright Assassin 120SE for $35 are perfectly fine.

 

[beyondlogic]

I'd bet a "typical" 4-5 heatpipe 120mm tower cooler could handle a 5700X3D just fine.

Remember that going "bigger" doesn't mean you have to spend lots of $$. Something like the ID-Cooling A620 Pro SE = $30. Or the Thermalright Assassin 120SE for $35 are perfectly fine.

I've tried a 6 heatpipe cooler on a loaded 105 chip and personally wouldn't recommend it.

Was hitting 85c to 95c multiple remounts to test if it was a bad mount.

Thermalright Phantom Spirit 120 EVO, 7 would be fine.​

Though I have heard that the 5700x3d runs a bit cooler then 5800x3d. Unsure what voltage it was at but anything above 1.3v made temps shoot up.

 

[CelicaGT]

I'd bet a "typical" 4-5 heatpipe 120mm tower cooler could handle a 5700X3D just fine.

Remember that going "bigger" doesn't mean you have to spend lots of $$. Something like the ID-Cooling A620 Pro SE = $30. Or the Thermalright Assassin 120SE for $35 are perfectly fine.

Correct, it can and does. I'm doing it right now on a 5800X3D. The stated need for liquid cooling was overcautious on AMDs part, and fully debunked not long after the the original X3D release. The 3DV-cache and interposer forms a thermal break that renders exotic cooling solutions mostly unnecessary. It's also just doesn't run that hot. I cooled mine with a basic Hyper 212, I only upgraded to my current Assassin Mini purely for aesthetics, it performs roughly the same.

 

[logainofhades]

I've tried a 6 heatpipe cooler on a loaded 105 chip and personally wouldn't recommend it.

Was hitting 85c to 95c multiple remounts to test if it was a bad mount.

Thermalright Phantom Spirit 120 EVO, 7 would be fine.​

Though I have heard that the 5700x3d runs a bit cooler then 5800x3d. Unsure what voltage it was at but anything above 1.3v made temps shoot up.

I ran a Vetroo V5 on my 5800x, which has a history of being a hot running chip. I never hit those kinds of temps, when gaming. A CPU-Z stress test would get me in that range, but never regular gaming use.

 

[CelicaGT]

I've tried a 6 heatpipe cooler on a loaded 105 chip and personally wouldn't recommend it.

Was hitting 85c to 95c multiple remounts to test if it was a bad mount.

Thermalright Phantom Spirit 120 EVO, 7 would be fine.​

Though I have heard that the 5700x3d runs a bit cooler then 5800x3d. Unsure what voltage it was at but anything above 1.3v made temps shoot up.

The X3D CPUs use a bit less juice compared than their non X3D brethren and as such don't hit the high temps expected. Mine only hits 105W to 115W, and plateaus around 60C above ambient in sustained all core loads at just under 4.4Ghz. This is under the Peerless Assassin Mini in my sig. As I stated just after your post I previously cooled it using a Hyper212, which dated back to a then new i7 4790K build and it performed similarly. These little buggers are probably the best CPU AMD has ever released, far surpassing everyone's (even their) expectations.

 

[beyondlogic]

The X3D CPUs use a bit less juice compared than their non X3D brethren and as such don't hit the high temps expected. Mine only hits 105W to 115W, and plateaus around 60C above ambient in sustained all core loads at just under 4.4Ghz. This is under the Peerless Assassin Mini in my sig. As I stated just after your post I previously cooled it using a Hyper212, which dated back to a then new i7 4790K build and it performed similarly. These little buggers are probably the best CPU AMD has ever released, far surpassing everyone's (even their) expectations.

I'm going to assume that's because there not clocked with a higher voltage. Probly between 1.25-1.3.v

I noticed when voltages went as high as 1.4 the cooler struggled and yes that was under a stress test. On a non x3d.

Some air coolers may pull it off but not all are equal. Thermal right are decent and not bad priced.

Hate the mounting of the 212 had a faulty unit and had another that came from factory without a mounting kit lol

 

[CelicaGT]

I'm going to assume that's because there not clocked with a higher voltage. Probly between 1.25-1.3.v

I noticed when voltages went as high as 1.4 the cooler struggled and yes that was under a stress test. On a non x3d.

Some air coolers may pull it off but not all are equal. Thermal right are decent and not bad priced.

Hate the mounting of the 212 had a faulty unit and had another that came from factory without a mounting kit lol

Honestly I can't remember the all core voltage, and, iirc it's typically lower per core that the single core peak when benchmarking. I'm currently running MSFS which is lightly threaded and really does like to bludgeon the one, chosen core and it's steady at 1.2ish Volts

Oh, and that mounting for the 212, I just remounted it, recycled yet again onto a family members PC with my old 5600X....it's like a punishment meted out by the old gods.

 

[m3city]

Thank you for all the comments, @tennis2 for a nice explanations "how it works", and for those concerning cooling. I completely forgot about it, assumed that I would use stock 5600g cooler. Seems not.

 

[tennis2]

The stock 5600G cooler (like pretty much all stock coolers) is serviceable, but it's loud and/or hot. (I ran the stock cooler on my 5600G for a month when I first got mine)
A $20-30 Thermalright cooler will be SIGNIFICANTLY better. The BA120 is a good cheap option.