CPU/Mobo Worth upgrading to Ryzen 5000 series from 2700?

ajish65

Patron
So I currently have a Ryzen 2700 on the B450I Gaming Plus AC. The board's CMOS battery is gone and replacing it is pointless as the replacement also runs out in like a day. But it works fine otherwise, just have to set the XMP profile on every startup. Slightly annoying but I can live with it.

I was planning to get the 3080 in a few months but no plans to change the mobo or CPU. However, I've recently come into possession of the Asrock B550-ITX/ac, which obviously does not support the 2700.

So I'm wondering if I should keep the current CPU/mobo and sell the B550 or buy a new 5000 series Ryzen and sell the old CPU/mobo combo.

I wanna keep this system for gaming, movies/shows and light office work for the next few years. So what do you guys suggest? Which is the better way forward?

Thanks for reading.
 
I've recently come into possession of the Asrock B550-ITX/ac, which obviously does not support the 2700.
Have you tried it? It might just work.

I bought a MSI A520 recently and it clearly didnt have support for Ryzen 2000 series processors but my 2700X worked fine on it. That said upgrading to a 5000 series processor is also warranted. You can change just the motherboard for now if it works and plan for 5000 processor later.
 
Really depends on your use case. I have 2700X, B450M mortar max, 32gb RAM and 3070FE system and I use it for casual gaming (non-competitive), Virtual machines and web browsing/entertainment. It's perfectly fine and I have no plans to upgrade further as of now.

Get a Duracell CR2032 battery from amazon. It lasts for years, I've never seen these batteries die in a few days. Mine's running on old mobos since 5+ years

Now If the B550 mobo you came across is for free somehow then perhaps upgrade by selling the old one but if you've paid for it then it's like building a whole new PC for IDK what reason you have in mind. As I said it depends on use-case.

Lastly you're asking our opinion on selling the old combo or keeping it for a secondary PC which honestly should be your decision depending upon your needs.
 
Definitely sell the 2700/B450I and get a 5000 series chip. Try to get a used or open box chip which has warranty and you'll be set.
Thanks. Any reccos as to which chip? I tried to look for used 5700 but those are hard to come by. Saw a couple of 5600s but didn't want to come down to the 5 series from a 7 series... And don't want to spend 20K+ on a new chip as that will affect 3080 budget also.

Have you tried it? It might just work.

I bought a MSI A520 recently and it clearly didnt have support for Ryzen 2000 series processors but my 2700X worked fine on it. That said upgrading to a 5000 series processor is also warranted. You can change just the motherboard for now if it works and plan for 5000 processor later.
I did try it and no it does not work unfortunately.
 
Any reccos as to which chip? I tried to look for used 5700 but those are hard to come by. Saw a couple of 5600s but didn't want to come down to the 5 series from a 7 series... And don't want to spend 20K+ on a new chip as that will affect 3080 budget also.
A 5600/5600X will match or exceed the multicore of the 2700(X) while having drastically higher single core scores. The single threaded performance is like 30% higher on average and in some special cases as much as 50%. I would suggest either of these but if you don't want to step down from 8 cores, 5700X is a VFM option. Having upgraded from a 5600X to a 5900X, I can confidently tell you that the extra cores barely make a difference in most games.

Prosenjit might be listing some new open box chips and updating the prices tonight. Keep an eye out on his thread in the dealer classifieds.
 
Last edited:
Saw a couple of 5600s but didn't want to come down to the 5 series from a 7 series.

Some related/unrelated thoughts:

Single threaded performance on the 6c 4500 is faster than the 8c 2700 but multi-threaded is about the same. And the 6c 3600 is faster than the 6c 4500. And the 6c 5600 is in a league of it's own, there's not a single real-world or synthetic benchmark that has the 5600 slower than or matched by the 2700.

I've personally seen that a workload that used 40% on the 2700 barely touched 5% on the 12c 5900. The 5000 series are pretty amazing for daily use but their price of entry is much higher than Intel 12th gen, the 4c 12100F is an amazing value. In some multi-threaded workloads like Handbrake, the 12100F beats the 2700 & 4500 and it's only a quad core.
 
Some related/unrelated thoughts:

Single threaded performance on the 6c 4500 is faster than the 8c 2700 but multi-threaded is about the same. And the 6c 3600 is faster than the 6c 4500. And the 6c 5600 is in a league of it's own, there's not a single real-world or synthetic benchmark that has the 5600 slower than or matched by the 2700.

I've personally seen that a workload that used 40% on the 2700 barely touched 5% on the 12c 5900. The 5000 series are pretty amazing for daily use but their price of entry is much higher than Intel 12th gen, the 4c 12100F is an amazing value. In some multi-threaded workloads like Handbrake, the 12100F beats the 2700 & 4500 and it's only a quad core.
Thanks, that's quite exhaustive. And yes, logically I know the 5600 beats the 2700 hands down, but I just seem to have this stupid fixation over cores and 'stepping down'. Guess it's just a matter of convincing myself. But thank you for those comparisons. Really help.
 
Thanks, that's quite exhaustive. And yes, logically I know the 5600 beats the 2700 hands down, but I just seem to have this stupid fixation over cores and 'stepping down'. Guess it's just a matter of convincing myself. But thank you for those comparisons. Really help.
It's not stupid and there's a difference between multi-threading performance and multi-tasking performance which most people are unaware of as benchmarks don't show that and I don't wanna go into that debate as it could last as long as eternity.
I'd advise against stepping down in cores or even buying the same no. of cores. If you do want to upgrade then actually upgrade with a step up in cores as well as IPC.
 
It's not stupid and there's a difference between multi-threading performance and multi-tasking performance which most people are unaware of as benchmarks don't show that and I don't wanna go into that debate as it could last as long as eternity.
Actually it would be helpful if you could give some examples. I can think of cases where you might need to dedicate cores like for VMs or some other specific tasks as as a reason not to step down.

Otherwise, if the core/thread count is not too far apart between the compared CPUs, the latency from juggling threads won't too much of an overhead. But lets take the case that there is significant overhead when multitasking, With say 2700X vs 3600X I might agree with you that its "stepping down". But if the IPC increase and latency improvements are drastic like the 5000 series, it should be enough to overcome those problems easily.

I'm not trying to be argumentative bud, I'm just genuinely curious as I've seen you post similar views in some other threads as well.
 
Last edited:
Actually it would be helpful if you could give some examples. I can think of cases where you might need to dedicate cores like for VMs or some other specific tasks as as a reason not to step down.
Okay, I won't go into much details because I really don't wanna type a research paper tbh :p

Simple example is that if you see a AAA game benchmark on 3700x and say a 5600X then the CPU usage across the two might be 50% and 75% but definitely most of the time you'll see the 6 core chip swamped to near 80% usage. Sure the faster 6 core chip is pushing more FPS but there's little room for background tasks or for the tasks that you wanna do simultaneously. Same goes for VMs as you said. Keeping multiple VMs running and doing web browsing etc. will take more usage on the 5600X then the 3700X in comparison.

Sure 5600X might do/process stuff faster then the 3700X but it can't do multiple stuff together without reaching it's limits. So a previous generation high core count chip has more room for multitasking but at less speed while newer generation lower core count chip can achieve better performance than previous higher core count one but has less room for doing multiple stuff at once.

Benchmarks show how fast is a chip (multi-threading) not how good it's in multi-tasking (using many processes at once while keeping from maxing the CPU out). That's the difference.
 
Last edited:
I’m not quite convinced bud. Now let’s say we have a 6 core (A) and 8 core processor (B), and we have 8 separate processes (Not threads).

Now B can dedicate 1 core each to each process, whereas A needs to keep switching processes on some cores. This switching is going to add some overhead in the form of extra latency (waiting time) and performance (reloading contextual information needed for the process at hand).

If each core of A is fast enough such that the tasks get completed much faster in a slice of time allocated to a process compared to B, then even though sometimes the processes sit idle in A, it is overall faster. If A is not sufficiently faster than B, then what you said holds true.

Also, CPU usage metrics like in task manager are deeply flawed. They are not at all an accurate representation of how hard a modern CPU is working. There are multiple reasons for this, you can just google it. A quick and dirty explanation is that the metric is based on the percentage time the processor is running the idle thread. If you have a program which cannot load all cores on the compared CPUs, the one with more cores is going to have more idle cores. The utilization is going to be an average reading of the time each core spends in idle. If each core of A does some X amount of work in a slice of time and B does X/2 and there are 4 processes running. Also, the penalty for switching (lost work) is X/10. The utilization of A is going to be 66% and B is 50% but even if you tack on 4 more processes, A will still be faster even accounting for the overhead. This is NOT reflected in the reported utilization. If you went by the reported values, then B should have been faster.

A 5600X seemingly getting hammered (going by its utilization in task manager) still has much more in reserve than you think. You might be pleasantly surprised. Of course, the caveat being you do not dedicate cores to process which cannot be used for other tasks (Again like VMs).

I've had to deal with the headaches of multitasking when I was writing special RTOS implementations for low powered, low latency hardware with networking capabilities. So I do get where you're coming from but it's not an absolute that more cores are better for multitasking. At least in this case (2700 vs 5600X) there are no drawbacks outside of not being able to reserve dedicated cores for certain tasks.
 
Last edited:
I waited a few days for the workloads to normalize but I've got some data here on 6 cores vs 8 cores as hypervisors:

2700x.png

4500.png

Synthetic benchmarks like Passmark put the 4500 as being 10% slower in multithreaded workloads than the 2700x which is what we're seeing here somewhat.

These two machines have nearly identical workloads. The minimum CPU utilization for the 4500 is 25% and for the 2700x it's 16%.
 
I upgraded from a 2700x to a 5900x on my gaming PC. There's a leap in performance. It's matched with a Strix X570-E and a 3090FE. You can expect about a 30% increase in certain tasks. Just make sure none of your other components create a bottleneck. Otherwise you may not see the performance upgrade you expect.
 
Back
Top