Which CPU should i choose

[Daham]

I'm planning on building a new pc.I cant decide which processor to choose. Ryzen 5 3600 or Intel i5 10400f. I'm planning to use the pc for gaming streaming and little bit of editing. I can find the Ryzen 3600 around $218 and i5 10400f around $148. which processor should I choose? I'm also on a budget.

 

 

 

[Lord Szechenyi]

Intel

 

Unless you're going to edit 8k footage with lots of effects, there's really no point of going with the ryzen 3600.

I mean, if you want to use more money, then sure, buy it, but for what you need it for you won't see any performance increase

[SupaKomputa]

$60 is pretty big. you can get a motherboard or ram.

If you planning only gaming, no overclocking, Intel wins it.

[Electronics Wizardy]

Id probalby go with a 11400 here if you can get one. igpu can be nice to have and not much extra. The 11th gen is slightly faster too.

[Lord Szechenyi]

1 minute ago, Electronics Wizardy said:

Id probalby go with a 11400 here if you can get one. igpu can be nice to have and not much extra. The 11th gen is slightly faster too.

this is an F, no IGPU

[Chris Pratt]

Be advised that the 10400F is only competitive with the 3600 when paired with 3600Mhz RAM (which you'd also want for the 3600) and a mobo that supports it (B460/X490). In other words, either is good, but don't cheap out on the RAM or mobo.

 

[-rascal-]

1 minute ago, Lord Szechenyi said:

this is an F, no IGPU

 

@Electronics Wizardy is suggesting a i5-11400, not the i5-10400F.

From reviews, compounding with the insane AMD prices right now, the i5-11400 is actually a decent chip (much better than the joke i9-11900K).

 

 

 

[StDragon]

1 minute ago, Chris Pratt said:

Be advised that the 10400F is only competitive with the 3600 when paired with 3600Mhz RAM (which you'd also want for the 3600) and a mobo that supports it (B460/X490). In other words, either is good, but don't cheap out on the RAM or mobo.

 

The 11 series only make sense if you plan on using the iGPU or QuickSync. Otherwise, I believe the 10 series is equal or faster than 11 all while being cheaper too.

 

So 11 series for laptops (non-gaming), 10 for gaming with a DGPU.

[Lord Szechenyi]

6 minutes ago, -rascal- said:

 

@Electronics Wizardy is suggesting a i5-11400, not the i5-10400F.

From reviews, compounding with the insane AMD prices right now, the i5-11400 is actually a decent chip (much better than the joke i9-11900K).

 

 

 

oops, didn't see the 1 instead of the 0.

Anyway, doesn't matter because OP seems to have a choice between 2 CPUs

[BahnStormer]

There's a $70 gap... but you need to look at all the components.

 

Do you have a cooler already? It's included in the Ryzen 3's and 5's... so that could easily close the gap by $30-$40, possibly WAY more. The included Ryzen 5 cooler is VERY usable - NOT to be confused with the Intel included coolers of old you can even overclock on the stock coolers.

 

Also factor in motherboard costs: I think the 10th gen Intel boards are mostly >$100, certainly if you want one to get the most out of the 10400... but you can probably get a very decent B450 board for <$80 (ASRock B450M-HDV @ $72).

 

Now I've just made it harder - 'cos by my reckoning that balances the prices out perfectly and there's just about nothing to separate those CPU's in terms of performance.... both are 6C/12T and Ryzen only leads by a few percent: https://cpu.userbenchmark.com/Compare/Intel-Core-i5-10400F-vs-AMD-Ryzen-5-3600/4079vs4040 

 

They're both very capable CPU's for decent gaming and light workstation loads - I'm sure you'll be happy with either.

 

Personally, when I had that choice (my bias was slightly more for the workstation capabilities), so I went with an even older Ryzen 7 2700X (slightly slower than the R53600 for gaming, but more cores)... as I liked the idea of a potential upgrade path to a 3900X later and AMD had just announced that the B450 platform would also support the next gen (5000-series).... but that only really comes into it if you think there's a chance you'd want to upgrade on the same platform.

[-rascal-]

17 hours ago, BahnStormer said:

Now I've just made it harder - 'cos by my reckoning that balances the prices out perfectly and there's just about nothing to separate those CPU's in terms of performance.... both are 6C/12T and Ryzen only leads by a few percent: https://cpu.userbenchmark.com/Compare/Intel-Core-i5-10400F-vs-AMD-Ryzen-5-3600/4079vs4040 

 

They're both very capable CPU's for decent gaming and light workstation loads - I'm sure you'll be happy with either.

 

 

UserBenchmark results are quite skewed; would recommend looking at actual reviews to get a more accurate comparison.

 

Interesting how the i9-10900K sits WAY BELOW the i9-11900K, i7-11700K, and even under the i5-11600K. Reviews show the older i9-10900K performs ABOVE the new i9-11900K.

 

[Daham]

Thank you guys for your support

[BahnStormer]

On 4/13/2021 at 6:39 PM, -rascal- said:

 

UserBenchmark results are quite skewed; would recommend looking at actual reviews to get a more accurate comparison.

TBH - "actual reviews" are the ones I'd be worried about unless they're reviewing your exact workload.

 

Even on Userbenchmark, I'd be at least a little wary of ANY of those 11th gen benchmarks / reviews given the current sample sizes.... looking at Userbenchmark: most 11th gen < 40 samples, most 10th gen >40k. So the 11th gen scores are likely to only be professional builds / reviewers who are getting the most out of the system.... just wait for the proper USER scores to get them to balance out the scores a little more.

 

If you know exactly what you want to do with the system, then you need to drill into the results a bit more: Best example is the gaming vs workstation scores - you may not care if a system gets marked down on 8 core load (i.e. most Ryzen 5 / i5 gaming CPU's). Conversely even an older 32C/64T Threadripper will thrash an i5 on the 64core synthetic and the memory bandwidth scores and probably score higher, despite being less capable at gaming.

 

Even having a review focussing on "gaming" is problematic as any selection of benchmarks from a reviewer can be tweaked and rigged to favour certain CPU's / GPU's over others (different brands or generations) - just look at the propaganda material that nVidia/AMD/Intel spew with every release.... and then ask their favourite influencers to regurgitate the same blinkered view to the consumers.

 

Taking no sides here - they've all taken turns with this trick to pedal underwhelming products over the last few years, but I would wager that the recent Intel 11th gen releases have asked any reviewer to reduce the focus on multi-core performance (or power/thermals).... likewise recent AMD CPU reviews are focussing on IPC and not Instructions per second because Intel is winning the now somewhat defunct Ghz race.... to be fair AMD are actually doing pretty well on both scores.... but they're outright winning on IPC  Likewise nVidia want everything reviewed with RayTracing and DLSS2 enabled, while AMD want you to focus on native res + rasterisation only. Both views are right/wrong... .depending on your preferences and use-case.

 

So yes, Userbenchmark is flawed, but so is every suite of benchmarks that doesn't exactly replicate your personal requirements. I don't place absolute faith in it, but do like the fact that it takes the same intentionally broad spectrum of tests for every CPU/GPU, usually has VERY high sample sizes (better show of real-world scores) and gives you a easy way to share it with people.

 

See my memory scorecard below: that is a good example of the spread of results... you'll see ANY 3200Mhz memory kit running XMP2 settings will be in the top third, because it also shows all the people who bought 3200Mhz and then run them at factory SPD settings instead of the XMP2 option.

 

Even my 3200Mhz running at 3000Mhz get in at 96.5% of the likely peak score because I was having problems with running 4 sticks at 3200Mhz XMP2 settings, so was experimenting with 3000Mhz (with fairly aggressive timings):

 

I'm also responsible for one of the ~66% scores when I forgot to apply XMP2 (running at 2133Mhz, CL18) and another of the ~110% scores when I only had two sticks, at slightly over 3200Mhz AND running more aggressive timings than the rated XMP2 profiles.