Top 5 Best CPUs: Intel Core vs. AMD Ryzen

HardReset

Posts: 1,109   +688
To remind just one year before the 3700X from the same manufacturer was 350€ and even the highly specced 3800X was 400€.

I guess some people here will defend AMD nevertheless...
That $1000 CPU (i7-5960X) is around 25% faster (because of higher core count) and has double price https://ark.intel.com/content/www/u...5930k-processor-15m-cache-up-to-3-70-ghz.html

5800X is around same 25% faster than 3700X at same core count(!) and still is only 33% more expensive.

Your points are simply absurd 🤦‍♂️
 

Irata

Posts: 1,442   +2,323
To be boring is your cheerleading AMD. You don’t need a better PSU for a 10700F. The PSU choice is influenced by the VGA you are going to use. It is absolutely ridiculous what you wrote here.
You are using a 1660Ti or a 3060? A good 550W is good for both. A 3070 ? Use a 650W for both. A 3080 ? The recommended size is 750W.
Those raw numbers are influenced by the GPU power needs, not the CPU.

And again, no one half competent is going to use the crappy included cooler you can find in a 5600X or in a 10700F.
A decent cooler (even less expensive than the dark rock Pro 4) would be enough for both.
The PSU should be chosen according to the entire system‘s max power consumption, including CPU, GPU and all other components. I personally like to keep a little reserve so that it never reaches 100% utilization even under a theoretical worst case.

So if the 10700‘s max power consumption is 214W and the 5600X‘s is 76W, that means I‘ll need to budget for an additional 138W, given that the PSU and power delivery circuitry don‘t have a 100% efficiency, 150W to be on the safe side.

So if a 550W PSU is sufficient for my 5600X build, I‘d need to get a 700W PSU for the same build with a 10700.

As for the cooler - you can use AMD‘s included cooler - I do so for my 2700X, although admittedly an aftermarket cooler would be nicer. Still it‘s doing its job and temps + noise are still OK and there is no throttling. A 10700 will come with a cooler that can handle 65W but not a permanent PL2 boost state, so you have to get an aftermarket cooler if you want to run it unlocked.

Even then, a cooler that can handle 100W will be fine for a 5600X build, but you should get one that can at least handle 200W for your 10700 build. Those two don‘t cost the same.
 

AMN3S1AC

Posts: 57   +28
In my region (SE Asia), AMD products are way, way overpriced. AMD CPUs are in short supply and even the R3 3100 is selling at some USD 20-30 above the MSRP if you can find them.

So right now, I am building PCs with Pentium G6400s for beginners, i3-10100s for productivity users and i5-10400F for gamers.

what is a beginner?
 

131dbl

Posts: 41   +13
This won't be a problem..



If you're talking about Tim from Hardware Unboxed, his test setup was flawed. I gave all the reasons why that test was flawed on Youtube.

First, you can't test load times using THEORETICAL gen4 drives to test gen4 interface, with drives that are only operating at gen3 speeds. FAIL. More than anything else, that's the biggest fail.

To do THAT particular testing, anyone who wanted to test that theory needed to WAIT until there were ACTUAL gen4 drives that were faster than gen3 drives, and not just sustained data transfers, but with different types of IO. There were none on the market when he did that test, and I think the 970 Pro at gen 3 speeds could beat out any of those drives with almost any type of load.

So, go find a video, if one exists yet, where a person tests the Samsung 980 Pro, the WD SN850, and the Sabrant Rocket 4 Plus, against gen 3 drives, AND, instead of using the CRAP system that was full of latency, use a LOW latency test setup, with either a Zen 3 CPU, USING 3600MHz memory, and not the 3200MHz memory they use for their test systems which is higher latency. Oh wait, they REFUSE to accept that skews results. It does, and other reviewers have shown that.

If you want to test the best GPUs on the market, you build a test setup that can give it the best results. Right now that's a Zen 3 platform. Well, if you want to test the best NVMe drives on the market, you first actually have to HAVE the best NVMe drives on the market, AND the best test setup. That video failed at both. If he would have waited two months to do that testing he would have had REAL gen4 drives that can run at gen 4 speeds, and he should have switched out the Zen 2 test platform with either an Intel 10th gen system which is lower latency, OR used Zen 3 which is also low latency.

I don't know what the hell Tim tested, but it's not what he said it was. He produced a totally useless video. Faster drives load games faster, this is simple FACT, as long as the game engine uses the CPU correctly and uses all the cores that it has. This also means, when the game engine is programmed correctly, that the more cores you have, the faster a game loads too. What this also means is, if you have a slow CPU, then yes it will be true that a faster NVMe probably won't make any difference, but considering a lot of loading time is actually decompression, then a Zen 3 5950X paired with a Samsung 980 Pro AND paired with 3600MHz CL14 memory, which most people don't want to spend the money on, will all make a difference.
 

Lionvibez

Posts: 2,298   +1,806
If you're talking about Tim from Hardware Unboxed, his test setup was flawed. I gave all the reasons why that test was flawed on Youtube.

First, you can't test load times using THEORETICAL gen4 drives to test gen4 interface, with drives that are only operating at gen3 speeds. FAIL. More than anything else, that's the biggest fail.

To do THAT particular testing, anyone who wanted to test that theory needed to WAIT until there were ACTUAL gen4 drives that were faster than gen3 drives, and not just sustained data transfers, but with different types of IO. There were none on the market when he did that test, and I think the 970 Pro at gen 3 speeds could beat out any of those drives with almost any type of load.

So, go find a video, if one exists yet, where a person tests the Samsung 980 Pro, the WD SN850, and the Sabrant Rocket 4 Plus, against gen 3 drives, AND, instead of using the CRAP system that was full of latency, use a LOW latency test setup, with either a Zen 3 CPU, USING 3600MHz memory, and not the 3200MHz memory they use for their test systems which is higher latency. Oh wait, they REFUSE to accept that skews results. It does, and other reviewers have shown that.

If you want to test the best GPUs on the market, you build a test setup that can give it the best results. Right now that's a Zen 3 platform. Well, if you want to test the best NVMe drives on the market, you first actually have to HAVE the best NVMe drives on the market, AND the best test setup. That video failed at both. If he would have waited two months to do that testing he would have had REAL gen4 drives that can run at gen 4 speeds, and he should have switched out the Zen 2 test platform with either an Intel 10th gen system which is lower latency, OR used Zen 3 which is also low latency.

I don't know what the hell Tim tested, but it's not what he said it was. He produced a totally useless video. Faster drives load games faster, this is simple FACT, as long as the game engine uses the CPU correctly and uses all the cores that it has. This also means, when the game engine is programmed correctly, that the more cores you have, the faster a game loads too. What this also means is, if you have a slow CPU, then yes it will be true that a faster NVMe probably won't make any difference, but considering a lot of loading time is actually decompression, then a Zen 3 5950X paired with a Samsung 980 Pro AND paired with 3600MHz CL14 memory, which most people don't want to spend the money on, will all make a difference.
Tim from hardware unboxed?

my post was a response to avro arrow's and the video shows everything.
 
Something which may have been overlooked - AMD 5600X at similar price bracket performed slightly better than the Intel 10700 recommended here for best Gaming CPU. Refer to Techspot's own findings at 11 game average: https://www.techspot.com/review/2135-amd-ryzen-5600x/
Not to mention the superior efficiency.

Recommendation should disregard arbitrary bias based on conventional core counts disregarding actual metrics that matter (performance and dollars). Otherwise it renders testing moot.