The Best CPUs 2019: This is what you should get

[Julio Franco]

The world of CPUs has been notoriously busy the past few years. Since our last update AMD’s 3rd generation Ryzen and Intel Cascade Lake-X parts have hit the market. After all the extensive testing you are familiar with, TechSpot's CPU buying guide means to answer that question for you in a few easy recommendations you can trust and follow.

Read the full article here.

https://www.techspot.com/bestof/cpu/

 

[Irata]

Very nice summary by category giving reasons why you ranked a CPU and the potential alternatives.

Thanks for the article.

 

[vt8235]

"destroys", "smokes"....hmmm the article smells of ...fire and brimstone.
and btw why did the author feel the need to say the i5-9400F has no integrated graphics?! Perplexed reader here as the AMD also has none.

 

[neeyik]

and btw why did the author feel the need to say the i5-9400F has no integrated graphics?!

Probably because the majority of Intel CPUs currently available for purchase do have integrated GPUs - taking the 9th generation series of Core processors, it pans out like this:

i9 - 6 models, 1 with no GPU
i7 - 12 models, 3 with no GPU
i5 - 14 models, 3 with no GPU
i3 - 11 models, 2 with no GPU

So that's 43 processor versions, with 9 having no GPU. Compare that to the 8th generation, and it's 62 models but all of them have a GPU (integrated or on-module); the same is true for the 7th and 6th generations.

That means a reader not fully knowledgeable of the 9th gen Intel processor may not be aware that there is a much large percentage of the range without a GPU, compared to previous revisions.

AMD's Ryzen range, with the exception of the APU models, have never had an integrated GPU; thus there's far less of need to point this out, when we're on the 3rd generation of that CPU.

 

[krizby]

When will ryzen ever reach 5ghz like they promise ? I'm not even saying all core, just single core 5ghz...

 

[Puiu]

When will ryzen ever reach 5ghz like they promise ? I'm not even saying all core, just single core 5ghz...

They never promised 5GHz. For this (or something close to it) you might have to wait for the 4000 series on 7nm+.

 

[quadibloc]

Of course, now some people will be kicking themselves for not waiting for Threadripper, given that it is expected to go from four-channel memory to eight-channel memory, on top of allowing more PCIe 4.0 ports.

The only thing I will fault AMD for is emphasizing the boost clocks of those CPUs, as if they, not the base clocks, are the "real" clock speeds of the CPUs. That's hardly true if the CPUs don't even always manage to attain the boost clocks! I think they should probably adjust the specs of those chips before they become the target of another frivolous lawsuit like the Bulldozer one they recently settled.

 

[gamerk2]

When will ryzen ever reach 5ghz like they promise ? I'm not even saying all core, just single core 5ghz...

They never promised 5GHz. For this (or something close to it) you might have to wait for the 4000 series on 7nm+.

I doubt even then. You're fighting the same physics Intel found insurmountable during the P4 era. While smaller nodes will allow higher clocks if no other changes are made, it's likely extra additions to the die will leave clocks tapped out at the high 4's.

Simply put, don't expect any major clockspeed advances.

 

[yeeeeman]

No need for 5Ghz. Speed can be gained from IPC, better memory access speeds/latency, etc, etc.

 

[bluetooth fairy]

Although, I'm not sure in praising 3rd-gen R5 thanks to WinRAR internal benchmark (in "Best All-Round Value CPU" section). Some other tech sources, that use their own data for compression in WinRAR, didn't come to TechSpot's conclusions.

I think the best part of the R5 3600 is that it's on par with 6c/12t Intel CPU, i7-8700k, in applications and almost matches it in gaming when paired with a mid-range graphics card.

 

[Puiu]

I doubt even then. You're fighting the same physics Intel found insurmountable during the P4 era. While smaller nodes will allow higher clocks if no other changes are made, it's likely extra additions to the die will leave clocks tapped out at the high 4's.

Simply put, don't expect any major clockspeed advances.

If they manage an extra 200MHz because of the power headroom then it's all good. 7nm+ should theoretically improve perf by about 10% (or 15% improved power efficiency), so if AMD's Zen3 architecture manages to tap into at least half of that then they are golden.

 

[Puiu]

Of course, now some people will be kicking themselves for not waiting for Threadripper, given that it is expected to go from four-channel memory to eight-channel memory, on top of allowing more PCIe 4.0 ports.

The only thing I will fault AMD for is emphasizing the boost clocks of those CPUs, as if they, not the base clocks, are the "real" clock speeds of the CPUs. That's hardly true if the CPUs don't even always manage to attain the boost clocks! I think they should probably adjust the specs of those chips before they become the target of another frivolous lawsuit like the Bulldozer one they recently settled.

AMD stated clearly that those are boost clocks and they also advertised the base clocks along side them. I don't get why people even thought for a second that these weren't boost clocks. Nobody was "confused" with Zen1 and Zen2.

 

[jbc029]

Of course, now some people will be kicking themselves for not waiting for Threadripper, given that it is expected to go from four-channel memory to eight-channel memory, on top of allowing more PCIe 4.0 ports.

The only thing I will fault AMD for is emphasizing the boost clocks of those CPUs, as if they, not the base clocks, are the "real" clock speeds of the CPUs. That's hardly true if the CPUs don't even always manage to attain the boost clocks! I think they should probably adjust the specs of those chips before they become the target of another frivolous lawsuit like the Bulldozer one they recently settled.

AMD stated clearly that those are boost clocks and they also advertised the base clocks along side them. I don't get why people even thought for a second that these weren't boost clocks. Nobody was "confused" with Zen1 and Zen2.

Agreed on the point of base vs boost clearly stated. Disagree on the point that it seems clear that AMD was rounding to claim a higher threshold than they were guaranteeing. I looked through the MB vendor breakdown and yes there does seem to be some culpability on them for their BIOS implementations, but frankly, they're working with what AMD gave them. The BIOS updates should not have been this raw. AMD needs to get their act together and give their hardware vendors enough time to implement quality updates.

 

[AIC1Drew]

Kinda surprised the i7-9700K didn't get a mention when referring to the best gaming CPU.

 

[Evernessince]

When will ryzen ever reach 5ghz like they promise ? I'm not even saying all core, just single core 5ghz...

They never promised 5GHz. For this (or something close to it) you might have to wait for the 4000 series on 7nm+.

I doubt even then. You're fighting the same physics Intel found insurmountable during the P4 era. While smaller nodes will allow higher clocks if no other changes are made, it's likely extra additions to the die will leave clocks tapped out at the high 4's.

Simply put, don't expect any major clockspeed advances.

It doesn't even really matter so long as IPC is increasing. In fact IPC is better then clock speed as it keeps the power requirements low.

 

[brkbeatjunkie]

Kinda surprised the i7-9700K didn't get a mention when referring to the best gaming CPU.

I agree since their own review publications show the 9700k a tiny bit faster than the 9900k in some metrics due to extra OC headroom.

 

[picka]

Kinda surprised the i7-9700K didn't get a mention when referring to the best gaming CPU.

Exactly, it's significantly cheaper for pretty much the same performance.

It (and the 9900k) also benefit from the fact they produce as much heat as a small nuclear reactor so heating bills in winter will be cheaper.

 

[Ludak021]

9700K has 95W TDP, no more if you don't overclock it. That's pretty much standard for intel K CPUs since forever. It doesn't even reach that unless you stress load it. They are easy to cool tbh.

 

[Indekkusu]

AMD has finally given Intel something to worry about, but AMD is still not priced low enough. And now we learn this:
https://www.tomshardware.com/news/a...ing-advertised-boost-speeds-survey,40291.html

Intel for all their wrongs, doesn't let AMD off the hook. I don't want to wait to till a decade later to finally see the justice done:
https://www.tomshardware.com/news/a...ion-lawsuit-eight-cores-settlement,40256.html

AMD can mitigate all this by pricing lower today!

 

[jbc029]

9700K has 95W TDP, no more if you don't overclock it. That's pretty much standard for intel K CPUs since forever. It doesn't even reach that unless you stress load it. They are easy to cool tbh.

As long as you don't OC it, sure. But then why by a K sku if you arent going to OC it. If you do OC it, it will pull a lot more than 95 watts and you will need a very good air cooler or an AIO.

 

[Jimster480]

When will ryzen ever reach 5ghz like they promise ? I'm not even saying all core, just single core 5ghz...

Maybe sometime soon. There are rumors of ES that hits 5Ghz already... so maybe there will be another wave of releases later on as the process matures.

 

[Indekkusu]

Maybe sometime soon. There are rumors of ES that hits 5Ghz already... so maybe there will be another wave of releases later on as the process matures.

AMD's marketing must be working really hard to get believers like this. How about show me the money! Lower prices I can see and don't need to wish upon a star for 5GHz.

 

[bluetooth fairy]

As long as you don't OC it, sure. But then why by a K sku if you arent going to OC it. If you do OC it, it will pull a lot more than 95 watts and you will need a very good air cooler or an AIO.

Because when you buy a K model it has greater TDP (95W vs 65W) and much higher base clock (3GHz for 9700 vs 3.6GHz for 9700K). It's valid for every top-model in i3, i5, i7 and i9 line-up.

 

[Irata]

9700K has 95W TDP, no more if you don't overclock it. That's pretty much standard for intel K CPUs since forever. It doesn't even reach that unless you stress load it. They are easy to cool tbh.

Isn't the 95W TPD for Intel CPU specified at base clock speed, I.e. at 3.6 Ghz for the i9-9900k ?

So even when not overclocking the CPU, it will use more than that once it turbos *and* loads a higher number of cores.

 

[Ludak021]

It is mostly (95W tdp) on all K cpus. Look at the similar non-K variants, you will usually see lower rated TDP (yes, it's for default clocks, but also by default even with overclock you can't exceed it unless you change it in the BIOS). And yes, CPU will not reach 95W tdp under boost clocks at stock.
The thing is, under something like, say, cinebench, all cores test, you will not see tdp go over ~60W. Under something like prime95 all core maximum heat, it will go over 90, and when it goes over 95W it will lower clocks (you fix this by going to BIOS and change max TDP from auto to whatever, like 150W, screw it ). But this is only happening in synthetic environments that are meant to stress CPU hard and don't leave any empty cycles. I doubt you will ever see tdp go over ~120ish on high overcloks even, under stability load tests. Meaning, in games 95W tdp will be fine even on 5GHz, provided thermals are in check.

edit: even 9400F in prime95 small ffts goes over 65W and lowers clocks to try and maintain 65W. But in render test, like I mentioned above, cinebench, it floats at around ~45W. Going into BIOS and setting max tdp to 110W solves the "issue" with dropping clocks when it goes over 65W.