AMD Ryzen 5 2600 review

Posted by: Hilbert Hagedoorn on: 05/01/2018 02:58 PM [ 30 comment(s) ]

It is time to have a look at the non-X model Ryzen 5 2600. The 'regular' six-core proc is a few tenners cheaper compared to the 2600X model, but really, it's the same stuff. It's just that the X is better binned and has better default clock frequencies. However, if you are willing to tweak a bit yourself, you can save them few tenners and retrieve the very same performance. The Ryzen 5 2600 has six-cores and twelve threads and is ready 'n waiting for you at a pretty spectacular value price. A perfect piece of silicon for proper threading and gaming? Well, let's find out!

Read article

Advertisement

Tagged as: amd, ryzen

pages 1 2

Embra
Senior Member

Posts: 839

Posted on: 05/01/2018 03:16 PM

Wow... really nice $200 chip. @ 4.2 it really zooooms!

Great review HH!! Thank you.

waltc3
Senior Member

Posts: 930

Posted on: 05/01/2018 04:12 PM

Nice review, as usual, HH...

I agree with you--if I'm going to go with a new cpu (to replace my 1600 @ 3.8GHz) then I should go with the 2700x, or stay put. THe one thing I've noticed from MSI is that the x470 boards all support the 1.0.0.2a AGESA, whereas the x370s have not been updated beyond 1.0.0.1a AGESA. I'm wondering what the difference is, if any. It would be nice if MSI would include that information in its bios update descriptions.

jortego128
Senior Member

Posts: 101

Posted on: 05/01/2018 04:18 PM

Agree with HH the non-X procs are still a great value, but the X versions are the better deal overall, esp. the 2700X-- totally opposite situation from the 1000 series launch last year.

IceVip
Senior Member

Posts: 658

Posted on: 05/01/2018 07:30 PM

Poor 2600x users, should've gone with the cheaper model.

Hilbert Hagedoorn
Don Vito Corleone

Posts: 35322

Posted on: 05/01/2018 07:42 PM

Poor 2600x users, should've gone with the cheaper model.

Yeah, I have been looking into that, looks like the latest FC5 patch or drivers increased Quad HD perf a tiny bit (less than 2% though). That or something has been bugging the initial 2600X/2700X measurement. Checking things and will rerun the test (and update) with a 2600X and 2700X.

Updated.

H83
Senior Member

Posts: 2643

Posted on: 05/01/2018 07:53 PM

I think i prefer the non X models to the others. They offer basically the same clocks at a lower price point. Also the 2600 is great value for those looking to build a nice gaming rig without breaking the bank.

schmidtbag
Senior Member

Posts: 4036

Posted on: 05/01/2018 08:58 PM

Considering this trades blows with the 8700K (which has much higher boost speeds and costs $150 more) I'd say this is a pretty great choice. If AMD can just figure out how to fix the memory latency issue, I'm sure they'll have an all-around better product.

I agree with Hilbert's consensus though - the 2600X is a better choice overall. You'll get significant power savings without much loss in performance. Meanwhile, you can stick with the box cooler. If you got the 2600 and wanted to overclock it to at least 4.2 GHz, the money you saved will all have to go toward a better heatsink. So unless you really don't care about thermals and/or think you can pass 4.2GHz, the 2600X is the better choice.

I have a feeling AMD was trying to avoid the situation they were in with the 1st gen Ryzens, where everyone was buying the lowest-tier product since they would all OC to the same levels. But with the 2nd gen, the X models make for a more compelling value.

Nice review, as usual, HH...

Weird how even my Biostar X370 board has 1.0.0.2a.

chispy
Senior Member

Posts: 8759

Posted on: 05/01/2018 10:25 PM

Great review Hilbert , as you pointed out i too believe the 2600x it's a better value due to the strict binning of this cpus. Moarrr higher overclocks ...

mat9v9tam
Member

Posts: 35

Posted on: 05/02/2018 12:33 AM

Yes, looking at Aida64 memory test screenshot, the bios on the Hero VII board is an "ancient" one not even with AGESA 1.0.0.2a - I wonder how it impacts the scores, there were some important speed-ups in newer bioses, not to mention it does not provide AMD microcode updates....
And yes, for example even cheap Asrock X370 Pro4 have 1.0.0.2a AGESA, so does Asus Prime X370 Pro - both had them for more then 2 weeks.

tunejunky
Senior Member

Posts: 875

Posted on: 05/02/2018 03:44 AM

just in time for graduation season. i have a few cheeky buggers to sort out, and this is the chip to do it with...and with the RX 580 back to its proper price the boys will have new console killers with free sync.

Arbie
Senior Member

Posts: 161

Posted on: 05/02/2018 03:57 AM

sverek
Senior Member

Posts: 5020

Posted on: 05/02/2018 03:57 AM

Thanks Boss. This CPU makes the wait for Zen2 very hard. I really like it performs in gaming and price is just too tempting.

2600 is basically 8700(non-K) for 100$ less.

Killian38
Senior Member

Posts: 312

Posted on: 05/02/2018 04:47 AM

If I plan on buying an AIO cooler Why not buy the 2600/2700 non "X". Same thing minus the $$. Not 100% sure about what I just said. I need ya'lls input.

Neo Cyrus
Senior Member

Posts: 9137

Posted on: 05/02/2018 04:51 AM

Thanks Boss. This CPU makes the wait for Zen2 very hard. I really like it performs in gaming and price is just too tempting.

2600 is basically 8700(non-K) for 100$ less.
Just look at the retarded RAM prices and you'll be less tempted to upgrade. I was going to buy a Ryzen 2600 + 470 board + 16GB DDR4... then I saw the DDR4 prices... then I saw the 470 board prices with those that have 8+ SATA ports. Nope nope nope nope nope.

Robbo9999
Senior Member

Posts: 1246

Posted on: 05/02/2018 07:29 AM

Is it not still true that if you overclock Ryzen you lose the boost features? If so, that seems like too large a point to (always) omit in the discussion(s).

Personally, I have to watch the calories per hour and love the fact that my 1800X is smart enough to ramp up when it needs to and not otherwise. Sites such as this almost never address the fact that OC isn't always better. Sometimes - and I believe it's true for Ryzen - you should just leave things stock.
I have this impression too, I'm sure I read this somewhere in a review somewhere, but I'd love to be corrected wrong if someone knows any different - I'm referring to your point about overclocking losing the boost features and the supposed fact that overclocking means you end up with increased idle CPU power usage as well as of course increased power usage when under load? This would be another argument to get the 2600X version, & to then leave it at stock.

Fox2232
Senior Member

Posts: 9090

Posted on: 05/02/2018 07:40 AM

I have this impression too, I'm sure I read this somewhere in a review somewhere, but I'd love to be corrected wrong if someone knows any different - I'm referring to your point about overclocking losing the boost features and the supposed fact that overclocking means you end up with increased idle CPU power usage as well as of course increased power usage when under load? This would be another argument to get the 2600X version, & to then leave it at stock.

Robbo9999
Senior Member

Posts: 1246

Posted on: 05/02/2018 07:50 AM

So, it actually eats around same as my old Sandy i5 which was allowed to downclock. If I forced 4.5GHz and all 4 cores, then idle power consumption used to be ~35W.
I may provide power values in case CPU is allowed to downclock at evening...
So this is your 2700X overclocked to 4.225Ghz and not allowed to downclock, and you have CPU power usage varying from 18W to 63.88W over the period of one hour - what were you doing with the PC, what kind of loads? They're certainly quite reasonable power values, not too high (but depends on what you were doing with your PC during that one hour). Will be interesting to see your power values when your CPU is allowed to downclock, while I assume still overclocking?

Fox2232
Senior Member

Posts: 9090

Posted on: 05/02/2018 09:02 AM

So this is your 2700X overclocked to 4.225Ghz and not allowed to downclock, and you have CPU power usage varying from 18W to 63.88W over the period of one hour - what were you doing with the PC, what kind of loads? They're certainly quite reasonable power values, not too high (but depends on what you were doing with your PC during that one hour). Will be interesting to see your power values when your CPU is allowed to downclock, while I assume still overclocking?

sverek
Senior Member

Posts: 5020

Posted on: 05/02/2018 09:13 AM

I had issue with my 2500k not downclocking when idling, not sure if relevant to AMD:
https://forums.guru3d.com/threads/2500k-oc-whats-your-idle-wattage.414665/

Robbo9999
Senior Member

Posts: 1246

Posted on: 05/02/2018 02:40 PM

Well, average for that time was just under 24W. So I was doing close to nothing most of the time. Regular morning online checks weather forecast, forums. And Apparently Windows doing its own morning stuff.

There are two things I think about on this CPU. When boosting is allowed and therefore downclock+downvolt works, CPU may go from my permanent 1.425V down to 1.1V or less. That may reduce power consumption of CPU cores themselves, but they eat already very little at idle.
That will affect "CPU Core Power" draw (~15W at idle), but not SoC (~8,6W at idle) as that has separate voltage. So those 15W may go down to 11.5W. That's just 3.5W.
So, CPU being able to downclock may drop down to ~20W.
Yes, so given that you can see that idle power usage is not gonna be greatly affected, it would be interesting to see a comparison during a measured (reproducible) everyday load. Something like PCMark8 runs perhaps that take the CPU through scripted everyday scenarios - would be interesting to see if power & performance differences would be great between static overclock and stock CPU with downclocking. (I'm assuming that you can't overclock your CPU and have it downclock too.).

waltc3
Senior Member

Posts: 930

Posted on: 05/02/2018 04:24 PM

Weird how even my Biostar X370 board has 1.0.0.2a.

Yes, it's strange. Perhaps MSI is simply confused by the AGESA version numbering scheme--I surely do not understand it. I updated to 1.0.0.6a long ago, but the latest versions are 1.0.0.1a and then 1.0.0.2a. I guess what they did is reset the AGESA numbering scheme when they introduced the 12nm +. That at least would add some rhyme and reason to it.

Fox2232
Senior Member

Posts: 9090

Posted on: 05/02/2018 05:49 PM

As promised:

toyo
Member

Posts: 65

Posted on: 05/02/2018 05:58 PM

Your CPU-Z score is weird. I get 520 single core on a fully patched OS/BIOS. You got 485 for the 8700K. That should not be a possibility unless you run stuff in the background or have some settings wrong. And I get lower speed memory as well, on a cheap motherboard.

https://valid.x86.fr/bench/g3egu4

Robbo9999
Senior Member

Posts: 1246

Posted on: 05/02/2018 07:19 PM

As promised:

But it is not this nice in reality. Upon start of system steam was downloading update for some game and moving files. While one thread was used, it put CPU to 4.35GHz on one core and rest somewhere between 4.1~4.25GHz.
So, to get actual idle readings, I had to reset statistics. In time I wrote this, average is 14.6W.
Minimum 9W, Maximum 69W. And yes, that means a bit of browser fun with upload of image and looking for this thread. (8minutes after taking screenshot.)

But voltage difference is really nice. But I am disappointed, minimal clock is 2GHz, but only on one core at time. Rest are on 2.2GHz. I do not get this at all. Maybe controlled by BIOS same way as XFR and power limits?

Secondly XFR itself puts all CPU cores only to 3.9GHz in Cinebench R15 as I did set Voltage to auto instead manually increasing it.
When load is on all cores, auto voltage for this 3.9GHz is 1.331V. In Single threaded Test, voltage goes to 1.512V @ 4.35GHz.
So it looks like you're saving about 10W on average during light load/idle situations when comparing with your other screenshot, I mean that's worthwhile but you're sacrificing top end performance if your max Cinebench frequency is only 3.9Ghz when at stock, in contrast to your 4.2Ghz overclock. Is it possible to overclock & retain XFR and Cool & Quiet? I suppose the Intel equivalent would be just running an Offset Voltage and increasing the multipliers, and then allowing it to downclock, etc, at idle.

(I've personally found with my Skylake 6700K that C-states enabled, Offset Voltage with an overclock, and Windows High Performance Profile which results in locked frequency at 4.7Ghz is the best combination of Power savings & performance, no performance sacrifice, but low idle power consumption: C-states enabled is alone enough to lower idle power consumption, so no need to drop the frequency too).

nizzen
Senior Member