AMD Ryzen 7 2700X Review: Redefining Ryzen

[Rogue Leader]

Why aren't these problems discussed more openly? Not everyone is a Windows gamer.

A very valid point but keep in mind this is a consumer-oriented level tech website. Also keep in mind that AMD has never targeted the professional level server business as Intel started doing with their Xeon line from some 20 years ago. While AMD's Threadripper offers an affordable counterpart to Intel's Xeon for a server processor in lower tier use, it has a lot of limitations. There's a reason major professional server vendors like my former employer, Dell/EMC, only use Xeon processors.

EPYC is the server Ryzen processor, AMD is targeting it as of last summer. Dell/EMC will be offering multiple models

https://www.zdnet.com/article/amd-epyc-powers-new-dell-emc-poweredge-servers/

 

[Sakkura]

I agree the Ryzen 2700 is a hot chip, but it may not be for everyone. Last year I tried to configure a server with the 2700 and found it seemed incompatible with VMware - at least without hamburger helpers I was unwilling to use. So I put the 2700 on the shelf and switched to a Xeon based board. I have also heard (but had no experience) that there are 2700 problems with Linux. Why aren't these issues openly discussed? Not everyone is a Windows gamer - where the 2700 is undoubtedly a prime contender.

Uh... the 2700 was not available last year.

 

[10tacle]

[EPYC is the server Ryzen processor, AMD is targeting it as of last summer. Dell/EMC will be offering multiple models

https://www.zdnet.com/article/amd-epyc-powers-new-dell-emc-poweredge-servers/

Ah thanks for the correction. Good to know. Obviously I've been out of the industry for some years and haven't kept up with the latest trending news.

 

[Marlin1975]

When will you add ALL the patches to the intel system and re-test?

Why don't you ask that of the other five major hardware review sites that tested the same way? And I'll remind you that Anantech was the only major one that did that and they are reviewing their results and going to re-test.

PS: overclocking Intel's Coffee Lake more than makes up for the 5-7% performance loss with the patches. So don't get all giddy. That's been a dominating factor over AMD ever since the first generation of the Core i-series (i5 7xx, i7 8xx). Here's a nine year old memory refresher for you:

https://images.anandtech.com/graphs/intellynnfieldlaunch_090409000254/19910.png

It used to be the other way around.

Where did I say I did not? Go look at some of those sites and see I did ask the same.

Instead of ASSuming why not go look for yourself?

 

[10tacle]

Where did I say I did not? Go look at some of those sites and see I did ask the same.

Instead of ASSuming why not go look for yourself?

Awesome - you can capitalize three letters with an insult! I've read all reviews from major sites. Only one has decided to retest as I stated due to discrepancies against the other sites' results (as another mentioned here previously already).

BTW: your ignoring the point of overclocking Coffee Lake erasing the performance loss of the patches is duly noted.

 

[Saurabh Harwande]

Where did I say I did not? Go look at some of those sites and see I did ask the same.

Instead of ASSuming why not go look for yourself?

Awesome - you can capitalize three letters with an insult! I've read all reviews from major sites. Only one has decided to retest as I stated due to discrepancies against the other sites' results (as another mentioned here previously already).

BTW: your ignoring the point of overclocking Coffee Lake erasing the performance loss of the patches is duly noted.

Why should he even comment about the overclocking stuff anyways? It is you who is pulling the things out of the thin air and trying to argue about it. He just asked if and when are they going to test with latest patches. Just need to sit back and relax.
Clue: He did not say anything bad about intel.

 

[10tacle]

Why should he even comment about the overclocking stuff anyways? It is you who is pulling the things out of the thin air and trying to argue about it. He just asked if and when are they going to test with latest patches.

Well because that's the typical snarky response people (staunch AMD fanboys) have on the performance loss of Intel's patches, that's why. If that was not his intent then I owe him an apology - that's just the way I read it. Anyway my next build is going to be a Ryzen 7 2800X as I'm getting more into video rendering and less PC gaming. And running a 1440p monitor takes any gaming advantage Intel had, no matter the overclock, away from Zen (as we addressed for days in the thread history above). I'm just waiting to see if this second generation is a problem-free release unlike the first which was never fully addressed and fixed.

 

[Marlin1975]

Where did I say I did not? Go look at some of those sites and see I did ask the same.

Instead of ASSuming why not go look for yourself?

Awesome - you can capitalize three letters with an insult! I've read all reviews from major sites. Only one has decided to retest as I stated due to discrepancies against the other sites' results (as another mentioned here previously already).

BTW: your ignoring the point of overclocking Coffee Lake erasing the performance loss of the patches is duly noted.

What are you rambling on about now? No where did I say anything about overclocking which makes your statements look just weird and no where near the point.

Its simple. Anandtech ran 2 systems at factory specs and both systems were patched with all current patchs. They ran a apple to apple test.

Toms ran the AMD system with patchs but intels unpatched. This is a bad and flawed review. Its should be pulled till its fixed. As should any other "review" that is not offering a fair comparison.
That and overclocking is never guaranteed. Way to many variables. Only good review is one that shows both systems as default settings and updated. That simple.

 

[10tacle]

What are you rambling on about now? No where did I say anything about overclocking which makes your statements look just weird and no where near the point. Its simple. Anandtech ran 2 systems at factory specs and both systems were patched with all current patchs. They ran a apple to apple test.

I'm making a point Mr. 3-post commenter since 2007. Only ONE tech review website deemed it worthy of a re-do based on inconsistencies of other site reviews, and yep, that's Anandtech.

Toms ran the AMD system with patchs but intels unpatched. This is a bad and flawed review. Its should be pulled till its fixed. As should any other "review" that is not offering a fair comparison.
That and overclocking is never guaranteed. Way to many variables. Only good review is one that shows both systems as default settings and updated. That simple.

Nobody said overclocking was a fix, but Intel has a long history of owning AMD on overclocking overhead since Core 1 as mentioned previously. If Intel jacked things up, it's on them. But AMD has little wiggle room for improvement when it comes to overclocking headroom. There is a reason people (like me) have held on to older Intel chipsets and not needed to upgrade every generation thanks to overclocking headroom.

Oh by the way: I can dig up old reviews and comments bragging about overclocking AMD processors that would stomp a more expensive Intel CPU. And it was true. At the time, for gaming. Again, AMD owns Intel on productivity and is the better solution overall in a balanced gaming + productivity system.

 

[vawokil]

Techradar also had similar results to Anandtech, also with Meltdown+Spectre patches applied.

 

[logainofhades]

Despite your concern that the Z370 Extreme4 is not capable of supporting the 8700k, you seem awfully confident that those atrocious 3+1/4+1 VRM with cheap caps and non-existent or functional heatsink on those B350 can safely delivers ~100A to power a 4Ghz 8cores Ryzen for daily usage.

As of right now, all the announced ASROCK X470 have that config, all the MSI X470 either have that config or the 8+8 one, the top/high end ASUS and Gigabyte have that config; so surely more than 50% of the X470 out there.

The B350 Pro 4 isn't a 3+1/4+1 board, and has heatsinks.
https://www.newegg.com/Product/Product.aspx?Item=N82E16813157761

I wouldn't call the Asus x470 strix a low end board, and it doesn't have the 8+4. Spending a lot of money for maybe a 4.2 overclock, at best, if you are lucky, or be happy with 4.0 and use that money on something more worthwhile. Spending a lot of money, on an AMD board, makes 0 sense, for 99.99% of most users.

Z370 extreme4 is capable of handling an 8700k. I really don't get where you think that.

 

[tblitch78]

How can I get rid of this annoying ray tracing video that appears on every single article I read?

 

[FD2Raptor]

The B350 Pro 4 isn't a 3+1/4+1 board, and has heatsinks.
https://www.newegg.com/Product/Product.aspx?Item=N82E16813157761

I wouldn't call the Asus x470 strix a low end board, and it doesn't have the 8+4. Spending a lot of money for maybe a 4.2 overclock, at best, if you are lucky, or be happy with 4.0 and use that money on something more worthwhile. Spending a lot of money, on an AMD board, makes 0 sense, for 99.99% of most users.

Z370 extreme4 is capable of handling an 8700k. I really don't get where you think that.

Did you just try to argue that the B350 Pro4 is not a 3 phases Vcore:
https://youtu.be/vcqu-9Ejf60?t=16m52s

And go look at your own post where you call using an Z370 Extreme4 for the 8700k:

even that might be a bit low

And I couldn't care less about 4.2Ghz Ryzen. If 2018 is going to be like 2017, then I'm looking 30+ degrees Celsius for most of the year (last year, temps stayed in the mid 30 well into November; from last summer till now, temps has never fallen below 20 degrees). If reviewers, in testing environment of ~20-25 degrees Celsius, need liquid cooler to just manage to hold the 4.2-4.3Ghz OC at not critical temperatures, I'm never going to see that OC anyway.

The reason why I decided that I might try AMD this time around is exactly the promise of 4Ghz@1.1-1.2V that would, in all likelihood, mean the temperatures would be manageable at that OC without too much work/cost.

 

[logainofhades]

https://www.asrock.com/mb/AMD/AB350%20Pro4/#Specification

- 9 Power Phase design

https://www.asrock.com/mb/AMD/X370%20Pro4/index.asp#Specification

Has the same phase design as the B350 pro 4.

I am going strictly by what the manufacturer is claiming.

 

[cabbagemoth]

The Stilt over on Anandtech forums thinks your testing is screwed.

You've got 104.7W in Prime95 and say PSO is off

He's saying that the default power limit is actually 141.7W even with PSO off so you've messed with the default power limit to get that.

Also that you can't get a score of over 1800 without PSO so for that you DID have PSO on.

Page 74 and 75: https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-75#post-39396698

There's only you two that talk of PSO settings and the CPU power consumptions.

Who's doing it right?

 

[FD2Raptor]

https://www.asrock.com/mb/AMD/AB350%20Pro4/#Specification

- 9 Power Phase design

https://www.asrock.com/mb/AMD/X370%20Pro4/index.asp#Specification

Has the same phase design as the B350 pro 4.

I am going strictly by what the manufacturer is claiming.

Yeah and if all those "9 phases" are for powering the CPU core, how does other stuff like the memory controller or iGPU get their power?

 

[cryoburner]

Only ONE tech review website deemed it worthy of a re-do based on inconsistencies of other site reviews, and yep, that's Anandtech.

Tom's Hardware admitted right in their review to having a problem with their testing methodology, namely that they made the mistake of not realizing that their X470 motherboard was already patched for Spectre, so they didn't apply patched firmware to their Intel motherboard, potentially providing Intel's CPUs with an unfair advantage in their test results. Whether that made any notable difference to how those processors performed is yet to be seen, but it's certainly possible. Unfortunately, they tucked that information away on the test setup page instead of posting it somewhere more prominent, so many readers probably missed it...

Unfortunately, we were only made aware that Variant 2 mitigations are present in our X470 board's firmware just before launch, precluding us from re-testing the Intel platforms with patches applied. We're working on this now, and plan to post updated results in future reviews.

The lack of Spectre Variant 2 patches in our Intel results likely give the Core CPUs a slight advantage over AMD's patched platforms. But the performance difference should be minimal with modern processors.

 

[Rogue Leader]

https://www.asrock.com/mb/AMD/AB350%20Pro4/#Specification

- 9 Power Phase design

https://www.asrock.com/mb/AMD/X370%20Pro4/index.asp#Specification

Has the same phase design as the B350 pro 4.

I am going strictly by what the manufacturer is claiming.

Yeah and if all those "9 phases" are for powering the CPU core, how does other stuff like the memory controller or iGPU get their power?

Take a close look at the board you can physically see that 8 phases go to the CPU and 1 goes to the memory controller. Theres no discrepancy here, when a board manufacturer talks about power phases they are specifically talking about the ones that go to the CPU and Memory, they have nothing to do with any other functionality on the motherboard.

 

[FD2Raptor]

Yeah and if all those "9 phases" are for powering the CPU core, how does other stuff like the memory controller or iGPU get their power?

Take a close look at the board you can physically see that 8 phases go to the CPU and 1 goes to the memory controller. Theres no discrepancy here, when a board manufacturer talks about power phases they are specifically talking about the ones that go to the CPU and Memory, they have nothing to do with any other functionality on the motherboard.

Please.

When manufacturer generally reserve the use of high phase counts (8/6*2/etc) VRM for 300W load of high-end GPU...
When, for example, the top-of-the-line Gigabyte X470 Gaming 7 have a (5*2)+2 config, if you think this budget tier board actually have a true 8 phase for Vcore, man....

And really? Did you just refer to the "memory controller and iGPU" as "other functionalities on the motherboard"?

 

[Rogue Leader]

Yeah and if all those "9 phases" are for powering the CPU core, how does other stuff like the memory controller or iGPU get their power?

Take a close look at the board you can physically see that 8 phases go to the CPU and 1 goes to the memory controller. Theres no discrepancy here, when a board manufacturer talks about power phases they are specifically talking about the ones that go to the CPU and Memory, they have nothing to do with any other functionality on the motherboard.

Please.

When manufacturer generally reserve the use of high phase counts (8/6*2/etc) VRM for 300W load of high-end GPU...
When, for example, the top-of-the-line Gigabyte X470 Gaming 7 have a (5*2)+2 config, if you think this budget tier board actually have a true 8 phase for Vcore, man....

And really? Did you just refer to the "memory controller and iGPU" as "other functionalities on the motherboard"?

The iGPU is powered by the same phases as the CPU. There is no separate power system. The memory controller is powered by the same phases as the CPU as well, Its part of the CPU (in a CPU such as this and Intel current CPUs that has an onboard memory controller of course).

You're literally twisting the design around to fit your narrative that the boards power phases are crap, as if they would manually designate parts of the phases for parts of the processor. Thats not how it works, at all.

Open your eyes, you can physically see the phases around the CPU. In fact many manufacturers use variations of the same exact PCB designs for all their models. They would literally need to make a different design at great cost to make the board "cheaper".

 

[FD2Raptor]

The iGPU is powered by the same phases as the CPU. There is no separate power system. The memory controller is powered by the same phases as the CPU as well, Its part of the CPU (in a CPU such as this and Intel current CPUs that has an onboard memory controller of course).

Next you're going to tell people to increase their Vcore voltage to get better iGPU OC.

 

[Rogue Leader]

The iGPU is powered by the same phases as the CPU. There is no separate power system. The memory controller is powered by the same phases as the CPU as well, Its part of the CPU (in a CPU such as this and Intel current CPUs that has an onboard memory controller of course).

Next you're going to tell people to increase their Vcore voltage to get better iGPU OC.

Actually if you knew what you were talking about you'd know that vcore voltage can affect the iGPU, but not always in a good way.

On the Skylake chips there was a known issue where if the CPU voltage were changed even slightly to run high speed ram for example it was causing issues with the iGPU (lines on screen, flickering, etc). Most people didn't notice this because when they were running high speed ram, or overclocking, they weren't using the iGPU anyway.

I literally had this exact problem with my wife's system, instead of waiting for a fix from Intel I stuck a GTX 950 in there.

 

[FD2Raptor]

Actually if you knew what you were talking about you'd know that vcore voltage can affect the iGPU, but not always in a good way.

On the Skylake chips there was a known issue where if the CPU voltage were changed even slightly to run high speed ram for example it was causing issues with the iGPU (lines on screen, flickering, etc). Most people didn't notice this because when they were running high speed ram, or overclocking, they weren't using the iGPU anyway.

I literally had this exact problem with my wife's system, instead of waiting for a fix from Intel I stuck a GTX 950 in there.

https://www.tomshardware.co.uk/amd-ryzen-3-2200g-raven-ridge-cpu,review-34213-2.html

We also adjusted the VDDCR SoC voltage, a single rail that feeds the uncore and graphics domains, to 1.25V.

https://www.anandtech.com/show/12542/overclocking-the-amd-ryzen-apus-guide-results/6

The SoC and APU GFX voltage share the same rail for both uncore and graphics, so setting both is paramount to ensuring stability when overclocking integrated graphics on your APU; 1.25V is AMDs maximum recommended safe SoC voltage.

This is the PWM controller for all the B350 Pro4/GamingK4:
https://www.intersil.com/content/dam/intersil/documents/isl9/isl95712.pdf

The Core Voltage Regulators can be configured for 4-, 3-, 2-, or 1-phase operation

 

[Rogue Leader]

Actually if you knew what you were talking about you'd know that vcore voltage can affect the iGPU, but not always in a good way.

On the Skylake chips there was a known issue where if the CPU voltage were changed even slightly to run high speed ram for example it was causing issues with the iGPU (lines on screen, flickering, etc). Most people didn't notice this because when they were running high speed ram, or overclocking, they weren't using the iGPU anyway.

I literally had this exact problem with my wife's system, instead of waiting for a fix from Intel I stuck a GTX 950 in there.

https://www.tomshardware.co.uk/amd-ryzen-3-2200g-raven-ridge-cpu,review-34213-2.html

We also adjusted the VDDCR SoC voltage, a single rail that feeds the uncore and graphics domains, to 1.25V.

https://www.anandtech.com/show/12542/overclocking-the-amd-ryzen-apus-guide-results/6

The SoC and APU GFX voltage share the same rail for both uncore and graphics, so setting both is paramount to ensuring stability when overclocking integrated graphics on your APU; 1.25V is AMDs maximum recommended safe SoC voltage.

This is the PWM controller for all the B350 Pro4/GamingK4:
https://www.intersil.com/content/dam/intersil/documents/isl9/isl95712.pdf

The Core Voltage Regulators can be configured for 4-, 3-, 2-, or 1-phase operation

Every single thing you posted supports exactly what I said that the power system setup is the same among the chipsets. So your assertion that for some reason a cheap motherboard would distribute 9 phases differently than any other board is still wrong, especially when you can see the phases around the CPU (and the memory) there is NO benefit of the phases being assigned to different things than in a more expensive board. More expensive board can have more phases though which are assigned to the vcore for overclocking purposes.

 

[bigpinkdragon286]

Only ONE tech review website deemed it worthy of a re-do based on inconsistencies of other site reviews, and yep, that's Anandtech.

Tom's Hardware admitted right in their review to having a problem with their testing methodology, namely that they made the mistake of not realizing that their X470 motherboard was already patched for Spectre, so they didn't apply patched firmware to their Intel motherboard, potentially providing Intel's CPUs with an unfair advantage in their test results. Whether that made any notable difference to how those processors performed is yet to be seen, but it's certainly possible. Unfortunately, they tucked that information away on the test setup page instead of posting it somewhere more prominent, so many readers probably missed it...

Unfortunately, we were only made aware that Variant 2 mitigations are present in our X470 board's firmware just before launch, precluding us from re-testing the Intel platforms with patches applied. We're working on this now, and plan to post updated results in future reviews.

The lack of Spectre Variant 2 patches in our Intel results likely give the Core CPUs a slight advantage over AMD's patched platforms. But the performance difference should be minimal with modern processors.

Anandtech has published an article detailing their findings and it's really quite interesting. Hopefully the folks at Tom's and other review sites are familiar with, or become aware of why Anandtech's benchmark numbers went the way they did.

Anandtech update

The gist of the matter is, Spectre and Meltdown mitigations decrease the performance of IO operations on Intel platforms. The HPET (High Precision Event Timer) is mapped in such a way, accessing it constitutes an IO operation. Forcing Windows to use HPET as it's primary timer has a major impact on Intel's CPU performance in cases where the software makes use of the timer. The difference hasn't been quantified between pre- and post-patched machines yet, but it's not an unfair assessment to say there won't be a performance difference, and the final result won't change anyway. HPET being used as the primary timer in Windows with a patched Intel system seems to come with a very high performance penalty.

It would be a nice addition to any benchmark setup pages going forward to include the state of the HPET timer settings, both in UEFI and the Windows setting used. Anandtech also points out that installing software can change the HPET setting in Windows, so the step at which the Windows HPET setting is looked at and potentially modified by the reviewer should obviously be after all pertinent software has been installed.

No conspiracies or intentional biases seem to have been necessary for the interesting results found by the Anandtech reviewers.