AMD launches Zen+ 12nm Ryzen and X470 motherboards

[.vodka]

Computerbase reports that the X470 boards already shipped with the Spectre V2 mitigations and that there is no impact on performance, not even disk performance.

So, Zen isn't Meltdown vulnerable, Spectre v2 fix doesn't impact performance, and Spectre v1 fix through the OS doesn't seem to impact performance either.

Great! Love my 1700 more every day

---------

On the other hand and while we're at it, I just updated my old Sandy notebook with the latest microcode (as Lenovo would never release a new BIOS update), and is now Spectre v2 fixed. It does seem slower and less snappier than when just dealing with the Meltdown fix... oh well. My parents use it for basic web browsing, so they won't feel much of an impact.

 

[IRobot23]

Guys who like to have raid of dual m.2 slots (2x4Gb/s PCIE3.0 - combo) on X470. C7H allows that and it is working, but only in raid mode as far as I know. Otherwise you will get 2x2Gb/s PCIE3.0

 

[Kenmitch]

So, Zen isn't Meltdown vulnerable, Spectre v2 fix doesn't impact performance, and Spectre v1 fix through the OS doesn't seem to impact performance either.

Great! Love my 1700 more every day

---------

On the other hand and while we're at it, I just updated my old Sandy notebook with the latest microcode (as Lenovo would never release a new BIOS update), and is now Spectre v2 fixed. It does seem slower and less snappier than when just dealing with the Meltdown fix... oh well. My parents use it for basic web browsing, so they won't feel much of an impact.

You do know there are many Intel users feverishly looking for AMD exploits that will level the playing field once again....Don't toot the horn just yet. /s

 

[Dasa2]

You do know there are many Intel users feverishly looking for AMD exploits that will level the playing field once again....Don't toot the horn just yet. /s

marketing people maybe
most users don't care as there is no evidence that it makes any difference to most users

most people are happy for the competition driving forward progress after a long stagnation

 

[Whitestar127]

Adore's take:

 

[alexruiz]

Other than Anand's
Has anybody seen reviews of the 2700 and the 2600?
All the reviews I have been able to find are 2700X or 2600X...

 

[B-Riz]

Other than Anand's
Has anybody seen reviews of the 2700 and the 2600?
All the reviews I have been able to find are 2700X or 2600X...

https://www.techspot.com/review/1614-ryzen-2600/

AMD only sampled the X series this time, so they would get the attention.

Maybe when B450 comes out, non X will get some love.

 

[positivedoppler]

https://wccftech.com/review/amd-ryzen-7-2700-cpu-review-asrock-x470-asus-x470-motherboards/

 

[Jan Olšan]

Other than Anand's
Has anybody seen reviews of the 2700 and the 2600?
All the reviews I have been able to find are 2700X or 2600X...

https://nl.hardware.info/reviews/81...0x-ryzen-7-2700-a-2700x-+-6-x470-moederborden - has 2700. Note that it probably reuses older results from comparisons so the Intel CPUs in the graphs have no Meltdown/Spectre patches (somebody correct me if I am wrong and the state otherwise somewhere).
https://www.computerbase.de/2018-04/amd-ryzen-2000-test/ -- all four SKUs.

 

[alexruiz]

Thank you guys!
I need to seem how much better the 2700X is over the vanilla 2700.

 

[PhonakV30]

https://nl.hardware.info/reviews/81...0x-ryzen-7-2700-a-2700x-+-6-x470-moederborden - has 2700. Note that it probably reuses older results from comparisons so the Intel CPUs in the graphs have no Meltdown/Spectre patches (somebody correct me if I am wrong and the state otherwise somewhere).
https://www.computerbase.de/2018-04/amd-ryzen-2000-test/ -- all four SKUs.

https://nl.hardware.info/reviews/81...aming-benchmarks-rise-of-the-tomb-raider-dx12

How come 7600K is Top ? ( 1080p Medium )

 

[zinfamous]

Thank you guys!
I need to seem how much better the 2700X is over the vanilla 2700.

I think with this round, the prices between the vanilla and X versions are both narrower on the box, and now include the coolers, so it sorta makes the non-X versions far less relevant. We're talking a ~$50? overall cost difference between the two; where with Zen, the first, the gap was about $100 + cooler cost (let's say $30-$100+) for X versions.

EDIT: got the prices wrong: Zen+ is actually only $30 cost difference between X and non-X. I mean...why even bother with non-X? Seriously: it's just dumb, isn't it?

with Zen, the first, it was $70 difference between 1700 and 1700X, and $40 between 1600 and 1600X (and then of course the extra cooler costs on the X versions). Another way to look at the value proposition, is that with the first round, the 1700 seemed to be the real sweet spot, even compared to price/performance of the 1800X, priced at what, $150 more? I seem to recall that slapping a Noctua on a 1700, some managed to get to within 2 or 3%? of the 1800X, for roughly 80% of the cost? Or did I pull those numbers out of my petard?

 

[burninatortech4]

I just bought a 2700x to replace the 1700 in my Gigabyte AB350M- Gaming 3. I'm a little nervous the motherboard won't be able to handle it. I recently attached a 50mm fan to the VRM heatsink. We'll see how it goes.

 

[IEC]

I think with this round, the prices between the vanilla and X versions are both narrower on the box, and now include the coolers, so it sorta makes the non-X versions far less relevant. We're talking a ~$50? overall cost difference between the two; where with Zen, the first, the gap was about $100 + cooler cost (let's say $30-$100+) for X versions.

EDIT: got the prices wrong: Zen+ is actually only $30 cost difference between X and non-X. I mean...why even bother with non-X? Seriously: it's just dumb, isn't it?

with Zen, the first, it was $70 difference between 1700 and 1700X, and $40 between 1600 and 1600X (and then of course the extra cooler costs on the X versions). Another way to look at the value proposition, is that with the first round, the 1700 seemed to be the real sweet spot, even compared to price/performance of the 1800X, priced at what, $150 more? I seem to recall that slapping a Noctua on a 1700, some managed to get to within 2 or 3%? of the 1800X, for roughly 80% of the cost? Or did I pull those numbers out of my petard?

The 2700X also comes with a better cooler than the 2700. So the value-add for $30 makes the 2700X the better pick of the two.

I bought the 1800X at $499 at launch and a 1700 later for $280(?). The 1700 can OC to 4GHz at lower voltage than the 1800X, so go figure...

 

[dlerious]

Other than Anand's
Has anybody seen reviews of the 2700 and the 2600?
All the reviews I have been able to find are 2700X or 2600X...

Gamers Nexus has quite a few up
https://www.youtube.com/watch?v=GDggr3kt96Q

 

[dlerious]

The 2700X also comes with a better cooler than the 2700. So the value-add for $30 makes the 2700X the better pick of the two.

I bought the 1800X at $499 at launch and a 1700 later for $280(?). The 1700 can OC to 4GHz at lower voltage than the 1800X, so go figure...

I bought the 1800X at launch too, along with the Crosshair 6 Hero ($250). At current prices for new 1st gen, I'm not going to get much trying to sell used. Wouldn't mind the better clocks and lower power of the new chips, but I'll wait for the next gen.

 

[AtenRa]

Does anyone know what changed in the AT Ryzen 2 review on the Gaming benchmarks ??
Now 8700K is in front and in some games with a big margin.

https://www.anandtech.com/show/12625/amd-second-generation-ryzen-7-2700x-2700-ryzen-5-2600x-2600/15

 

[.vodka]

This changed:

https://www.anandtech.com/show/12678/a-timely-discovery-examining-amd-2nd-gen-ryzen-results

HPET strikes again. What a mess.

 

[AtenRa]

This changed:

https://www.anandtech.com/show/12678/a-timely-discovery-examining-amd-2nd-gen-ryzen-results

HPET strikes again. What a mess.

thanks

 

[MrTeal]

It's an interesting problem, that's for sure. I agree with Ian and Ryan's decision to benchmark the processors without HPET enabled as that is the default that most users would have set up, but that makes the benchmarks less useful to me.

I really wish AT would include both sets of benchmarks, as many of the people who would really care about minute differences in system performance would also overclock and likely have HPET enabled.

 

[PotatoWithEarsOnSide]

Good article.
Reminded me of the sleep mode bug related to the first gen Ryzen, where various benchmarks were showing higher than consensus results after going into sleep mode caused a timing issue on wake-up. The takeaway from that was that benchmarks utilising system timers were inherently flawed/exploitable.

 

[gregulator]

So is performance actually better, or is it just that the reporting software changes what it reports? There should be a third party way to verify benchmarks. I guess for render time type benchmarks you could use a stopwatch to compare. Not sure about fps tests (maybe there is a fancy monitor that can tell you when you get a new frame?)

 

[Paratus]

So is the deficit on the 8700k caused by the HEPT timer usage hitting IO more often which is uniquely impacted by the meltdown microcode patches?

 

[IEC]

So is performance actually better, or is it just that the reporting software changes what it reports? There should be a third party way to verify benchmarks. I guess for render time type benchmarks you could use a stopwatch to compare. Not sure about fps tests (maybe there is a fancy monitor that can tell you when you get a new frame?)

The software changes what it reports. A similar thing happened with Ryzen with HPET disabled and REFCLK changed from stock:
https://hwbot.org/newsflash/4335_ry...bias_w88.110_not_allowed_on_select_benchmarks

Some HWBOT benchmarks require HPET to be enabled to be considered a valid score for this reason.

 

[IllogicalGlory]

So is the deficit on the 8700k caused by the timer usage hitting IO more often which is uniquely impacted by the microcode patches?

The gaming results hadn't changed since launch until the latest update, so I'd expect that the microcode updates haven't affected anything. The effects of forcing HPET were visible back then too.

Like this result where, the i5-7400 looks like the fastest thing ever (from the 8700K launch review). I'm fairly confident that this is caused by HPET being forced, and the 8700K (before the results were removed), scored the same (or a percent or two higher) in the Ryzen 2000 review as it did at launch.