AMD launches Zen+ 12nm Ryzen and X470 motherboards

[Markfw]

Frys already has in stock just can’t sell them yet. They have the prices too.

So, what are the prices ?

 

[csbin]

https://www.chiphell.com/thread-1841382-1-1.html

2600X+DDR4 3660 CL14

 

[Mockingbird]

https://www.chiphell.com/thread-1841382-1-1.html

2600X+DDR4 3660 CL14

That's not good.

Core i7-6700K with DDR4-3600 @ 17-18-18-38 gets 41.8 ns for Latency

http://www.legitreviews.com/ddr4-me...nding-the-best-ddr4-memory-kit-speed_170340/3

 

[Gideon]

That's not good.

Core i7-6700K with DDR4-3600 @ 17-18-18-38 gets 41.8 ns for Latency

http://www.legitreviews.com/ddr4-me...nding-the-best-ddr4-memory-kit-speed_170340/3

It's certainly an improvement, but yeah, Intel would probably hit < 40 ns with those timings.

It seems that Ryzen 2 has a hard time going substantially below 60 ns just like original Ryzen had with 70 ns. An AGESA update that would shave another 5-10ns off of that (similar to the one we had a month after the launch of Ryzen) would really help and put it on parity with Skylake @ faster JEDEC frequencies.

I'm not holding my breath, but hopefully we'll see some improvements down the line.

 

[exquisitechar]

That's not good.

Core i7-6700K with DDR4-3600 @ 17-18-18-38 gets 41.8 ns for Latency

http://www.legitreviews.com/ddr4-me...nding-the-best-ddr4-memory-kit-speed_170340/3

Obviously, that wasn't going to change, AMD is going to have a hard time achieving parity with Intel in this area because of the way Ryzen CPUs are designed. But from what I've read, it's not always as simple as Ryzen having more latency, either. There are some workarounds at play here that help mitigate that in some workloads, such as the relatively large caches. Sub 60 ns memory latency is still really nice.

 

[PhonakV30]

That's not good.

Core i7-6700K with DDR4-3600 @ 17-18-18-38 gets 41.8 ns for Latency

http://www.legitreviews.com/ddr4-me...nding-the-best-ddr4-memory-kit-speed_170340/3

Seriously You expected 75ns => 40ns? while everyone knows that's Impossible to reach 40ns on Ryzen System.

 

[Gideon]

I was wondering why AMD chose to release this CPU that late (19th), despite being already sold by some overeager retailers since last week (Italy, several Central European countries, now Russia ...)

Usually you'll get similar leaks/accidental buys ... maybe a week or so in-advance, not 3 weeks. The only reason I could think of, is to get a new AGESA out and bioses stable, but I'm not sure if that's a case.

Interestingly though, A new AGESA (PI 1.0.0.2a) does seem to be in the works.

 

[IRobot23]

That's not good.

Core i7-6700K with DDR4-3600 @ 17-18-18-38 gets 41.8 ns for Latency

http://www.legitreviews.com/ddr4-me...nding-the-best-ddr4-memory-kit-speed_170340/3

How so? i7 7820X has some problems to reach below 60ns.

 

[french toast]

How so? i7 7820X has some problems to reach below 60ns.

Interesting, have you got a decent review you can link that shows this please? I'm more interested in Intel's non ring bus equivalent as this will give us a better idea of future positions.

 

[jpiniero]

Interesting, have you got a decent review you can link that shows this please? I'm more interested in Intel's non ring bus equivalent as this will give us a better idea of future positions.

You can get lower than 60 for sure but you would need to OC the mesh.

 

[coercitiv]

You can get lower than 60 for sure but you would need to OC the mesh.

The same way we OC IF on Ryzen?

 

[xblax]

Obviously, that wasn't going to change, AMD is going to have a hard time achieving parity with Intel in this area because of the way Ryzen CPUs are designed. But from what I've read, it's not always as simple as Ryzen having more latency, either. There are some workarounds at play here that help mitigate that in some workloads, such as the relatively large caches. Sub 60 ns memory latency is still really nice.

Ryzen 2000 has slightly lower L3 cache latency than Coffee Lake which also helps a bit.

But I really wonder why AMD is not allowing to adjust the Infinity Fabric frequency independent of the RAM speed. Running the Infinity Fabric at exactly twice the RAM frequency should in theory lower the latency dramatically while avoiding most of the penalties for crossing clock domains. Maybe that's going to be an optimization we seen in Zen 2.

 

[Mockingbird]

Obviously, that wasn't going to change, AMD is going to have a hard time achieving parity with Intel in this area because of the way Ryzen CPUs are designed. But from what I've read, it's not always as simple as Ryzen having more latency, either. There are some workarounds at play here that help mitigate that in some workloads, such as the relatively large caches. Sub 60 ns memory latency is still really nice.

Raven Ridge has higher latency than Coffee Lake-S, so it's not the CCX.

 

[Mockingbird]

Review: AMD Ryzen 5 2600 (Chipset X370)
BY IVÁN MARTÍNEZ
04/09/2018
CPUS & APUS


Following the leaks before the launch date, with us today we have the new AMD Ryzen 5 2600 , the younger brother of the Ryzen 7 2700X that we analyzed last week. Of course, like the Ryzen 5 1600 of the previous generation, offers 6 cores and 12 threads , which increase their frequencies to the base of 3.40 GHz , and can reach 3.90 GHz in Boost mode.

Let's see first its complete technical specifications :

AMD Ryzen 5 2600
Socket AM4
Fabrication process 12 nm
Cores / Threads 6 cores / 12 threads
Base Frequency 3.40 GHz
Turbo Frequency 3.90 GHz
Cache L3 2 × 8 MB
Compatible Memory DDR4-2933 Dual-Channel
PCIe lines 16 lines
TDP 65W
Index of contents [Ocultar]

• 1 Packaging and Accessories
• 2 AMD Ryzen 5 2600
• 3 Testing Equipment
• 4 Tests
• 5 Temperature, Overclocking and Consumption
• 6 Conclusion

Packaging and Accessories




The AMD Ryzen 5 2600 comes in a box similar to the previous generation design, showing the processor from the side and protecting it perfectly inside, as well as the heatsink.





In addition to the warranty card and a Ryzen 5 sticker, this model comes with the AMD Wraith Stealth dissipator, a model already seen in the first generation and which offers nothing new.







AMD Ryzen 5 2600


As we anticipated, the AMD Ryzen 5 2600 offers 6 cores , which in turn offer 12 threads that start from a base frequency of 3.4 GHz to reach 3.9 GHz in Turbo mode thanks to Precision Boost technology. It offers a total of 2 × 8 MB of L3 cache , as well as 6 x 512 KB of L2 cache and 6 x 32 KB + 6 x 64 KB of L1 cache, so it offers a cache distribution identical to that of the Ryzen 5 1600 .



As we can see in CPU-Z , the set of instructions supported has MMX (+), SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, SSE4A, x86-64, AMD-V, AES, AVX, AVX2, FMA3 and SHA, also the same as in the first generation of Ryzen .



Obviously it is a processor for the AM4 socket and therefore makes use of the " Pin Grip Array " connection system, where the pins are in the processor and not in the motherboard, highlighting the use of a total of 1,331 contacts.

 

[Mockingbird]

Test Equipment


To analyze the performance of the AMD Ryzen 5 2600 processor we have assembled the following test equipment:

• Motherboard ASRock X370 Killer SLI
• Memories G.Skill FlareX DDR4
• MSI GeForce GTX 1070 Gaming Z
• Be Quiet Font! Dark Power Pro 11 1200W
• Adata SU900 256GB SSD
• SSD Corsair LX 512 GB
• Operating System Windows 10 64 bits

Unlike the Ryzen 7 2700X that we already analyzed, this model has not allowed us to configure the memories above 2933 MHz , something that reminds us of what happened in the previous generation, although having no optimized drivers for the chipset was a possibility . In addition, the impossibility of playing on this platform is maintained , something that is sure to be solved at its official launch or with other motherboards.

Tests
Next, we show you the results of the performance tests performed on this and other processors in calculation benchmarks (wPrime), rendering (Cinebench), coding (x264), memories (Aida64) or the most popular focused on games (3D Mark ).















Temperature, Overclocking and Consumption
Under an ambient temperature of 20ºC, the AMD Ryzen 5 2600 offers temperatures of about 28ºC at rest and about 45ºC in its stress test, a really excellent value for such a compact heatsink. This is also not too noisy, around 40 dB at full performance, although easily improved by more advanced models.

On the consumption , say that it has improved notably remaining around 140W for the complete equipment in the stress test of Aida64 (without discs or GPU), a big difference compared to the 200W that the Ryzen 7 2700X has with the same configuration.



Overclocking is still not the strong point of these processors, because although they have increased their frequencies compared to the previous generation, we have only managed to take them up to 4.09 GHz with 1.292V . The Ryzen Master OC application has not been able to use it for not having it updated to give compatibility to these processors, but without problems from the BIOS.

conclusion


The AMD Ryzen 5 2600 supposes a revision of frequencies on the Ryzen 5 1600 , something that improves its performance around 10% . We speak of an improvement of the kind that Intel usually offers us, so the price is the one that will mark whether it compensates or not.

For now, the platform seems somewhat green , although we will have to wait to have X470 motherboards to check their performance in games, which is very important for the lower performance they offered in the previous generation against Intel.

However, it offers a very good overclocking capacity , since it exceeds 4 GHz being a model of the simplest, although it is something to be expected when other models allow it and this only seems a limitation of frequencies.

If the prices that we saw in Amazon Germany last week are true , it will go on sale at around 215 euros , competing against the 8th generation i5 , which also offer 6 cores , although not HyperThreading and therefore "only" 6 threads

Main advantages
+ Good performance / price ratio for professional environments
+ Excellent multi-core performance
+ Very good temperatures and consumption
+ 6 cores / 12 threads
+ Good overclocking capacity, for its factory frequencies
+ Compatible with previous generation base plates
Negative aspects
- The platform is still somewhat green
Other Aspects
* We do not know his performance in games
Therefore, from El Chapuzas Informático we awarded the Gold Award to this processor for its performance in tasks of calculation, rendering and coding, in which the brute force of its 12 threads matters, but the test in 3DMark presages that it does not will have the same luck in games .





Read more https://elchapuzasinformatico.com/2018/04/amd-ryzen-5-2600-review/

 

[Kenmitch]

@Mockingbird

Probably going to date me but whenever I see the CPU packaging sleeve my mind thinks VHS tape for some reason.

I find ryzen + interesting but would much rather see reviews on the new chipsets. Guess we'll have to wait for them till official release.

 

[Gideon]

@Mockingbird

Probably going to date me but whenever I see the CPU packaging sleeve my mind thinks VHS tape for some reason.

I find ryzen + interesting but would much rather see reviews on the new chipsets. Guess we'll have to wait for them till official release.

About the chipsets, there was a (quite credible sounding) rumor that these are exactly the same as B350 and X370. It's just that some requirements for the boards are stricter (e.g. VRM cooling, memory trace length). Rather than allowing some features on some X370 MOBOs and not on other AMD just decided to rebrand the chipsets with those requirements (so that everybody would be at equal footing).

 

[Christopher Bohling]

Their latency still seems wonky, right? Like the timings still aren't properly set up?

 

[.vodka]

Read more https://elchapuzasinformatico.com/2018/04/amd-ryzen-5-2600-review/

This review is worthless. It's more clickbait than anything.

They've had problems with memory, look at their AIDA64 results, they're running DDR4-3200 and these are DDR4-2133/2400 results... They also had issues running games, resulting in instability and BSODs

Not worth publishing but they did anyway. They used an asrock board, their PR BIOSes (v4.xx) are more trouble than it's worth.


If you want coherent and useful results at this time you have to look for testing done on ASUS boards, their BIOS is the most prepared right now for PR without much issues overall... Relative to results on Gigabyte and Asrock boards, it's night and day. In particular the C6H is the most prepared to deal with these new chips right now.


Someone posted that reddit thread where the russians already have the chips and are doing their testing, you can draw early conclusions from there.


In fact, you can expect new results since today ASUS released beta BIOS 0001 for the C6H with PR AGESA 1.0.0.2, that one's new. All testing so far was done on 6001/AGESA 1.0.0.0a

 

[Mockingbird]

This review is worthless. It's more clickbait than anything.

They've had problems with memory, look at their AIDA64 results, they're running DDR4-3200 and these are DDR4-2133/2400 results... They also had issues running games, resulting in instability and BSODs

Not worth publishing but they did anyway. They used an asrock board, their PR BIOSes (v4.xx) are more trouble than it's worth.


If you want coherent and useful results at this time you have to look for testing done on ASUS boards, their BIOS is the most prepared right now for PR without much issues overall... Relative to results on Gigabyte and Asrock boards, it's night and day. In particular the C6H is the most prepared to deal with these new chips right now.


Someone posted that reddit thread where the russians already have the chips and are doing their testing, you can draw early conclusions from there.


In fact, you can expect new results since today ASUS released beta BIOS 0001 for the C6H with PR AGESA 1.0.0.2, that one's new. All testing so far was done on 6001/AGESA 1.0.0.0a

I considered it a "preview" rather than a "review" and it's certainly a lot better than the CanardPC garbage (that tested a Ryzen 7 2700X [105W processor] on an AsRock A320M-HDV, which only supports processors up to 65W).

 

[DooKey]

It's looking like it's worth a buy if you want a Ryzen and don't have a Ryzen, but if you have a decent Ryzen CPU this is a waste of money. I don't have a decent Ryzen 1700X because I can only do 3.8 all core at 1.4v when running DC apps, but will do 4.0 same voltage gaming. If I can get a 2700X that will do 4.3 all core gaming and DCing then I'm onboard.

 

[IRobot23]

You can get lower than 60 for sure but you would need to OC the mesh.

Barely. You said like it is easy to do so.

I haven't seen anyone below 50ns.

it will still suffer poor L3 latency and bandwidth.

 

[Topweasel]

About the chipsets, there was a (quite credible sounding) rumor that these are exactly the same as B350 and X370. It's just that some requirements for the boards are stricter (e.g. VRM cooling, memory trace length). Rather than allowing some features on some X370 MOBOs and not on other AMD just decided to rebrand the chipsets with those requirements (so that everybody would be at equal footing).

Yeah. Honestly I hope people understand that AMD may come out with chipset "updates" just to refresh the product lines as most hardware and specially motherboards sell the best during the launch period and the mobo companies like having these drop in pin compatible upgrades for their cheaper stuff and a reason to design and even more super selection on the highend. But in the end there is really not a whole room for growth without a new socket. What can go into the chipset hasn't changed (no new connectivity options), and everything else is limited by the 3 things AMD can't change without a new socket. Zen 2 will more than likely use the same chipset as well.

Another key reason for the support is people buying new components can be rest assured of one thing, that without updating the BIOS that the Ryzen 2k series will work out of the box with a 400 series chipset.

 

[xblax]

Yeah. Honestly I hope people understand that AMD may come out with chipset "updates" just to refresh the product lines as most hardware and specially motherboards sell the best during the launch period and the mobo companies like having these drop in pin compatible upgrades for their cheaper stuff and a reason to design and even more super selection on the highend. But in the end there is really not a whole room for growth without a new socket. What can go into the chipset hasn't changed (no new connectivity options), and everything else is limited by the 3 things AMD can't change without a new socket. Zen 2 will more than likely use the same chipset as well.

Feature wise the 400 series chipset is obviously the same. But I still hope that it is produced on a newer process:
https://www.anandtech.com/show/1223...led-with-ryzen-apus-zen-on-12nm-vega-on-7nm/8

What we know so far is that the new X470 chipset is designed for lower power. How that is achieved, AMD would not go into at this point, however the fact that the X370 chipset was an outsourced product through ASMedia built on TSMC’s 55nm process would point to at least one way AMD could save power: building it on TSMC’s 40nm process for a start. We are told that the X370 and X470 chipsets are pin-compatible, which is not a limit to this theory, but an interesting point nonetheless.

At one point the 400 series promontory chipset was listed at pci-sig.com:
https://www.techpowerup.com/239987/...on-pci-sig-pcie-3-0-general-purpose-confirmed
I therefore assume that the x470 chipset is not exactly the same die. Hopefully some reviews will investigate further at launch.

Zen 2 might receive an updated chipset with more general purpose PCIe lanes. Pci-sig has listed a 12 lanes, 6 port PCIe 3.0 x4 switch (ASM2812) which could probably be repurposed for a future x570 chipset. AMD can't stick to 8 PCIe 2.0 general purpose lanes until Zen 4 if they wan't to stay competitive with Intel.

 

[jpiniero]

Barely. You said like it is easy to do so.
I haven't seen anyone below 50ns.

https://imgur.com/a/XU6na

Straight Mesh OC to 3200 and DDR4 4000 is 56.6. That seems easily achievable, although DDR4 that fast would obvs be quite pricey. This guy got it under 50 by also setting the tRFC to 300.