HKSpot - Powered by Discuz! Board

標題: [硬件] [AMD] Ryzen ("Zen") / AM4 Platform [打印本頁]

作者: dom 時間: 2016-12-13 08:46 標題: [AMD] Ryzen ("Zen") / AM4 Platform

http://www.amd.com/en-us/innovations/new-horizon
https://www.youtube.com/watch?v=4DEfj2MRLtA
[yt]https://www.youtube.com/watch?v=4DEfj2MRLtA[/yt]
- 8Core 16Threads

- 3.4GHz BASE CLOCE

- 95W TDP

作者: dom 時間: 2017-1-13 10:52

https://www.techpowerup.com/229540/amd-begins-sampling-entry-level-ryzen-chips-4-cores-with-smt-disabled

作者: qcmadness 時間: 2017-1-13 17:24

引用:

原帖由 dom 於 2017-1-13 10:52 發表
 https://www.techpowerup.com/2295 ... s-with-smt-disabled

New die先係新聞

作者: dom 時間: 2017-1-14 01:21

https://forums.anandtech.com/thr ... e-162#post-38678416

AM4 Chipset compare

Majord 以前係 XS 見佢 AMD 版出文

最新係 Rev F3->F4

A 仔寧可花多d 錢respin

[ 本帖最後由 dom 於 2017-1-14 01:22 編輯 ]

作者: dom 時間: 2017-2-1 02:18

https://twitter.com/Dresdenboy
http://imgur.com/gallery/qheFr

ASUS AM4 ROG board confirmed !!

作者: qcmadness 時間: 2017-2-8 23:43

http://www.linleygroup.com/mpr/article.php?id=11666

作者: Puff 時間: 2017-2-10 23:41

Retail Price?
http://www.shopblt.com/search/order_id=%21ORDERID%21&s_max=25&t_all=1&s_all=AMD+AM4&search=Search

source: link from S|A forum

作者: qcmadness 時間: 2017-2-11 14:11

引用:

原帖由 Puff 於 2017-2-10 23:41 發表
Retail Price?
http://www.shopblt.com/search/or ... 4&search=Search

source: link from S|A forum

Get the real price at launch

作者: qcmadness 時間: 2017-2-12 00:54

https://www.techpowerup.com/2305 ... ated-to-xfr-feature

作者: dom 時間: 2017-2-12 04:22

https://twitter.com/Dresdenboy/status/830446602720526336
For QC

作者: qcmadness 時間: 2017-2-12 16:10

引用:

原帖由 dom 於 2017-2-12 04:22 發表
 https://twitter.com/Dresdenboy/status/830446602720526336
For QC

1個CCX係66mm^2
即係Summit Ridge係~240-260mm^2

作者: qcmadness 時間: 2017-2-12 16:13

http://www.xtremesystems.org/for ... es-%28per-Reddit%29

作者: qcmadness 時間: 2017-2-13 21:38

http://www.xtremesystems.org/for ... requencies-Revealed

作者: Sandbo 時間: 2017-2-14 03:16

引用:

原帖由 qcmadness 於 2017-2-13 21:38 發表
 http://www.xtremesystems.org/for ... requencies-Revealed

Really not bad, I think Intel doesn't have any tricks yet.
If the performance is as shown in the leaks, AMD is really turning the table this time.

作者: 符碌 時間: 2017-2-15 16:03

目標1500

作者: qcmadness 時間: 2017-2-19 18:30

https://www.google.com.hk/url?sa ... st=1487585433374628

Est. die size ~ 200mm^2

作者: Puff 時間: 2017-2-20 13:22

引用:

原帖由 qcmadness 於 2017-2-19 18:30 發表
 https://www.google.com.hk/url?sa ... itecture-details.24 ...

More interested in how Zeppelin manages to be form a 4-chip MCM. They have a giant PCIe/SerDes PHY block, but I can't say for sure how many 16-lane links there would be. Link speed is unknown too (8 GT/s? 10 GT/s?). Ideally, one needs five 16-lane links to achieve glueless MCM plus 128 lane PCIe.

[ 本帖最後由 Puff 於 2017-2-20 13:24 編輯 ]

作者: qcmadness 時間: 2017-2-20 22:49

引用:

原帖由 Puff 於 2017-2-20 13:22 發表

More interested in how Zeppelin manages to be form a 4-chip MCM. They have a giant PCIe/SerDes PHY block, but I can't say for sure how many 16-lane links there would be. Link speed is unknown too (8 ...

I don't buy the suggestion of 4-chip MCM

作者: qcmadness 時間: 2017-2-22 22:30

http://www.anandtech.com/show/11 ... y-on-sale-march-2nd
http://techreport.com/news/31471 ... ad-the-ryzen-charge
http://www.tomshardware.com/news ... 0x-1800x,33702.html

作者: Sandbo 時間: 2017-2-23 05:41

引用:

原帖由 qcmadness 於 2017-2-22 22:30 發表
 http://www.anandtech.com/show/11 ... y-on-sale-march-2nd
http://techreport.com/news/31471 ... -chips-lead-the-ryz ...

If I haven't bought 6700k recently, I might be swapping the whole thing to Zen.
However, in the case of 4 threaded games, Kabylake probably still has an edge given the clock difference.

作者: pk1667 時間: 2017-2-24 17:00

https://www.facebook.com/capitalcomputer
AMD RYZEN處理器
YD1700BBAEBOX AMD Ryzen 7 1700｜R7 1700 3.0GHZ/ AM4 / 8核心16線程 / 65W / 不鎖倍頻
預訂價:$2899

YD170XBCAEWOF AMD Ryzen 7 1700X｜R7 1700X 3.4GHZ / AM4 / 8核心16線程 / 95W / 不鎖倍頻
預訂價:$3499

YD180XBCAEWOF AMD Ryzen 7 1800X｜R7 1800X 3.6GHZ/ AM4 / 8核心16線程 / 95W / 不鎖倍頻
預訂價:$4700

作者: kevinyuri 時間: 2017-2-24 23:43

引用:

原帖由 pk1667 於 2017-2-24 05:00 PM 發表
 https://www.facebook.com/capitalcomputer
AMD RYZEN處理器
YD1700BBAEBOX AMD Ryzen 7 1700｜R7 1700 3.0GHZ/ AM4 / 8核心16線程 / 65W / 不鎖倍頻
預訂價:$2899

YD170XBCAEWOF AMD Ryzen 7 1700X｜R7 1700X 3.4GHZ ...

amazon連運費仲平正都幾舊水

作者: q_p 時間: 2017-2-24 23:49 標題: 回覆 22# kevinyuri 的帖子

1700 & 1700X 都貴過amazon?

作者: dom 時間: 2017-2-25 00:20

引用:

原帖由 q_p 於 24-2-2017 23:49 發表
1700 & 1700X 都貴過amazon?

炒價嘛, 重貴過京東/淘寶天貓 AMD 官方店
好正常

作者: qcmadness 時間: 2017-2-26 03:32

http://www.xtremesystems.org/for ... upport-NVIDIA-s-SLI

作者: qcmadness 時間: 2017-2-26 13:51

http://www.tweaktown.com/news/56 ... ps-begin/index.html

引用:

Intel Core i7-6950X ($1599 US) - $300 Price Cut
Intel Core i7-6900K ($999 US) - $200 Price Cut
Intel Core i7-6850K ($549 US) - $150 Price Cut
Intel Core i7-6800K ($359 US) - $140 Price Cut
Intel Core i7-5820K ($319 US) - $100 Price Cut
Intel Core i7-7700K ($299 US) - $80 Price Cut
Intel Core i7-6700K ($259 US) - $140 Price Cut
Intel Core i7-4790K ($279 US) - $90 Price Cut
Intel Core i7-7700 ($289 US) - $50 Price Cut
Intel Core i7-6700 (259 US) - $90 Price Cut
Intel Core i5-7600K ($199 US) - $70 Price Cut
Intel Core i5-6600K ($179 US) - $$90 Price Cut
Intel Core i5-4690K ($189 US) - $70 Price Cut
Intel Core i5-7500 ($189 US) - $30 Price Cut
Intel Core i5-6500 ($179 US) - $50 Price Cut
Intel Core i5-4590 ($159 US) - $60 Price Cut
Intel Core i3-7350K ($159 US) - $20 Price Cut
Intel Core i3-7100 ($114 US) - $15 Price Cut
Intel Core i3-6100 ($109 US) - $20 Price Cut
Intel G4400 ($49.99 US) - $20 Price Cut
Intel G3258 ($49.99 US) - $27 Price Cut

作者: tat0801 時間: 2017-2-26 22:29

引用:

原帖由 qcmadness 於 2017-2-26 13:51 發表
http://www.tweaktown.com/news/56440/intel-rocked-core-over-ryzen-price-drops-begin/index.html

減得咁勁

作者: XT 時間: 2017-2-27 00:32

有燈

作者: 符碌 時間: 2017-2-28 03:45

1500個頻率其實係唔係低過1300?

作者: Sandbo 時間: 2017-2-28 08:01

引用:

原帖由 tat0801 於 2017-2-26 22:29 發表

減得咁勁

Almost like spelling it out how strong Ryzen actually is

作者: tat0801 時間: 2017-2-28 12:30

引用:

原帖由 Sandbo 於 2017-2-28 08:01 發表

Almost like spelling it out how strong Ryzen actually is

一定啦，今次AMD NDA 做得咁好？
INTEL 好似完全無預計過咁
精力去左做baseband想拎生果單

作者: qcmadness 時間: 2017-3-6 23:07

https://www.techpowerup.com/2312 ... mises?cp=2#comments

作者: XT 時間: 2017-3-7 00:54

引用:

原帖由 kevinyuri 於 2017-3-5 02:46 發表
 http://www.evolife.cn/computer/54351.html
呢個作者講解得好仔細,CCX之間latency唔細

Ref. from : https://bbs.hk-spot.com/viewthre ... p;page=2#pid1893812

作者: qcmadness 時間: 2017-3-7 22:27

http://www.tweaktown.com/article ... -preview/index.html

作者: Puff 時間: 2017-3-8 23:39

打算落搭 1700X
iMac 十世都唔出

過兩日落場睇吓有乜板同有冇貨先
唔記得之前 sandro 推果隻 modular 牛牛乜型號添

[ 本帖最後由 Puff 於 2017-3-9 01:11 編輯 ]

作者: Sandbo 時間: 2017-3-9 01:58

引用:

原帖由 Puff 於 2017-3-8 23:39 發表
打算落搭 1700X
iMac 十世都唔出

過兩日落場睇吓有乜板同有冇貨先
唔記得之前 sandro 推果隻 modular 牛牛乜型號添

Why not 1700?
All those three are 99% same, except the X series has so called XFR.
If you overclock them, the highest clock are all within 100 MHz.

作者: Puff 時間: 2017-3-9 03:00

引用:

原帖由 Sandbo 於 2017-3-9 01:58 發表

Why not 1700?
All those three are 99% same, except the X series has so called XFR.
If you overclock them, the highest clock are all within 100 MHz.

Had a game bound by single thread, and not interested in OC either. Thought OCing would disable many of the power mgmt features, wouldn't it?

Edit: It seems the idle states are still there. Hmm.
Edit 2: Eyeing for the MSI B350 boards.

[ 本帖最後由 Puff 於 2017-3-9 03:12 編輯 ]

作者: Sandbo 時間: 2017-3-9 06:49

引用:

原帖由 Puff 於 2017-3-9 03:00 發表

Had a game bound by single thread, and not interested in OC either. Thought OCing would disable many of the power mgmt features, wouldn't it?

Edit: It seems the idle states are still there. ...

https://www.pcper.com/reviews/Pr ... utm_medium=facebook

The modern OC (not really sure about AMD, but for Intel Skylake at least) does not alter the power saving feature;
it will still be the same idle power (given you use an adaptive core voltage instead of a fixed one).

If you refer to my link, 1700/1800X are just the same.......AMD binned the chips by voltage needed, it had certain effects on overclocking ability, but from all I have seen so far, 1700 and 1800X do not show much of a difference on how fast they can be clocked to.
The worst maybe you need to lower the RAM clock to to like 2666 MHz for 1700 to stay at 3.9-4 GHz, but it changes nothing....

I probably will get a 1700, overclock it to 3.8 GHz, but later.

PS: If single thread performance is your major concern, KabyLake is definitely better.
In this case you might want to wait for Skylake-X, which (hopefully) will be having a more reasonable price without losing the single thread stuff.

[ 本帖最後由 Sandbo 於 2017-3-9 06:51 編輯 ]

作者: Sandbo 時間: 2017-3-9 07:42

引用:

原帖由 tat0801 於 2017-2-28 12:30 發表

一定啦，今次AMD NDA 做得咁好？
INTEL 好似完全無預計過咁
精力去左做baseband想拎生果單

They will take back the lead with Skylake-X effortlessly; what they may have to compromise is just the price.

作者: Puff 時間: 2017-3-9 11:55

引用:

原帖由 Sandbo 於 2017-3-9 06:49 發表

 https://www.pcper.com/reviews/Pr ... utm_medium=facebook

The modern OC (not really sure about AMD, but for Intel Skylake at least) do ...

That load power though.

作者: Sandbo 時間: 2017-3-9 12:05

引用:

原帖由 Puff 於 2017-3-9 11:55 發表

That load power though.

This is even less than when I overclock my 6700k to 4.4GHz....... It's no big deal, if your cooler is better; the money you can save from 1700X is worth a very good water cooler.

And we are taking about 33% performance gain here, not the 10% Intel is sparing me.

[ 本帖最後由 Sandbo 於 2017-3-9 13:32 編輯 ]

作者: tat0801 時間: 2017-3-9 12:15

引用:

原帖由 Puff 於 2017-3-8 23:39 發表
打算落搭 1700X
iMac 十世都唔出

過兩日落場睇吓有乜板同有冇貨先
唔記得之前 sandro 推果隻 modular 牛牛乜型號添

iMac 點都拖到四月中尾嫁啦
我睇因RYZEN 出現，令到生果同INTEL 最後第二水供應pricing 未傾好，個價會唔會比而家iMac加係未知數
美國方面都應要加少少價

作者: Puff 時間: 2017-3-9 13:11

引用:

原帖由 Sandbo 於 2017-3-9 12:05 發表

This is even less than when I overclock my 6700k to 4.4GHz....... It's no big deal, if your cooler is better; the money you can save from 1700X is worth a very good water cooler.

And we are taking a ...

Hmm, interesting. That said at this price point KBL does seem very attractive out of the box, esp 7700K doesn't seem too far behind 8-core Ryzen in Linux compilation benchmarks, while I am not really into the places where Ryzen shines (media editing).

Decisions, decisions, decisions.

P.S. Moreover, dom seems getting 1700 soon.

[ 本帖最後由 Puff 於 2017-3-9 13:30 編輯 ]

作者: tat0801 時間: 2017-3-9 18:03

引用:

原帖由 Puff 於 2017-3-9 13:11 發表

Hmm, interesting. That said at this price point KBL does seem very attractive out of the box, esp 7700K doesn't seem too far behind 8-core Ryzen in Linux compilation benchmarks, while I am not really ...

doom effect in action

GGGGGGGGGGGG

作者: Puff 時間: 2017-3-9 18:27

seems like P1/P2 is still available in OC mode, hmm.

作者: Sandbo 時間: 2017-3-10 02:58

引用:

原帖由 Puff 於 2017-3-8 23:39 發表
打算落搭 1700X
iMac 十世都唔出

過兩日落場睇吓有乜板同有冇貨先
唔記得之前 sandro 推果隻 modular 牛牛乜型號添

If you were referring to me, I always suggest Corsair RMx series (make sure you don't buy the old RM series).
I think they have the best performance given the price (not efficiency wise, but ripple performance).
650W: https://www.techpowerup.com/reviews/Corsair/RM650i/10.html
1000W: https://www.techpowerup.com/reviews/Corsair/RM1000x/9.html
*RMx and RMi use the same components and design, except for RMi can be connected to the PC for doing some monitoring.

But if you aim higher, now we have the best from Seasonic (and if I am right, they made their warranty 12 years long......while Corsair's still has 10)
https://www.techpowerup.com/reviews/Seasonic/Prime_750/6.html

[ 本帖最後由 Sandbo 於 2017-3-10 03:07 編輯 ]

作者: Puff 時間: 2017-3-10 03:02

引用:

原帖由 Sandbo 於 2017-3-10 02:58 發表

If you were referring to me, I always suggest Corsair RMx series (make sure you don't buy the old RM series).
I think they have the best performance given the price (not efficiency wise, but ripple p ...

Yep, sorry for misremembering your username.
Leaning towards OC 1700 with B350, perhaps 3.7-3.8 GHz all cores. Just that the idea of burning more power when browsing web pages kinda bugs me. Not sure if idle downclocking still works in OC mode.

[ 本帖最後由 Puff 於 2017-3-10 03:12 編輯 ]

作者: Sandbo 時間: 2017-3-10 08:18

引用:

原帖由 Puff 於 2017-3-10 03:02 發表

Yep, sorry for misremembering your username.
Leaning towards OC 1700 with B350, perhaps 3.7-3.8 GHz all cores. Just that the idea of burning more power when browsing web pages kinda bugs me. Not sure ...

A quick search gave me this:
https://www.bit-tech.net/hardwar ... zen-7-1700-review/6
On overclocking, the idle power does go up a bit, but it is just on par with 7700k's idle......far from being a worry.
While it seems every 1700 you can buy now overclock pretty well, there is no guarantee yours is going to be the same, just bare that in mind

*And I stand corrected - the max load power is higher than my 6700k overclocked to 4.4 GHz, but not by far.
With this it means you really would like a water cooler (some Corsair H80 or above) to proceed with clocking it at 4 GHz.

[ 本帖最後由 Sandbo 於 2017-3-10 08:22 編輯 ]

作者: Sandbo 時間: 2017-3-10 11:28

So it seems those available AM4 boards are already supporting ECC RAM to a certain extent:
https://www.reddit.com/r/Amd/com ... _and_other/def6vs2/

After some confirmation later I will probably upgrade my first gen Xeon E3-1235 to Ryzen 1700.

[ 本帖最後由 Sandbo 於 2017-3-10 11:32 編輯 ]

作者: Puff 時間: 2017-3-10 13:01

引用:

原帖由 Sandbo 於 2017-3-10 08:18 發表

A quick search gave me this:
https://www.bit-tech.net/hardware/2017/03/08/amd-ryzen-7-1700-review/6
On overclocking, the idle power does go up a bit, but it is just on par with 7700k's idle......far ...

Probably not 4Ghz, since I heard hitting 3.9+ GHz needs a high-end board. I'd like somewhere around 3.7 and tune the voltage for perf-watt.
Saw some nice results in power consumption.

[ 本帖最後由 Puff 於 2017-3-10 16:08 編輯 ]

作者: Puff 時間: 2017-3-10 19:51

https://forum.beyond3d.com/posts/1970348/
DF: 32B bi-directional @ memclk

[ 本帖最後由 Puff 於 2017-3-10 20:06 編輯 ]

作者: Sandbo 時間: 2017-3-11 01:04

引用:

原帖由 Puff 於 2017-3-10 19:51 發表
 https://forum.beyond3d.com/posts/1970348/
DF: 32B bi-directional @ memclk

Not knowing much about this, but there is a thread about the interconnect being the bottleneck for some cases:
https://www.reddit.com/r/Amd/com ... g_like_a_dual_4c8t/
It might not be a serious issue given the task is distributed smartly by the OS.

作者: Puff 時間: 2017-3-11 02:10

引用:

原帖由 Sandbo 於 2017-3-11 01:04 發表

Not knowing much about this, but there is a thread about the interconnect being the bottleneck for some cases:
https://www.reddit.com/r/Amd/com ... ving_like_a_dual_4c ...

It is not really a "design flaw" but apparently an intentional design choice (no system LLC). Yes, it is a bottleneck, but only for implementations that assume a giant shared LLC for all cores, and ping-pong the working set between cores on top of the assumption. Otherwise, it is no different from any UMA chips without a shared LLC running at core clock.

作者: qcmadness 時間: 2017-3-11 15:59

引用:

原帖由 Puff 於 2017-3-11 02:10 發表

It is not really a "design flaw" but apparently an intentional design choice (no system LLC). Yes, it is a bottleneck, but only for implementations that assume a giant shared LLC for all cores, and p ...

The OS needs to be NUMA-aware to fully support Ryzen

作者: Puff 時間: 2017-3-11 16:54

引用:

原帖由 qcmadness 於 2017-3-11 15:59 發表

The OS needs to be NUMA-aware to fully support Ryzen

It needs to be aware of the cache topology, not NUMA.

作者: qcmadness 時間: 2017-3-11 17:13

引用:

原帖由 Puff 於 2017-3-11 16:54 發表

It needs to be aware of the cache topology, not NUMA.

Both

作者: Puff 時間: 2017-3-11 18:09

引用:

原帖由 qcmadness 於 2017-3-11 17:13 發表

Both

The chip is UMA, and doesn't statically partition the physical memory address space betweeb CCXs. So essentially apps just need to be aware of the topology, e.g. porting console CPU scheduling code to the PC. XB1 and PS4 have essentially the same situation.

NUMA awareness isn't a must. Let alone the fact that the chip is not advertising itself as two NUMA nodes, and thus NUMA aware apps relying on OS API would not treat it as two.

[ 本帖最後由 Puff 於 2017-3-11 18:21 編輯 ]

作者: Sandbo 時間: 2017-3-11 23:11

引用:

原帖由 qcmadness 於 2017-3-11 15:59 發表

The OS needs to be NUMA-aware to fully support Ryzen

It might end up locking everything to a single CCX and uses only four core that case, this isn't ideal.

There is a new article from PCProspective :
https://www.pcper.com/reviews/Pr ... utm_medium=facebook

They tested and claimed the Windows 10 scheduler wasn't to blame for swapping task among core, but I don't think it's very conclusive; people consistently found Windows 7 to perform better with Ryzen, there might be something else behind the OS.

[ 本帖最後由 Sandbo 於 2017-3-12 09:26 編輯 ]

作者: qcmadness 時間: 2017-3-11 23:35

引用:

原帖由 Sandbo 於 2017-3-11 23:11 發表

It might end up locking everything to a single CCX and uses only four core that case, this isn't ideal.

There is a new article from PCProspective :
https://www.pcper.com/reviews/Processors/AMD-Ryzen ...

There are other claims that Windows 10 misreports the cache size for Ryzen CPUs

作者: qcmadness 時間: 2017-3-11 23:39

引用:

原帖由 Puff 於 2017-3-11 18:09 發表

The chip is UMA, and doesn't statically partition the physical memory address space betweeb CCXs. So essentially apps just need to be aware of the topology, e.g. porting console CPU scheduling code t ...

Cat. it as NUMA simplifies the problem.
But don't treat NUMA is an advertisement as NUMA reduces efficiency and possibly performance.

作者: Sandbo 時間: 2017-3-12 12:58

引用:

原帖由 qcmadness 於 2017-3-11 23:35 發表

There are other claims that Windows 10 misreports the cache size for Ryzen CPUs

https://www.guru3d.com/news-stor ... back-amd-ryzen.html

作者: qcmadness 時間: 2017-3-12 13:25

http://www.xtremesystems.org/for ... eople-to-buy-Nvidia

第2日就走去接UAE d AMD stock

作者: dom 時間: 2017-3-12 15:31

引用:

原帖由 qcmadness 於 12-3-2017 13:25 發表
 http://www.xtremesystems.org/for ... eople-to-buy-Nvidia

第2日就走去接UAE d AMD stock

LOL

作者: Puff 時間: 2017-3-12 17:16

Things do seem kinda messy. The only thing which has correct readings is Ryzen Master. Even Windows reports a different operating frequency, and common tools like CoreTemp do not work at all. SpeedFan reports my fan spinning at 10630 RPM.

Burning 3.7 GHz @ 1.2V (stock: 1.1875V) with Prime95. The temp so far lies within 55-58C. I use Noctua NH-U12S, FYI.

Edit: 1.2V crashed after 5-10 minutes. Trying 1.225V now.

Edit: 3.7Ghz @1.215V is pretty stable. 3.8Ghz @1.25V seem to be stable, but somehow Firefox crashed.　3.9 GHz @1.275V crashed CB15 immediately. Gonna stick with 3.7.

[ 本帖最後由 Puff 於 2017-3-12 22:58 編輯 ]

作者: Sandbo 時間: 2017-3-12 23:33

引用:

原帖由 Puff 於 2017-3-12 17:16 發表
Things do seem kinda messy. The only thing which has correct readings is Ryzen Master. Even Windows reports a different operating frequency, and common tools like CoreTemp do not work at all. SpeedFan ...

Not really sure about the voltage, but from what I heard Ryzen tends to use a lot higher than the numbers you gave(PCPerspective applied 1.4V, and it went up to 7X degree C), however you may need a better cooler and mainboard to go any further.

But 3.7 GHz stable with only 55 C and just Noctua U12S is pretty good.
(My 6700k only got 4.4 GHz with 1.32 V(typical number for Skylake, even with NH-C14 and industrial 140mm 3000rpm, it can reach 82C with Prime95, though under normal load it's below 70C)

[ 本帖最後由 Sandbo 於 2017-3-12 23:35 編輯 ]

作者: Puff 時間: 2017-3-12 23:43

引用:

原帖由 Sandbo 於 2017-3-12 23:33 發表

Not really sure about the voltage, but from what I heard Ryzen tends to use a lot higher than the numbers you gave(PCPerspective applied 1.4V, and it went up to 7X degree C), however you may need a b ...

The F/vt curve is significantly steeper beyond ~3.7 GHz as experienced by many.

PCPer targeted 4.0 GHz for the 1700. 3.9-4.1 GHz requiring at least 1.35V is kinda expected (mentioned by AMD too), especially on 1700 which is presumably binned with parts performing worse at higher frequencies.

So far I didn't see any OC reviews on undervolting, except one who clocked 1700 at the same clock as mine (3.7). But I forgot which reviewer it is from, and the voltage. Sigh.

Edit: Here is a thread from HardOCP. https://hardforum.com/threads/ry ... ng-tidbits.1926296/

[ 本帖最後由 Puff 於 2017-3-13 00:07 編輯 ]

作者: Sandbo 時間: 2017-3-13 00:27

引用:

原帖由 Puff 於 2017-3-12 23:43 發表

The F/vt curve is significantly steeper beyond ~3.7 GHz as experienced by many.

PCPer targeted 4.0 GHz for the 1700. 3.9-4.1 GHz requiring at least 1.35V is kinda expected (mentioned by AMD too), es ...

Looks like 3.7 GHz is the sweet spot for 1700, at least for this batch.
With almost the same performance for $200 USD cheaper than 1800X, it kind of renders 1700X/1800X meaningless to get given how low cost and easy the overclocking can be achieved.

作者: Puff 時間: 2017-3-13 01:34

引用:

原帖由 Sandbo 於 2017-3-13 00:27 發表

Looks like 3.7 GHz is the sweet spot for 1700, at least for this batch.
With almost the same performance for $200 USD cheaper than 1800X, it kind of renders 1700X/1800X meaningless to get given how l ...

Yep, kinda glad I didn't get 1700X, esp. after verifying that the power saving states still being functional (7 cores idling at 1.x Ghz) when I am just redditing.

[ 本帖最後由 Puff 於 2017-3-13 01:37 編輯 ]

作者: BlackBird 時間: 2017-3-13 16:49

引用:

原帖由 Puff 於 2017-3-12 23:43 發表

The F/vt curve is significantly steeper beyond ~3.7 GHz as experienced by many.

PCPer targeted 4.0 GHz for the 1700. 3.9-4.1 GHz requiring at least 1.35V is kinda expected (mentioned by AMD too), es ...

同Deneb差不多

作者: qcmadness 時間: 2017-3-13 19:17

引用:

原帖由 BlackBird 於 2017-3-13 16:49 發表

同Deneb差不多

Deneb低頻好多

作者: dom 時間: 2017-3-13 22:03

3.7GHz for 8C16T pretty good

作者: Puff 時間: 2017-3-14 08:30

If you gonna build soon and OC, be aware of the fact that not all BIOS currently offers options to edit P-states. For example, ASUS C6H offers it, but the ASUS B350 board offer just a constant FID, DID and Vcore. IOW the core would not drop to lower clocks when idling.

OC done in Ryzen Master does not suffer from the issue, but it doesn't seem to persist after sleep or reboot.

[ 本帖最後由 Puff 於 2017-3-14 08:33 編輯 ]

作者: BlackBird 時間: 2017-3-14 09:22

引用:

原帖由 qcmadness 於 2017-3-13 19:17 發表

Deneb低頻好多

Frequency scaling 都差不多係3.7後就要瘋狂上電4G就玩完

作者: qcmadness 時間: 2017-3-14 09:26

引用:

原帖由 BlackBird 於 2017-3-14 09:22 發表

Frequency scaling 都差不多係3.7後就要瘋狂上電4G就玩完

3.7要好後期的Deneb

作者: Puff 時間: 2017-3-14 23:34

引用:

原帖由 qcmadness 於 2017-3-14 09:26 發表

3.7要好後期的Deneb

P2 980 嘛
我記得喺 Istanbul 出先到佢.

作者: dom 時間: 2017-3-15 00:55

https://community.amd.com/commun ... system?sf62307686=1

So AMD pust this link on their Facebook explain some finding of their own test

引用:

Finally, as part of AMDs ongoing development of the new AM4 platform, AMD will increase support for overclocked memory configurations with higher memory multipliers. We intend to issue updates to motherboard partners in May that will enable them, on whatever products they choose, to support speeds higher than the current DDR4-3200 limit without refclk adjustments.

Damn it .. now DDR4-3000 is not officially supported , I should have bought
DDR4-3200 when it was on sales :(

The blog also suggested use High Performance Power Profile with Windows 10 for optimal performance

[ 本帖最後由 dom 於 2017-3-15 00:56 編輯 ]

作者: Puff 時間: 2017-3-15 08:26

引用:

原帖由 dom 於 2017-3-15 00:55 發表
 https://community.amd.com/commun ... system?sf62307686=1

So AMD pust this link on their Facebook explain some finding of their own test ...

Gonna stick with 2400 for a while.

作者: Sandbo 時間: 2017-3-15 08:43

引用:

原帖由 dom 於 2017-3-15 00:55 發表
https://community.amd.com/community/gaming/blog/2017/03/14/tips-for-building-a-better-amd-ryzen-system?sf62307686=1

So AMD pust this link on their Facebook explain some finding of their own test ...

It doesn't make a practical difference beyond 2666 MHz

作者: Puff 時間: 2017-3-21 01:35

My Ryzen tower is nice. But the attempt to dump macOS (?) severely ruined my daily workflow. For god's sake I can't even find a client which supports CalDAV and CardDAV properly while not sucking at UX and UI.

Hope that we will see P-state editing software soon (which auto-applies settings at launch/wake) from the community, though probably it will take some time since AMD hasn't even released a single developer manual for 17h.

[ 本帖最後由 Puff 於 2017-3-21 01:39 編輯 ]

作者: Sandbo 時間: 2017-3-21 10:21

Extremely insightful and detailed analysis on Ryzen's gaming performance:
https://www.youtube.com/watch?v=TId-OrXWuOE

In short, something is very funny with the core utilization;
Disabling 2 cores and all those CPU hungry games lost performance only marginally.

作者: Puff 時間: 2017-3-21 15:11

引用:

原帖由 Sandbo 於 2017-3-21 10:21 發表
Extremely insightful and detailed analysis on Ryzen's gaming performance:
https://www.youtube.com/watch?v=TId-OrXWuOE

~11:05 is a big "told ya" moment for me. People around the Internet really enjoys the "inter-CCX is flawed" bandwagon while (understandably) not looking into the details. Could I blame clickbait journalism?

引用:

原帖由 Sandbo 於 2017-3-21 10:21 發表
In short, something is very funny with the core utilization;
Disabling 2 cores and all those CPU hungry games lost performance only marginally.

Got an exception for you: Guild Wars 2. Setting core affinity to half of the cores yields a considerable FPS drop (up to 50%), despite the game being known for CPU taxing and not scaling well (like any MMO).

[ 本帖最後由 Puff 於 2017-3-21 15:23 編輯 ]

作者: Puff 時間: 2017-3-26 20:44

Got stuck on 2667 with my 2400 2x 8GB kit + 1700 + ASUS B350M-A + 0509 BIOS even with way looser timings and way higher VDDQ/VTT. Anyway ~95% of stock 1800X (1600-1650) in Cinebench R15 is pretty impressive.

Edit: It seems that locking at full speed doesn't hurt the idle power consumption at all...

[ 本帖最後由 Puff 於 2017-3-26 22:54 編輯 ]

作者: Sandbo 時間: 2017-3-27 10:37

引用:

原帖由 Puff 於 2017-3-26 20:44 發表
Got stuck on 2667 with my 2400 2x 8GB kit + 1700 + ASUS B350M-A + 0509 BIOS even with way looser timings and way higher VDDQ/VTT. Anyway ~95% of stock 1800X (1600-1650) in Cinebench R15 is pretty impr ...

From what I know 2666 MHz for ram is the maximum unless you use ASUS's Crosshair;
this is a limitation of clock generation of the CPU or something, where ASUS's board has an independent clock generator for it.
I can be wrong though. Anyway 2666 MHz is very good already, no observable gaming difference can be seen beyond it baiscally, instead you may play with the CL timings.

作者: Sandbo 時間: 2017-3-27 11:55

Just another article justifying your choice:
http://www.eurogamer.net/article ... 00x-vs-1800x-review

[ 本帖最後由 Sandbo 於 2017-3-27 11:56 編輯 ]

作者: Puff 時間: 2017-3-27 13:50

引用:

原帖由 Sandbo 於 2017-3-27 10:37 發表

From what I know 2666 MHz for ram is the maximum unless you use ASUS's Crosshair;
this is a limitation of clock generation of the CPU or something, where ASUS's board has an independent clock generat ...

IIRC External CLKGEN is for >3200 MHz, since AMD doesn't offer multipliers higher than 32x.

作者: cheungmanhoi 時間: 2017-3-29 10:38

low b 問題
其實係咪用唔番am3果d heatsink

作者: Sandbo 時間: 2017-3-29 10:55

引用:

原帖由 cheungmanhoi 於 2017-3-29 10:38 發表
low b 問題
其實係咪用唔番am3果d heatsink

Many of the old ones can be used if a new kit is provided (Corsair and Noctua), check with the manufacturers.

[ 本帖最後由 Sandbo 於 2017-3-29 11:22 編輯 ]

歡迎光臨 HKSpot (https://bbs.hk-spot.com/)