[TechPowerUp] [Updated] Der8auer: Only Small Percentage of 3rd Gen Ryzen CPUs Hit Their Advertised Speeds - Page 2 - Overclock.net - An Overclocking Community

Forum Jump: 

[TechPowerUp] [Updated] Der8auer: Only Small Percentage of 3rd Gen Ryzen CPUs Hit Their Advertised Speeds

Reply
 
Thread Tools
post #11 of 88 (permalink) Old 09-02-2019, 10:29 AM
New to Overclock.net
 
The Stilt's Avatar
 
Join Date: May 2013
Posts: 2,426
Rep: 834 (Unique: 318)
Quote: Originally Posted by ozlay View Post
Its just an AGESA issue. The earlier AGESA allowed higher boosts. It was buggy but it boosted higher.
Not really.
The new behavior, which in many cases results in reduced boost frequencies is caused by the changes made to the system management unit (SMU) firmware.
These changes were first introduced in a firmware released on 6/21/2019 (version 46.37.0) and have been in place ever since.

All of the four official review boards (ASRock, ASUS, GIGABYTE and MSI) were provided with a NPRP (AMD validated & recommended for media) bios, which was based on AGESA 1.0.0.2CA version.
1.0.0.2CA AGESA specifically includes the 46.37.0 SMU version, which features the new boost behavior. Despite the fact that there are 1.0.0.3x AGESA versions, which use an older SMU FW version.

Based on that, the consumers were never supposed to see the old behavior.
The Stilt is offline  
Sponsored Links
Advertisement
 
post #12 of 88 (permalink) Old 09-02-2019, 11:02 AM
 
The Robot's Avatar
 
Join Date: Mar 2013
Posts: 2,293
Rep: 128 (Unique: 81)
I agree, AMD shouldn't advertise boost speeds that are never achieved out of the box and provide clear technical explanation to enthusiasts why they reduced the boost on new firmware.

Main
(17 items)
Nintendo DS
(8 items)
CPU
6700K
Motherboard
Gigabyte Z170X-Gaming 3
GPU
MSI GTX 1080 Gaming X
RAM
G.Skill Ripjaws V 16GB 3000
Hard Drive
Samsung 850 Evo 500GB
Hard Drive
WD Blue 3TB
Power Supply
EVGA 650 G2
Cooling
Noctua NH-D15S
Cooling
Nanoxia Deep Silence 140mm
Cooling
Nanoxia Deep Silence 120mm
Case
Corsair 400Q
Operating System
Windows 10 Enterprise
Monitor
ViewSonic XG2703-GS 1440p
Keyboard
Leopold FC750 (MX Brown)
Mouse
Logitech Performance Mouse MX
Audio
Mayflower Objective2 + ODAC Rev. B Combo
Audio
Audio-Technica ATH-A990Z
CPU
ARM946E-S 67.028 MHz
CPU
ARM7TDMI 33.514 MHz
RAM
4 MB
Hard Drive
256 kB
Power Supply
850 mAh
Operating System
DS OS
Monitor
3" 256×192 18-bit
Monitor
3" 256×192 18-bit
▲ hide details ▲
The Robot is offline  
post #13 of 88 (permalink) Old 09-02-2019, 11:45 AM
Waiting for 7nm EUV
 
tpi2007's Avatar
 
Join Date: Nov 2010
Posts: 11,307
Rep: 893 (Unique: 502)
Quote: Originally Posted by The Stilt View Post
Not really.
The new behavior, which in many cases results in reduced boost frequencies is caused by the changes made to the system management unit (SMU) firmware.
These changes were first introduced in a firmware released on 6/21/2019 (version 46.37.0) and have been in place ever since.

All of the four official review boards (ASRock, ASUS, GIGABYTE and MSI) were provided with a NPRP (AMD validated & recommended for media) bios, which was based on AGESA 1.0.0.2CA version.
1.0.0.2CA AGESA specifically includes the 46.37.0 SMU version, which features the new boost behavior. Despite the fact that there are 1.0.0.3x AGESA versions, which use an older SMU FW version.

Based on that, the consumers were never supposed to see the old behavior.

If I'm reading the timeline correctly, then we should have never seen the slides that advertised the boost clocks that we all know, but by June 21st it was too late for that as we had already seen the specs on slides on multiple occasions, the latest on June 10 at E3, and now AMD may have a serious problem on their hands. Why didn't they take action when they realized the boost would be lower? Bad PR? They thought they could make a workaround in the meantime?

They still have time to correct the specs for the 3950X and that mighty 4.7 Ghz boost clock before they release it. It seems like a far away dream that it will hit 4.7 Ghz on even half a core.

Is your CPU bottlenecking your GPU ? Find out: CPU and GPU usage along with FPS in-game
Read my reviews here.
Clubs (founder): The rare / unusual CPU club
Clubs (member): Corsair Professional HX / AX Series PSU Owners Club || The Official Cooler Master HAF X/932/922/912(+) Club
CPU
Core i7-3820
Motherboard
Asus Sabertooth X79
GPU
MSI GTX 1060 6 GB Gaming X
RAM
16 GB Corsair DDR3 1866 Mhz Dominator
Hard Drive
Samsung SSD 830 128GB + WD Caviar Black 2TB
Optical Drive
Sony Optiarc DVD-RW
Power Supply
Corsair AX750 Professional Modular 80 Plus Gold
Cooling
Corsair A70 + Noiseblocker M12-P
Case
Cooler Master HAF 912 Plus
Operating System
Windows 7 Home Premium 64-bit
Monitor
BenQ RL2455HM
Keyboard
Cooler Master Octane
Mouse
Cooler Master Octane
▲ hide details ▲


tpi2007 is offline  
Sponsored Links
Advertisement
 
post #14 of 88 (permalink) Old 09-02-2019, 12:48 PM
New to Overclock.net
 
Imouto's Avatar
 
Join Date: Mar 2012
Posts: 1,942
Rep: 219 (Unique: 102)
Quote: Originally Posted by tpi2007 View Post
and now AMD may have a serious problem on their hands.
By missing the target single core boost by 2~3%?

You should read out loud what you type because it is honestly hilarious.

#EnthusiastLivesMatter
Imouto is offline  
post #15 of 88 (permalink) Old 09-02-2019, 12:51 PM
9 Cans of Ravioli
 
The Pook's Avatar
 
Join Date: May 2006
Posts: 16,543
Rep: 1067 (Unique: 629)
early AGESA allowed advertised boost clocks, later AGESA doesn't ... so blame AMD? don't really follow.

My 9900K is advertised as a "5.0 boost" CPU but nothing ever will let it boost to 5.0 since nothing loads one core only. The only time it hits 5.0 on any core is for milliseconds while basically idle.

Intel i9 9900K Coffee Lake @ X8 5309mhz (+47.5%)
AMD FX-8320 Vishera @ X4 5022mhz (+43.5%)
Intel i5 4690K Haswell @ X4 5013mhz (+43.2%)
AMD X4 960T Zosma @ X6 4870mhz (+62.3%)
Intel i7 6700 Skylake @ X4 4709mhz (+38.5%)
Intel i5 6400 Skylake @ X4 4588mhz (+69.9%)
2 OP pls nerf
(24 items)
NUC U
(10 items)
CPU
i9 9900K
Motherboard
AsRock Z390 Taichi
GPU
MSI GTX 1080 Ti
RAM
G.Skill DDR4 4133 32GB
Hard Drive
Mushkin Pilot NVMe 1TB
Hard Drive
HP EX950 NVMe 2TB
Hard Drive
Samsung 860 Evo 1TB
Hard Drive
Sandisk Ultra II 480GB SSD
Hard Drive
Toshiba X300 5TB
Hard Drive
Western Digital EZRZ 3TB
Power Supply
Seasonic Focus Plus Gold 750w
Cooling
Corsair H115i Pro
Case
Fractal Design Meshify S2
Operating System
Windows 10 Education x64
Monitor
Acer XF270HU
Monitor
Dell E2311H
Keyboard
CoolerMaster MasterKeys Pro S
Mouse
Logitech G502
Audio
SMSL SA50 + FiiO E10K
Audio
Micca RB42
Audio
Polk Audio PSW111
Audio
Philips Fidelio X2
Audio
Philips SHP9500S
Audio
Audio Technica ATH-M50X
CPU
i3 8109U
Motherboard
Intel NUC8i3BEH
GPU
Iris Plus 655
RAM
Kingston 8GB DDR4 2400
Hard Drive
Dierka K1 120GB
Hard Drive
ADATA SU800 128GB M.2
Operating System
Windows 10 Professional
Operating System
MX Linux
Keyboard
Rii RKM709 Wireless
Mouse
Logitech M510
CPU
i7 860
Motherboard
Advantech AIMB 280 ITX
GPU
EVGA GTX 950 2GB
RAM
Kingston 2GB DDR3 1333
Hard Drive
Seagate Momentus 2.5" 500GB
Hard Drive
Western Digital Blue 2.5" 320GB
Power Supply
Silverstone SFX-L Gold 500w
Cooling
Arctic Alpine 11 GT 2
Case
Silverstone Milo Z ML07B
Operating System
Windows XP x86 SP3
Keyboard
Velocifire TKL01 Outemu Browns
Mouse
Logitech M512
▲ hide details ▲


The Pook is online now  
post #16 of 88 (permalink) Old 09-02-2019, 01:10 PM
✾ ✿ ❀ ❁
 
Alex132's Avatar
 
Join Date: Dec 2009
Posts: 8,342
Rep: 342 (Unique: 274)
Quote: Originally Posted by The Pook View Post
early AGESA allowed advertised boost clocks, later AGESA doesn't ... so blame AMD? don't really follow.

My 9900K is advertised as a "5.0 boost" CPU but nothing ever will let it boost to 5.0 since nothing loads one core only. The only time it hits 5.0 on any core is for milliseconds while basically idle.
I wouldn't trust task manager for frequency, especially when trying to see it on a per-core basis. CPU-Z is better to see per-core clocks for that kind of thing.



I think people forget what AGESA is:
Quote:
AMD Generic Encapsulated Software Architecture, is a procedure library developed by Advanced Micro Devices, used to perform the Platform Initialization on mainboards using their AMD64
If Intel pushed a BIOS update to the motherboard vendors which disallowed all but <15% of 9900Ks to boost to advertised speeds I'm sure there would be a big outcry.

| This Cannot Continue | We Are Become As Gods | This Cannot Continue |

Vehicles:
Current: None.
Ex: '07 Fiat Palio, '07 Honda Accord Type-S
Illya
(26 items)
Yuki
(23 items)
CPU
Intel i9 9900K @ 5.2Ghz
Motherboard
Gigabyte Aorus Master
GPU
EVGA 1080 Ti FTW3
RAM
G.Skill Trident Z 32GB 4000Mhz
Hard Drive
Samsung 970 Pro 512GB
Hard Drive
Samsung 850 Evo 1TB
Hard Drive
WD Red 8TB
Power Supply
Seasonic Prime Titanium 850W
Cooling
EK Velocity CPU Block
Cooling
EK 1080 Ti FTW3 GPU Block
Cooling
HardwareLabs GTX420
Cooling
HardwareLabs GTS280
Cooling
HardwareLabs GTS140
Cooling
EK D5 140 Glass Pump/res
Case
Phanteks Evolv X
Operating System
Windows 10 Pro
Monitor
Asus ROG Swift PG279Q 165Hz 1440p
Monitor
I-INC 1920x1200 TN 27"
Keyboard
Ducky One 2 Midnight
Mouse
Logitech G502
Mousepad
CoolerMaster Swift-RX XL
Audio
Sennhesier HD650
Audio
Schiit Jotunheim Amp/DAC
Audio
Samson C01U Microphone
Audio
Edifier R1700BT
Other
Oculus Rift
CPU
Intel 2500K 5Ghz
Motherboard
ASUS P8P67 Pro
GPU
EVGA 1080 Ti FTW3
RAM
G.Skill RipJaws X 2133Mhz 16GB
Hard Drive
120GB Corsair Neutron GTX
Hard Drive
Samsung 850 Evo 1TB
Hard Drive
WD Red 8TB
Power Supply
EVGA G3 850W
Cooling
Hyper 212
Cooling
EK Vardar fans
Case
Corsair 270R w/TG mod
Operating System
Windows 10
Monitor
Asus ROG Swift PG279Q
Monitor
I-Inc 1200p
Keyboard
Razer BlackWidow 2013
Mouse
Logitech G502 Proteus Spectrum
Mousepad
CoolerMaster Swift-RX XL
Audio
Sennheiser HD650
Audio
Schiit Jotunheim Balanced DAC + Amp
Audio
Edifier R1700BT Speakers
Audio
Samson C01U Microphone
Other
Oculus Rift
Other
PS4 Controller
▲ hide details ▲


Alex132 is online now  
post #17 of 88 (permalink) Old 09-02-2019, 01:13 PM
9 Cans of Ravioli
 
The Pook's Avatar
 
Join Date: May 2006
Posts: 16,543
Rep: 1067 (Unique: 629)
Quote: Originally Posted by Alex132 View Post
I wouldn't trust task manager for frequency, especially when trying to see it on a per-core basis. CPU-Z is better to see per-core clocks for that kind of thing.

I'm not using task manager to view CPU frequency.

Intel i9 9900K Coffee Lake @ X8 5309mhz (+47.5%)
AMD FX-8320 Vishera @ X4 5022mhz (+43.5%)
Intel i5 4690K Haswell @ X4 5013mhz (+43.2%)
AMD X4 960T Zosma @ X6 4870mhz (+62.3%)
Intel i7 6700 Skylake @ X4 4709mhz (+38.5%)
Intel i5 6400 Skylake @ X4 4588mhz (+69.9%)
2 OP pls nerf
(24 items)
NUC U
(10 items)
CPU
i9 9900K
Motherboard
AsRock Z390 Taichi
GPU
MSI GTX 1080 Ti
RAM
G.Skill DDR4 4133 32GB
Hard Drive
Mushkin Pilot NVMe 1TB
Hard Drive
HP EX950 NVMe 2TB
Hard Drive
Samsung 860 Evo 1TB
Hard Drive
Sandisk Ultra II 480GB SSD
Hard Drive
Toshiba X300 5TB
Hard Drive
Western Digital EZRZ 3TB
Power Supply
Seasonic Focus Plus Gold 750w
Cooling
Corsair H115i Pro
Case
Fractal Design Meshify S2
Operating System
Windows 10 Education x64
Monitor
Acer XF270HU
Monitor
Dell E2311H
Keyboard
CoolerMaster MasterKeys Pro S
Mouse
Logitech G502
Audio
SMSL SA50 + FiiO E10K
Audio
Micca RB42
Audio
Polk Audio PSW111
Audio
Philips Fidelio X2
Audio
Philips SHP9500S
Audio
Audio Technica ATH-M50X
CPU
i3 8109U
Motherboard
Intel NUC8i3BEH
GPU
Iris Plus 655
RAM
Kingston 8GB DDR4 2400
Hard Drive
Dierka K1 120GB
Hard Drive
ADATA SU800 128GB M.2
Operating System
Windows 10 Professional
Operating System
MX Linux
Keyboard
Rii RKM709 Wireless
Mouse
Logitech M510
CPU
i7 860
Motherboard
Advantech AIMB 280 ITX
GPU
EVGA GTX 950 2GB
RAM
Kingston 2GB DDR3 1333
Hard Drive
Seagate Momentus 2.5" 500GB
Hard Drive
Western Digital Blue 2.5" 320GB
Power Supply
Silverstone SFX-L Gold 500w
Cooling
Arctic Alpine 11 GT 2
Case
Silverstone Milo Z ML07B
Operating System
Windows XP x86 SP3
Keyboard
Velocifire TKL01 Outemu Browns
Mouse
Logitech M512
▲ hide details ▲


The Pook is online now  
post #18 of 88 (permalink) Old 09-02-2019, 01:54 PM
Overclocker
 
JackCY's Avatar
 
Join Date: Jun 2014
Posts: 9,898
Rep: 337 (Unique: 240)
Well using software tools to detect peak frequencies of a CPU is silly. The CPU switches clocks way faster than one will be able to detect via software polling.
As far as I understand it the clocks on Ryzen are max stock clocks, not average clocks. So if a CPU runs <5% average of it's max, that's not that bad. The clocks are opportunistic and dynamic. It's the same as modern GPUs, the clocks are no longer fixed. But on CPUs people are used to the old slow clocks from Intel CPUs and OC locking their clocks too. Where as on more modern CPUs the clocks don't get OC locked and out of the box it auto clocks itself.

Maybe AMD has a way of monitoring the clock more reliably, at best with a hardware probe.

Sure would be nice if the advertised clocks were average multi threaded clocks but that's not the case even for Intel.
Even Haswell/DC will tank to base clock on heavy load when not OCed, won't hold the minimal turbo under full load, but most people won't load their CPU that much to ever see it.
JackCY is offline  
post #19 of 88 (permalink) Old 09-02-2019, 02:01 PM
Invalid Media
 
J7SC's Avatar
 
Join Date: Dec 2018
Posts: 887
Rep: 22 (Unique: 12)
Quote: Originally Posted by The Pook View Post
early AGESA allowed advertised boost clocks, later AGESA doesn't ... so blame AMD? don't really follow.

My 9900K is advertised as a "5.0 boost" CPU but nothing ever will let it boost to 5.0 since nothing loads one core only. The only time it hits 5.0 on any core is for milliseconds while basically idle.

That's a very important (highlighted) point. Apart from that, I was wondering a bit why PBO was excluded by DerBauer, having read several reviews of Ryzen3K (which I incidentally don't own) that seemed to suggest that PBO was the way to go compared to manual oc. There's also the 'werbevideo' (German for 'advertising video') carried throughout the vid. This is not suggesting anything nefarious by DerBauer btw, but at the end of the day, I rather stick with Silicon Lottery's recent Ryzen3K boost info. The reason being that an online survey is not an ideal tool for a variety of reasons, including differing methodologies, but results obtained by the same tester, with the same methodology and conditions are preferable.

All that aside, I personally think that AMD's marketing department got a bit too enthusiastic re. 'ideal scenario' results, and it is always better to err on the side of caution (engineering > marketing)
Attached Thumbnails
Click image for larger version

Name:	RyzenGN_SilLot.jpg
Views:	39
Size:	200.9 KB
ID:	292998  

J7SC is online now  
post #20 of 88 (permalink) Old 09-02-2019, 02:05 PM
New to Overclock.net
 
deepor's Avatar
 
Join Date: Feb 2013
Posts: 4,729
Rep: 474 (Unique: 320)
Quote: Originally Posted by Imouto View Post
By missing the target single core boost by 2~3%?

You should read out loud what you type because it is honestly hilarious.

There was a similar problem with Bulldozer that was just about how things were worded in their specs, and AMD was sued for that. AMD lost that Bulldozer lawsuit recently. They'll be sued again for this new problem, and they'll lose again.

This whole new problem was completely avoidable. It's just about words. They should have just put safe numbers into the specs that are a bunch of 100MHz below of what the average CPU can boost.

The actual CPUs are great, they are the best choice for most use-cases. There really was no reason for them to artificially create this problem for themselves.
deepor is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 3 (0 members and 3 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off