• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Kaveri 7850K and APUNB scaling (how much?:) )

Joined
May 18, 2009
Messages
406 (0.07/day)
Location
czech republic
System Name AMD forever, AMD overclocker
Processor Athlon 3000+ Venice, Athlon x2 4600+ EE Windsor,x4 955 BE@3.9GHz AIR, X4 965 BE, x4 970 BE, x6 1090T
Motherboard Asus C5F-Z, Asus C5F, Asus C4E, Asus C4F, Asus RIIIE, Asus R4E, Asus M5F, Asus M6F, Asus M7H,
Cooling Corsair H100, Swiftech H220, CM Hyper 212, Xigmatek 1283 DK+Ultra Kaze, CM V6GT, Noctua NHD14
Memory Kingston Hyper X, 1600 A-Data 2000x, Corsair Dominator GT 2000 MHz, GSkill TridentX 2400MHz, HyperX
Video Card(s) HD3870 512MB GDDR4, HD5770 1024MB GDDR5, ATI HD4870, HD 6870 GDDR5, HD 7870,Radeon R9-270X
Storage 2x 320GB WD+Samsung, 1x 500 GB Samsung, SSD X-25, SSD HyperX, SSD Seagate, SSD Corsair GT
Display(s) BenQ 24" 24XL
Case HAF 922, Aspire X-Cruiser, benchtable Wroom
Power Supply Seasonic 500W and Seasonic 650W, Corsair AX 1200W, Zalman Goldrock 750, Galaxy OC edition 1200W
Software x64 win 7 and x86 Win XP SP3
Benchmark Scores you rather I do not want to see .. :-D !!!Thuban coming soon, new secret Gigabyte mobo too !!!
I tested few days ago new ROG board Asus Crossblade Ranger and with my alone A10-7850K chip I would like how good is 7850K with different APUNB frequency. Ussually are Kaveri lazy go higher than 1950 MHz at APUNB... I tested it in Superpi 32M, Cinebench R11.5 and in 3D Mark11. So...

CPU 4400 MHz APUNB 1800 MHz (stock), iGPU 800 MHz
15m03s / 4.20pts/ 2758pts






CPU 4400 MHz APUNB 1900 MHz, iGPU 800 MHz
14m 56s/ 4.21pts/ 2768pts






CPU 4400 MHz APUNB 2000 MHz, iGPU 800 MHz
14m 51s/ 4.23pts/ 2783pts







The 2100 MHz was not possible get stable for 32M, so I must set differents values near the 4400 MHz. Of course, the iGPU is now at higher clocks than before (816 MHz). CPU a bit lower frequencies.
CPU 4385 MHz APUNB 2040 MHz, iGPU 816 MHz
14m 46s/ 4.19pts/ 2835pts






So we can see nice improvement in SUperpi 32M. +100 MHz at APUNB is similar as 400 MHz APUNB at Richland! In last case with lower CPU clock is 32M again better than before! +240 MHz at APUNB bring you -17s at least. For CInebench are little extra points - with exactly same the CPU clock +0,03 points and in 3D Mark (only three results, not the last one) +23 points.

For fun, my average chip and 32M run:


This CPU is very good in 32M, some 6.1 GHz LN2 chip could hit around 10min 32M or less!
 
Joined
Dec 3, 2009
Messages
1,298 (0.25/day)
Location
The Netherlands
System Name PC || Acer Nitro 5
Processor Ryzen 9 5900x || R5 2500U @ 35W
Motherboard MAG B550M MORTAR WIFI || default
Cooling 1x Corsair XR5 360mm Rad||
Memory 2x16GB HyperX 3600 @ 3800 || 2x8GB DDR4 @ 2400MTs
Video Card(s) MSI RTX 2080Ti Sea Hawk EK X || RX 560X
Storage Samsung 9801TB x2 + Striped Tiered Storage Space (2x 128Gb SSD + 2x 1TB HDD) || 128GB + 1TB SSD
Display(s) Iiyama PL2770QS + Samsung U28E590, || 15,6" 1080P IPS @ 100Hz Freesync enabled
Case SilverStone Alta G1M ||
Audio Device(s) Asus Xonar DX
Power Supply Cooler Master V850 SFX || 135Watt 19V OEM adaptor
Mouse ROG Pugio II
Software Win 11 64bit || Win 11 64bit
nice write-up! Thanks for sharing

I honestly did not expect to see much if any difference at all, cool to see some benchies! Also interesting to see the NB having that low of a frequency compared to vishera chips.
 

Virssagòn

New Member
Joined
Aug 19, 2014
Messages
4 (0.00/day)
Nice Write-up enjoyed it alot! Kaveri offers some new performance in different areas as I can see!
 
Joined
May 18, 2009
Messages
406 (0.07/day)
Location
czech republic
System Name AMD forever, AMD overclocker
Processor Athlon 3000+ Venice, Athlon x2 4600+ EE Windsor,x4 955 BE@3.9GHz AIR, X4 965 BE, x4 970 BE, x6 1090T
Motherboard Asus C5F-Z, Asus C5F, Asus C4E, Asus C4F, Asus RIIIE, Asus R4E, Asus M5F, Asus M6F, Asus M7H,
Cooling Corsair H100, Swiftech H220, CM Hyper 212, Xigmatek 1283 DK+Ultra Kaze, CM V6GT, Noctua NHD14
Memory Kingston Hyper X, 1600 A-Data 2000x, Corsair Dominator GT 2000 MHz, GSkill TridentX 2400MHz, HyperX
Video Card(s) HD3870 512MB GDDR4, HD5770 1024MB GDDR5, ATI HD4870, HD 6870 GDDR5, HD 7870,Radeon R9-270X
Storage 2x 320GB WD+Samsung, 1x 500 GB Samsung, SSD X-25, SSD HyperX, SSD Seagate, SSD Corsair GT
Display(s) BenQ 24" 24XL
Case HAF 922, Aspire X-Cruiser, benchtable Wroom
Power Supply Seasonic 500W and Seasonic 650W, Corsair AX 1200W, Zalman Goldrock 750, Galaxy OC edition 1200W
Software x64 win 7 and x86 Win XP SP3
Benchmark Scores you rather I do not want to see .. :-D !!!Thuban coming soon, new secret Gigabyte mobo too !!!
Thanks guys.

Yes, Kaveri has lower NB than Vishears, but scaling very well - with lower increment of MHz very good scores.
 
Joined
Mar 19, 2005
Messages
302 (0.04/day)
Location
Chicago
System Name Big Beast
Processor AMD Ryzen 7 5800x3D
Motherboard Asus Strix x570-I Gaming
Cooling NZXT AIO
Memory Patriot Viper 16GB 3666MHz
Video Card(s) AMD Radeon 7900XTX
Storage Lot's of it
Display(s) Samsung CHG70 27"
Case NZXT H1
Power Supply 850watt
Software Windows 11 Professional x64
The last NB speed doesn't seem worth it if you already have it at 2000MHz and if you have a discrete GPU. When using the IGP, the CPU would just downclock to 3.0Ghz so in that case I could see it being worth it to overclocking the bus versus multiplier. Also look into lower the voltages, I have the CPU voltage at 1.3975 (something like that) and the 'APU' voltage at 1.275. http://valid.x86.fr/cjqvfe
 
Joined
May 18, 2009
Messages
406 (0.07/day)
Location
czech republic
System Name AMD forever, AMD overclocker
Processor Athlon 3000+ Venice, Athlon x2 4600+ EE Windsor,x4 955 BE@3.9GHz AIR, X4 965 BE, x4 970 BE, x6 1090T
Motherboard Asus C5F-Z, Asus C5F, Asus C4E, Asus C4F, Asus RIIIE, Asus R4E, Asus M5F, Asus M6F, Asus M7H,
Cooling Corsair H100, Swiftech H220, CM Hyper 212, Xigmatek 1283 DK+Ultra Kaze, CM V6GT, Noctua NHD14
Memory Kingston Hyper X, 1600 A-Data 2000x, Corsair Dominator GT 2000 MHz, GSkill TridentX 2400MHz, HyperX
Video Card(s) HD3870 512MB GDDR4, HD5770 1024MB GDDR5, ATI HD4870, HD 6870 GDDR5, HD 7870,Radeon R9-270X
Storage 2x 320GB WD+Samsung, 1x 500 GB Samsung, SSD X-25, SSD HyperX, SSD Seagate, SSD Corsair GT
Display(s) BenQ 24" 24XL
Case HAF 922, Aspire X-Cruiser, benchtable Wroom
Power Supply Seasonic 500W and Seasonic 650W, Corsair AX 1200W, Zalman Goldrock 750, Galaxy OC edition 1200W
Software x64 win 7 and x86 Win XP SP3
Benchmark Scores you rather I do not want to see .. :-D !!!Thuban coming soon, new secret Gigabyte mobo too !!!
Some guys telling its some issue in CPU, but with my testing with Asus I had no problems in 3D Marks and in Sniper Elite or UT3 benchmark.
 
Last edited:

zlobster

New Member
Joined
Jun 18, 2012
Messages
2 (0.00/day)
Nice! Thanks for sharing!

Anyway, may I suggest some tests that actually benefit from the increased NB clocks? Something like the new Prime95 (28.5, or newer), LuxMark, Kishonti's stuff, etc.? I believe these will tell an entirely new story.

The LuxMark, OpenCL and many game tests revealed a whole new story when I bumped the iGPU of my 7850K to 960MHz. It was done on an A88X-PRO via the provided 'Extreme' iGPU OC setting.
 
Top