Kaveri vs Trinity Clock to Clock

[AtenRa]

AIDA64 Memory & Cache and GPGPU added in the OP

 

[Erenhardt]

Were you able to OC past 4.5 ghz and get additional performance doing it?

It was running on with mild OC (4.4GHz?) established on stock voltage. Somehow the software overclocking decided to give it more Volts. Thankfully I checked power usage as soon as stress test started, so I killed it quickly.

I am assuming this set-up has an aftermarket cooler?

If so, I wonder what happens if the user has the stock cooler?

How much CPU performance can we expect (under full GPU load) under those circumstances?

This was with DeepCool Gammax s40

or Silentium PC Ferra 2 He1224

Im not sure which.

 

[AtenRa]

PC Mark 8 - Creative - Accelerated + Energy Consumption added in the OP

 

[Sequences]

Just out of curiosity, where do you guys think Richland sits between those scores?

 

[DrMrLordX]

It was running on with mild OC (4.4GHz?) established on stock voltage. Somehow the software overclocking decided to give it more Volts. Thankfully I checked power usage as soon as stress test started, so I killed it quickly.

Eek. Good thing you picked up on that quickly. The reason why I ask is that there's been an observed "cap" on performance (at least under air or water) at 4.5 ghz. Basically, if you clock past 4.5 ghz, the chip refuses to produce superior performance. Don't think it's been an issue under phase/LN2 though.

 

[SlowSpyder]

Thanks AtenRa. Looks like they were able to make jumps in CPU and GPU performance. Does the A10 actually use less power when benching, or are the power savings simply from it being a better performer and hence finishing the bench suite quicker?

 

[AtenRa]

Just out of curiosity, where do you guys think Richland sits between those scores?

Somewhere in between the 7700K and 5800K, little faster than 5800.

 

[AtenRa]

Thanks AtenRa. Looks like they were able to make jumps in CPU and GPU performance. Does the A10 actually use less power when benching, or are the power savings simply from it being a better performer and hence finishing the bench suite quicker?

Default or at 4GHz ??

 

[SlowSpyder]

Default or at 4GHz ??

The benches you added in the OP, 4GHz / 800MHz.

 

[Headfoot]

Looks like that DeepCool Gammax S40 still loses out to the trust Hyper 212+, despite looking like it should be better.
http://www.frostytech.com/articleview.cfm?articleid=2721&page=4

 

[SPBHM]

Somewhere in between the 7700K and 5800K, little faster than 5800.

at the same clock it should be exactly the same, not a little faster!?

perhaps a difference in power usage and that's all.

 

[Erenhardt]

Looks like that DeepCool Gammax S40 still loses out to the trust Hyper 212+, despite looking like it should be better.
http://www.frostytech.com/articleview.cfm?articleid=2721&page=4

Its pretty small cooler. If you look closely the 12cm fan is bigger than the heatsink /OT

 

[AtenRa]

Edit 28-07-2014 :
I will have to apologize, i have mistakenly used the Default settings score for the A10-5800K in PC MARK 8. I have included both 4GHz and Default scores now so you can see the difference.
Interestingly at Default, the A10-5800K is a little faster, probable because it has turbo at 4.2GHz.

 

[AtenRa]

The benches you added in the OP, 4GHz / 800MHz.

Kaveri is using less power throughout the Benchmarks. For example, in PC MARK 8 (Gaming part 2), Total system power (peak) was 91W for Kaveri and 120W for Trinity.

 

[AtenRa]

Cenebench 11.5 and R15 with Total system Peak power consumption added in the OP. I will post the pics here .

Cinebench 11.5

Total System Peak Power Consumption

Kaveri Single Thread = 62W
Kaveri Multi Thread = 106W

Trinity Single Thread = 81W
Trinity Multi Thread = 126W


Cinebench R15

Total System Peak Power Consumption

Kaveri Single Thread = 63W
Kaveri Multi Thread = 103W

Trinity Single Thread = 76W
Trinity Multi Thread = 126W

 

[AtenRa]

Edit 29-07-2014 :
x264, Pov-Ray, 7zip , SVPmark3, CL Benchmark and LuxMark added in the OP.

 

[SlowSpyder]

Kaveri is using less power throughout the Benchmarks. For example, in PC MARK 8 (Gaming part 2), Total system power (peak) was 91W for Kaveri and 120W for Trinity.

Wow, nice.

 

[inf64]

Great work AtenRa .

Overall pretty solid IPC bump for Kaveri. This time the uarch and process node changes didn't "brake" the chip . Too bad they didn't manage to put out 3M version. There is a glimmer of hope XV based Carrizo may end up with 3M variant eventually, but it will go up against Skylake i3 and i5 parts, a very tough competition.

 

[AtenRa]

Great work AtenRa .

Overall pretty solid IPC bump for Kaveri. This time the uarch and process node changes didn't "brake" the chip . Too bad they didn't manage to put out 3M version. There is a glimmer of hope XV based Carrizo may end up with 3M variant eventually, but it will go up against Skylake i3 and i5 parts, a very tough competition.

Yea pity they didnt released an 8-core AM3+ Steamroller FX SKU. From what is see from Kaveri, a 125W TDP 8-Core would be nice to have for upgrade or new build option, especially next year with Excavator.

Excavator with HDL would be very nice to have in AM3+. Release a 6-core 65W and 8-core 95W configurable TDP and let people play with them.

At least it would be nice to have a 6 or 8 core iGPU-less SKUs in FM2+ but im not sure if the socket can support it.

 

[AtenRa]

OK and now some games,

Remember both APUs are at 4GHz CPU, 800MHz for the iGPU, 2133MHz memory with Catalyst 14.4 on Win 8.1 64bit

I have used the most playable settings with the highest image quality possible at 720p and 1080p. I have also included total system peak power consumption in the same graphs to be easier to evaluate the efficiency.

Edit 7-8-2014 : 5800K default added in all Gaming Benchmarks.

 

[AtenRa]

Games Continue

Edit 7-8-2014 : 5800K default added in all Gaming Benchmarks.

Civilization coming soon

 

[PPB]

Games Continue



Civilization coming soon

If you can, try to bench LoL in ARAM. That map is a CPU hog and would be nice if AMD CPU perf improved in that scenario (I dont even ask for Summoner's Rift because that one plays swell).

 

[monstercameron]

If you can, try to bench LoL in ARAM. That map is a CPU hog and would be nice if AMD CPU perf improved in that scenario (I dont even ask for Summoner's Rift because that one plays swell).

on my a10 its pretty much ~80fps. Maxed out.

 

[PPB]

on my a10 its pretty much ~80fps. Maxed out.

No sub 60fps slowdons on big TFs? That's nice to hear.

 

[monstercameron]

No sub 60fps slowdons on big TFs? That's nice to hear.

https://www.youtube.com/watch?v=1jjA4t-6D9o

This videos isn't representative of performance due to me streaming to youtube but look at the dips in the team fight.