Kaveri turns out to be one of the best products AMD produced in the last 3-5 years.
I think I'll wait until proper reviews come out to say anything.
That forum users has no NDA signed, he just bought the chip from some retail store.
New AIDA gpgpu benchmarks:
It doesn't matter since he overcloclocked them at ~same level. Richland is just higher clocked Trinity.Why are all the tests done in comparison with trinity and not with the A10 -6X00 richland series ? That would make a better comparison, yes ?
Why are all the tests done in comparison with trinity and not with the A10 -6X00 richland series ? That would make a better comparison, yes ?
The 720p benchmarks indicate some pretty impressive CPU performance gains. Why oh why can't we get an 8 core chip? C'mon AMD!
Here is .FlanK3r graph from incoming review concretely: Wprime 2.09, 1024M. Compare vs. AMD A8-3870K 3.0GHz "Liano" (Q2/2011) is still too bad
Actually according to the wprime chart, 7850K beats all previous gens 4-6C models, Llano included.
Well given the fact that Llano's 4 cores each have 4 FP units with 3 pipelines (12 FP pipes in total) while 2M/4T Kaveri has 2 FP units each having 2 pipelines(4 FP pipelines in total), the very fact Kaveri is in front of llano while having just 25% higher clock speed is a testament to its efficient design.maybe his face palm is related to the difference not that it is better
Yeah, it looks pretty good here from that perspective. But on the other hand, it does have half the FPUs of a comparable quad core. Especially when it still has a 128-bit width, floating point performance is going to be very lacking compared to Haswell and Broadwell.Well given the fact that Llano's 4 cores each have 4 FP units with 3 pipelines (12 FP pipes in total) while 2M/4T Kaveri has 2 FP units each having 2 pipelines(4 FP pipelines in total), the very fact Kaveri is in front of llano while having just 25% higher clock speed is a testament to its efficient design.
Here is .FlanK3r graph from incoming review concretely: Wprime 2.09, 1024M. Compare vs. AMD A8-3870K 3.0GHz "Liano" (Q2/2011) is still too bad
Considering that the A10-7850 is only 245mm² with the iGPU utilizing ~47% of the die area, whereas the FX-8350 is 315 mm², AMD should be able to add two more Steamroller cores and still have die that is smaller than the FX-8350. It is unfortunately that AMD has decided to leave the entirety of the enthusiast CPU market :|
Going from 8 to 10 cores would have a negligible benefit even for most enthusiasts.
Lanno (K10) has 4 dedicated FP units per 4 separate cores. Each core gets its own dedicated 3-issue(3 pipelines) FP co-processor. That is 12 FP pipelines in total.Llano else have 4FPU, but 128bit. and Bulldozer/Piledriver/Steamroller have in CU two 128bit FMAC (=> 2CU APU = 4)!
Well given the fact that Llano's 4 cores each have 4 FP units with 3 pipelines (12 FP pipes in total) while 2M/4T Kaveri has 2 FP units each having 2 pipelines(4 FP pipelines in total), the very fact Kaveri is in front of llano while having just 25% higher clock speed is a testament to its efficient design.
Wprime is a floating point benchmark..
Edit : The square root extraction step can indeed be an FP op but as i already pointed it Steamroller has an efficient radix 8 based hardware divider and root extractor wich can substancialy spare cycles , hence the good scores.wPrime is a leading multithreaded benchmark for x86 processors that tests your processor performance by calculating square roots with a recursive call of Newton's method for estimating functions, with f(x)=x2-k, where k is the number we're sqrting, until Sgn(f(x)/f'(x)) does not equal that of the previous iteration, starting with an estimation of k/2. It then uses an iterative calling of the estimation method a set amount of times to increase the accuracy of the results. It then confirms that n(k)2=k to ensure the calculation was correct. It repeats this for all numbers from 1 to the requested maximum.