AMD Launches Three Kaveri APU SKUs in February 2014 – Feature Set For A10 and A8 APUs Detailed

[humbug]

AMD On Kaveri APU x86 Performance Against Intel Haswell,”We’ll Lose” – Flagship APU To Cost Around $150

Read more: http://wccftech.com/amd-kaveri-apu-...ell-lose-flagship-apu-cost-150/#ixzz2m8M2CPSb

No one expected it to beat Intel in x86, just to have a marked improvement on what Piledriver is.

so...

A10-7850K which has been detailed here and features a total compute power of 856 Gigaflops. Compared to Intel Core i5-4670K that’s a definite leap in performance but when compared in x86 performance alone, AMD’s x86 enabled Steamroller cores would loose in the performance battle.

Thats not saying a lot, what does that actually mean? whats 856 Gigaflops when its at home and how many does the 4670K have?

Its missing key info. its a no news story

Edit - Ooohh Santa Slimy ^^^^

 

[CAT-THE-FIFTH]

I like how the tech tabloids forget something that Intel makes called the Core i3,which funnily enough is the same price as the AMD A10 CPUs.

 

[humbug]

I like how the tech tabloids forget something that Intel makes called the Core i3,which funnily enough is the same price as the AMD A10 CPUs.

I know right? Reviewers do this a lot with CPU's, "oh its not as good as the Intel" and fail to mention the i5 they used costs close to twice as much.

 

[gpuerrilla]

and on the rare occasion the FX83 do well they compare it to an i7 which again is loads more...

 

[joeyjojo]

Thats not saying a lot, what does that actually mean? whats 856 Gigaflops when its at home and how many does the 4670K have?

856 GFLOPS is a lot. The theoretical peak of the Intel range is 32 times the clock speed in GHz (using AVX vectorisation), so the 4670K at stock at 3.4 GHz = 108.8 GFLOPS. That's a theoretical limit - stuff like linpack gets closest and usually only 80-90% of peak.

As you'd expect, the difference comes from the GPU-like region of the chip, so you'll only get close to that performance using highly-paralellised code.

 

[Boomstick777]

I'm not sure how it's even a fair comparison to put Haswell against Kaveri in CPU performance terms, like they mention in that article. Kaveri is a about an all in one budget chip. I really hope AMD launch a desktop CPU only (No igp) Steamroller chip with more cores and cache. That at least could potentially compare favorably against the Haswell i5's. Maybe not in raw ipc but overall performance, where extra cores would pull ahead in encoding etc. Very tempted to move to an AMD 8 core setup if AMD do ever release Steamroller CPU for desktop.

 

[teppic]

I think Kaveri will compete very well with the 4670K as part of an overall system. It's not fair to compare just on raw CPU single thread performance and ignore the rest. With the 20% improvements and an overclock it's going to make the i3 absolutely pointless, and should stand up pretty well against the 4670K in newer games on a pure CPU level.

If we believe DICE and AMD, Kaveri with Mantle is going to be very impressive.

 

[CAT-THE-FIFTH]

Some details about the clockspeeds of the desktop Kaveri CPUs:

http://prohardver.hu/teszt/mit_tudhat_a_kaveri_gpu-ja/hd_7750-bol_kaveri_igp-t.html

It seems the A10 7850K has a maximum clockspeed of 4GHZ. For the 20% projected improvement in performance over Richland,that would mean a decent boost in IPC(if you compare it to the A10 6800K),since the A10 7850K has 3.7GHZ to 4GHZ clockspeeds and the A10 6800K,4.1GHZ to 4.4GHZ clockspeeds. Even if we look at the earlier AMD claims of 10% to 15% performance improvement per generation, IPC probably has improved around 20% to 30% it seems.

 

[d_brennen]

Worth remembering that this iteration of Kaveri is a year late, so not likely to fit in the 10-15% IPC per generation claim

 

[Martini1991]

AMD's slide wasn't IPC gain, it was performance.
If that's what you're referring to.

 

[BurntProc]

AMD's slide wasn't IPC gain, it was performance.
If that's what you're referring to.

Does it matter? All I care about is performance

 

[Martini1991]

Does it matter? All I care about is performance

Of course it matters.
PD's "10-15%" over BD came mainly from the higher default clocks, if it had been 10-15% IPC improvement, the end gains would be much better.

As it's comparing stock for stock, so you could get one of these new chips being clocked higher, yet having the same headroom, it's faster, but it's end performance when overclocked won't be as high as the quoted figure in that case (As an example)

 

[CAT-THE-FIFTH]

The Kaveri chips are clocked lower than Richland but looking at the AMD slides the performance boost is meant to be around 15% to 20% more. This means the IPC is probably a decent boost,considering the 10% lower clockspeeds for the A10 7850K when compared to the A10 6800K.

 

[Martini1991]

It has potential, but we'll wait and see really.
Don't want to get too hyped really

 

[teppic]

I've just ordered an FM2+ board

 

[Martini1991]

Queue Kaveri being on a super secret FM3 socket

 

[teppic]

Heh.

 

[teppic]

Ordered this, only ~£55:

https://www.asus.com/Motherboards/A88XMPLUS/

It has a heatsink on the VRMs, unlike the Gigabyte boards.

My 6800K that's sitting in its sealed box will finally get some use until Kaveri.

 

[Boomstick777]

Ordered this, only ~£55:

https://www.asus.com/Motherboards/A88XMPLUS/

It has a heatsink on the VRMs, unlike the Gigabyte boards.

My 6800K that's sitting in its sealed box will finally get some use until Kaveri.

Looks like a nice mobo, especially for that price.

 

[teppic]

Yup, I'm putting together a Micro ATX Steambox and HTPC system. When Kaveri's out hopefully it'll have enough grunt to run games at 1080p