[WCCF] AMD Kaveri Mobile APUs Vs. ULV Haswell Benchmarks

[rainy]

I'm taking those results with a grain of salt, however they look very promising.
http://wccftech.com/amd-kaveri-mobile-apus-vs-ulv-haswell-benchmarks/

 

[strata8]

GPU could be about right.

But the CPU in the FX-7500 should be between the i3 and i5 in single-threaded and just below the i5 in multithreaded.

 

[AtenRa]

Last year there was only a single AMD SKU at 19W TDP. This year there are 3-4 SKUs at 19W TDP, clearly outlining AMDs intentions in Mobile.

Lets see the reviews coming,

 

[frozentundra123456]

Need power consumption and battery life measurements. The has well is only 15 watts tdp vs 19 for kaveri.

 

[rainy]

Need power consumption and battery life measurements. The has well is only 15 watts tdp vs 19 for kaveri.

I hope you're not expecting a miracle - Intel using "only" 22nm while AMD 28nm.
FinFET is optimized and as result especially effective at lower clocks.

 

[frozentundra123456]

I hope you're not expecting a miracle - Intel using "only" 22nm while AMD 28nm.
FinFET is optimized and as result especially effective at lower clocks.

Whatever process node the chips are on doesnt really concern me. The performance per watt and low power states "are what they are". If AMD is only on 28nm, and that has not even come out yet for mobile, and intel will be on 14nm soon in mobile, that is not my problem. That is just the way it is. What counts is the product that comes to market.

That said, mobile Kaveri might be nice based on price vs Haswell/Broadwell, but all we have seen so far, both for Beema/Mullins and mobile kaveri are performance benchmarks, with no battery life because there are no shipping products yet.

 

[witeken]

Need power consumption and battery life measurements. The has well is only 15 watts tdp vs 19 for kaveri.

Does AMD's 19W include the PCH? Haswell is 15W with PCH while Ivy Bridge was 17W + 3W = 20W.

 

[Insert_Nickname]

Does AMD's 19W include the PCH? Haswell is 15W with PCH while Ivy Bridge was 17W + 3W = 20W.

I'd expect not. Kaveri still has an off-die PCH/FCH, rated at 4.7W TDP (A70M). I think I've seen -somewhere- that it uses ~0.75-1W idle. Can't remember where though, so don't quote those numbers.

 

[bullzz]

doesnt the 3d mark 11 score put FX7500 on par with HD5000?

 

[monstercameron]

for reference:

Kaveri^

http://www.anandtech.com/show/7072/intel-hd-5000-vs-hd-4000-vs-hd-4400
http://www.notebookcheck.net/Intel-HD-Graphics-5000.91978.0.html
Intel hd ^

http://www.anandtech.com/show/7106/...t-1-the-apu-and-radeon-hd-8650g-performance/2
http://www.notebookcheck.net/AMD-Radeon-HD-8650G.87916.0.html
richland ^ can score up to ~1500

http://www.anandtech.com/show/6993/intel-iris-pro-5200-graphics-review-core-i74950hq-tested/16
intel hd 5200^

intel hd 5100 scores ~1300
http://www.notebookcheck.net/Review-Asus-Zenbook-Infinity-UX301LA-Ultrabook.103027.0.html

the nvidia 820m or the 540m really scores 1500
http://www.3dmark.com/3dm11/2842301

 

[Enigmoid]

Last year there was only a single AMD SKU at 19W TDP. This year there are 3-4 SKUs at 19W TDP, clearly outlining AMDs intentions in Mobile.

Lets see the reviews coming,

Richland low power SKUs.

A4-5145M-2 core - 17W
A6-5345M-2 core - 17W
A8-5545M-4 core - 19W
A10-5745M -4 core - 25W

Plenty on richland low power SKUs. Nice to see quad core moving deeper into the ULV territory.

for reference:

Kaveri^

http://www.anandtech.com/show/7072/intel-hd-5000-vs-hd-4000-vs-hd-4400
Intel hd ^

http://www.anandtech.com/show/7106/...t-1-the-apu-and-radeon-hd-8650g-performance/2
richland ^

the nvidia 820m or the 540m really scores 1500
http://www.3dmark.com/3dm11/2842301

With a 19W tdp 3dmark scores are not completely relevant as clocks will change when the the CPU cores are heavily loaded in a game. The a10-4600m had the same problem where in 3dmark it did disproportionately well compared to discrete GPUs. Intel's HD series are also suspect; IVB did disproportionately poorly while HW does unrealistically well.

 

[strata8]

Don't forget Beema as well. A6-6310 @ 15W scores 780 points.

 

[metalliax]

Does AMD's 19W include the PCH? Haswell is 15W with PCH while Ivy Bridge was 17W + 3W = 20W.

It doesn't include PCH, however last I checked, Intel doesn't stay within their respected TDP ceiling very well anyway.

This is from a 15W Haswell (MacBook Air review):

 

[witeken]

It doesn't include PCH, however last I checked, Intel doesn't stay within their respected TDP ceiling very well anyway.

This is from a 15W Haswell (MacBook Air review):

Quote Anand:

"Remember, with Sandy Bridge Intel introduced Turbo Boost 2.0 that effectively allowed for two separate power limits - one equal to the processor's TDP (PL1) and one higher than the processor's TDP (PL2) that could be hit as long as the die temperature doesn't get too high."

 

[Enigmoid]

Yeah the 45W quads has a 56W turbo. Lasts about 10-25 seconds. This is why short benchmark runs are not indicative of gaming performance.

 

[IntelUser2000]

I'm taking those results with a grain of salt, however they look very promising.
http://wccftech.com/amd-kaveri-mobile-apus-vs-ulv-haswell-benchmarks/

Kaveri does very well on PCMark 8: http://www.extremetech.com/wp-content/uploads/2014/01/Kaveri-Perf.jpg

 

[Enigmoid]

Kaveri does very well on PCMark 8: http://www.extremetech.com/wp-content/uploads/2014/01/Kaveri-Perf.jpg

OpenCL acceleration.

Again not sure what they are trying to do. If PCmark is trying to show a measure of general performance then OpenCL acceleration is not helpful at all, given the rather large dearth of OpenCL consumer apps. Because pretty much every consumer workload (sans gaming) is going to be faster on the i5. Media on HD 4600 is fine, its just gaming and OpenCL that's poor.

 

[mrmt]

Again not sure what they are trying to do.

That's AMD value proposition, isn't it? AMD is trying to sell Kaveri to people who do care enough about gaming or OpenCL but don't care too much about CPU or power consumption, and is cash strapped enough to not consider a dGPU. Those benchmarks are exactly to showcase this scenario for the targeted consumers. It's the same old strategy that is making them lose market share for the last the last three years. I can't really see Kaveri changing anything.

 

[kagui]

these cpu are about the AIO, the low budget, on those machines people doesnt care about single thread performance or even power consumption , it cares about flash games, youtube , minecraft, firefox, chrome, they want their low budget machine to work and these cpu work very well on those task and are cheaper and better

 

[AtenRa]

The PCMark 8 Creative benchmark test contains the following workloads:

-Web Browsing
-Photo Editing
-Batch Photo Editing
-Video Editing
-Media To Go
-Mainstream Gaming
-Video Group Chat

I believe those can cover 90% or more of the market usage. And there are already applications that utilize OpenCL or GPGPU among those categories.

Of course you can run Cinebench and show how fast the Intel CPU is but i doubt 1% of the market will ever use it.

 

[monstercameron]

wow, now you guys are bashing opencl -read, an apple initiative.

opencl is in wide spread adoption for alot of video editors [which you would need a more beefy apu for], even adobes creative suite supports opencl in a few of its programs.

Also opencl will be a driving force in the industry going forward.

 

[ShintaiDK]

That's AMD value proposition, isn't it? AMD is trying to sell Kaveri to people who do care enough about gaming or OpenCL but don't care too much about CPU or power consumption, and is cash strapped enough to not consider a dGPU. Those benchmarks are exactly to showcase this scenario for the targeted consumers. It's the same old strategy that is making them lose market share for the last the last three years. I can't really see Kaveri changing anything.

Actually you forgot, IGP only. Since even AMD can outcompete most of the Kaveri line in performance/$ with its own offerings if you add a dGPU.

 

[Erenhardt]

Actually you forgot, IGP only. Since even AMD can outcompete most of the Kaveri line in performance/$ with its own offerings if you add a dGPU.

Speaking about dgpu. Pair the APU with mobile R7 250 and step up your performance to the next tier.

 

[ShintaiDK]

Speaking about dgpu. Pair the APU with mobile R7 250 and step up your performance to the next tier.

More cost not to mention CF problems that can be avoided with an even better dGPU on the alternatives. Again, Kaveri looses flat out to AMDs own product line.

 

[mrmt]

wow, now you guys are bashing opencl -read, an apple initiative.

opencl is in wide spread adoption for alot of video editors [which you would need a more beefy apu for], even adobes creative suite supports opencl in a few of its programs.

Also opencl will be a driving force in the industry going forward.

I didn't see OpenCL bashing here, just that openCL performance is by no means representative of what you can get in the overall market.