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

[stulid]

And I got 112W - http://forums.overclockers.co.uk/showpost.php?p=25743641&postcount=816

So I call BS on that power consumption article

 

[Legion]

Probably M-ATX as I'm very limited on space atm. Using something like the corsair obsidion 350D as the case.
Not really a lot of options for decent FM2+ m-ATX boards atm.

I will be running it with an R9 290, I'm sure some of you will say that's a waste, but due to a lack of space and limited funds this seems to be a viable way for me to make a second pc that I can game on.

Indeed, I'd class all the current mATX boards as budget boards including the Asus A88XM-PLUS I have in my HTPC.

I'm sure you know about the MSI Gaming boards due to be released soon, and I have my eye on the mATX board (A88XM Gaming)

That's what I'd be considering as the best candidate for your proposed build.
I just hope the reviews are good and they clock as good as they look.

 

[Gazb]

Indeed, I'd class all the current mATX boards as budget boards including the Asus A88XM-PRO I have in my HTPC.

I'm sure you know about the MSI Gaming boards due to be released soon, and I have my eye on the mATX board (A88XM Gaming)

That's what I'd be considering as the best candidate for your proposed build.
I just hope the reviews are good and they clock as good as they look.

Yeah, I'm keeping an eye on the MSI A88XM Gaming board, though there doesn't appear to be any details about actually being available to buy anywhere, either in the UK or US.

 

[Legion]

Yeah, I'm keeping an eye on the MSI A88XM Gaming board, though there doesn't appear to be any details about actually being available to buy anywhere, either in the UK or US.

Sometime this month.

 

[Electroshadow]

Stulid which version of Guiminer are you running? I am yet to manage to get the GPU to mine. Through CG or GUI.

 

[CAT-THE-FIFTH]

Interestingly I use Gigabyte in my current main build(plus an old Zotac in another),so I honestly I don't see why people have an issue with believing another company might make a better motherboard for certain uses. However,it seems some fellow Gigabyte users seem desperate though - not really sure why?? Its almost like a sales pitch TBH. WEIRD.

However for people who missed the last few pages of this thread(or CBA),all of this stemmed from this comment I made:

This one looks OK:

http://www.asrock.com/mb/AMD/FM2A88X-ITX+/?cat=Specifications

Power consumption is meant to be decent.

Which eventually lead to this:

Just wait till the horror stories of the burning hot VRM's and throttling on the Asrock in tiny cases start appearing.

and this:

Its not just the cheap build quality of asrock that puts me off. The warranty is worse too.

This is despite me mentioning the 65W TDP A8 7600 specifically,not one of the 100W jobbies,and knowing quite a few people using the earlier version of the board,for 6+ months with 100W TDP Trinity and Richland CPUs which consume MORE power under load.

On top of this I also linked to a review which analysed the VRM section of the A85X mini-ITX AsRock motherboard in detail saying it used decent components and that was Hardwaresecrets which is known for good technical analysis of hardware parts,especially PSUs and motherboards. Their PSU reviews are even considered decent by the people over on Jonnyguru.

If the roles were reversed in the hardware.info review I bet there would have been no issue with saying the Gigabyte was better for the uses I suggested.

However,this has turned into some side snipe brand justification contest,and I should have known better where it would lead.

So here's something for people who've been "misinformed" with the power consumption of the Gigabyte GA-F2A88XN-WIFI mITX motherboard.

I brought my power meter home over the weekend to test this, as I was calling BS on the article that was showing abnormally high power consumption figures.
They used an A10-6800K Processor in that review and while that is a little less power efficient than Kaveri, there's definitely something up with their numbers.
I'm testing in an Overclocked environment and will do the same tests as shown in the graph further up the page. (I'll do some stock tests at some point)
I'll also include some real world results of my own.

Test setup
AMD A10-7850K Overclocked to 4.2ghz CPU, 1000mhz iGPU
Thermalright AXP-200 cooler
Gigabyte GA-F2A88XN-WIFI motherboard (F4a Bios)
2x4gb Team Xtreem LV DDR3 2400 Cas 10, 12, 12, 31, 2T
256gb Samsung 830 SSD
500gb HGST 2.5" Storage Drive 7200rpm
160w Pico Psu /w 192w power brick
Silverstone Milo ML06B case
USB mouse and keyboard

Windows 8.1 Pro 64 bit
iGPU on AMD catalyst 14.1 beta
AMD Chipset driver 13.20

All C states enabled and APM enabled
Onboard WIFI enabled (using even more power)
Figures given are total system power consumption measured at the socket with my trusty Kill A Watt meter.

Here's the tests to Compare with the graph, although it's not clear on their methodology for their 3DMark Firestrike testing, I'd concur that scaled up, my overclocked testing would be similar to their results clock for clock if they'd posted proper numbers and not some skewed kind of average.

Idle power consumption (after 5 mins idle) = 28w

3DMark Firestrike
Graphics Test 1 = 103.8w Peak
Graphics Test 2 = 102.4w Peak
Physics Test = 101.7w Peak
Combined Test = 114.3w Peak

Cinebench 11.5 = 104.7W Peak

Somehow their CPU intensive test (Cinebench 11.5) vastly differs from mine, like I've said previously Kaveri is a little more power efficient than Trinity, but there is something badly wrong with their result.
This particular test result is the one responsible for the "Misinformation", they didn't even test the board with a Kaveri APU.

Real world results and observations of my own.

Borderlands 2 1080p, Medium Settings, No AA
Co-op play with 3 other friends
Buttery smooth, no jitters
Complete system power draw ~125w with the odd peaks of ~130w.

Starcraft 2 1080p Medium Settings No AA
Single player game, jitter free, scrolls nicely
Complete system draw 117w

1080p 10gb Blu-ray rips, played back through XBMC (DXVA)
Complete system power draw = 62w

Got Guiminer working, 112k hashrate (at 1000mhz iGPU)
Complete system power draw = 96w (shows it's working the iGPU nicely)

Stock clocks testing and Underclocking / Undervolting / cTDP function to be tested some time soon

Those are from Hardware.fr with an Asus A88X Plus. They are the biggest computer hardware site in France. The A10 6800K can consume much more power than an A10 7850K.

Edit!!

Another thing. Major sites tend to use normal PSUs,not pico-PSUs which hit 90% efficiency at low loads,especially if you choose the correct power brick. The power brick alone makes a big difference to power consumption.

 

[krooton]

Well I have the same mobo, running a 6800k and 7750.

Idle: 50w from the wall
Prime: 120w from the wall
Gaming at 1080p (Smite): 150w from the wall

This is with a 350w SFX psu.

 

[CAT-THE-FIFTH]

KitGuru measures 155W at the wall using the A85X version of the Gigabyte under Prime95 with an A10 6800K:

http://www.kitguru.net/components/m...e-f2a85xn-wifi-mini-itx-motherboard-review/5/

 

[Legion]

Interestingly I use Gigabyte in my current main build(plus an old Zotac in another),so I honestly I don't see why people have an issue with believing another company might make a better motherboard for certain uses.

If the roles were reversed in the hardware.info review I bet there would have been no issue with saying the Gigabyte was better.

Those are from Hardware.fr with an Asus A88X Plus. They are the biggest computer hardware site in France.

Edit!!

Another thing. Major sites tend to use normal PSUs,not pico-PSUs which hit 90% efficiency at low loads,especially if you choose the correct power brick. The power brick alone makes a big difference to power consumption.

Nobody is saying any board is better.
I'm saying that after buying, owning and using previous incarnations of a "certain" board, I wouldn't put my money on that brand again.
I have no issue with whatever someone else chooses to spend their money on, it's their money right?

I'm also saying that the Ukhardwareinfo "comparison" cinebench 11.5 test "result", using the 6800k in the Gigabyte ITX board is flatout wrong.
There's something way off the mark, with their result.
It can't pull that much power in the "CPU" intensive part of the test (at stock settings)

The Hardware france Luxmark result showing the 142w draw is a combined test hammering the CPU and iGPU both at the same time,
Ukhardwareinfo's Cinebench test result is CPU only. (or if it isn't, it's unclear in their test description)


Fire up Cinebench 11.5 Krooton and show him.

 

[CAT-THE-FIFTH]

Nobody is saying any board is better.
I'm saying that after buying, owning and using previous incarnations of a "certain" board, I wouldn't put my money on that brand again.
I have no issue with whatever someone else chooses to spend their money on, it's their money right?

I'm also saying that the Ukharwardwareinfo "comparison" cinebench 11.5 test "result", using the 6800k in the Gigabyte ITX board is flatout wrong.
There's something way off the mark, with their result.
It can't pull that much power in the "CPU" intensive part of the test (at stock settings)

The Hardware france Luxmark result showing the 142w draw, is a combined test hammering the CPU and iGPU both at the same time,
Ukhardwareinfo's Cinebench test result is CPU only.


Fire up Cinebench 11.5 Krooton and show him.

But your statements are based on ONE board(as with the other statement). The fact it had a 3+1 phase VRM alone was why I never recommended it.

I linked to a review which analysed the VRM design of the current AsRock mini-ITX motherboards and a well known site said it was a decent design.

However, I also know loads of people with their A75 mATX and A85 mini-ITX who are fine,and all with 100W TDP Trinity and Richland CPUs.

At least one other person in this thread has said the same and has the A88X AsRock mini-ITX motherboard and an A10 7850K. I was talking about an A8 7600 which consumes less power.

I even linked to the overclock.net AMD motherboard failure list - it is mostly MSI motherboards and only that single AsRock A75 mini-ITX listed.

I have yet to see any mass issues with any of the other FM2+ motherboards AsRock did,even after looking at the massive AMD section on that site.

Gigabyte had throttling issues with their AM3+ range in the first six months. Asus is the only company which seems to have consistently been OK on the AMD side for BOTH AM3+ and FM2/FM2+ but they take ages to put out BIOS updates for older motherboards.

Regarding the hardware.info review,its not average power consumption but max for the CB,and ultimately even then the AsRock consumes less power than the Gigabyte during the Firestrike run. It was the Firestrike run I was referring too specifically.

Why do you think I am interested in getting an A8 7600??

Its because I want a small rig in an ISK110/or a small case which I can run games on and 3DMark Firestrike is a worse case scenario for gaming IGP power consumption,and I would be running the CPU in 45W TDP mode.

Do I think the Gigabyte would be better for overclocking - yes I do and TBH I tend to like their motherboards. But,ultimately it is not what I am interested in.

Moreover,the point of the other figures was to show the A10 7850K can consume far less power than the A10 6800K.

This means the A8 7600 is better off especially in 65W and 45W TDP mode. Hence the AsRock won't be going puff with an A8 7600 IMHO,especially with the horizontal type coolers being used. I specifically mentioned the A8 7600.

Edit!!

I think we are going around in circles. I think we will need to agree to disagree on AsRock and leave it at that.

 

[Legion]

/snip

I agree Cat, it's horses for courses here.
You want low power and aren't into overclocking.
I'm an overclocker and a gamer and for my needs, the Gigabyte board is a better choice for me.

I suffer from bouts of insomnia and my little rig is used on nights when I can't sleep and I just game or watch a few movies in bed in the man cave.

I despise laptops and tablets.
An overclockable, almost silent, sub 150w (dGPU less) pc box running games at 1080p is a great little toy to have.
To get half decent frame rates though with medium settings at 1080p, you have to clock the iGPU as much as you can and give it all the memory bandwidth (tighter timings the better) you can throw at it.
I'm not interested in gaming at lesser resolutions and this is what it takes to get smooth gameplay in newer games at 1080p.
Some heavy AAA titles are still unplayable at these settings, but if I want to play those I've still got the big rig.

 

[CAT-THE-FIFTH]

My main rig is mini-ITX too. I moved from a modified Shuttle to a Cubitek Mini Cube which has been the biggest case I had for years. Its nice enough but I do have a hankering for a Shuttle R series system due to their smaller volume(despite their many flaws and high price).

Something like an A8 7600 would be a nice general purpose/LAN rig, especially with a smaller montor like a 19" display which should make it reasonably portable too. I knew the A10 5700 could fit into some of the smaller Streacom cases,so I hope the A8 7600 in 45W TDP mode can fit into the ISK110 or something similar. It might not,but if it could that would be fantastic.

I get the impression the A10 7850K due to its 512 shaders is probably more bandwidth starved with regards to the IGP than the A8 7600. It probably will benefit more from overclocking IMHO and higher bandwidth RAM.

 

[Gazb]

I posted this in the mantle feedback thread but it's relevant here too.

Originally Posted by MjFrosty View Post
http://www.pcper.com/reviews/General...medium=twitter

Some interesting questions put forward in this piece, including some info on shader translations

I could be wrong but the answer to one of the questions sounds like we won't be seeing an APU with dedicated GPU card gaming setup using mantle were the physics functions etc are being handed off to the APU's iGPU while the dGPU handles the rendering of the game itself.

i'm refering to this point:

Q: How will Mantle tie in with HSA features? For instance, can both halves of an APU collaborate while discrete GPUs are busy drawing? For example, during AI update code which blends serial (logic) and parallel (pathfinding and visibility) tasks?

[Guennadi] This is something we'll need to evaluate in the future. Right now HSA and Mantle try to solve different sets of problems, but there is certainly room for overlap at some point.

 

[Legion]

The original article here:
http://www.pcper.com/reviews/Genera...uer?utm_source=twitterfeed&utm_medium=twitter

I could be wrong but the answer to one of the questions sounds like we won't be seeing an APU with dedicated GPU card gaming setup using mantle were the physics functions etc are being handed off to the APU's iGPU while the dGPU handles the rendering of the game itself

This was always going to be the case.
It's going to take a long time for HSA to start getting coded into mainstream games and programs.

Mantle also needs HSA coding into it.
Mantle is also limited to less than a handful of games for most of this year.
The only feature you can currently take advantage of with Mantle is to reduce "some" of the CPU bottleneck.

2 Mantle enabled games (by the end of Feb) BF4 and Thief.
In the vast majority other games you'll still be heavily CPU bound, strangling that top end dGPU. (making that purchase of a top end gfx card a very hard swallow)

Until the software catches up, I'd be looking at other options.

 

[CAT-THE-FIFTH]

The main thing to watch is not the games but the engines they are on. Mantle has to be part of the engine IIRC,and many titles are launching with CryENGINE 3 and Frostbite 3 this year. Star Citizen uses Mantle.

 

[Gazb]

The original article here:
http://www.pcper.com/reviews/Genera...uer?utm_source=twitterfeed&utm_medium=twitter




This was always going to be the case.
It's going to take a long time for HSA to start getting coded into mainstream games and programs.

Mantle also needs HSA coding into it.
[snip]

Until the software catches up, I'd be looking at other options.

I thought mantle did have HSA stuff like this built into it already.
I mean AMD are very much focusing on APU's rather than raw CPU's, so it makes sense for them to have their mantle protocol be able to leverage the full ability of an APU even when it's being used with one of their dedicated GPU cards.

 

[tbyeah]

The new APUs just don't have the CPU power to be competitive other than in very small gaming PCs without room for a dedicated GPU.

 

[Legion]

I thought mantle did have HSA stuff like this built into it already.
I mean AMD are very much focusing on APU's rather than raw CPU's, so it makes sense for them to have their mantle protocol be able to leverage the full ability of an APU even when it's being used with one of their dedicated GPU cards.

Scroll down to the HSA part on the link, might explain it better than I'm trying to.
http://www.anandtech.com/show/7729/...rivers-now-available-mantle-frame-pacing-more

This will also help make things clearer:
http://www.extremetech.com/gaming/164817-setting-hsail-amd-cpu-gpu-cooperation

 

[joeyjojo]

Interesting, quite technical, article for those of us interested in compute performance on Kaveri relative to Haswell:

Floating point peak performance of Kaveri and other recent AMD and Intel chips

It seems Haswell is still the way to go for DP, but in SP Kaveri has the advantage over all but the GT3e Haswell chips (-R suffix, unobtanium) on Windows.

On Linux Kaveri is the only option as Intel will probably never release a OpenCL driver.

 

[humbug]

In HSA Kaveri destroys any Intel Chip.

But I don't get why AMD are seemingly ignoring FP performance. its as if they are pretending it does not exist.