A10-7850K Performance Review

I am disappointed that the A8-7600 isn't released. I want to build a new HTPC and I was going to buy a 65 watt Kaveri chip and motherboard on day 1 if released and now it isn't. I have money from Christmas burning a hole in my pocket and I need a few other PC parts so I'll probably end up spending it on something else. Then I know what will happen; as soon as I do they will release the A8-7600 the next day.
 
Shark974 said:
(thank goodness I didn't listen to H forums and get a higher clocked dual core instead as they advised).
Click to expand...

That's what I did when I went I picked up the E8400 Wolfdale. Still regretting it, woulda been cool to still be pimpin' a Q6600. I also missed out on Nehalem (a platform I was waiting for since I first heard about it in '05). The E8400 was getting old for me around the beginning of Q2 in '11. I was holding out for the Bulldozer launch. I guess I can laugh about it now.
 
htpc_user said:
I am disappointed that the A8-7600 isn't released. I want to build a new HTPC and I was going to buy a 65 watt Kaveri chip and motherboard on day 1 if released and now it isn't. I have money from Christmas burning a hole in my pocket and I need a few other PC parts so I'll probably end up spending it on something else. Then I know what will happen; as soon as I do they will release the A8-7600 the next day.
Click to expand...

HotHardware did a review of the A8-7600;
http://hothardware.com/Reviews/AMD-Kaveri-Arrives-A87600-APU-Review/
The AMD A8-7600 will arrive at a price point of $119, sometime this quarter. The A10-7700K and A10-7850K should be available right away, however, at $152 and $173 price points, respectively. Though we can't speak definitively on the performance of the higher-end models yet, the A8-7600 seems to be priced aggressively. It is significantly less expensive than the Core i3-4330, despite offering competitive CPU performance in multi-threaded workloads and much better graphics performance.

Kaveri doesn't change the game for AMD today, but it is a major step forward for the company and lays the foundation for a number of future advances. If software developers get on board and leverage Kaveri's heterogeneous compute capabilities to their fullest potential though, then the future could be bright.
Click to expand...

2014-01-15_07-56-00.png
 
FordFanboy said:
That's what I did when I went I picked up the E8400 Wolfdale. Still regretting it, woulda been cool to still be pimpin' a Q6600. I also missed out on Nehalem (a platform I was waiting for since I first heard about it in '05). The E8400 was getting old for me around the beginning of Q2 in '11. I was holding out for the Bulldozer launch. I guess I can laugh about it now.
Click to expand...

Everyone makes mistakes - I replaced my Q6600 with X6 1055T instead of getting cheap xeon with HT for 1156 and then i waited for Bulldozer thinking it might be drop in replacement for my less that year old am3 mobo :D

At least I learned to never make future proof buys and to get what is best here and now.
 
NinjaGnat said:
Look, it's been 4 years. That's practically forever in tech. The fact we're even comparing the Deneb core with Kaveri should be recognized as the sham it is.
Click to expand...

If only it was meant to compete in the same market segment and it is not.....

wonderfield said:
From a total combined FLOPS/watt perspective, it's actually a good evolution. You just have to contort the hell out of things to get that to really show up in any performance test, and theoretical max FLOPS is even harder to hit.

The actual computational power is there, though. Whether that means anything to you or not is another matter.
Click to expand...


Sadly for games that are cpu bound you are right. However if EA starts using Mantle in Frostbite 3 there will be very few games not running like champ on a Kaveri.

Bur it is typical AMD before a good deal of software is using HSA,Huma and HQ it is launching a new cpu.
Luckily there is more to HSA then just AMD;
 
mesyn191 said:
Hahaha passive aggressive much.


Aaaand logical fallacy. All kinds of useless crap and even straight up scams exist. Just because its there doesn't mean its legit or useful.
Click to expand...

The other reviews aren't showing anything even close to a 20% increase in IPC either. Also Encoding isn't a synthetic benchmark.

Overall I find the release very disappointing since there is no software support now or even really on the horizon for HSA, etc...
 
TomGreen321 said:
Overall I find the release very disappointing since there is no software support now or even really on the horizon for HSA, etc...
Click to expand...

But you are wrong about this Libre office will use some of the features. Also this month will see the Mantle patch for BF4.

And in the near future EA will have Frostbite 3 version using Mantle this is rather good news.
 
Libre Office going HSA is fine and dandy, but the use case they show the acceleration in is, to be fair, niche. And for the 6 people in the world that A) Use complex-as-fuck tables B) Don't use MS OFFICE for it, Kaveri is good.

For this to truly take off, it has to be built into every (and I mean every) major compiler, which so far it is not. On Linux for example, if GCC can optimize for HSA, a vast many things might show massive gains. This could annihilate intel on server apps if for example, apache and SQL can leverage the massively parallel compute without any action from the devs.

But back to windows: There are no real HSA apps yet. That will take years, as devkits were not in the pipeline for long at all for HSA, and usable hardware is only now available to slot into dev pipelines and test rigs.

Mantle may well be a killer app for kaveri, especially on the <$100 end. If that A8 can hold it's own, ON IT'S OWN, in every major AAA game for the next few years, I don't see a need for dGPU's for a while, short of high res, high setting, game play.

For my uses, it'll be fantastic, and I'm waiting for 35w parts at this rate to get massive gains on my A10-5757M at the same power envelopes..

For my 6800K rig, I'll keep that, since Kaveri has no gains at that level *yet*.
 
AMD sells enough server chips that software designers should take notice and incorporate the fastest tech into their next batch of server software.
 
cageymaru said:
It would be nice if AMD sold enough server chips that software designers should take notice and incorporate the fastest tech into their next batch of server software.
Click to expand...

Fixed it for you. AMD's server market share has sadly been declining steadily. Maybe HSA can change that, but as it is right now they don't have the weight to throw around anymore. Hell, if they were back into double digits of the market share that would be a huge improvement.
 
there will be low power server variants of Kaveri. really hopeful that OEMs will be getting them
 
omega-x said:
Libre Office going HSA is fine and dandy, but the use case they show the acceleration in is, to be fair, niche. And for the 6 people in the world that A) Use complex-as-fuck tables B) Don't use MS OFFICE for it, Kaveri is good.

For this to truly take off, it has to be built into every (and I mean every) major compiler, which so far it is not. On Linux for example, if GCC can optimize for HSA, a vast many things might show massive gains. This could annihilate intel on server apps if for example, apache and SQL can leverage the massively parallel compute without any action from the devs.

But back to windows: There are no real HSA apps yet. That will take years, as devkits were not in the pipeline for long at all for HSA, and usable hardware is only now available to slot into dev pipelines and test rigs.

Mantle may well be a killer app for kaveri, especially on the <$100 end. If that A8 can hold it's own, ON IT'S OWN, in every major AAA game for the next few years, I don't see a need for dGPU's for a while, short of high res, high setting, game play.

For my uses, it'll be fantastic, and I'm waiting for 35w parts at this rate to get massive gains on my A10-5757M at the same power envelopes..

For my 6800K rig, I'll keep that, since Kaveri has no gains at that level *yet*.
Click to expand...

To be fair either you are a computer user or someone that knows how to flip the on and off switch.

Should I care what people use if they are stuck in the past with their software products you could argue as well why they would need a new apu also.

Reality is that there just a "few" that know what software features,do people don't even care as long as it runs the program, is it taking 10 seconds longer but the computer is in the colour they like , you guessed it already......

You are mistaken about HSA HSA is not platform bound it is used and supported by many of the industry, will it take time before we see more of it of course but all of the companies on board are very serious about this.
 
Last edited:
I might take back what I said about potential 35W gains.

power-peak.gif


45W T parts vs the new "45" watt part.
 
omega-x said:
I might take back what I said about potential 35W gains.

power-peak.gif


45W T parts vs the new "45" watt part.
Click to expand...

But 33.7% faster.

x264.gif


Also, hold long was peak power held.

power.gif


Not saying good or bad, just that ya need more info!
 
It's clearly violating peak TDP, FX-style, which will matter a lot on mobile devices.
 
It'd be better to see the numbers during a burn-in test than a race-to-idle test, but still.
 
Michaelius said:
Everyone makes mistakes - I replaced my Q6600 with X6 1055T instead of getting cheap xeon with HT for 1156 and then i waited for Bulldozer thinking it might be drop in replacement for my less that year old am3 mobo :D

At least I learned to never make future proof buys and to get what is best here and now.
Click to expand...

Yeah things happen.

I had a LGA1366 Shuttle SFF case with a Core i7-920 in it. Good CPU but I was chafing at the lack of expansioin in that little case., and the motherboard was built in in that model and couldn't be replaced.

I learned my lesson about not going SFF unless you absolutely have to as future expanson needs always seem to grow more than you'd expect.

When I finally couldn't wait any longer I sold it here on the hard forums, and built a rig with a Phenom II x6 1090T on an AM3+ motherboard as soon as they came out, planning to sell the 1090T and get a bulldozer on launch...

When the bulldozer reviews were so abysmal, I just said eff it, and picked up a LGA2011 Core i7-3930k and it has been fantastic . Kicking ass for more than two years now, and nowhere close to needing to be replaced. It was expensive, but well worth the investment IMHO.
 
xorbe said:
But 33.7% faster.

x264.gif


Also, hold long was peak power held.

power.gif


Not saying good or bad, just that ya need more info!
Click to expand...

Hmm...

I had planned on getting a quiet Corsair H55 to cool my A10-7850k when I pick it up, but judging by these charts, I am starting to wonder if it will be enough...
 
omega-x said:
It's clearly violating peak TDP, FX-style, which will matter a lot on mobile devices.
Click to expand...

Mmm yeah for a 45W part that seems a bit hungry, but it is 80W "system power consumption" whatever that means. If you account for psu efficiency and dual channel ram, it's not a off as one might think. A lot of cpus don't touch their tdp, but I bet this one is sitting 45W on the nose.
 
Last edited:
omega-x said:
It's clearly violating peak TDP, FX-style, which will matter a lot on mobile devices.
Click to expand...

Possibly, but I'd imagine that power management algorithms will be optimized very different on Mobile parts as compared to desktop parts, so judging one by the other is probably not very telling.
 
xorbe said:
Mmm yeah for a 45W part that seems a bit hungry, but it is 80W "system power consumption" whatever that means. If you account for psu efficiency and dual channel ram, it's not a off as one might think. A lot of cpus don't touch their tdp, but I bet this one is sitting 45W on the nose.
Click to expand...

I'm not so sure, if the 45W richland is on the same test-bed with dual channel ram and other components being the same. Unless richland simply isn't using the entire TDP.

It does need more science but it is not looking "great" as such, for the same TDP labeling.
 
Isn't the 45w A8 just a downclocked version of the 65w part? I wouldn't be too surprised if it wasn't clicking at exactly 45w for all workloads.
 
omega-x said:
Unless richland simply isn't using the entire TDP.
Click to expand...

Yeah that's what I was pondering.

NinjaGnat said:
Isn't the 45w A8 just a downclocked version of the 65w part? I wouldn't be too surprised if it wasn't clicking at exactly 45w for all workloads.
Click to expand...

I just noticed today on cpu-world.com that different A parts have different MHz/V steps, so it's not quite just a down-clock (possibly).
 
cageymaru said:
AMD sells enough server chips that software designers should take notice and incorporate the fastest tech into their next batch of server software.
Click to expand...
AMD is down to less than 4.5% of the x86 server market, and has a slightly higher share than that in HPC. The market share or whether developers will support it isn't what's important, it's whether those chips were chosen for a particular application and if workloads would benefit from custom programming.
 
good.

i have an asus dcII 7790 i bought for £80 on standby for a kaveri htpc.
 
cheers, but i think that combo will struggle enough to run 1080p without adding AA to the mix.
 
You must log in or register to reply here.
Back
Top