Sunday, October 7th 2012

A10-5800K Cracks 7.30 GHz in Overclocking Feat

Apparently, AMD wasn't exaggerating with its 6.50 GHz over LN2 claim for its A10-5800K APU. Overclocker GASBK_TW achieved 7317.74 MHz clock-speed for the A10-5800K, with a base clock of 118.03 MHz, multiplier of 62x, and a staggering 1.956V core voltage. The APU was cooled by liquid nitrogen. Other key components include Biostar Hi-Fi A85X motherboard, 4 GB of G.Skill DDR3-1600 MHz memory, and discrete Radeon HD 7700 series GPU. The CPUID validation can be found here.

Source: PC Games Hardware
Add your own comment

48 Comments on A10-5800K Cracks 7.30 GHz in Overclocking Feat

#4
repman244
by: Konceptz
Wow, thanks for this man...glad I hadn't purchased anything yet.
Seeing your CPU runs at 4.3GHz, an A10 would probably need around 5.1GHz to even match it.
Until AMD dramatically improves IPC (or increase the stock clocks even more but with same/lower power consumption) they can't even compete with their own Phenom II which is bad for all of us in the end.
Posted on Reply
#5
cdawall
where the hell are my stars
by: repman244
Seeing your CPU runs at 4.3GHz, an A10 would probably need around 5.1GHz to even match it.
Until AMD dramatically improves IPC (or increase the stock clocks even more but with same/lower power consumption) they can't even compete with their own Phenom II which is bad for all of us in the end.
You do understand the entire reasoning behind moving away from a monolithic core right?
Posted on Reply
#6
repman244
by: cdawall
You do understand the entire reasoning behind moving away from a monolithic core right?
Yes, and I even understand that there was nothing to improve about Phenom II (which is basically an improved K8). That is evident from Llano's CPU (32nm) side which gained almost nothing compared to Phenom II (45nm). A new design offers room for improvements.
It doesn't take away the fact that something went horribly wrong, and only time will tell if they are able to improve it (Steamroller/Excavator). Don't expect much from Piledriver since it doesn't include drastic changes in the design itself (like it's planned for Steamroller).
Posted on Reply
#7
Konceptz
by: cdawall
You do understand the entire reasoning behind moving away from a monolithic core right?
My main goal is to get off the ddr2/775 setup. After what I've seen I guess i'll wait for the FX-6300. Ideally I'd like to move to a 6 core setup. Its going to be AMD whatever I choose, I just can't justify Intel's pricing anymore, that and Im interested in seeing what and AMD platform can do, its been years since I've ran AMD.
Posted on Reply
#8
cdawall
where the hell are my stars
by: repman244
Yes, and I even understand that there was nothing to improve about Phenom II (which is basically an improved K8). That is evident from Llano's CPU (32nm) side which gained almost nothing compared to Phenom II (45nm). A new design offers room for improvements.
It doesn't take away the fact that something went horribly wrong, and only time will tell if they are able to improve it (Steamroller/Excavator). Don't expect much from Piledriver since it doesn't include drastic changes in the design itself (like it's planned for Steamroller).
The cores are completely and utterly different in design vs phenom II. In applications that take advantage of the new core structure we have seen massive improvements in performance. Llano is a modified Athlon II that for the most part outperformed Phenom II even though it lacked an L3 cache. The new Piledriver based units we see now in mobile and desktop PC's has almost zero related to Llano and from the looks of it very little in common with Piledriver minus the core hierarchy. The addition of the GPU core is quite a change up and the physical usage of resources is 100% different.

It's a good thing you understand the entire concept so well :slap:

by: Konceptz
My main goal is to get off the ddr2/775 setup. After what I've seen I guess i'll wait for the FX-6300. Ideally I'd like to move to a 6 core setup. Its going to be AMD whatever I choose, I just can't justify Intel's pricing anymore, that and Im interested in seeing what and AMD platform can do, its been years since I've ran AMD.
Honestly stock for stock the A10 would perform better but with both overclocked in single IPC the 775 setup will win. With proper coding and usage of AVX coding the AMD will win. It's a sideways move for you to go with an APU. The new piledriver based chips are a slightly different story in theory...
Posted on Reply
#9
trickson
OH, I have such a headache
by: BazookaJoe
Yes, OK, I get that it is cool that we have pushed silicon this far - but am I the only one who just does not care anymore about bu11$h!t overclocking feats that are accomplished under completely unsustainable and downright ridiculous conditions, that could not possibly be used outside of an overclocking contest and bear no fruit for any other application what so ever?

Again I get it - woo! something functionally useless ran really fast for a very short time in an utterly unsustainable environment, and does not in any way represent how the equipment will or ever even COULD actually perform in the real world.

Like land speed testing a rocket car - yay it went fast - it's still completely useless...

...Unless one day the key to stopping the alien invasion is moving a rely small object across a Nevada salt flat in 20 seconds...
Now this is just the plain truth. Big Deal can it play crysis! Useless! More FLUFF from AMD.
Posted on Reply
#10
repman244
by: cdawall
The cores are completely and utterly different in design vs phenom II. In applications that take advantage of the new core structure we have seen massive improvements in performance. Llano is a modified Athlon II that for the most part outperformed Phenom II even though it lacked an L3 cache.
Never said that Phenom II was the same as BD/PD :confused: (lot's of people wanted a 32nm Phenom II) so I pointed out that there is no use and a new design is the way to go (even if it cannot deliver for the first time) however if they don't manage to improve it then it's another thing.

But is that advantage only due to new instructions? Besides, I don't see that many applications that can use AVX right now (AVX2 should be more widely supported), you need to have a plan for the future but design for the present.
Example: I do a lot of rendering on my PC, and looking at what AMD has to offer I have no choice but to push on with my Thuban (4 module BD is a little better in rendering but at the expense of power consumption which is high as it is anyway).
But if a run an analysis for airflow or similar, some of those parts are running on one core and I know that is the weak spot for the new design, so why would I pay 250€+ for a new CPU and MB that runs slower - so like I said it's not ready and needs many improvements.

Llano does not outperform a Phenom II (in total (IPC+clock)). And AFAIK overclocking Llano isn't that good as overclocking a Phenom II.
http://www.anandtech.com/bench/Product/81?vs=399

You can see the IPC increse in Llano but when you put an x4 980BE against it the clock destroys it.
The new Piledriver based units we see now in mobile and desktop PC's has almost zero related to Llano and from the looks of it very little in common with Piledriver minus the core hierarchy. The addition of the GPU core is quite a change up and the physical usage of resources is 100% different.
Again, I never said it's the same and I understand the whole modules concept which enables easier scaling and adding components (like a GPU).
I see Llano as a test of the APU design, and future versions will only get better (GCN GPU, SR cores and shrink to 28nm...).
It's a good thing you understand the entire concept so well
I don't quite get what you were trying to say with this line, but we are only discussing here and AFAIK I didn't say anything wrong in my post.
Posted on Reply
#11
cdawall
where the hell are my stars
People who need and use avx have access wide spread support doesn't make a difference fact is it is there and works well.

As for llano it performs better clocker clock while lacking an l3 it's a huge overall departure from phenom.

Your original argument was bulldozer sucked. It doesn't it needs proper usage of its design to work. They are already getting looked at for the server market it works better than Intel's offerings in certain server apps thanks to the design.
Posted on Reply
#12
repman244
by: cdawall
Your original argument was bulldozer sucked. It doesn't it needs proper usage of its design to work. They are already getting looked at for the server market it works better than Intel's offerings in certain server apps thanks to the design.
Not really no, I stated they have a hard time competing with their own design (Phenom II) which is a fact. I'm not ignoring another fact which is that it does have a better IMC, more instructions and offers a fresh design to build upon.
But when I buy a new CPU I would expect it do be better in pretty much everything (at least on par with the old one in some things) and not sacrifice one workload to get a bit more performance in other.
Would you call a new car (with same HP) that does less km/l than the previous model a success? I wouldn't.

For server purposes BD does indeed have some sweet spots I just hope that PD lowers power consumption which is a critical factor in most servers (apart from high performance servers where you need as much performance as you can buy).

IMO if they managed to get stock clocks to 4.5/4.6GHz with same TDP it would be a great CPU. But a question arises here: what exactly is holding the clocks down, is it the silicon itself, design, or the manufacturing process itself.
Posted on Reply
#13
trickson
OH, I have such a headache
Fluff. Means nothing.
Posted on Reply
#14
3870x2
by: repman244
Not really no, I stated they have a hard time competing with their own design (Phenom II) which is a fact. I'm not ignoring another fact which is that it does have a better IMC, more instructions and offers a fresh design to build upon.
But when I buy a new CPU I would expect it do be better in pretty much everything (at least on par with the old one in some things) and not sacrifice one workload to get a bit more performance in other.
Would you call a new car (with same HP) that does less km/l than the previous model a success? I wouldn't.

For server purposes BD does indeed have some sweet spots I just hope that PD lowers power consumption which is a critical factor in most servers (apart from high performance servers where you need as much performance as you can buy).

IMO if they managed to get stock clocks to 4.5/4.6GHz with same TDP it would be a great CPU. But a question arises here: what exactly is holding the clocks down, is it the silicon itself, design, or the manufacturing process itself.
Your argument is flawed. It would be like better luxury features but less km/l, or faster but less km/l.
Posted on Reply
#15
repman244
by: 3870x2
Your argument is flawed. It would be like better luxury features but less km/l, or faster but less km/l.
Yeah, I realized that later that I should have included some good points as well so I stand corrected ;):)
Posted on Reply
#16
cdawall
where the hell are my stars
Pentium 4 was still hyped when it couldn't outperform p3 in any aspects...bulldozer shares some of that nonsense. I see monolithic core as a thing of the past this time a mere stepping stone.
Posted on Reply
#17
trickson
OH, I have such a headache
by: cdawall
Pentium 4 was still hyped when it couldn't outperform p3 in any aspects...bulldozer shares some of that nonsense. I see monolithic core as a thing of the past this time a mere stepping stone.
No the P4 was a good replacement for the P3. The Athlon CP's were better.
This is not about what CPU is better though. This is about an oc that is while cool still has no meaning at all in real time applications.
Posted on Reply
#18
eidairaman1
by: trickson
No the P4 was a good replacement for the P3. The Athlon CP's were better.
This is not about what CPU is better though. This is about an oc that is while cool still has no meaning at all in real time applications.
clock for clock the P3 beat the P4
Posted on Reply
#19
trickson
OH, I have such a headache
gay.
Posted on Reply
#20
cdawall
where the hell are my stars
by: trickson
No the P4 was a good replacement for the P3. The Athlon CP's were better.
This is not about what CPU is better though. This is about an oc that is while cool still has no meaning at all in real time applications.
No not even a little bit go look into the original 1.4ghz P4's and compare them directly to the 1.0-1.4ghz P3's and come back to post again. More so the Pentium M's that used the P3 architecture vs the netburst architecture destroyed desktop chips hence the implementation of socket 479 to socket 478 converters so people could use Banais and Dothan based chips.

by: eidairaman1
clock for clock the P3 beat the P4
Clock for clock almost everything beats P4.
Posted on Reply
#21
eidairaman1
by: cdawall
No not even a little bit go look into the original 1.4ghz P4's and compare them directly to the 1.0-1.4ghz P3's and come back to post again. More so the Pentium M's that used the P3 architecture vs the netburst architecture destroyed desktop chips hence the implementation of socket 479 to socket 478 converters so people could use Banais and Dothan based chips.



Clock for clock almost everything beats P4.
ive done a compare of the Intel atom to my Athlon XP machine and the Athlon Was faster. Hell my P4 laptop was faster than the Atom
Posted on Reply
#22
blibba
by: faramir
Not exactly an upgrade to move from QX9770 to A10-5800K, is it ? Unless of course you left your poor brother with HD5550 or a similar graphics card wannabe ;) Then again you mentioned he doesn't play games at all so I guess graphics is irrelevant.

The purchase makes sense for somebody who doesn't have a computer already or who has something completely obsolete (Pentium4 or something like that) but you'll be spending $300 to get exactly same level of performance ...
Depends how much the 9770 can be sold for.
Posted on Reply
#23
EpicShweetness
by: eidairaman1
ive done a compare of the Intel atom to my Athlon XP machine and the Athlon Was faster. Hell my P4 laptop was faster than the Atom
A Texas Instruments Graphing calculator has more power then a atom, my friend would even debate a toaster being more powerful :roll:
In all reality this overclock "record" crap has no implication, even at that clock as much as I try to support AMD, a Intel chip at 4GHZ would smear it's lunch all over it's face.
Posted on Reply
Add your own comment