A8-3850 Has Ineffective BClk Multiplier

[Maban]

http://www.pcper.com/news/Graphics-...eries-APU-Overclocking-and-Gaming-Performance

Of course it made a difference. They upped the base clock, not the multiplier.

 

[[H]@RD5TUFF]

Ok, is it just me or is this a bit erm, dumb? Why don't they just hard lock it if there is no point in increasing it anyway?

This is AMD we are talking about . .. and that would make sense . . . so obviously the opposite must be done. . . . . . ..

Joking aside I wonder the same thing myself.

 

[Sihastru]

LoL.

 

[cheesy999]

Why should they hard lock it? if it doesnt do anything

 

[JATownes]

Of course it made a difference. They upped the base clock, not the multiplier.

I just checked out too. That was all done with bclk, not multiplier. So, not relevant to this article.

 

[yogurt_21]

I thought we already knew this?

 

[TheLaughingMan]

I thought we already knew this?

We did.

 

[chew*]

This news is not entirely true

Almost 100% (repectively cpu test ) performance increase anyone?

http://hwbot.org/forum/showpost.php?p=120967&postcount=2

 

[Nesters]

Of course it made a difference. They upped the base clock, not the multiplier.

Which means that memory was OCed as well, increasing GPU performance on which these tests are based.

 

[qubit]

"Empty overclocking" - sounds like "empty calories". Don't give you nothin' and makes you fat.

Damn, this is a really stupid bug and one which could bring their products into disrepute, especially with uninformed and ignorant users and reviewers who will spread misinformation. Also, unsrcrupulous sellers will try to hoodwink clueless buyers.

Fix it quick, AMD and give Intel some competition for once.

 

[TheLaughingMan]

Which means that memory was OCed as well, increasing GPU performance on which these tests are based.

That depends. Some memory will not take a higher clock and its multiplier (which does work) may have been adjusted to get a higher bus.

 

[brandonwh64]

THEN OCED

 

[Steven B]

yea but you need a special BIOS that is not released yet, and prob wont be.

 

[chuchnit]

Brandonwh64 how about posting the source of those screenshots and give chew* credit for his work?

 

[chew*]

yea but you need a special BIOS that is not released yet, and prob wont be.

This is not achieved via some super secret bios

It can be with an option but this is all via registers and thus can be done in software.

My only concern is making it easy for the majority and it's in the works.

 

[Steven B]

ahh i see why not make an executable file that will overwrite the registry?

i am with you on that churchnit.

BTW i am sin0822, if you are wondering.

 

[chew*]

ahh i see why not make an executable file that will overwrite the registry?

i am with you on that churchnit.

BTW i am sin0822, if you are wondering.

Well here's how it works.

It changes the default multi string from 16x-47X ( IIRC thats the lowest to highest) to 32X-63X.

That is 31 possible multies. And the default is 29x which is 13 multis above the lowest.

Now if it was applied, the multipliers would shift to the high range but it would get applied once again to 13 multis above lowest which is 32X+13 = 45X so 4500 mhz which would certainly crash the system.

So that said it's a little more complicated as the range and lowest base multiplier both must be applied at same time.

32x aka 3200mhz should apply fine however even with no voltage applied.

There may be some other issues to however unless users purchased locked parts, that I can't help with.

 

[Steven B]

yea, but will a normal A3850 be able to do it?

that is pretty cool btw, i hope AMD takes note and makes it standard, it will help them with APU sales definitely. Multiplier OCing like sandybridge with the same high frequencies?

Honestly i really enjoyed playing with the APu system, but other reviewers hated it. I think they were just expecting to much.

 

[chew*]

yea, but will a normal A3850 be able to do it?

that is pretty cool btw, i hope AMD takes note and makes it standard, it will help them with APU sales definitely. Multiplier OCing like sandybridge with the same high frequencies?

Honestly i really enjoyed playing with the APu system, but other reviewers hated it. I think they were just expecting to much.

If a normal A3850 comes with an unlocked multi yes.

Not sure about hitting sandy speeds at the same temps but yes on the extreme side of things and with the process refined more who knows.

 

[Thefumigator]

Overclocking is a madness.

AMD Llano costs 130$ and something. while the Core i7 2600k costs around 300$ and can't even come close in terms of gaming performance. Don't get me wrong, we all know what a core i7 is, a great cpu. But in a laptop, I would take a llano over the other if I'm considering gaming and battery life.

But talking about the desktop llano, I would really love:

1-Black Edition
2-Six cores model. And why not, eight core model.
3-integrated Dual GPU (or bigger GPU, yeah I think it will work way better)
4-keep price under 200 for all the above.

 

[chew*]

Overclocking is a madness.

AMD Llano costs 130$ and something. while the Core i7 2600k costs around 300$ and can't even come close in terms of gaming performance. Don't get me wrong, we all know what a core i7 is, a great cpu. But in a laptop, I would take a llano over the other if I'm considering gaming and battery life.

But talking about the desktop llano, I would really love:

1-Black Edition
2-Six cores model. And why not, eight core model.
3-integrated Dual GPU (or bigger GPU, yeah I think it will work way better)
4-keep price under 200 for all the above.

1= feasible
2= not realistic especially due to TDP and heat.
3= not realistic part 1, and 2 the gpu is actually very strong, the problem is TDP.
4= N/A due to 2 and 3.

 

[btarunr]

This news is not entirely true

Almost 100% (repectively cpu test ) performance increase anyone?

http://hwbot.org/forum/showpost.php?p=120967&postcount=2

Engineering sample is a lot different. Llano ES chips which were circulated by AMD has an actual upwards unlocked BClk multipler, which could actually increase clock speeds.

The news article covers retail A8-3850, which do not (and are not supposed to) have an upwards unlocked multiplier. Even if the BIOS allows you to set multi above 29x, the chip doesn't actually respond to it, and keeps running at 29x.

So no, the news is entirely true.

 

[Steven B]

the retail A3850 i had could to upwards multipliers it would report to be at those multis but not preform at them. The fact that in CPU-Z it would show the higher clocks does give hope.

 

[btarunr]

the retail A3850 i had could to upwards multipliers it would report to be at those multis but not preform at them. The fact that in CPU-Z it would show the higher clocks does give hope.

CPU-Z is just telling you whatever the BIOS reports. Yet the CPU isn't actually running at what the BIOS reports to you.

 

[Steven B]

yea its not, all i am saying is that i have faith in chew, as he is the AMD king and what he says usually proves to be correct.