First Socket FM2 APUs Spotted in Russian Store

[btarunr]

Ahead of their October 1 launch, upcoming second-generation AMD A-Series APUs are beginning to surface on retailers. Russian online store Regard.ru listed most of these chips, albeit in their OEM (chip-only) packaging. According to the prices at hand, the quad-core A10-5800K and A10-5700 are listed for 4,130 RUB (US $132.6). The A8-5600K and A8-5500 quad-core APUs go for 3,540 RUB (US $114). The cheapest of the lot, dual-core A4-5300 is priced at 1,940 RUB (US $62.2).



Source: Overclockers Ukraine

[mediasorcerer]

Thank's for the update, these apu,s that amd are making quite fascinate me, i nearly went that route for this latest build, with the thought of adding a dedicated gpu later etc.

Perfect for the budget lounge room media machine really, and from what ive read, although in the last release the cpu performance side of things could be approximated to one third the power of a core i5, the igpu was quite a bit better than the intel hd4000, so for the prices, i think they offer good price/perf characteristics for those who haven't big $ to spend!

I get the feeling down the track these chips will excell, and i hope they do, some even oc!

[eidairaman1]

not enough FM2 boards out

[Atom_Anti]

Originally Posted by mediasorcerer

the igpu was quite a bit better than the intel hd4000, so for the prices, i think they offer good price/perf characteristics for those who haven't big $ to spend!

The GPU performance is quite a lot better than HD4000, even the weak A6 mobile version outperform easily in most of the game titles. And I guess these desktop versions will have faster GPU than mobile versions....

[NC37]

Originally Posted by eidairaman1

not enough FM2 boards out

Exactly. Another reason I say Trinity has been a lackluster release. For months there was less than a handful of laptop makers using it. So-so configurations. High prices....Llano was adopted much faster than this and it had production issues. I finally went with an i7 and nVidia. Got tired of waiting for that inexpensive Crossfire Laptop like I saw with Llano.

Just hope desktop isn't going to sit around with as few board makers for months. Competition is good!

[ramcoza]

Originally Posted by Atom_Anti

The GPU performance is quite a lot better than HD4000, even the weak A6 mobile version outperform easily in most of the game titles. And I guess these desktop versions will have faster GPU than mobile versions....

Actually A6 5400K is slower than Last gen A8 3850 in games.
http://www.tomshardware.com/reviews/...k,3224-17.html
http://www.tomshardware.com/reviews/...k,3224-15.html

[Dent1]

Originally Posted by ramcoza

Actually A6 5400K is slower than Last gen A8 3850 in games.
http://www.tomshardware.com/reviews/...k,3224-17.html
http://www.tomshardware.com/reviews/...k,3224-15.html

That is expected. The Trinity A6 5400K is a dual core processor. The Llano A8 3850 is a quad core processor.

[Atom_Anti]

Originally Posted by NC37

Llano was adopted much faster than this and it had production issues. I finally went with an i7 and nVidia. Got tired of waiting for that inexpensive Crossfire Laptop like I saw with Llano.

During the Llano time laptop makers did not enjoyed high priority over desktops, therefore chips could go earlier into desktops. Trinity's production yield is good since the beginning, however the supply still not enough for laptop makers. Some manufacturer like Dell did not even get Trinity chips yet. How could AMD supply desktops than?

Originally Posted by ramcoza

Actually A6 5400K is slower than Last gen A8 3850 in games.
http://www.tomshardware.com/reviews/...k,3224-17.html
http://www.tomshardware.com/reviews/...k,3224-15.html

And the old mobile A8-3550MX also faster than the new A6-4400M. However I talked about Intel HD4000...

[Sampsa]

€ prices are listed on Finnish price comparison engine: http://hintavertailu.muropaketti.com

Muropaketti news post:

http://translate.google.com/translat...ntavertailussa (Google Translate=

AMD A10-5800K: 130,25 €
AMD A8-5600K: 108,25 €
AMD A6-5400K: 72,05 €
AMD A5-5300: 58,15 €

Usually 10-20 € cheaper in Germany

[Aquinus]

Can't wait to see the performance of the final product. I'm not expecting it to break records but I bet it's going to be pretty cost effective while still giving you a lot.

[Dent1]

Originally Posted by Atom_Anti

During the Llano time laptop makers did not enjoyed high priority over desktops, therefore chips could go earlier into desktops. Trinity's production yield is good since the beginning, however the supply still not enough for laptop makers. Some manufacturer like Dell did not even get Trinity chips yet. How could AMD supply desktops than?



And the old mobile A8-3550MX also faster than the new A6-4400M. However I talked about Intel HD4000...

It isn't clock for clock faster. What you are seeing is dual core vs quad core.

The Trinity A6-4400M is a dual core, the is Llano A6-4400M is a quad core.

[hat]

Originally Posted by mediasorcerer

Thank's for the update, these apu,s that amd are making quite fascinate me, i nearly went that route for this latest build, with the thought of adding a dedicated gpu later etc.

Perfect for the budget lounge room media machine really, and from what ive read, although in the last release the cpu performance side of things could be approximated to one third the power of a core i5, the igpu was quite a bit better than the intel hd4000, so for the prices, i think they offer good price/perf characteristics for those who haven't big $ to spend!

I get the feeling down the track these chips will excell, and i hope they do, some even oc!

One third? What is this, a stock FM2 cpu compared to an overclocked 3570k?

[ramcoza]

Originally Posted by Dent1

That is expected. The Trinity A6 5400K is a dual core processor. The Llano A8 3850 is a quad core processor.

But it's an IGP comparison isn't it? So dual core or quad core is out of argument. That's why I mentioned "In Games".

[jrs3000]

Originally Posted by ramcoza

But it's an IGP comparison isn't it? So dual core or quad core is out of argument. That's why I mentioned "In Games".

Dual or quad is in the argument. The cpu is holding the gpu due to the fact that it's dual. If they took the same igp and used it in a quad it is stronger. If you want to compare it right though then dual vs dual would be llano a4. Trinity a6 not equal to llano a6.

[faramir]

Originally Posted by ramcoza

But it's an IGP comparison isn't it? So dual core or quad core is out of argument. That's why I mentioned "In Games".

There are few (if any at all) 3D-heavy games that will not bottleneck such a low-end CPU. Games that don't employ 3D will be even worse off, obviously, as CPU limitations kick in even harder.

Regarding Trinity performance: think of Trinity "modules" as a single hyper-threaded core. A "quad-core" (two "module") Trinity such as A10-5800K should perform along the lines of Intel's i3 Sandy or Ivy bridge chips (due to come out soon) in multithreaded tasks, lose in single-threaded tasks (only Crapple still does the single-threaded shit in 2012 ...) and win in iGPU-based gaming, while costing about the same.

As most people don't need *that much* single-threaded performance to notice any difference they will be better off with a Trinity chip than with i3 Sandy?ivy Bridge (both of which are a dead end with socket 1155, while Trinity's successor Kaveri *might* still be FM2 compatible).

[Atom_Anti]

Originally Posted by Dent1

It isn't clock for clock faster. What you are seeing is dual core vs quad core.

The Trinity A6-4400M is a dual core, the is Llano A6-4400M is a quad core.

Oh, I'm not talking about CPU performance, but GPU. Trinity A6-4400M is significantly faster in games than the good old Llano A6-3400M. Of course (especially an overclocked) A6-3400M was much faster for CPU part...

[Dent1]

Originally Posted by ramcoza

But it's an IGP comparison isn't it? So dual core or quad core is out of argument. That's why I mentioned "In Games".

An IGP still require CPU horsepower to drive it.

Generally speaking if the CPU isn't powerful enough it will bottleneck the IGP or even a dedicated GPU.

The Llano A8-3550MX makes up for having a slower CPU by having two additional cores, thus driving its IGP more efficiently and performing better in modern muilti-threaded games.

Originally Posted by Atom_Anti

A6-3400M was much faster for CPU part...

I'm not sure that is accurate. The only way you could measure which CPU is faster is to run them both at the same clock speed and disable two cores on the A6-3400M. But I haven't seen any review do that.

[Atom_Anti]

Originally Posted by Dent1

I'm not sure that is accurate. The only way you could measure which CPU is faster is to run them both at the same clock speed and disable two cores on the A6-3400M. But I haven't seen any review do that.

To find out clock to clock, core to core speed, I've tested A10-4600M Trinity against 2.4 GHz A8-3550MX, but Llano was still faster for CPU cores. Fortunately the much faster GPU really makes up this difference and overall Trinity has a big lead in gaming. You are right with saying CPU have to feed GPU and this is well visible in GTA4. Please see my tests in my signature.

[Atom_Anti]

Originally Posted by Dent1

Then how come almost every non-gaming review shows the Trinity outperforming the Llano in all areas from encoding, file compression, general productivity etc?

The review in your signature shows gaming tasks which isn't a pragmatic way to isolate just CPU performance.

That is because they've used a stock clocked A8-3500M at 1.5GHz, while I've used an overclocked A8-3550MX at 2.4GHz. Note, even the 1.4GHz A6-3400M were able easily run at 2.4GHz and sometimes on higher speed without heating too much. And I've also have CPU tests in my review.

[Dent1]

Originally Posted by Atom_Anti

but Llano was still faster for CPU cores.

Then how come almost every non-gaming review shows the 4 core Trinity outperforming the 4 core Llano in all areas from encoding, file compression, general productivity etc?

The review in your signature shows primarily gaming tasks which isn't a pragmatic way to isolate just CPU performance.

Even in your review, the only two non-gaming tasks are handbrake and Winzip which basically shows a 2 core Trinity vs a 4 core Llano, Trinity is unfairly handicapped as it's only a dual core. Your own review confirms this.

The review in Atom_Anti signature:

Handbrake use all available CPU cores, therefore not surprising quad Core APUs are much faster than twin core A6. No OpenCL acceleration yet!

Again show me a review of a 4 core Trinity vs a 4 core Llano and lets see whom is faster

[Atom_Anti]

Originally Posted by Dent1

Even in your review, the only two non-gaming tasks are handbrake and Winzip which basically shows a 2 core Trinity vs a 4 core Llano, Trinity is unfairly handicapped as it's only a dual core. Your own review confirms this.

The review in Atom_Anti signature:

Again show me a review of a 4 core Trinity vs a 4 core Llano and lets see whom is faster

There is something misunderstanding from you, because A10-4600M has two FPU and four integer cores which works like 4 core CPU. Do not mix this up with Intel's Hyper Threading, two different thing!

[Dent1]

Originally Posted by Atom_Anti

There is something misunderstanding from you, because A10-4600M has two FPU and four integer cores which works like 4 core CPU. Do not mix this up with Intel's Hyper Threading, two different thing!

As much as I like independent reviews (sometimes better than commercial ones) and I appreciate the time it took to conduct the tests and do a write up. One thing which your review lacks is a decent sample of tests. You only have Handbrake and WinZip so it makes it impossible to conclude anything. If you had 10 or 20 non-gaming benchmarks we can look at trends. But we can't trend on 2 tests, so nothing can be determined. The review is actually void.

Every other commercial reviewer from Tom's Hardware, Hexus, Trusted Reviewer, Hot Hardware, Legit review and many others all show the Trinity 4 core outperforming the Llano 4 core both in gaming and non gaming tasks. They all have 10+ benchmarks unfortunately I'm more inclined to take their results more seriously.

If you are willing to review the desktop Trinity on release and are able to give us 10+ bechmarks (gaming and non-gaming) I will be interested to read it.

[Atom_Anti]

What kind of CPU tests would you desire, which are non-synthetics and commonly used CPU-task by people?

[cadaveca]

Originally Posted by Atom_Anti

What kind of CPU tests would you desire, which are non-synthetics and commonly used CPU-task by people?

Take any CPU limited-game?

CPU is made up of CPU core, Cache, NorthBridge, and GPU.


So there are four parts, and you need to find tests that shows differences for each part. Why does each part matter? That is what your tests should show.

I run on average 4-8 benchmarks on top of the normal benchmarks that I am constantly posting in my reviews, and I use these other benchmarks simply for the purpose of to see if they fit in reviews, and convey what I want to show. They change ALL THE TIME.

Iguess what I'm saying is that asking what benchmarks people want to see may not exactly fit the bill...you may under up running a bunch of useless benchmarks that show no differences at all...

[Dent1]

Originally Posted by Atom_Anti

What kind of CPU tests would you desire, which are non-synthetics and commonly used CPU-task by people?

For me the most important things are:

All CPUs reviewed are at the same clock speed and the same number of cores.


For CPU-dependant synthetics, I like to see a varied range. Between 10-20 different benchmarks. Wprime, Cinebench, Pass Mark Vantage: Productivity score, 7Zip, WinRar, TrueCrypt, x264 HD, Handbrake. Just to name a few. There are many more good ones.

Another important thing is to do each benchmark twice. Once in single threaded mode and once in multi threaded mode. (usually in the options you can customise the threads/cores utility)

For testing the gaming performance of the APU, it's going to be difficult to separate the affect the CPU is having on the GPU by benchmarking just in game FPS. To isolate the GPU portion use 3D Mark 2011 and Heaven DX11 - noting to only take the GPU score (not CPU, physic or overall score).