AMD Trinity FM2 APU Preview

[cadaveca]

Today we've got an early look at what AMD's latest and greatest desktop products, namely AMD FM2 APUs. Soon to launch into the retail space, we take an early look at what's coming from AMD, in the form of the AMD A10-5800K processor.

Show full review

 

[TRWOV]

Excellent. I'll take a look at the A10s for some thin gaming PC.

So the GPU side of the APU has undergone much more drastic changes, now adopting the VLIW 4 shader design. Previous Llano chips saw 400 shader cores and a VLIW 5 shader design, while these new chips see just 384 shader cores. This change is very similar in going from a 5-series AMD Desktop GPU, to a 6-series AMD Desktop GPU. Also added is the ability to clock the GPU via AMD OverDrive, a feature that was sorely missing from previous FM1 parts.

VLIW4 was used only in Barts (HD6900 series). The rest of the 6-series were VLIW5.

 

[cadaveca]

VLIW4 was used only in Barts (HD6900 series). The rest of the 6-series were VLIW5.

..um Barts was 6800. Cayman is 6900's.

 

[TRWOV]

brain fart

 

[DannibusX]

Thanks for the preview, Dave. I look forward to the full review.

 

[mypg0306]

Thanks for the preview, Dave. I look forward to the full review.

Yes, great review. Waiting for the full review especially on A10-5800K in terms of gaming.

 

[HumanSmoke]

Strange that you're saddled with an A75 board and preview, yet Anand has an A85 and review...you forget to send Christmas cards to Sunnyvale ?

"AMD let us know that we could go ahead and post a preview"

Didn't happen to get the same email as Scott Wasson by any chance ?

 

[HammerON]

Thanks Dave for the preview
Look forward to the full reviews/

 

[qubit]

Nice preview. Look forward to AMD saying "yes" and letting you loose with the full review.

 

[dj-electric]

AMD will say yes to the whole wide world at the same second
OB 3D Performance seem great.

 

[Hustler]

Well given just how much the FM1 CPU's gaming scores increased when using faster memory, I'd expect a similar improvement with these FM2 ones, scores of 2000+ should be possible in 3dMark11

But...as usual for CPU limited tasks, these CPU's,whilst an improvement over Bulldozer, still suck. For a supposedly '4' core (lol) CPU, Intel's dual core i3 will still most likely outperform them.

And I still HATE the idea of not having any L3 Cache.

 

[BigMack70]

Wish you guys had held out and not played ball with AMD's nonsense!
http://techreport.com/blog/23638/amd-attempts-to-shape-review-content-with-staged-release-of-info

 

[Dent1]

Wish you guys had held out and not played ball with AMD's nonsense!
http://techreport.com/blog/23638/amd-attempts-to-shape-review-content-with-staged-release-of-info

Like if Intel hasn't been doing the same thing for YEEEARRRS.

 

[BigMack70]

Like if Intel hasn't been doing the same thing for YEEEARRRS.

If intel did this I'd hope the site would hold out from that as well... review sites shouldn't let companies hide the weaknesses of their product - half the point of the review is to find/expose weaknesses.

 

[W1zzard]

review sites shouldn't let companies hide the weaknesses of their product - half the point of the review is to find/expose weaknesses.

we did a preview today, giving you a rough idea about the product.

a proper review will follow later, you may choose to not read the preview and wait till the review is up

 

[BigMack70]

we did a preview today, giving you a rough idea about the product.

a proper review will follow later, you may choose to not read the preview and wait till the review is up

TPU comes before techreport on my daily routine of checking tech sites - already read the preview before I knew AMD was trying to hide their x86 performance .

I know you guys gotta do what you gotta do it just bugs me when companies try to unduly manipulate things in their favor.

 

[eidairaman1]

TPU comes before techreport on my daily routine of checking tech sites - already read the preview before I knew AMD was trying to hide their x86 performance .

I know you guys gotta do what you gotta do it just bugs me when companies try to unduly manipulate things in their favor.

Intel has been doing it for years

 

[BigMack70]

Intel has been doing it for years

How so (honest question)? I'm not as well versed in CPUs as GPUs so I haven't followed them as closely over the years, but I can't remember seeing reviews of Intel CPUs where Intel only allowed some types of tests and not others.

 

[Dent1]

How so (honest question)? I'm not as well versed in CPUs as GPUs so I haven't followed them as closely over the years, but I can't remember seeing reviews of Intel CPUs where Intel only allowed some types of tests and not others.

I've read the past, Nvidia and Intel would refuse to give reviewers engineering samples or access to samples in general if the review wasn't bias in their favour. I can remember one tech website was banned from the sample list for refusing to fix the review. I can't remember the name of the tech site. I'm sure someone here will remember and post a link or two.

Back in the day all the reviewer fixing we common. I can remember seeing Nvidia reviews where every page was a Phyx supported game. Obviously the reviewer knows that AMD products will be unfairly handicapped.

 

[BigMack70]

Yeah I knew that Nvidia had done some sleazy stuff like that in the past - didn't know Intel had done it. Was that during the time of their netburst debacle?

 

[DannibusX]

How so (honest question)? I'm not as well versed in CPUs as GPUs so I haven't followed them as closely over the years, but I can't remember seeing reviews of Intel CPUs where Intel only allowed some types of tests and not others.

That post at TR is red-herring. AMD is allowing some information about their product to be released early. The NDA's for the full product performance review are still in effect. They will not limit any information from being included on product reviews.

Wording is key here. AMD is allowing a "preview"; TPU is posting this information in accordance with AMD's wishes. They could have chosen not to. I'm glad they did.

It's no different then when information used to "leak" out of AMD. I'm pretty sure this practice isn't exclusive to AMD either.

 

[W1zzard]

banned from the sample list for refusing to fix the review

if that is true (which i dont know), then what did they do wrong by banning someone incompetent from working with their products, who'd otherwise give them a bad reputation.. especially if the author is not willing to fix his mistake?

if you come to me and say "look w1zz, x in your review is wrong, here is my data showing y. i suspect that z is happening" i'd be the first to start investigating, solving such puzzles is lots of fun, you'll learn something and if you can go to the vendor and tell him "here is my data, now i know what went wrong, actually the underlying issue is here, maybe look into this suggested improvement for the future", you will instantly gain tons of credibility with companies and they might send you stuff earlier than anyone else because they value your feedback.

 

[Mathragh]

Apart from all the buzz about it being a dirty move from AMD(which I think is really not a big deal at all, since everything is out in the open about the reasonings for this decision), I think this is actually a good sign. It shows that there is finally something happening in the AMD marketing department.

I'm not saying they should actually start manipulating the press, but compared to the Nvidia and Intel (and actually a lot of the other companies), they can use a bit more ingenuity and agression in the marketing department.

 

[Dent1]

if that is true (which i dont know), then what did they do wrong by banning someone incompetent from working with their products, who'd otherwise give them a bad reputation.. especially if the author is not willing to fix his mistake?

Sorry. I'm an English native so I forget that others here it's their second language.

When I said fix I meant "to rig" or "falsify" the reviews.

There was a few situations in the past a few reviewers confessed to Intel or Nvidia blackmailing them. Saying no samples if you don't falsify the benchmarks in their favour.

It's ashame I can't find any links, this is going back maybe 5-12 years so the Google links are removed and broken.

 

[jigar2speed]

If intel did this I'd hope the site would hold out from that as well... review sites shouldn't let companies hide the weaknesses of their product - half the point of the review is to find/expose weaknesses.

The author at TR is playing a bit of bias game here, he has done the preview for Intel -

http://techreport.com/review/9538/intel-conroe-performance-previewed/1

Plus they were not even given the permission to touch the system. :shadedshu

EDIT: Before you judge me, i am a long time reader at TR (2006) and have more than 4000+ post counts in their forum.