AMD Trinity news (looks, its near to launch?)

[informal]

And the reviewers still compare 5xxxK to CPUs that are clearly way above its price segment...

[tbone8ty]

One more review and I'm going to sleep - this time is GB mobo but with Trinity onboard:

http://www.back2gaming.com/review/gi...4-motherboard/

fix yur unigine heaven 3.0 bar chart i think is should read (higher is better) instead of (lower is better)

[Maxforces]

tbone8ty review isn't his

[tbone8ty]

tbone8ty review isn't his

lol ok soory

but am i right chart is wrong?

[Maxforces]

yep chart is wrong

[tbone8ty]

http://www.xbitlabs.com/picture/?src...-4500-1086.png

[god_43]

more reviews.


http://www.guru3d.com/articles_pages...iew_apu,1.html

http://hexus.net/tech/reviews/cpu/46...k-trinity-apu/

[SKYMTL]

Bottom line: Trinity has serious issues with resource allocation when Turbo CORE is enabled.

[tbone8ty]

Bottom line: Trinity has serious issues with resource allocation when Turbo CORE is enabled.

link to example possible? just curious what you mean by that...

like this?

http://www.legitreviews.com/article/2047/19/

resource sharing?

[tbone8ty]

http://vr-zone.com/articles/amd-trin...ht-/17271.html

nice and quick overview

hopefully jim keller of amd is looking at this and taking a red pen to their arch

[god_43]

Bottom line: Trinity has serious issues with resource allocation when Turbo CORE is enabled.

if review sites would use windows 8 i dont think it would have those issues. are they using cinebench r14 (has support for new cpus)? also why is everyone comparing this to i7's and high end i5's. this is an entry level cpu, its targeted between i3 and i5. i hope we start seeing comparisons between windows 8 and windows 7 for apu benches. fact is, by the end of the month, the only thing you can buy will be windows 8.

[SKYMTL]

link to example possible? just curious what you mean by that...

Sure. Basically, under some load situations when all cores are loaded, the architecture struggles against the Llano APUs:

http://www.legitreviews.com/article/2047/14/

http://www.pcper.com/reviews/Process...D-Mark-Vantage

I have spent hours diagnosing it. At first I thought it was the mobo, memory, etc. Turns out that certain situations cause Trinity's architecture to really struggle, even against the Llano APUs. Usually it entail tests that put a high amount of stress upon the memory controller. I'm still going back and forth with AMD about it but they have confirmed that some applications and synthetic benchmarks will see Trinity boasting less x86 performance than the A8 3870K and A8 3850.

[SKYMTL]

if review sites would use windows 8 i dont think it would have those issues. are they using cinebench r14 (has support for new cpus)? also why is everyone comparing this to i7's and high end i5's. this is an entry level cpu, its targeted between i3 and i5. i hope we start seeing comparisons between windows 8 and windows 7 for apu benches. fact is, by the end of the month, the only thing you can buy will be windows 8.

And exactly what huge difference will Win 8 make? This sounds like the excuses so many people were making when BD was launched: "But, but, Windows doesn't have the proper support for it!!". Look how that turned out.

And plus, Cinebench r14 isn't even publicly available.

[god_43]

chinese vr-zone.

http://translate.google.com/translat...ew-10022012%2F

[god_43]

And exactly what huge difference will Win 8 make? This sounds like the excuses so many people were making when BD was launched: "But, but, Windows doesn't have the proper support for it!!". Look how that turned out.

And plus, Cinebench r14 isn't even publicly available.

your comment was about resource utilization, windows 8 properly manages bulldozer resources. huh well there were a bunch of ppl in the amd forums using it, i just assumed it was out. ok how did it turn out? i have not seen bulldozer benched on windows 8? link to benches please? i know my bd runs much better with windows 8, this is what the review sites should figure out, is there a difference? if so by how much? if not then the issue can be put to rest.

[jjj]

@ SKYMTL: Llano had "proper" cores not modules,can't it be just that (yet to see in what types of workloads it struggles but you should be able to figure it out).

[god_43]

http://www.anandtech.com/show/6347/a...desktop-part-2


good review by anand.

[R101]

Sure. Basically, under some load situations when all cores are loaded, the architecture struggles against the Llano APUs:

http://www.legitreviews.com/article/2047/14/

http://www.pcper.com/reviews/Process...D-Mark-Vantage

I have spent hours diagnosing it. At first I thought it was the mobo, memory, etc. Turns out that certain situations cause Trinity's architecture to really struggle, even against the Llano APUs. Usually it entail tests that put a high amount of stress upon the memory controller. I'm still going back and forth with AMD about it but they have confirmed that some applications and synthetic benchmarks will see Trinity boasting less x86 performance than the A8 3870K and A8 3850.

As repeated time and time again - Bulldozer's two-thread module has the 160% performance of single thread in said module. The cinebench single/multithread result is almost perfectly in line with the info we have had for years now.

Only optimization that windows can do is use one thread per module until it runs out of modules..

[lizardmech]

What I don't understand with bulldozer modules is the size of them. Why did they think it would be ok when a bulldozer module is roughly 2x the size of a single core. If a module is 200% as big but only offers 160-180% the performance of 2 seperate cores how could it ever work out well?

[informal]

First of all it's obviously not 2x the size of single core( since you have one FP unit shared' this alone means that in other case,replicating FP unit's hw and accompanying logic would make "true" bulldozer dual/quad core considerably larger). In most of the tests ,MT performance suffers from the shared floating point unit and slowish L1D cache design. AMD said they are working on these issues with SR core and we have to wait and see what comes out of that. When they designed BD's FP unit they were relying on recompiled code(FMA) in order to utilize the max potential of it. For desktop,unfortunately, this won't happen so the legacy performance is subpar in some occasions.

[EniGmA1987]

CPU's and boards in stock at Newegg

[Particle]

No ITX boards in stock anywhere yet that I see. Someone post if you see one.

[tbone8ty]

http://www.guru3d.com/articles_pages...ew_apu,19.html

Sata performance looks good!

[demonkevy666]

Sure. Basically, under some load situations when all cores are loaded, the architecture struggles against the Llano APUs:

http://www.legitreviews.com/article/2047/14/

http://www.pcper.com/reviews/Process...D-Mark-Vantage

I have spent hours diagnosing it. At first I thought it was the mobo, memory, etc. Turns out that certain situations cause Trinity's architecture to really struggle, even against the Llano APUs. Usually it entail tests that put a high amount of stress upon the memory controller. I'm still going back and forth with AMD about it but they have confirmed that some applications and synthetic benchmarks will see Trinity boasting less x86 performance than the A8 3870K and A8 3850.

some of it is memory because any cache miss in the Bulldozer architecture is told to go to main memory for a miss.

if you want to see why it struggles then I suggest you use Intelburntest While running cinebench 11.5r singles thread.
Set affinity to one core in taskmanger, and set IntelBurnTest to 1 thread and one in affinity taskmanger too. Make sure they're set to run in [b]one module[.b]

see The Modules can Work as "Dual cores" or as single cores When running as "dual cores" they effectively drop from 4 way decoders to 2 way decoders. 33% less single thread for both cores. that's what I usually see with my FX.

[FlanK3r]

So, there are all reviews, what I read . I think, tomorow will be here next 10-20...

http://www.computerbase.de/artikel/p...e-im-cpu-test/
http://www.anandtech.com/show/6347/a...desktop-part-2
http://translate.google.com/translat...ew-10022012%2F
http://www.overclockers.ru/lab/49852...K_Trinity.html
http://www.hartware.de/review_1534.html
http://www.guru3d.com/articles_pages...iew_apu,1.html
http://pctuning.tyden.cz/hardware/pr...-mereni-vykonu
http://hexus.net/tech/reviews/cpu/46...k-trinity-apu/
http://www.legitreviews.com/article/2047/1/
http://www.pcper.com/reviews/Process...0K-Break-Cover 4.4 GHz
http://www.hardwareheaven.com/review...roduction.html
http://www.hitechlegion.com/reviews/...-amd-a10-5800k
http://hothardware.com/Reviews/AMD-A...U-Performance/
http://www.svethardware.cz/art_doc-E...600561F71.html
http://uk.hardware.info/reviews/3202...y-for-desktops
http://www.xbitlabs.com/articles/cpu...a10-5800k.html
http://www.pureoverclock.com/Review-...-5800k-review/
http://www.tomshardware.com/reviews/...ency,3315.html
http://pclab.pl/art51168.html
http://www.planet3dnow.de/cgi-bin/ne...?id=1349152485
http://www.hardware.fr/articles/878-...-deuxieme.html
http://www.hardwarecanucks.com/news/...es-apu-launch/
http://www.bjorn3d.com/2012/10/amd-v.../#.UGrlzlH6nAE
http://www.techpowerup.com/reviews/AMD/FM2_APU_Review/
http://extrahardware.cnews.cz/testy/...-vs-ivy-bridge
http://www.vortez.net/articles_pages..._review,1.html
http://www.pcworld.fr/processeur/tes...2,532041,1.htm