AMD CPU speculation... and expert conjecture

Page 167 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

Oh well, we just gotta adapt to the new world of LOOKATMYSUPEREXTRAAWESOME4770KAT5GHZWITHA$500WATERBLOCK
PS: Of all things, you got the 4200+ :ouimaitre: I had to resort to the 3800+@2.4 with the abit AN8 32X (Has SLI stuck in my mind) and a 7900GT, with the ZALMAN CNPS9500.
 


Nah, my computer still sits in a giant beige case and I can hear the fans all the way up the stairs if it is really working. I also have an original 1987 IBM Model M clicky keyboard with the PS/2 detachable cord. It's not terribly fast as it only has low-midrange CPU and 16 GB of RAM. But it is EXTREMELY reliable. That's the one thing about this machine that is very not old school- it simply runs like a tank. The only thing I did to it in three years besides blow the dust out of it is trade off the initial pair of CPUs which lacked Turbo CORE to a buddy for a newer single CPU which did have it, because the Magny-Cours CPUs were SO slow in anything not massively multithreaded. I used to have to tweak with machines much more often back in the day but not this one. I hadn't even restarted it in over a year until I lost power during some storms about a month ago. 380 days of uptime- not too shabby. I'll eventually make it a beast with 128-256 GB of RAM, four 3+ GHz Opteron 6300s (probably 6328s or 6384s), 2-3 4K monitors, and one or two really nice GPUs to round it out. That will be a big improvement over my 16 GB of RAM, one 6234, a very old GTS250, a Dell 2001FP that's starting to go out, and an odd duck of a Dell SP2309W with its "we thought it would take off" 2048x1152 resolution. But that won't happen until those parts become cheap on eBay or I pay off my mortgage of student loans- I'd bet on the eBay route myself.

PS: Of all things, you got the 4200+ :ouimaitre: I had to resort to the 3800+@2.4 with the AN8 32X and a 7900GT, with the ZALMAN CNPS9500.

The 7900GT was a very worthy adversary to the x1900GT. I remember the CNPS9500 being one of the better heatsinks. The AN8 32X was a better board than my KN8 SLI as the AN8 had a full 16+16 SLI whereas my KN8 SLI had an 8+8. I would have loved to have two real PCIe x16 ports because my GPU didn't fit really well in the lower slot which must be filled with a GPU or the shadow card. Gee, that's another thing from the past, shadow cards. Ranks right up there with the continuity RIMMs in i820/i850 chipsets and SCSI terminators in the obsolete tech file. Folks today would probably laugh if you had to put a shadow card in a PCIe slot to allow the other slot to turn on.
 
Heh... The good times... Athlon64 X2 4400+ (Toledo) @2.6Ghz here with the DFI LanParty nF4 SLI-D, 4x512MB DDR1 400 C2 and a X1800XL (which I still have in my closet). Actually, the GeminII used went to another build, using an FX4300 for a friend of mine; it keeps it at 4.3Ghz under 55°c. And the Athlon64 was given away to charity with the MoBo and RAM. That was my second and my first was the PIII 850 @1.1Ghz and 512MB SDR-100 on the 440BX, plus a Radeon8500 (from the TNT2 Vanta); man that was a rock solid build.

Anyway, as much as I'd like to recall those good times, I'd say we still have better times ahead of us... At least, I want to think about it that way, haha.

Come on AMD. Make a comeback and make this fight more interesting.

Cheers!
 

Cazalan

Distinguished
Sep 4, 2011
2,672
0
20,810


That's what I was afraid of. The sky high numbers AMD put out were for synthetic benchmarks. Real world games got 4% improvement if that. Ouch!


EDIT: SA has numbers up too. http://semiaccurate.com/2013/06/04/amd-brings-richland-to-the-desktop-the-a10-6800k/


 


Reread it VERY carefully and you'll notice something.

http://www.overclock3d.net/reviews/cpu_mainboard/amd_a10-6800k_richland_review/2

4GB Kingston HyperX 1600MHz @ 1333MHz

The review is essentially trash. Trinity requires DDR3-1866 at a minimum, Richland's beefier GPU means it would need DDR3-2133 to really hit it's stride. Running it at 1333 just results in memory limiting the iGPU, mind as well insert a Trinity or Llano at that point in time.
 

griptwister

Distinguished
Oct 7, 2012
1,437
0
19,460
palladin9479: NICE CATCH!

So, My question is (it might be a noob question) Will Steamroller APUs be a modification or Richland and the FX line a modification of Piledriver? Or did I just mis-understand something with the naming scheme?
 

Cazalan

Distinguished
Sep 4, 2011
2,672
0
20,810


Good point. The SA review used faster memory but not 2133.

"8 GB DDR3 1866Mhz kit from AMD"

 

8350rocks

Distinguished


Steamroller will supposedly be a die shrink, so not only a refinement, but also a new process (from 32nm to 28nm)
 


Considering how slow Trinity Mobile is, its not hard to improve.

Richland desktop is a pointless upgrade for me as at present I have about as good as it gets.

Prodigy Orange
Bitfenix Alchamy Orange cable extenders
Corsair CX430M
ASRock FM2A85X ITX
Avixir DDR3 2800@2400/NB-1.290v (to get 2400 stable is a mission 2800 is impossible)
A10 5800K@4ghz/1.295v
Thermaltake Water performa

I have a few GPU's which I tested to see which gives the lowest cost and best performance and I believe its the HD7790 when you consider the GTX650ti boost is R3000 and the R7790 is R1600 the 650ti boost only 5% faster with new drivers and the added cost makes it a bit of a let down but thats the local market, importing is different but obviously the RMA risks and shipping, the 7790 is maxing everything but Crysis 3 at 1080 so its a pretty solid deal at under 2K, but the uneducated crowd don't buy it because of the 128bit interface.

 



heh. just as i suspected. :pt1cable: 6800k isn't being tested with 2133 ram. if amd themselves don't supply appropriately specced ram, what can reviewers do? as if amd wants 6800k to underperform in reviews. ironically, amd recently announced ddr3 2133 'gaming' ram iirc.
there's a gap here though. i dunno what amd told in their richland reviewers' guide, but nothing should stop reviewers from at least trying to oc the 1866 ram... :whistle:
i bet it's intel's fault again, they hijacked amd review kits midway and replaced 2133 ram with lower specced ram and edited the review guide! corporate espionage is at play. brace yourselves people! :pt1cable:
 
richland round up and some jaguar-related stuff:
http://semiaccurate.com/2013/06/04/amd-brings-richland-to-the-desktop-the-a10-6800k/
http://vr-zone.com/articles/amd-rolls-out-richland-based-desktop-apus/34806.html
http://www.techpowerup.com/reviewdb/Processors/AMD/A-Series-Richland/
http://www.bjorn3d.com/2013/06/amd-richland-june-5-2013-1201am-edt-nda/

"forward compatibility to socket fm2+...."?
http://www.techpowerup.com/185067/amd-extends-graphics-and-compute-leadership-with-2013-elite-a-series-desktop-apu.html

Gigabyte, Quanta show AMD Temash Windows 8 tablets
http://techreport.com/news/24905/gigabyte-quanta-show-amd-temash-windows-8-tablets

Sapphire Unveils Three APU-based Innovations
http://www.techpowerup.com/185077/sapphire-unveils-three-apu-based-innovations.html

ECS KBN-I AMD "Kabini" SoC Motherboard Smiles for the Camera
http://www.techpowerup.com/185040/ecs-kbn-i-amd-kabini-soc-motherboard-smiles-for-the-camera.html

AMD hopes to bridge gap between Xbox, PlayStation and PCs
http://www.pcworld.com/article/2040804/amd-hopes-to-bridge-gap-between-xbox-playstation-and-pcs.html

kaveri's physical existence revealed? :)
http://www.pcworld.com/article/2040763/amd-fires-back-at-intels-haswell-with-its-a-series-desktop-processors.html
At Computex, AMD showed off the next generation of the technology in a chip known as "Kaveri," reportedly due in the second half of 2013.

 




Huh .... did you read the S/A article? They specifically mentioned that all they had on hand in the lab was DDR3-1833 and that they really wished they had DDR3-2166 for testing. AMD didn't provide them with memory, they used their own.

The rest of your post stinks of an attempt at trolling certain folks on here, I would ask you to please restrain yourself from such actions.
 
i did read the s/a article. that's why i wondered why the reviewers didn't just use a.m.p. or x.m.p. to overclock the memory... or at least try or say why it didn't work.

yeah... the last bit was a bit... well.. i am definitely going to tone my post down. although i am sure it wasn't at certain folks - that much i can verify.
 
Productivity and Power is where Richland does well, as for gaming I believe current APU's have hit the bandwidth wall so it is expected that the gaming performance increase was going to be minor. I will be in the market for a A8 6500 which looks the sound option.
 

mlscrow

Distinguished
Oct 15, 2010
71
0
18,640


Yes, Steamroller APU's will be a modification of Richland, but much more so than Richland was of Trinity and still even moreso than Trinity was of Llano. Steamroller will be the biggest change in the Module architecture in it's history. Hopefully, this will result in the performance increase that so many of us are anticipating.

My biggest hope is in the duplicating of the decode hardware in each module so that each core has it's own 4-wide instruction decoder, which will run in parallel rather than alternating cycles. This means that a single Steamroller Module (which I compare to a single Intel core) can potentially have a rate of 8 instructions/cycle as opposed to the previous 4. This should technically put Steamroller finally on par with Intel in that area, actually beyond with 8 cores vs Intel's 6, and hell, it should technically put Steamroller finally beyond Phenom II, which has outperformed anything AMD has released in that area since (at least beyond single core performance). Here, I edited a slide from one of Anand's articles, adding in a Steamroller column...

steamrollerdecodingg.png

 


Am I the only one who thinks that it would be in AMD's best interest to ship the $100 worth of faster RAM to reviewers to get better reviews? I mean, WTH is AMD thinking?

Still, CPU benchmarks basically show little change; still about 30% slower then IB, and 40% slower then HW. So it will be interesting to see how much faster RAM helps the GPU side of the house.

Though I do find it funny the people who were claiming 40% improvements in a single generation haven't chimed in yet. Just saying, again, never make assumptions based on "best case" benchmarking.
 
All I can say about Richland so far is...

610x.jpg

DAT PERFORMANCE

The darn thing is still too hot, though. I'll wait for more reviews, but that thing alone turns me a little down to be honest. Otherwise, a good step up.

Cheers!
 

amdfangirl

Expert
Ambassador
I am very excited about Kaveri - AMD finally getting decent single threaded performance.

Add that to insane overclocking and we might have a slightly power in-efficient chip that harks back to the days of the Phenom II.
 
Well it was 40% faster in some synthetics, but synthetics as we know are not very useful other than equating value to hypothetical circumstances.

I have looked at the numbers again and the productivity numbers signify the bigger improvement, that being said despite being faster on the CPU side it is limited in how far it can push the iGPU simply down to bandwidth limitations. Take a HD7970 slap on GDDR3 and its about a 3rd of its performance with GDDR5.

Over the weeks Richland started to appear more as a refresh than a upgrade but its hard to say whether I want or need it, I mean I may get about 2-3 FPS more in BF3 which is all the gaming I do (or F1 2012) but for everything else it may actually be faster so Im in a pickle now. If Richland is 10% rounded I was hoping more for 20% where its more tangible but that all said nobody seems delighted nor do they seem disappointed so its kinda like expected. The real talking point is Richland mobile where AMD claimed 40% faster than HD4000 mobile and this is where I think it will make a bigger impact.....sounds more like Intel now (focused on mobile over us DT junkies :D)

What I am looking forward to is the Availability of Gigabytes new FM2 itx setup, Asrock made no sense with 7 SATA's simply no iTX case has enough space for 6 drives :D, that Gigabyte board looks delicious and being a 3D bios virgin and having the oportunity to test out a gigabyte board I was really impressed with how much they improved considering me and Gigabyte are not very good friends.
 
After reading this http://www.pcworld.com/article/2040804/amd-hopes-to-bridge-gap-between-xbox-playstation-and-pcs.html posted above. I thought could AMD release a chip with say 10 jaguar cores and 7870 like graphics with only 2 cores running a custom linux and the rest of the cores and the GPU dedicated purely to games so it could run almost direct ports from the Xbox one & PS4 at the same performance and Valve could have there steam box running all non exclusive titles + whatever they can add. Could this happen & is it likely or even a good idea?
 
^^ if amd wants, there might be an opteron sku with 2x 4 core jaguar-based opteron in the future, like the interlagos cpus, possibly aimed at microservers. 10 cores might also be possible.
as for an soc like the consoles, may be at 16-14 nm.
 


Nope. Remember, you have totally different threading, memory, and graphical backends on consoles, so the only way to run direct ports is via emulation, which is VERY expensive to do [especially when you need bus-accurate timings]. So you are never going to be able to just pop in a PS4 disk into a PC and run it without an emulation layer.
 
Status
Not open for further replies.