Upcoming AMD Richland A10-6700 - Radeon HD 8670D tested!

[Ajay]

28nm should really increase the xtor budget AMD had for the design engineers.

The clockspeeds should be improved, 28nm being higher Idrive than 32nm, and I'd expect dynamic power to be lowered over that of 32nm as well. But I'd expect static leakage to be higher (or at best, no worse) than 32nm because they won't have SOI to assist them in lowering substrate leakage at 28nm.

So depending on what they do with their power-consumption budget and xtor budget, we'll either get something on the order of 20% higher IPC but no faster clockspeed (they throw all the xtors towards bolstering IPC), or they'll keep it lean and clean and push for 5GHz+ clockspeeds (small die, IPC not much better than PD, but higher clocks from the power-budget).

The problem is they are coming out late even for 28nm with Steamroller. 28nm has been out now for what, 18 months? AMD really needs steamroller to be on TSMC 20nm IMO with a roadmap to porting it to 16nm w/FinFet. At GloFo they simply have no future.

Really, higher Idrive with a narower gate?
Do bulk xtors pack better than SOI, or are we talking about an ~25% gain in xtors?

I can't imagine AMD going for that high a clock; SS would need to be very steep and hence off current too high (although putting 5GHz! on a box would be good for mass marketing - 'think old school' might work). Then again, we haven't gotten much in the way of details on IPC improvements except for MT.

Agree with the last paragraph 900%; having an Albatross tied to a sinking ship would give any captain heartburn!

 

[Ajay]

Like magic :awe: We need an IDC-signal installed on the roof or something.

Yeah, AT should pay him a stipend

 

[inf64]

SR core looks to be pushing both IPC and core count. We already saw AMD lists Kaveri as "up to 3 control unit" part. This implies that they went from minimum of 1 module part as a lowest/cheapest die, to 2 module variation. This in turn implies that instead of 8C(or better thread) parts on desktop, we might see 12C(Thread) parts in 2014 . There is a chance these will come on FM2 socket to cover the "mainstream/performance segment" in desktop and 2P servers while the MCM variation(24C) will be on a separate new socket AMD will introduce just for 4P configurations.

Also by upping the core count and improving the IPC(as an effect of drastic core redesign/changes) we may very well see some clock regression, probably by 10-15%. This is not a big deal as this will only apply to full load on all cores while light loads might very well be addressed by some new Turbo Core which will cover north of 4Ghz (just like PD does now on 32nm). "Stock clock" these days is not a constant when both intel and AMD are in question. Modern CPUs hardly run at their lowest clocks and new Turbo function is always looking to boost when possible,even for longer periods of time if there is headroom.

 

[Blandge]

SR core looks to be pushing both IPC and core count. We already saw AMD lists Kaveri as "up to 3 control unit" part. This implies that they went from minimum of 1 module part as a lowest/cheapest die, to 2 module variation. This in turn implies that instead of 8C(or better thread) parts on desktop, we might see 12C(Thread) parts in 2014 . There is a chance these will come on FM2 socket to cover the "mainstream/performance segment" in desktop and 2P servers while the MCM variation(24C) will be on a separate new socket AMD will introduce just for 4P configurations.

Also by upping the core count and improving the IPC(as an effect of drastic core redesign/changes) we may very well see some clock regression, probably by 10-15%. This is not a big deal as this will only apply to full load on all cores while light loads might very well be addressed by some new Turbo Core which will cover north of 4Ghz (just like PD does now on 32nm). "Stock clock" these days is not a constant when both intel and AMD are in question. Modern CPUs hardly run at their lowest clocks and new Turbo function is always looking to boost when possible,even for longer periods of time if there is headroom.

Actually, modern CPUs almost always run at their lowest clock frequency, but I understand what you really mean. Better phrasing next time IMHO.

 

[inf64]

Yeah sorry,I obviously meant their "stock" clock ,not clock in idle state. My bad.

 

[Centauri]

6700 supports 2133Mhz RAM so its GPU ,even at 5800K's GPU clocks, will perform a lot better. Couple that with more Mhz on GPU and better CPU Turbo when needed and you have much better perf./watt chip (65W vs 100W,better performance).

Which isn't really that amazing since pretty much every FM2 board supports 2133 RAM. All you have to do is install some and set the speed.

 

[inf64]

One thing is unofficial support , the other official support. 5800K does not support 2133Mhz spec on the memory side while Richland does. Richland will probably be able to run RAM at higher clocks than Trinity today(via OCing of course).

 

[Centauri]

It doesn't matter anyways as far as OEMs see it; they're still shipping Trinity systems with 1600 RAM.

 

[TheReaWarmonger]

Thanks, I`ve seen you at overclock before

One and the same, OCN is full of corrupt mods. It's too bad, as I was one of their more knowledgeable members.

 

[jacktesterson]

Which isn't really that amazing since pretty much every FM2 board supports 2133 RAM. All you have to do is install some and set the speed.

Mine supports 2400 or 2600 Ram - and I'm using 2400 Ram - Just had to increase the voltage slightly to get it to work 100% stable (1.67v vs 1.65 stock)


The Kit I got was only $60 - "Team" - Got it on sale - But it works. I think its still only 69.99

 

[jacktesterson]

I will upgrade to the A10-6800K on release

I use my Desktop for 95% HTPC duties and maybe 1-2 hours of some light gaming a week.

APU's suit me great. Done spending tons of money on high end stuff with no time to use it.

 

[tweakboy]

Gosh soo many bunk video cards!!!!! Who the fudge cake on Anand uses their iGPU .

They simply have a dedicated video card that is 100x faster then the other trash. gl

 

[Headfoot]

It is there, write in black in their annual report:



In typical AMD shaddy style, the WSA information is spread around a lot of topics, there isn't a section with the entire relevant information, you need to read the entire report to get a picture of the agreement. Other relevant information, AMD had to pay 700 million to get Kabini to TSMC, and only during a limited time frame.



They weren't. The WSA wasn't disclosed until sometime later and with a lot of redacted clauses, and there were the amendments that changed a lot of those redacted clauses... The only surprise here is that nobody ever filled a lawsuit against the BoD for breach of fiduciary duty.

Yeah, I was going to say the same. Sounds like someone should've gone for a 10(b)(5) complaint

 

[bgt]

my 5700 won't run on 2133mhz. It will crash. 1866 is ok.

 

[Rvenger]

Mine supports 2400 or 2600 Ram - and I'm using 2400 Ram - Just had to increase the voltage slightly to get it to work 100% stable (1.67v vs 1.65 stock)


The Kit I got was only $60 - "Team" - Got it on sale - But it works. I think its still only 69.99

Adam, do you think its safe to go beyond 1.65v on Trinity? I wanted to do the same as well.

 

[NTMBK]

I will upgrade to the A10-6800K on release

I use my Desktop for 95% HTPC duties and maybe 1-2 hours of some light gaming a week.

APU's suit me great. Done spending tons of money on high end stuff with no time to use it.

I'd say wait for Kaveri, to be honest. That's going to be the real upgrade- Richland is just a tweak.

 

[jacktesterson]

Adam, do you think its safe to go beyond 1.65v on Trinity? I wanted to do the same as well.

In my reading I've seen lots of people using 1.7v with Trinity without issue for memory.

To be honest, I use the ASRock O/C Profiles - I run my system 100% stock (I run memory at 1866 @ 1.5v - downclock it) when using it for its main purpose - a HTPC.

When I do play a few games with it on my TV, I apply my O/C profile - so I only ever run it at the Specs in my Sig for maybe 2-3 hours for an average week.

I've been fine so far.


I'm running my CPU @ 1.45v when O/C'd, and Northbridge @ 1.375v.

I can't get my iGPU stable anything higher than 1085 MHz - and My CPU likes to throttle if I go beyond 4.3 GHz as I'm using a HTPC case and a $25 cooler.

 

[jacktesterson]

I'd say wait for Kaveri, to be honest. That's going to be the real upgrade- Richland is just a tweak.

GPU Tweak - which is the main reason I'll prob grab it.

Looks like 20-30% boost in GPU performance, correct? Should be a cheap upgrade after I sell the 5800K

Since I'm actually using the APU to game @ 1360x768 on a TV, this will help out.


I'm finishing up Sleeping Dogs right now - I was 75% done then moved on but have come back to it.

I'm running it at 1360x768 with High Textures and High AA (not Ultra), everything else on Normal - Average 45 FPS at these settings with 29.8 FPS Minimums and high around 58 - Very playable - Using a 360 controller on my couch.

Move AA to Normal and I can get 60FPS locked but I find it fine with 45.

a 20% boost would be nice for little $$.

 

[Rvenger]

I'd say wait for Kaveri, to be honest. That's going to be the real upgrade- Richland is just a tweak.

Actually the A10-5800k fetches a decent buck on Ebay. He could probably sell it and be out only $15 with a new Richland.

 

[Arkadrel]

I can't get my iGPU stable anything higher than 1085 MHz - and My CPU likes to throttle if I go beyond 4.3 GHz as I'm using a HTPC case and a $25 cooler.

Still not bad running the iGPU at 1085mhz, thats like a 36% overclock.

I will upgrade to the A10-6800K on release
I use my Desktop for 95% HTPC duties and maybe 1-2 hours of some light gaming a week. APU's suit me great.

^ sadly I find myself doing more and more of that...
ei spending less and less time playing games.

In part I blame it on the decline of RPGs for the PC, and my enjoyment of FPS style games droping over time.



All I can say is, Im surpised by IGPs leap in performance these last few years.
You see people with IGPs actually being able to run a few games at 1920x resolutions now (ofc not with all settings max'ed ect, and not at max AA... but still).


http://hwbot.org/submission/2318901_froxic_3dmark06_radeon_hd_7660d_12066_marks
http://hwbot.org/submission/2317779_froxic_3dmark_vantage___performance_radeon_hd_7660d_7920_marks

^ guy with a 5800k@4.5ghz & igp@1150 mhz (~43% Igp oc)

He scores 12,066 in 3Dmarks 06 (thats better than most Nvidia GTX 640 mobile gpu's)
He Scores 7,920 in 3Dmark Vantage performance (640 mobile Gpu's are around 9k)


At some point I believe IGPs will squeeze the markets so much you wont really see many mobile GPUs.

 

[Gikaseixas]

A10-6800K here i come. I have a Llano 3850 @ 3.4ghz but didn't mess with the igp cuz it games perfectly well on 720p but my new smart tv is 1080p and will need more horsepower to drive it.

 

[Abwx]

There was this in the link provided by Centauri
in the AMD finance thread......

5) Testing and projections develop by AMD Performance Labs. The score for the 2012 AMD A10-4600M on the Cinebench Multi CPU benchmark was 2.0 and the 2012 AMD A8-4555M was 1.30, while the "Richland" 2013 AMD A10-5750M was 2.35 and the 2013 AMD A8-5545M was 1.55.

Nice bump at same TDP given that it s the same CPU....

 

[Enigmoid]

GPU Tweak - which is the main reason I'll prob grab it.

Looks like 20-30% boost in GPU performance, correct? Should be a cheap upgrade after I sell the 5800K

Since I'm actually using the APU to game @ 1360x768 on a TV, this will help out.


I'm finishing up Sleeping Dogs right now - I was 75% done then moved on but have come back to it.

I'm running it at 1360x768 with High Textures and High AA (not Ultra), everything else on Normal - Average 45 FPS at these settings with 29.8 FPS Minimums and high around 58 - Very playable - Using a 360 controller on my couch.

Move AA to Normal and I can get 60FPS locked but I find it fine with 45.

a 20% boost would be nice for little $$.

Well its a 5% boost in gpu clock speed on the desktop. On mobile its going to be more (can actually hit higher boost clocks).

 

[inf64]

Mobile Richland tested in dual GPU mode (against GT650M too):
http://big5.xinhuanet.com/gate/big5/news.xinhuanet.com/info/2013-04/18/c_132318291.htm

 

[mikk]

Mobile Richland tested in dual GPU mode (against GT650M too):
http://big5.xinhuanet.com/gate/big5/news.xinhuanet.com/info/2013-04/18/c_132318291.htm

Useless as long as AMD didn't solve their micro stuttering mess.