Kaveri refresh name "Godavari" official?

[Phynaz]

I haven't heard anything "official" about the refresh in months now....siiiigh. AMD be AMDing. xD

Another Month Delay

 

[monstercameron]

Another Month Delay

And what is so happy about that?

 

[erunion]

And what is so happy about that?

look at the first letter of each word

 

[monstercameron]

look at the first letter of each word

Ok

 

[svenge]

I thought the official acronym was either "Always Massively Delayed" or "Again Missed Deadline"...

 

[cbn]

Does anyone know what a HS78 Class (D3) heatsink is? See footnotes on that table which indicates A10-8850K requires that heatsink.

 

[PG]

Does anyone know what a HS78 Class (D3) heatsink is? See footnotes on that table which indicates A10-8850K requires that heatsink.

I was curious about that too so I did some searching. The only thing I found was this PDF: http://support.amd.com/TechDocs/43375.pdf

Inside that there is only one mention of the HS78. Right above it, there is a mention of thermal profile AE. Then if you search for AE you find that thermal profile under certain cpus, like the HDT55TFBK6DGR and HDT90ZFBK6DGR. Now if you search cpu-world for those cpus it says those have a 125W TDP. They came with the well known AMD heatspipe heatsink.
My guess is that this new top of the line Godavari APU will come with that Phenom II heatpipe heatsink as well.

 

[ShintaiDK]

Asrock have leaked details again:
http://www.asrock.com/mb/AMD/FM2A88X%20Extreme4+/?cat=CPU
http://techreport.com/news/28186/asrock-leaks-specs-for-refreshed-kaveri-apus

 

[Shehriazad]

Asrock have leaked details again:
http://www.asrock.com/mb/AMD/FM2A88X%20Extreme4+/?cat=CPU
http://techreport.com/news/28186/asrock-leaks-specs-for-refreshed-kaveri-apus

AMD managed to put some higher clock on the 65W models.
But I think it's pretty much 1:1 Kaveri with some better binning.

Now I would like to see if those manage to overclock any better...especially on their Athlon counterparts. x)

 

[Abwx]

A micro review of the 7870K...

http://www.overclock.net/t/1555615/amd-kaveri-refresh-a10-7870k-apu-arrives

 

[DrMrLordX]

Yeah, I saw that. That particular sample looks really good! I'll bet he could hit 4.9-5.0Ghz with elbow grease.

That's on an A88x-Pro though. I'd like to know if the same chip would hit those speeds AND get full performance scaling on a cheaper board. My guess is that it wouldn't.

 

[Shehriazad]

Yeah, I saw that. That particular sample looks really good! I'll bet he could hit 4.9-5.0Ghz with elbow grease.

That's on an A88x-Pro though. I'd like to know if the same chip would hit those speeds AND get full performance scaling on a cheaper board. My guess is that it wouldn't.

Why would it fail on a cheaper board? If it got decent phases the chipset itself wouldn't matter as in the FM2+ case it only affects the amount of USB/Sata/m.Sata connectors...no?

 

[Abwx]

Yeah, I saw that. That particular sample looks really good! I'll bet he could hit 4.9-5.0Ghz with elbow grease.

That's on an A88x-Pro though. I'd like to know if the same chip would hit those speeds AND get full performance scaling on a cheaper board. My guess is that it wouldn't.

Planet3D got 4.2 at the 1.35V stock voltage of their 7650K with a A68H MB, same voltage as this one they should hit 4.6-4.7 without problem, although this seems to me extreme overclocking.

4.0-4.2 is more reasonable thermaly and comsumption wise, for the low power users the low cost 4C Kaveris allow very good thermals, sub 35W levels at 3.3GHz for the 7650K...


http://www.planet3dnow.de/cms/14312...ubpage-leistungsaufnahme-und-energieffizienz/

 

[Erenhardt]

I run 7850k @4.0GHz with only 1.27V. I would expect a refresh to beat that easily. Kaveri hit a wall around 4.5GHz. Where addding more voltage didn't enabled higher clocks.

 

[DrMrLordX]

Why would it fail on a cheaper board? If it got decent phases the chipset itself wouldn't matter as in the FM2+ case it only affects the amount of USB/Sata/m.Sata connectors...no?

cTDP throttling. Some boards will throttle performance imperceptibly if you're trying to push too much power through the socket. Some boards won't. This guy went bonkers trying to fix the problem on his 7850k:

http://www.overclock.net/t/1495168/breaking-down-the-wall-yawa-vs-the-7850k-throttling-issue/0_100

Had he switched motherboards, he would have had no such trouble. His mistake was assuming it was merely the VRMs . . .

Planet3D got 4.2 at the 1.35V stock voltage of their 7650K with a A68H MB, same voltage as this one they should hit 4.6-4.7 without problem, although this seems to me extreme overclocking.

If they can do 4.7 with an A68H then more power to em. I wouldn't count on it. Hell my 7700k will do ~4.3 at that voltage, but 4.7 takes 1.5125v, go figure.

I run 7850k @4.0GHz with only 1.27V. I would expect a refresh to beat that easily. Kaveri hit a wall around 4.5GHz. Where addding more voltage didn't enabled higher clocks.

You got that right. I had to do surgery to get past 4.5 .

 

[cbn]

Planet3D got 4.2 at the 1.35V stock voltage of their 7650K with a A68H MB, same voltage as this one they should hit 4.6-4.7 without problem, although this seems to me extreme overclocking.

Here are the motherboards listed in that test:

http://www.microsofttranslator.com/...8-7650k-der-letzte-kaveri/subpage-testsystem/

Though I am confused on what cooler was used for the A68H overclock testing? Scythe Mugen (Tower cooler) or SilverStone Nitrogon NT06 evolution (non-tower cooler that blows air downward)

The A68 board is a ASUS A68HM-Plus (five phase VRM without heatsink on VRM):

http://www.asus.com/ca-en/Motherboards/A68HMPLUS/

http://www.newegg.com/Product/Product.aspx?Item=N82E16813132406

I am kinda thinking they might have used the Scythe Mugen in that test because a previous Kaveri test had the Gigabyte A88X board paired with SilverStone Nitrogon NT06 evolution

http://www.microsofttranslator.com/...ms/13907-apu-amd-a6-7400k/subpage-testsystem/

 

[cbn]

Regarding overclocking on cheap boards (without heatsink on VRMs), I do wonder if the next FM2+ APU flagship comes with the AMD 125 watt cooler:

http://forums.anandtech.com/showpost.php?p=37344069&postcount=32

And how that would turn out compared to using a tower cooler?

 

[Shehriazad]

Regarding overclocking on cheap boards (without heatsink on VRMs), I do wonder if the next FM2+ APU flagship comes with the AMD 125 watt cooler:

http://forums.anandtech.com/showpost.php?p=37344069&postcount=32

And how that would turn out compared to using a tower cooler?

Not only was it rumored to come with the FX series cooler...if you look at the package shot of the 7870K above...You can tell that this thing is a lot bigger than the other Athlon and Kaveri boxes.

The only reason this thing would be bigger is if it was housing a bigger heatsink/fan.




That said...I'm STILL waiting for the official 870K... Those leaks have still not been corrected and sites still want to tell me that it's gonna be a 3.5 Ghz base @ 95W.

Which just makes 0 sense. 3.5 Ghz APUs already have 65W TDP for Kaveri. Godavari comes with a new stepping and all and the new top model comes @ 3.9 base.

I want to see if they either copy their 7870K or go straight up 4 Ghz base.


I mean the 860K already beats a FX4350 which clocks at 4.2 Ghz base...I still fail to see how people say AMD hasn't made a step forward in years. xD
3.7/4.0 chip without L3 cache @ 95W beats a 4.2/4.3 chip with L3 cache @ 125W. Sure...it's still miles away from Intel...but still...FM2(+) actually tries to go ahead even though it uses a failed architecture.

Now if the new stepping allows me to push the 870K anywhere near 5 Ghz at sub $100...I will be a super happy camper. (I'd probably expect the chip to cost $80 since 860K currently costs $70 here. I just want to see FM2+ hit 5Ghz...for the decadence of it.

 

[Abwx]

Not only was it rumored to come with the FX series cooler...if you look at the package shot of the 7870K above...You can tell that this thing is a lot bigger than the other Athlon and Kaveri boxes.

I mean the 860K already beats a FX4350 which clocks at 4.2 Ghz base...I still fail to see how people say AMD hasn't made a step forward in years. xD
3.7/4.0 chip without L3 cache @ 95W beats a 4.2/4.3 chip with L3 cache @ 125W. Sure...it's still miles away from Intel...but still...FM2(+) actually tries to go ahead even though it uses a failed architecture.

The FX4350 reach barely 80W under prime 95 and 62W under MT load, actualy it s the best voltage binned regular FX, it has 200MHz advantage at same voltage than the 8350, right that the 860K get better perfs though...

 

[Shehriazad]

That the heatsink of the Godavari? Because that looks like what is used for FX. (I got an unused FX + a FM2+ heatsink/fan in my "parts" box...and that heatsink looks just like FX.)

I mean...in the end 7870K is just a 7850K with updated stepping, mild CPU overclock and not-so-mild GPU overclock....the mini Kaveri solution would not work at all...the thing would legitimately throttle because of heat...luckily AMD didn't change the socket mounting for like the past 6 mainstream socket generations. Good guy AMD.

 

[Abwx]

That the heatsink of the Godavari? Because that looks like what is used for FX. (I got an unused FX + a FM2+ heatsink/fan in my "parts" box...and that heatsink looks just like FX.)

That s what is displayed with the rest of the gear, look to be the stock cooler :

http://www.overclock.net/t/1555615/amd-kaveri-refresh-a10-7870k-apu-arrives

I mean...in the end 7870K is just a 7850K with updated stepping, mild CPU overclock and not-so-mild GPU overclock....the mini Kaveri solution would not work at all...the thing would legitimately throttle because of heat...luckily AMD didn't change the socket mounting for like the past 6 mainstream socket generations. Good guy AMD.

Kaveri is quite frugal in normal use, the numbers i generaly quote are extracted from Hardware.fr 7850K review using the worst retail sample they bought, this pointed that there are two savour of 7850K.

One is at 1.36V at stock and the other at 1.3125V stock, likely that the 7870K manage 3.9 base at the former voltage with the same stability the 7850K has at 3.7.(by stability read voltage margin).

 

[Shehriazad]

That s what is displayed with the rest of the gear, look to be the stock cooler :

http://www.overclock.net/t/1555615/amd-kaveri-refresh-a10-7870k-apu-arrives



Kaveri is quite frugal in normal use, the numbers i generaly quote are extracted from Hardware.fr 7850K review using the worst retail sample they bought, this pointed that there are two savour of 7850K.

One is at 1.36V at stock and the other at 1.3125V stock, likely that the 7870K manage 3.9 base at the former voltage with the same stability the 7850K has at 3.7.(by stability read voltage margin).

Either way...I'll be the first to snatch a 870K (well...I'll try) and as soon as I got it, push it toward 5Ghz as hard as I can....they're cheap enough so I can afford messing around. Kaveri seems to hit a brick wall @ 4.7ish Ghz (Mine could also go there...but everything above was suicidal.) on average...refresh has got to be better, pretty please. I haven't had a 5 Ghz chip since my FX8350.

 

[Abwx]

Though I am confused on what cooler was used for the A68H overclock testing? Scythe Mugen (Tower cooler) or SilverStone Nitrogon NT06 evolution (non-tower cooler that blows air downward)

ASUS A68HM-PLUS / Gigabyte GA-F2A88XM-D3H

Scythe Mugen / SilverStone Nitrogon NT06 Evolution

http://www.planet3dnow.de/cms/14312-amd-a8-7650k-der-letzte-kaveri/subpage-testsystem/

The Nitrogon was used in a previous review for the 7850K/6800K numbers that are displayed in this review, according to their power numbers the chip should be at about 57-58W at 4.2 with their X264 test, so the cooler is not of big importance in this case.

 

[Abwx]

Stock voltages and tweaked voltages posted by The Stilt for a 7870K.

Pb0 (4100MHz) = 1.43750V vs. 1.1960V
P0 (3900MHz) = 1.38750V vs. 1.1530V
P1 (3800MHz) = 1.35000V vs. 1.1180V
P2 (3700MHz) = 1.30000V vs. 1.0960V
P3 (3500MHz) = 1.23750V vs. 1.0640V
P4 (3000MHz) = 1.08750V vs. 0.9630V

http://www.overclock.net/t/1555615/amd-kaveri-refresh-a10-7870k-apu-arrives/60#post_23963448

The 1.1960V value at 4.1GHz is remarkable as it s at 10% higher frequency than Hardware early retail sample that was capable of 3.7 at the same voltage, this chip has 20% better efficency at the same frequency within the same stability margin.

 

[VirtualLarry]

Newegg's Godavari promo page:

http://promotions.newegg.com/amd/15-0992/index.html?icid=317876