AMD Bristol/Stoney Ridge Thread

[AtenRa]

At this time they better release Desktop BristolRidge at the same time as ZEN in January and only keep it for only 6 months like Ritchland, or release RavenRidge earlier in Q2 2017.

But because of AMDs low marketing budget, I wouldnt be surprised if they will skip Desktop BristolRidge all together. They simple dont have the budget to spend for Marketing in retail and are focusing in OEMs mostly + Consoles and GPUs. Currently nobody buys any AMD CPUs in desktop, simple because everyone is waiting for AM4.

 

[hojnikb]

Are there any stoney ridge motherboards in the pipeline ?

 

[Insert_Nickname]

At this time they better release Desktop BristolRidge at the same time as ZEN in January and only keep it for only 6 months like Ritchland, or release RavenRidge earlier in Q2 2017.

But because of AMDs low marketing budget, I wouldnt be surprised if they will skip Desktop BristolRidge all together. They simple dont have the budget to spend for Marketing in retail and are focusing in OEMs mostly + Consoles and GPUs. Currently nobody buys any AMD CPUs in desktop, simple because everyone is waiting for AM4.

I think a January top-to-bottom DIY launch is the most likely scenario at this time. I am glad however that AMD is focusing were the money is, because they need the volume and I don't think there is that much of that in retail DIY Bristol Ridge, with everyone waiting for Zen and Raven Ridge.

As an enthusiast that sucks of course, as my HTPC is about due for an upgrade, and an AM4 system could be fun to toy with. But I can understand why. Besides, my Athlon 845 is quite fun to play with already, so I get my Excavator fix that way...

 

[ET]

I don't think there is that much of that in retail DIY Bristol Ridge, with everyone waiting for Zen and Raven Ridge.

Self fulfilling prophecy. The longer Bristol Ridge is postponed, the more people will wait for Raven Ridge.

 

[Insert_Nickname]

Self fulfilling prophecy. The longer Bristol Ridge is postponed, the more people will wait for Raven Ridge.

Precisely. The question is whether or not this is a bad thing. The excavator core, while impressive for what it is, aren't really geared to tackle SKL/KBL head-on. The only advantage is then the IGP, which makes BR a though sell in DIY circles. So, again, I can see why AMD is chasing OEM, rather then DIY sales.

 

[NostaSeronx]

Self fulfilling prophecy. The longer Bristol Ridge is postponed, the more people will wait for Raven Ridge.

Don't tell no one, what if Raven Ridge never comes out?

14nm FinFETs is a bit more expensive than historic 28nm Bulk.

232 mm² FinFET = $199-$239
212 mm² Bulk = $329-$349

While the GPUs can eat those margins, the CPU/APU side can't.

 

[NTMBK]

Don't tell no one, what if Raven Ridge never comes out?

14nm FinFETs is a bit more expensive than historic 28nm Bulk.

232 mm² FinFET = $199-$239
212 mm² Bulk = $329-$349

While the GPUs can eat those margins, the CPU/APU side can't.

Something something FDSOI

 

[majord]

Some more pic's, incl BIOS SS from the A320M-C and A12-9800 combo fitted to the japanese Aerostream system I linked earlier

source: https://twitter.com/momomo_us

 

[AtenRa]

If this board is $50-60 and can take a 95W TDP 8C 16T ZEN, then we have something to talk about.

 

[AtenRa]

hmm, 29% faster in Cinebench MT at 65W TDP over the Caveri 95W TDP A10-7870K.

http://pc.watch.impress.co.jp/img/pcw/docs/1028/597/html/g12.png.html

 

[mikk]

hmm, 29% faster in Cinebench MT at 65W TDP over the Caveri 95W TDP A10-7870K.

Look closer to the bottom, this is the power consumption, 29% higher power consumption. Performance is unchanged in Cinebench MT: http://pc.watch.impress.co.jp/img/pcw/docs/1028/597/html/g01.png.html

Pretty poor I would say.

 

[AtenRa]

Look closer to the bottom, this is the power consumption, 29% higher power consumption. Performance is unchanged in Cinebench MT: http://pc.watch.impress.co.jp/img/pcw/docs/1028/597/html/g01.png.html

Pretty poor I would say.

Power consumption in MT load is equal with the BristolRidge being 29% faster, that is 29% higher perf/watt. Pretty amazing as both Chips are made on 28nm.

Single threat perf/watt is lower because of the 28nm HDL having worse perf/watt at high clocks. But you already knew that didnt you ??

 

[NostaSeronx]

@AtenRa different power supplies;
500W (定格450W) for A12-9800
SilverStone SST-ST30SF(300W 80PLUS BRONZE) for A10-7870K

 

[mikk]

Power consumption in MT load is equal with the BristolRidge being 29% faster, that is 29% higher perf/watt. Pretty amazing as both Chips are made on 28nm.

Just wrong.

Cinebench R15 MT

A12-9800= 323 +0.6%
A10-8770K= 321

Cinebench R15 MT Power consumption

A12-9800= 128 watts +19,6%
A10-8770K= 107 watts


Based on this, perf/watt is down by almost 20%, not 29% higher! Indeed pretty amazing.

 

[hojnikb]

Unless power is measured straight from the cpu, they are most defenetly not comparable.

You got power supply losses (if you measure at wall) and different motherboards have different power consumption themselves.

Pretty pointless tests really.

 

[AtenRa]

Just wrong.

Cinebench R15 MT

A12-9800= 323 +0.6%
A10-8770K= 321

Cinebench R15 MT Power consumption

A12-9800= 128 watts +19,6%
A10-8770K= 107 watts


Based on this, perf/watt is down by almost 20%, not 29% higher! Indeed pretty amazing.

Heh my apologies, it seems i was looking something else, you are right they have the same perf but BR has higher consumption. But as Nosta said above, they used different PSUs so we have to wait and see whats going on.

 

[dark zero]

We need to see GPU performance.

 

[LTC8K6]

If you show power consumption, it makes it seem like that's the best feature.

Why is the voltage seemingly so high?

 

[FlanK3r]

Where are the CPUs? I cant find it in eshops

 

[AtenRa]

Where are the CPUs? I cant find it in eshops

If the results above are indicative, i wouldnt bet on AMD releasing BristolRidge for the DIY market. perhaps the 35W TDP may be something better in iGPU than the competition and a new Desktop tier in AMDs APU lineup that we havent had before.

 

[ET]

I agree. I'd wait for better comparative benchmarks, but so far BR looks pretty bad. Losing significantly on some 3DMark tests is certainly not something I had expected, given the higher max GPU clock, more memory bandwidth, and more modern architecture.

I'd love to see a power and performance comparison between the A12-9800 and A10-7860K.

 

[richaron]

It appears AMD is once again using the "strategic" approach of delaying and/or cancelling products. No word of a lie I would own both a GCN 1.2 based GPU and and an 8~core CPU atm if I hadn't got my knickers in a knot over AMD's past strategies.

In this case we know Carrizo had low power 'optimizations' in design, and while it may ramp up OK to desktop level power/performance.. It still might be best if AMD try to avoid another Bulldozer-esk "loLz OMg 1t c4nT eveN BEAt 0LD APUz!!1"

 

[Dresdenboy]

Just wrong.

Cinebench R15 MT

A12-9800= 323 +0.6%
A10-8770K= 321

Cinebench R15 MT Power consumption

A12-9800= 128 watts +19,6%
A10-8770K= 107 watts


Based on this, perf/watt is down by almost 20%, not 29% higher! Indeed pretty amazing.

This is an endless loop it seems.

See right part of the chart:

 

[USER8000]

Well the A12 chip is also using a rather rubbish cooler,too.

I wonder if it is having issues actualling hitting its rated clockspeeds??

 

[The Stilt]

That's a standard HS55 heatsink, with 0.45°C/W rating.
CZ / BR is a mobile design, so it has a higher TjMax than the normal desktop designs (100°C vs. 57.1 - 72°C).
In case of CZ / BR that heatsink can dissipate 128.66W of power in the maximum allowed tAmbient (42°C), while keeping the temperature below the below 100°C (99.9°C).

So if it throttles that's due the power limit.