AMD - Piledriver chitchat

[Biscuit]

Terrible's a bit harsh; the cooler for my Semron did the job brilliantly and quietly (and was somewhat smaller than the stock cooler for the 3670K), and I think some of the higher-end chips still come with the heat-pipe based coolers which I thought were OK? Plus in a tower case under a desk the stock cooler would've been fine on the 3670K - it's because I was using it for an HTPC that I wasn't happy with the noise. The performance was surprisingly good. They just spec the absolute minimum that will keep the chip under 60C in normal conditions, so those of us with more demanding expectations tend to be disappointed.

Can't say Intel are an awful lot better mind, the stock Q6600 struggled to keep temps below 70C under load. It's only now they've got TDPs under tighter control that their stock coolers have been able to cope better with the heat output...

I guess everyones perspective of what is good and bad is different. The cooler that came with my 955 performed ok but, my word, the noise!

I spend a fair amount of time and money making my PC as quiet as it can be though, so i have reasonably high standards!

intel coolers seem to have been a bit of a rollercoaster. I remember having collection of different styles when i was building a lot fo PCs for friends and the differences in fan noises and performance depending on what chip you got was bizzarley varied.

[CAT-THE-FIFTH]

It seems the Resonant Clock Mesh found in Trinity is absent in Vishera.

[Biscuit]

It seems the Resonant Clock Mesh found in Trinity is absent in Vishera.

What affect does that have?

[watercooled]

The main one, if implemented, is lower power consumption. However, AFAIK it would also make overclocking potentially harder.

[Biscuit]

The main one, if implemented, is lower power consumption. However, AFAIK it would also make overclocking potentially harder.

As in when its implimented thats its affect?

[watercooled]

Yeah, I could be wrong about it making overclocking harder though.

[scaryjim]

I think it would make overclocking by adjusting the FSB harder, or at least prone to massively increase power consumption, as the RCM has to be tuned at a specific frequency to carry the standard clock through the chip, so if you messed with the standard clock generation the mesh wouldn't work properly. It certainly shouldn't affect overclocking by changing the multiplier though.

I have a sneaking suspicion that we won't see RCM on an AM3+ platform, Trinity (as did Llano before it) runs off a 100MHz base clock, so presumably that's what the RCM is tuned to. Don't know how hard it would be to tune one to 200MHz, but given the AM3+ processors won't be finding derivations heading into laptops, I doubt they'll worry about implementing a relatively minor power-saving feature in them. After all, Trinity draws quite a lot of juice at full load on the desktop too, so it obviously doesn't improve efficiency at high power draws that much...

[DanceswithUnix]

Overclocking Vishera certainly doesn't seem hard:

http://en.expreview.com/2012/10/26/a...ghz/24814.html

1.932V though, and all 8 cores enabled. What sort of power must that thing be pulling through the VRM!!

[CAT-THE-FIFTH]

I think it would make overclocking by adjusting the FSB harder, or at least prone to massively increase power consumption, as the RCM has to be tuned at a specific frequency to carry the standard clock through the chip, so if you messed with the standard clock generation the mesh wouldn't work properly. It certainly shouldn't affect overclocking by changing the multiplier though.

I have a sneaking suspicion that we won't see RCM on an AM3+ platform, Trinity (as did Llano before it) runs off a 100MHz base clock, so presumably that's what the RCM is tuned to. Don't know how hard it would be to tune one to 200MHz, but given the AM3+ processors won't be finding derivations heading into laptops, I doubt they'll worry about implementing a relatively minor power-saving feature in them. After all, Trinity draws quite a lot of juice at full load on the desktop too, so it obviously doesn't improve efficiency at high power draws that much...

I have a feeling it must be having an effect with lower TDP models,as the A10-5700 only has a 10% drop in clockspeeds,for a 35% drop in TDP. The A8-3800 had a 20% drop in clockspeed for a 35% TDP reduction in TDP over the A8 3850. Also,IIRC,the Resonant Clock Mesh could also be used to increase clockspeeds without increasing TDP. the FX4300 actually runs at lower clockspeeds than the A10-5800K.

[Biscuit]

Overclocking Vishera certainly doesn't seem hard:

http://en.expreview.com/2012/10/26/a...ghz/24814.html

1.932V though, and all 8 cores enabled. What sort of power must that thing be pulling through the VRM!!

Yeah i mean all you need is a selection of cherry picked CPUs and a liquid nitrogen cooler... easy peasy

[DanceswithUnix]

Yeah i mean all you need is a selection of cherry picked CPUs and a liquid nitrogen cooler... easy peasy

Exactly, now will one of the real overclockers please crack open the liquid helium.

After all, SuperPi is important to me

[CAT-THE-FIFTH]

There seems to be two other CPUs which were not released to retail:

http://www.cpu-world.com/CPUs/Bulldo...20FX-8300.html

http://www.cpu-world.com/CPUs/Bulldo...20FX-6350.html

I still think the FX6300 is the best of the bunch!

[DanceswithUnix]

So the 8300 is an 8 core cpu designed to go on cheap 95W OEM motherboards? Hmm...

[CAT-THE-FIFTH]

I suspect it will be around FX8150 level performance but better for games.

[DDY]

So what's the practical difference between the 8300, 8320 and 8350 if I can just change the multi?

[CAT-THE-FIFTH]

So what's the practical difference between the 8300, 8320 and 8350 if I can just change the multi?

The FX8350 probably can overclock a bit better. The 95W CPUs might be binned for lower leakage.