AMD Richland

[Willzzz]

I guess that lower consumption is coming from the H80i cooling, I'm not sure I could justify that for a low end CPU :/

[CAT-THE-FIFTH]

Richland also overclocks better,and the A10 6700 has slightly better performance than the A10 5800K and consumes less power than either the A10 5700 or A10 5800K. The H80 is used for all the CPUs in that review too. Plus Richland is still using the same 32NM process node as Trinity and Llano. Kaveri is the bigger step,with SR cores,GCN shaders,hardware optimisation and a change to a 28NM node.

Edit!!

Pricing could be lower though.

[Willzzz]

While the H80i is used for all processors it may well be that it is more necessary when the 6800K is boosting to 4.4ghz, whereas at Trinity speeds the stock cooler may be adequate.

The puzzle is that with basically the same chip on the same process node surely it should be cheaper to produce by now?

The GTX770 was in a similar position, relaunched with higher clock speeds and minor tweaks, imagine the uproar if that had launched at a higher price than the 680.

[CAT-THE-FIFTH]

I suspect they want to make sure all the Trinity CPUs are sold first - after all Richland launched first for laptops. It seems to have much bigger gains there. I expect the prices to go down once,Trinity stock start getting depleted,and they are about £15 to £20 more than I would want them to be.

Regarding the cooler,all the CPUs tested by Bit-tech use either the H80 or H100i,so the same applies to all the Intel CPUs too. Haswell runs hot anyway,so better cooling would also affect its Turbo clockspeeds for both the CPU and IGP,and especially in the latter case as the Turbo is very high and very agressive compared to AMD IGPs. If anything Richland seems to run slightly less hotter than Trinity,so I expect in real life it will be more likely to boost higher anyway.

[Willzzz]

Bit-tech are quite sensibly only comparing Richland to the cheap i3s which have very low TDPs and don't turbo very high.

But regardless of which sub £100 CPU I pick I'm certainly not going to want to add an £80 water cooler to it.
At this end of the market I'm really looking at using the stock cooler.

[watercooled]

What makes you think an £80 cooler is necessary?

Ricchland is, for the desktop side at least, a minor improvement over Trinity - you probably wouldn't want to upgrade from Trinity, but when pricing levels out it makes sense to buy it for a new build/upgrade from an older system. I don't think anyone was expecting much else?

[Willzzz]

I don't think it is necessary, that's why they should have tested without it.

[watercooled]

Unless you're comparing load temps, I don't see why that's relevant? If anything, a water cooler will use more power than an air cooler, and the increased cooling capacity should allow the CPU to use higher turbo states for longer.

Core temperature can have a marginal impact on leakage, therefore power draw, but as long as it's a level playing field I don't see why it's a problem. Either way, a few watts is within testing error margin, sample variance and shouldn't be taken too seriously when making a purchasing decision.

[DanceswithUnix]

Are they still shipping that junk lump of aluminium?

I just bought a 965BE which came with the new Coolermaster heatpipe cooler which looks quite nice. Now cooling my FM1 socket server 3800K instead of the nasty thing that came with.

The thing is, if you want to build a uATX machine using an AMD cpu, then the choice of motherboards for AM3+ is currently awful so you would probably end up with Richland. In that case, an aftermarket cooler would be on a lot of people's shopping list.

Benches with stock cooler vs something decent would be very interesting to see.

[Willzzz]

I'm just saying reviewers should use common sense when choosing components.

No one in their right mind is going to buy an A10-6800K and an H80i.
You could get a FX-6300 and 7750 for £30 less.

[DanceswithUnix]

I'm just saying reviewers should use common sense when choosing components.

Well that would be nice, but I won't hold my breath.

To some degree I think they use whatever review samples they have been sent, hence the stupidly high end PSUs and motherboards they tend to stick with.

[watercooled]

It's probably just a case of shipping something which is barely suitable for the TDP of the CPU, expecting most people won't fully load it. I'm not sure what the current ones are like but the fans on AM2 and FM1 boxed coolers I've used are fairly horrid.

I agree about the strange lack of choice for AM3+ motherboards too - I was looking for a decent uATX board for a spec a while back and they're almost non-existent now! ATX AM3+ also seem to do fairly poorly in terms of power efficiency too in my experience, which also heavily skews results when comparing to other CPUs. Not that it's an unfair comparison - you can't forgo a motherboard, but more choice would be nice.

There's nothing wrong with using OTT hardware in a test rig as long as they're open about it and it doesn't unfairly impact results. Tailoring the review system to every component being tested, aside from costing money and being a pain, would make comparisons between components essentially impossible. Using a 1600W PSU to test idle power of a modern system isn't ideal but it still allows you to make useful comparisons between other systems tested, provided, again, you can recognise and disregard testing fluctuations/sample variance.

A few additional real-world measurements would be nice though, I agree.

[Willzzz]

And of course sometimes it is the other way around.

Several of the top end 4770K reviews used the stock cooler!

[watercooled]

If that's against other chips under a better cooler then that is unfair. If they used stock for everything though, I see no problem? If the stock cooler is incapable of adequately cooling the CPU then it's important potential buyers are aware of it, not everyone will be using an aftermarket cooler, although I see little point in buying a k series Haswell CPU unless you're going to OC the pants off it.

[Willzzz]

Indeed, testing the 4770 with stock cooler would make sense.
The 4770K manages with the stock cooler but it would be utterly pointless because you couldn't overclock.

Really comparing anything with anything is fair as long as you take into account the total price.

[bonn4lolly]

Thinking of using the A4 4000 in a HTPC - can someone just confirm that it will handle Blu ray playback, 1080 streaming and recording from TV to HD via XBMC please? Wee bit worried at the lack of reviews.

THAT SAID, £32 minus the £25 you can Ebay SimCity for... no brainer at £7-ish!