Haswell Refresh Tempuratures

[zir_blazer]

AnandTech reviewed several Haswell Refresh Engineering Samples and power consumption is at the exact same spots than the Haswell that they replace were. All of them are a few Watts higher than current Haswells due to the extra 100 MHz or so, otherwise identical. It even seems weird considering that bin and Heatspreader contact variations could drastically the end result, so samples should be of equivalent quality for those results to be so close.
Assuming AnandTech is right, Tweaktown did something wrong and they results are flawed. They had a very good Haswell Refresh and a very bad Haswell, otherwise it doesn't make sense.

 

[Magic Carpet]

Yeah, looks like there is no 'magical' reduction of power consumption. No free cheese guys. One more review is up.

 

[ShintaiDK]

Its a refresh, a speedbump. Ofcourse nothing changes.

But there is no drama in that. Intel didnt cheat current Haswell owners and didnt want to milk everyone for the "fixed" refresh.

The conspiracy crowd loses again, as always.

 

[angevil]

How does the conspiracy crowd lose ? I thought that intel will milk us by releasing very small incremental upgrades. I was wrong that it is not sold at a higher price, but it is a 3% increase in performance in 1 year at the same power consumption at best.

And "engineering sample" is a fancy word for this cpu that has no improvement. They could have released this the same day haswell was released. I am disappointed, i thought they would at least improve thermals and power consumption, even if it is 10%, but it is something.

And again regarding "speed hump" and "refresh", that term is generous for a 3% difference in clock speed.

 

[IntelUser2000]

I find it interesting Anandtech review is showing at the worst, 100% scaling with clock speed differences of the 4790 vs 4770K. Some are even showing 10% differences, with mere 3% clock changes.

 

[mikeymikec]

How does the conspiracy crowd lose ? I thought that intel will milk us by releasing very small incremental upgrades.

That makes no sense. If what they've released doesn't constitute a significant enough upgrade for you, then don't bother buying it.

If you (and other Haswell owners) don't buy it, it can hardly be considered to be a business model (in the context of your accusation), can it?

- edit - I'm just having a look over the review on AT, and it seems to be that more Haswell processors have been released and that's it. Looking at the i3, it's 100MHz faster than the i3-4330 and the benchmarks seem to reflect that exactly. It's like what AMD used to do in say the Ph2 days, initially there were only 3 different Ph2 CPUs, by the time the generation was done there were probably 12-15. What's to say about that, "oh no, they're giving us more to choose from!"?

 

[tolis626]

Well, I don't think there is anyone stupid enough to upgrade to Haswell Refresh from Haswell just because it's newer. But for someone who builds their rig now, why not? It's nothing groundbreaking, but still better than nothing, especially if Devil's Canyon brings improvements for overclocking...

@ShintaiDK

If the specs you posted about DC are true, then I'm afraid we'll have to deal with a somewhat shocking price point. If it ends up in the same price as the current 4770k, then yeah, it seems fishy.

 

[SlowSpyder]

Its a refresh, a speedbump. Ofcourse nothing changes.

But there is no drama in that. Intel didnt cheat current Haswell owners and didnt want to milk everyone for the "fixed" refresh.

The conspiracy crowd loses again, as always.

I don't understand how people feel they got cheated or Intel was somehow not giving them their money's worth. When you buy a CPU it is guaranteed to work at a certain speed within a certain power envelope. Anything you do to go above the speed Intel sells the CPU at is a bonus and nothing more. Looking at it from Intel's perspective, there was nothing to fix, the CPU's worked as advertised and allowed for tweakers to get extra out of the CPU. Any change to the TIM with the clock bumps and new models is just a bonus.

 

[StrangerGuy]

Well Intel broke this in first place, by going for TIM instead of solder, so I don't know why you think consumers should be grateful?

Stockholm Syndrome is a powerful thing, bro. Even without overclocking why would anyone *not* want lower temperatures.

Even with proper TIM this is the most sleep inducing PC hardware "release" for a long time.

 

[angevil]

Well, I don't think there is anyone stupid enough to upgrade to Haswell Refresh from Haswell just because it's newer. But for someone who builds their rig now, why not? It's nothing groundbreaking, but still better than nothing, especially if Devil's Canyon brings improvements for overclocking...

In the last few years we got down to 10% improvement per year, how about that from now on we get a 3% improvement per year. It is good for who builds a new right now, right ?

If it was not for ARM, intel might as well stop all R&D and get free profits from anybody getting a new CPU after it breaks using. Using this line of thought, i have an idea for intel: drastically decrease the quality of components used in cpu, so that it never lasts more than the warranty.

Combined with the previous idea of never spending anything for R&D, you get people replacing their CPU ever 3 years because it breaks down, which results in very low costs and very high profits. Intel could advertise this as a "feature" too, that CPU degradation over 3 years makes it use 5% more power, so you have to be "green" and buy a new CPU constantly.

 

[Pheesh]

Stockholm Syndrome is a powerful thing, bro. Even without overclocking why would anyone *not* want lower temperatures.

Even with proper TIM this is the most sleep inducing PC hardware "release" for a long time.

So the 4.0stock/4.4ghz i7-4790k for a refresh is sleep inducing? I think you forget what typical refreshes are like.

 

[zephyrprime]

Help refresh my memory for a moment...
I thought that IDC's delidding experiment showed that the major improvement in temps wasn't actually substituting the TIM for something better; he said Intel used good quality stuff), but the reduction in the distance between the core and the heatspreader, after sanding down the heatspreader a little and/or scraping off the black rubber glue. He speculated that Intel might've wanted abit more looseness, maybe to avoid another soldergate (Nvidia), given Intel's more careful nature.

Looseness would make solder problems more likely, not less. Note that XBOX RROD fix kits relied in part on increases the force pressing down on the heatsink.

 

[Ajay]

Looseness would make solder problems more likely, not less. Note that XBOX RROD fix kits relied in part on increases the force pressing down on the heatsink.

Depends are on where the actual shear force is. If compression can reduce movement under shear, then great - but Intel probably didn't spec out a real high compression level for CPU heatsinks. They weren't going to issue a 'fix' for a product that, in the vast majority of scenarios, worked fine within specs.

 

[OCGuy]

Devil's Canyon are the only chips claimed "re-engineered' by Intel, so anything other than a speed bump on non-K products wasn't even on the table.

Why would they waste any extra time or energy on a non-enthusiast SKU that operates perfectly fine? The people reading this thread are not the target market for these chips.