Haswell Refresh Tempuratures

[Wall Street]

http://www.tweaktown.com/articles/6...h-cpu-and-z97-performance-preview/index5.html

It looks like Haswell Refresh uses 15 watts less power and is 10 degrees C cooler than original Haswell. Let's see if Devil's Canyon is similar or if it is even better. This looks like the new TIM is paying off, let's hope all of the chips run this cool instead of the lottery effect.

 

[24601]

Over 2 years to fix TIM (Likely not even fluxless solder, maybe we'll get that in 2 more years).

Then sells fixed TIM CPU as a "Refresh."

I wish I didn't call this 2 years ago.

It's just too depressing.

 

[videogames101]

Over 2 years to fix TIM (Likely not even fluxless solder, maybe we'll get that in 2 more years).

Then sells fixed TIM CPU as a "Refresh."

I wish I didn't call this 2 years ago.

It's just too depressing.

selling upgraded (fixed if you like) product at same price, how depressing

 

[24601]

selling upgraded (fixed if you like) product at same price, how depressing

Breaking something to fix later is the worst kind of planned obsolescence.

It's downright lazy. And a total dick move to anyone that bought Haswell (And Ivy Bridge) when it released.

 

[GreenChile]

Over 2 years to fix TIM (Likely not even fluxless solder, maybe we'll get that in 2 more years).

Then sells fixed TIM CPU as a "Refresh."

I wish I didn't call this 2 years ago.

It's just too depressing.

Yes, shame on Intel for listening to the enthusiast community and giving them what they want. I too am depressed by this unfortunate development.

 

[24601]

Yes, shame on Intel for listening to the enthusiast community and giving them what they want. I too am depressed by this unfortunate development.

They planned this from the start.

That's why we had arguments with Intel engineers on these very forums where the engineers were stating something along the lines of "Intel doesn't want to spend the money validating anything better than the crap TIM under the IHS for consumer desktop use right now. Be happy we cater to you at all."

The fact that they didn't do this in the first place was purely because they didn't care to spend the extra money validating something better to start with, nothing more.

It wasn't some "magical engineering breakthrough!!!!!!"

They're doing this to throw us a bone while Desktop Broadwell gets delayed purely because Intel wants it's 14nm wafers to go to "more profitable sectors" like mobile.

 

[seitur]

Yes, shame on Intel for listening to the enthusiast community and giving them what they want. I too am depressed by this unfortunate development.

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?

 

[Haserath]

No "Yay Intel" from me. Overclockers pay more for their chips, so they should at least put a decent(better) implementation on the k chips.

It shouldn't cost them that much and it'll be like a half node improvement! :/

 

[BSim500]

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?

I have to agree with this. I don't think fluxless solder cost Intel much at all. Even budget Clarkdale i3's / old Sandy Bridge Pentium's had them, pre Ivy Bridge and they never complained about the cost then...

Also, is there any more detailed review out there? Saying "This chip uses 15w less load than a 4770k" is meaningless unless you compare voltages at same freq (with same heatsink & fan speeds). Is one a "good" chip with a stock 0.95v VID vs another's "bad" chip with a 1.2v stock VID, etc? Exactly the same i5 can differ from up to 20w depending on stock voltage / VID tables. Also it makes no sense that it pulls 31w more idle than an Ivy Bridge 3770K given the 800MHz vs 1600MHz difference...

A few more questions if anyone knows the answers:-

1. What exactly is the difference in "improved TIM"? Back to solder? Or simply better quality control regarding the "gap" found during delidding due to having too much glue?

2. Is "improved TIM" just on i7's, or on all new Haswell Refresh chips, ie, i5-4690, i3-4360, etc?

3. Do they still do that dumb +0.1v thing with AVX instructions?

 

[Homeles]

Over 2 years to fix TIM (Likely not even fluxless solder, maybe we'll get that in 2 more years).

Then sells fixed TIM CPU as a "Refresh."

I wish I didn't call this 2 years ago.

It's just too depressing.

So you mean to say that Intel's been playing the long con, starting with Ivy Bridge, just so they could make a quick buck and be able to "refresh" Haswell by using solder instead?

Also, if the product is working 100% okay, it's not broken. What you're saying could easily be classified as trolling.

Kids say the darndest things...

 

[24601]

So you mean to say that Intel's been playing the long con, starting with Ivy Bridge, just so they could make a quick buck and be able to "refresh" Haswell by using solder instead?

Also, if the product is working 100% okay, it's not broken. What you're saying could easily be classified as trolling.

Kids say the darndest things...

Don't make me pull the quotes from the Intel Engineers on this forum to make you look silly. I got banned so very many times to get those quotes originally too.

 

[ShintaiDK]

From the same website where a 3770 performs like a 4790 in tests?

Lets try another review site before making any judgements.

 

[jj109]

TIM or not doesn't really matter to me. Based on OCN's Haswell overclock spreadsheet and my own experience, the limit is voltage instead of temperature except in AVX2 loops that fit in the cache.

What would be interesting is if Haswell Refresh has a new stepping and responds better to voltage.

 

[Homeles]

Don't make me pull the quotes from the Intel Engineers on this forum to make you look silly. I got banned so very many times to get those quotes originally too.

I highly doubt that you've got "quotes" that support your statements.

Both of you need to play nice. I'm not going to entertain a match between the two of you.
-ViRGE

 

[24601]

I highly doubt that you've got "quotes" that support your statements.

Gimme a few days, gotta crawl through the last thousand or so threads in my history.

EDIT: Most of this thread.

http://forums.anandtech.com/showthread.php?t=2323720

Most of this thread.

http://forums.anandtech.com/showthread.php?t=2326941

Both of you need to play nice. I'm not going to entertain a match between the two of you.
-ViRGE

 

[frozentundra123456]

I would not make any judgement yet. That is a terrible test. The gaming test is totally gpu limited. And is it a misprint, or did they really test an 8150?

 

[DominionSeraph]

They planned this from the start.

Sure, let's get out the tinfoil to surmise that this was some grand conspiracy that for some reason only targeted the tiny enthusiast community, ignoring that the TIM works perfectly fine for OEM chips which constitute the bulk of Intel's sales and for whose volume a cheaper material would constitute considerable savings.

i5-4430: 3.0GHz. $180
i5-4440: 3.1GHz. $184
i5-4570: 3.2Ghz. $200
i5-4670: 3.4GHz. $205
i5-4670k: 4.5GHz OC. $230

Which of these doesn't belong?

A 4.5GHz Haswell i5 should be selling for more than double that price. That Intel didn't give it a TIM allowing Sandy k-series overclocks is irrelevant because the Sandy k-series was massively underpriced as well.
You're complaining that the steal you were given wasn't good enough. Get over it.

 

[bononos]

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.

 

[PPB]

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.

Well, now this is Intel tightening more the IHS to the die, so there is no gap between them. Or just making a more vigorous checking of the IHS installing process, becase not all Haswell chips were so bad with temps, although there was an even more pronounced "silicon lottery" than with IB, or should we call it "IHS gap lottery"?

 

[IEC]

Yes, shame on Intel for listening to the enthusiast community and giving them what they want. I too am depressed by this unfortunate development.

I'm sitting this round out. After "upgrading" from Sandy Bridge to Ivy Bridge it felt like I'd downgraded. Worse thermals and worse OC.

 

[Mand]

They planned this from the start.

Source?

 

[coffeejunkee]

A few more questions if anyone knows the answers:-

1. What exactly is the difference in "improved TIM"? Back to solder? Or simply better quality control regarding the "gap" found during delidding due to having too much glue?

2. Is "improved TIM" just on i7's, or on all new Haswell Refresh chips, ie, i5-4690, i3-4360, etc?

3. Do they still do that dumb +0.1v thing with AVX instructions?

1. The first time I saw that slide I thought 'yes, back to solder!' but on second thought 'if it was solder why would they call it tim and not solder?'.

2. It was a DC specific slide so I'm thinking only on DC cpu's.

3. Hard to say, 4790 uses less power and runs cooler than 4770. But it still uses more power than 3770. My guess is it still does the AVX bump, it's just a refresh after all. But despite using more power it still runs cooler than 3770 so something must have improved.

 

[Pheesh]

Guys, this non-K chip doesn't even have the TIM/package tweaks that devil's canyon does, so any improvement you are seeing here is due to something else. The reason devil's canyon is coming later is because it's running on it's own process.

 

[ShintaiDK]

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.

Correct, but that never stopped the conspiracy crowd.

Its also one of the reasons why the benefit of dellid is so wide in range. Some got a IHS with a thight fit, others less tight.

 

[Pheesh]

Gimme a few days, gotta crawl through the last thousand or so threads in my history.

EDIT: Most of this thread.

http://forums.anandtech.com/showthread.php?t=2323720

Most of this thread.

http://forums.anandtech.com/showthread.php?t=2326941

Both of you need to play nice. I'm not going to entertain a match between the two of you.
-ViRGE

Okay, so I just went through all 6 pages of that first thread and there was absolutely nothing outside of you having some sparring matches with other forum members. There's supposed to be a quote in there from an intel engineer? It really just seemed like you convinced yourself of a certain point of view and then chastised anyone else for disagreeing.